Bottlehead Forum
General Category => Technical topics => Topic started by: L0rdGwyn on March 09, 2020, 09:41:58 AM
-
Hi Bottleheaders,
I owe a lot to Bottlehead for starting me on my DIY journey, having started off with the Crack + SB and Crackatwoa (this was my Crackatwoa build log: https://forum.bottlehead.com/index.php?topic=10708.0). Hard to believe that was only two years ago.
I've started working on my own personal from-scratch designs and builds. I hope it is okay to ask about personal DIY projects on the forum? The first is nearly complete, a MH4/REN904 MOSFET CCS loaded input, 6A5G SET output design using Lundahl iron (mid-build photo attached).
I am now working on my second design. Nothing innovative, but at a high level, it is a MOSFET CCS loaded 6J5 input and MOSFET CCS loaded 45 parafeed output headphone amp. I am currently working through the various quirks of a parafeed design ;D
I have schematics and whatnot, but I will not bore you with the details unless asked. I really just wanted to ask the forum if my own conceptual understanding of a MOSFET CCS loaded parafeed stage is accurate, if you would be so kind (gosh there are a lot of parafeed conceptualizers out there lately, huh?). These are somewhat bouncing off of the questions asked here by Deke609.
My parafeed stage cascode CCS uses the IXYS IXTP08N100D2 (top) and IXTP08N50D2 (bottom), which from my reading provides an 500Mohm or more AC impedance at low frequencies (I haven't worked it out myself). With that being said, my understanding is that the load the tube "sees" can essentially be treated as the OPT primary impedance alone. No matter how heavily the secondary is loaded, the magnitude of the CCS AC load will dwarf the OPT primary in parallel. The CCS is basically a brick wall to AC signal at all frequencies. I made this assumpton when drawing the load lines for my design.
With the heavily loaded OPT primary provided by a high-impedance headphone (I use 300ohm) and the relatively high inductance provided by a parafeed OPT primary (137H of the Sowter 8665 I am considering), the low frequency response does not seem to be a cause for concern from my manual calculations and verified using Bode plot of my circuit model in LTSpice.
While this advantage of CCS loading the output tube is very appealing, there are two sacrifices: required B+ and power dissipation. I am using a low-voltage bias point for my 45 tubes of 180V, so a reasonable 400V power supply can be used. However, the 45 CCS top device must dissipate 6W at a 180V 31mA bias point, which is a bit of a challenge to dissipate.
The last sacrifice is determining the parafeed capacitor value with a CCS load. I have seen anode choke calculations out there, some provided by Paul, but there does not seem to be an accurate way to calculate the cap size when CCS loading. Is the best approach to get a swatch of different high voltage electrolytic capacitors, pop them in the circuit, get a Bode plot to see the affect on frequency response, then substitute an equal value film cap when the proper capacitance is determined? Using LTSpice, I can view the effect of the parafeed cap value on the low frequency reponse and resultant subsonic LC resonance with the OPT primary, but I can't say whether or not it is a reliable way to determine the value...
The last practical concern for this design is not so much related to parafeed as it is the 45 tube. I have been told that because the 45 can have high amounts of internal noise, a high-turns-ratio output transformer should be used. I was originally looking at the Sowter 8665, wired in either 12:1 or 6:1, but it was suggested to me to use a full-sized speaker parafeed OPT like the Sowter 8983 with a 25:1 or 17:1 winding ratio. Otherwise, this internal noise would be an issue with headphones. This high turns ratio results in a flat load line for the output tube with a 300ohm load, and power output from 30-60mW. Is the internal noise of the 45 something anyone can comment on from personal experience? If a high-turn-ratio OPT was used, an option would be to put a low value resistor in parallel with the headphone output, say 9.1ohm, to lower the load seen by the tube to 8ohms.
Thank you kindly for input :)
-
The 5K:8/16 transformer recommendation is a good one. I just finished a similar project and found that the high step-down ratio was a good idea. It wasn't heater hum or power supply buzz that was audible, but rather tube rush. Since you're on our forum, I'll plug our SEX Iron ugprade package for this, with is a 40H plate choke and 4K OPT that would be perfect for this project.
Yes, CCS dissipation is pretty brutal at 6W. You'd want a big heatsink, maybe 7C/W or lower to whisk that heat away. You could also load the 45 with a pentode to move that heat above the chassis plate, which would help with the lifetime of all the components inside the amp.
I would recommend buying a few values of Solen caps to try out. 4.7uF, 6.8uF, 8.2uF, and 10uF would be worth trying with headphones and speakers. My intuition says that anything between 5 and 10uF would probably work just fine.
-
Thanks for the recommendation on the iron upgrade kit, I will definitely check it out.
That is a beautiful amplifier, really like the color scheme and accents, has a sort of "steam punk" vibe. Is it getting a production run :) how fortunate you had the same experience with the transformers - duly noted, I suppose it is all but certain then I will us a high-turns-ratio OPT then. Maybe a ~17:1, the high turns will help my output impedance too for lower impedance headphones.
For the CCS dissipation, my options are A) large TO-220 PCB mount heatsink (Ohmite has one that does 3C/W, if I can fit it inside) paired with a reasonable amount of ventilation for the amplifier chassis, B) a large fin type heat sink mounted to the side of the chassis (ground isolated) with a drilled hole to mount the device directly onto it, or C) the pentode CCS. The design of a pentode CCS is not something I am very familiar with, but I suppose I have to consider it. I would really prefer to follow through with my MOSFET CCS, but like you said, I might be cooking my capacitors...
And absolutely brilliant suggestion on the Solen caps, I forgot they were so reasonably priced. I'm sure I will find some use for the extras when the project is completed too. Thanks for knocking out a few of my questions, PB.
-
As I said in my reply to Deke609 (posted March 8 at 04:45:05 PM), the parafeed output transformer is not a simple inductance. To model it accurately, you need to know a number of things, like eddy currents in the laminations and hysteresis of the core material. These things are nearly impossible to find out with sufficient accuracy to rely on simulations. Best bet is to measure and listen with several capacitances.
Here's what I can say about CCS-fed parafeed outputs that are lightly loaded:
The simple analysis gives a capacitance of L / R-squared, but the R is the tube's plate resistance rather than the primary impedance. This usually calls for very large caps. With more usual capacitors, you would expect a substantial resonance. I tried, but I couldn't measure that, because the transformer parasitic losses damp that resonance.
-
You're very welcome for the advice. On the Ohmite heatsink, be extra careful that they aren't giving you a 3C/W rating with some given amount of airflow. I've nearly been tricked into taking those ratings as gospel without noticing that they are calling for a hurricane to be blowing over the heatsink in operation.
The pentode CCS requires setting a stable screen to cathode voltage (zener diodes are fine for this), then you can set the G1 to cathode voltage to set your standing current. The challenge with these is that the cathode of that pentode will be at several hundred volts, so you will need to bias up the heater winding quite a bit, and this may preclude using the heater winding you would use on the driver tube with the pentode CCS.
-
As I said in my reply to Deke609 (posted March 8 at 04:45:05 PM), the parafeed output transformer is not a simple inductance. To model it accurately, you need to know a number of things, like eddy currents in the laminations and hysteresis of the core material. These things are nearly impossible to find out with sufficient accuracy to rely on simulations. Best bet is to measure and listen with several capacitances.
Here's what I can say about CCS-fed parafeed outputs that are lightly loaded:
The simple analysis gives a capacitance of L / R-squared, but the R is the tube's plate resistance rather than the primary impedance. This usually calls for very large caps. With more usual capacitors, you would expect a substantial resonance. I tried, but I couldn't measure that, because the transformer parasitic losses damp that resonance.
I see. In that case, it seems an experimental approach really is best. Thank you Paul, that is very helpful. I'll get an array of Solen capacitors and alligator clips to swap the caps and be prepared to take measurements, both subjective and objective. I'll just have to budget out an extra, extra large space for the capacitors, just in case. I certainly hope larger the 10uF is not necessary.
You're very welcome for the advice. On the Ohmite heatsink, be extra careful that they aren't giving you a 3C/W rating with some given amount of airflow. I've nearly been tricked into taking those ratings as gospel without noticing that they are calling for a hurricane to be blowing over the heatsink in operation.
The pentode CCS requires setting a stable screen to cathode voltage (zener diodes are fine for this), then you can set the G1 to cathode voltage to set your standing current. The challenge with these is that the cathode of that pentode will be at several hundred volts, so you will need to bias up the heater winding quite a bit, and this may preclude using the heater winding you would use on the driver tube with the pentode CCS.
Good point on the airflow. The Ohmite datasheet for their largest R series heat sink states 3C/W at "natural convection", so that's promising. They have some heat dissipation vs. airflow data as well. I see the dilemma with the high cathode voltage on the pentode CCS. I think my first approach is going to be assessing the feasibility of the heatsinked MOSFET. If all else fails, I will get serious about using the pentode. I'm never against adding more tube feng shui, but it does add a bit more complexity and footprint. I'll do some reading regardless, even if they aren't used in this instance, I'm sure I will utilize them in the future.
-
Here's a set of pentode curves for the E55L with the screen grid 125V above the cathode. For what you're proposing to do, you could use a pair of 60V diodes in series connected to the cathode, then a dropping resistor from B+ to the zeners to get yourself a screen supply that's reasonably well regulated with respect to the cathode. Screen current is about 4mA, and you could look at the zener datasheets to see what kind of current they would like to maintain decent regulation in order to calculate the value for your dropping resistor.
40mA of current corresponds to 3.25V of bias on the E55L. Screen current plus plate current is 44mA, so that's a 73 ohm resistor that you would place between the #45 plate and the E55L cathode. Then you would connect the #45 plate to E55L G1. The 73 ohm resistor could be a 100 ohm pot that you could adjust to get your desired plate current with some precision.
If you wanted to be a little lazier, if you have 400V of B+ and 180V on the plate, then you'd want the E55L screen at about 300V. 400V-300V=100V, and we have about 4mA of screen current, so that's a 25K resistor between B+ and the screen (throw in a 1uF cap from screen to cathode as a bypass). This gives you your pentode CCS with two resistors and one capacitor, and you'll have an impedance of about 30K, so not too shabby. I picked the E55L because the cathode can be 200V above the heater according to the datasheet, so you can heat it from the 6J5 winding.
The E55L would require a minimum of 25V of compliance, so there is an advantage there with the solid state CCS.
-
Here's a set of pentode curves for the E55L with the screen grid 125V above the cathode. For what you're proposing to do, you could use a pair of 60V diodes in series connected to the cathode, then a dropping resistor from B+ to the zeners to get yourself a screen supply that's reasonably well regulated with respect to the cathode. Screen current is about 4mA, and you could look at the zener datasheets to see what kind of current they would like to maintain decent regulation in order to calculate the value for your dropping resistor.
40mA of current corresponds to 3.25V of bias on the E55L. Screen current plus plate current is 44mA, so that's a 73 ohm resistor that you would place between the #45 plate and the E55L cathode. Then you would connect the #45 plate to E55L G1. The 73 ohm resistor could be a 100 ohm pot that you could adjust to get your desired plate current with some precision.
If you wanted to be a little lazier, if you have 400V of B+ and 180V on the plate, then you'd want the E55L screen at about 300V. 400V-300V=100V, and we have about 4mA of screen current, so that's a 25K resistor between B+ and the screen (throw in a 1uF cap from screen to cathode as a bypass). This gives you your pentode CCS with two resistors and one capacitor, and you'll have an impedance of about 30K, so not too shabby. I picked the E55L because the cathode can be 200V above the heater according to the datasheet, so you can heat it from the 6J5 winding.
The E55L would require a minimum of 25V of compliance, so there is an advantage there with the solid state CCS.
Thanks for the example, PB. Let me take a day to dig into the pentode CCS topic, I will be back!
-
Hey PB,
Been doing some reading on pentode CCS. My big take away is the major advantage of using a pentode lies in the effect of greatly increased internal resistance due to leaving Rk unbypassed. It should be equal to:
r(ccs) = ra + Rk*(mu+1)
So if I am understanding correctly, I would expect the resistance of the E55L CCS to be higher? Maybe I am overlooking something, let me know if this makes sense:
mu = gm * ra = 0.045A/V*20,000ohm = 900
r(ccs) = 20,000ohm + 73ohm*(900+1) = 85,773ohm
I've been looking over datasheets for some other potential pentodes for this circuit targeting ~31mA g2+cathode current, came up with what I think might be the best option (with some drawbacks, of course), courtesy of Morgan Jones, the EL83. Attached are the pentode curves at 170V on g2, would target -3V on g1 for a 28mA bias, with Ig2 at ~4ma, would give me close to 31ma out. Check out these flat curves!
If my understanding/math above is accurate, here is what I would expect for a EL83 r(ccs):
mu = 130,000ohm * 0.010A/V = 1,300
Rk = 3V / 0.028A = 107ohm cathode resistor
r(ccs) = 130,000ohm + 107*(1300+1) = 269K
Obviously a lot less than what is achievable with a MOSFET device. With a 5K:16 wiring and a 300ohm secondary, would give a reflected primary of 93,750ohm. A 16ohm resistor in parallel with the 300ohm headphone on the output would drop that back down to 5K, which is an option. Still, the EL83 has a Pa of 9W, so should be able to handle the voltage drop and get that heat above the chassis.
Two drawbacks with the EL83 are the requirement for elevated heaters (Vkf max is 100V), and the drop out voltage of around 70V. If my load lines are accurate, I think the 70V dropout will work with my 400V B+. As an example, I've attached a load line for a 5K primary 8ohm secondary. The max voltage on the negative grid swing is limited by the symmetrical 0V grid point on the positive swing. With a 400V B+ and 31V on the 45 grid, overhead is:
400-266-31 = 103V
Also attached a schematic of what the CCS might look like, need to figure out the value of the g2 to cathode bypass cap. I included grid stoppers, and the heaters would need to be elevated at least 100V.
Gotta do my due diligence and see if this would work, but if my numbers are accurate, the performance hit is pretty massive vs. the cascode MOSFET solution. Have to weigh it against figuring out effective cooling. More tubes is always fun
though :)
Edit: fixed the orientation of the schematic below.
-
In the model I worked through, the impedance was pretty dominated by that screen dropping resistor and bypass cap. These components would still be in parallel with your CCS, but of course the screen bypass cap is reactive so the CCS impedance becomes reactive as well.
IMO there is definitely a point of diminshing returns when you have something like a #45 with an RP of 1.6K, which is relatively low. I wouldn't get too excited about going much beyond 16K at 20Hz, which would have performance like a 125H plate choke. I've also attached some measurements of a random output transformer showing what impedance is actually reflected when unloaded vs. loaded vs. shorted. This is the end of the world as we know it when a pentode is driving the output transformer, but not so much for a triode that has an amplification factor that's rather stable with changing loading impedance. It also illustrates that your 5K:16 transformer won't reflect a 100K load in the real world.
When you move away from the ideal transformer model, you have to deal with all the abnormalities and limits in the transformer design itself that determine the shape of that unloaded curve and the shorted curve, as those are operational boundaries.
-
In the model I worked through, the impedance was pretty dominated by that screen dropping resistor and bypass cap. These components would still be in parallel with your CCS, but of course the screen bypass cap is reactive so the CCS impedance becomes reactive as well.
IMO there is definitely a point of diminshing returns when you have something like a #45 with an RP of 1.6K, which is relatively low. I wouldn't get too excited about going much beyond 16K at 20Hz, which would have performance like a 125H plate choke. I've also attached some measurements of a random output transformer showing what impedance is actually reflected when unloaded vs. loaded vs. shorted. This is the end of the world as we know it when a pentode is driving the output transformer, but not so much for a triode that has an amplification factor that's rather stable with changing loading impedance. It also illustrates that your 5K:16 transformer won't reflect a 100K load in the real world.
When you move away from the ideal transformer model, you have to deal with all the abnormalities and limits in the transformer design itself that determine the shape of that unloaded curve and the shorted curve, as those are operational boundaries.
Very interesting...thanks for the measurements, that does put the transformer limitations in context. And of course the bypass cap and dropping resistor are in parallel, I overlooked it. I suppose they limit the achievable performance of a pentode CCS in this configuration. Well good performance is still very achievable with the EL83, especially if aiming for 16K at 20Hz.
Well thanks for humoring my questions, I'll leave you be. Maybe I'll make a post here when the amp is complete and update on the design I ultimately went with. Much appreciated!
-
Sure, keep us updated on your project!
-
Hey PB and Paul - I have been digging deep into the pentode CCS topic for a few days, there is surprisingly little information out there on using them. Anyhow, I have all but committed to using the EL83 as my parafeed CCS. Hoping I can ask you a question about it first, have a little picture of it below.
Before my question, relating to PSRR at 120Hz I have found this design has 31dB of attenuation. This will work for my power supply since I will have about 2mV rms ripple at the top of the EL83, that will be knocked down to around 55uV before it hits the OPT, so that's okay.
My question is on AC impedance vs frequency. As you pointed out, with the 1uF bypass cap, the CCS is reactive, with the impedance rising as frequency decreases, sort of the opposite problem than the choke loaded parafeed design. With a large g2 dropping resistor on a small signal pentode CCS, the contribution from the cap to the overall impedance might be pretty small, but not so much the case here.
For example, at 20Hz, the impedance is ~61.6K, at 100Hz ~58K, and things start to level off around 200Hz. A friend of mine did a Spice sim of PSRR vs frequency, shown below that shows the trend.
Okay finally my question: will this increasing impedance at low frequencies have an audible effect on the frequency response? Or a subtle shift on the AC load line with higher impedance at low frequencies? I guess I am just wondering if this will be an audible issue, or if I should forget about it and move forward. Increasing the value of the bypass cap would flatten things out a bit, but I am sure that will have some other consequence.
Trying to figure all of this stuff out ain't easy, a lot of respect for what you guys do, every new design change opens a host of new issues to tackle! Thanks guys.
-
Okay finally my question: will this increasing impedance at low frequencies have an audible effect on the frequency response?
The math for gain of a grounded cathode amplifier with a cathode bypass cap is:
Av = (mu*RL)/(RL+rp) where rp is the plate impedance of the tube an RL is the load seen by the tube.
In your case you can get a snapshot of gain at a given frequency to make this determination. Your RL will be the CCS impedance in parallel with the parafeed cap reactance plus the OT reactance. The output of this is in terms of amplification factor, so then you can take the log of that number and multiply it by 20 to find the gain. The mu of the #45 is 3.5 and I would use 1600 ohms as the #45 rp, which will be close enough (to save you some time looking it up).
Av = (3.5*RL)/(RL+1600) To go down 3dB from an infinite load, you would need to drop from that infinite load to about a 4K load.
-
Okay, let me give this a shot...
My calculated impedance at 20Hz is ~61.71K, with a 5K primary, corresponds to a RL of 4625, gain of 8.30dB
Calculated impedance at 1kHz is ~57.13K, with a 5K primary, corresponds to a RL of 4597, 8.28dB
So a boost of 0.02dB at 20Hz relative to 1kHz......yeah, this doesn't matter at all. Thanks for showing me, PB, I'll save for formula for next time.
-
I am enjoying this thread as well. Plant some seeds for a future project...John
-
Thanks for showing me, PB, I'll save for formula for next time.
Thanks for doing the leg work!
-
Hey Paul/PB - hope I can bug you guys again. My 45 parafeed headphone amp is pretty close to being build, just waiting on some transformers to arrive. I have a bunch of Solen films to figure out the value of the parafeed cap.
When it comes to deciding the parafeed cap value with a CCS load, what other measurements are useful other than a Bode plot / frequency response? I'll be doing subjective listening tests as well.
Once the cap value is determined, I'll probably upgrade to something fancy, I've carved out a sizable chunk of space in the chassis for some huge capacitors. Surprisingly, the Solens are quite small.
Thanks!
-
I would sweep the amp with a variety of headphone impedances (try to get a 16 or 32 ohm headphone and a 300 or 600 ohm headphone) and look at what the amp spits out. There's probably some middle ground value that will make the most sense.
-
Awesome, thanks, I'll let you know how it goes!
-
Yeah, do post your results, it's a useful reference.
-
Will do, I will screenshot the plots with supporting information and post them here.
-
Okay PB, I have been quite the busy bee...here is where we are at with the parafeed caps in this amp, which is otherwise complete :)
I took FR sweeps of five different capacitances: 3.3uF, 4.7uF, 6.8uF, 8.2uF, and 10uF. Not sure how to insert attachments on the BH forum, but I have included them in a GIF format attached below, easier to see the relative change of low end response this way, but let me know if you would like the individual plots.
As you can see the effect on low end response from 20-30Hz is pretty subtle, we are talking about a difference of ~0.3dB when tripling the capacitance from 3.3uF to 10uF.
*I am copy/pasting/reformatting from another forum, subjective impressions were done in two different sittings.
Part 1:
Okay, so we've established there is a very small difference in frequency response from 20-30Hz. But how does the sound change subjectively? I first decided to compare at the extremes, 3.3uF and 10uF. The results were somewhat unexpected...
I much preferred the sound of the 3.3uF. Even if the low-frequency extension is marginally improved with the 10uF caps in, there is a very noticeable loss of soundstage, airiness, and detail. The treble sounds a bit more harsh to my ears as well with 10uF. The 3.3uF sound is much more spacious and cohesive. Up to this point I had mostly been listening with the higher capacitance caps (6.8uF and 10uF), but the sound is much improved with 3.3uF. In fact, the bass sounds better! Better definition even if the extension is objectively slightly worse.
So there are clearly two competing factors in play here that will need to be balanced when assessing the other three caps. I can say without a doubt I will not be using the 10uF, bigger does not necessarily appear to be better.
This is an interesting finding since using a lower value cap opens up many more possibilities in terms of the quality of caps that can be used. I had somewhat assumed I would need a value higher than 5.6uF, so my top choice was the Rike Audio S-Cap 2 (aluminum foil paper in oil) as these caps are the appropriate size, value, and are well-reviewed for their "neutral" sound. They are also quite affordable.
So, there is more listening to be done.
Part 2 (after the suggestion was made that differences heard between 3.3uF and 10uF may not have been all due to the capacitance - try the caps in parallel for lower ESR/ESL):
Worked on getting the parafeed capacitance nailed down this morning. I started by comparing the 4.7uF and 3.3uF in parallel vs. the single 8.2uF cap. I can say without a doubt the two parallel caps sounded better. Better sounstage, air, and detail. Similar to what I said about the 3.3uF vs. 10uF, the singular higher capacitance cap causing something of a collapse of soundstage and more harsh treble.
Next, I compared the 4.7uF cap to the 4.7uF and 3.3uF in parallel to compare a singular lower capacitance cap with the higher capacitance combination. Again, I felt the two parallel caps gave better detail, soundstage separation, and this is the first time I felt the bass was significantly improved between two sets of caps. Fuller, tighter, more body.
Next, I compared two sets of parallel caps with different capacitances. With what I had on hand, I went for the two lowest capacitance combos for the sake of practicality, 4.7uF + 3.3uF and 6.8uF + 3.3uF. This is where differences started to get more difficult to detect...however, I did feel vocals started to get smeared a bit with the higher capacitance and focusing on other aspects of my test tracks revealed a loss of microdetail.
So, with the parallel combination of around 8uF, I was getting the best sound. But what about the relative values of the two capacitors? I then compared the 4.7uF + 3.3uF combination with an 8.2uF + 0.22uF combination. Not an exact capacitive match, but close enough with the 5% tolerances. I felt the two caps of similar size gave better nuance and vocal detail.
If two paralleled caps is good, is three better? I put the 0.22uF cap in parallel with the 4.7uF and 3.3uF for a total of 8.22uF. Adding this 0.22uF cap undeniably reduced the fullness and quality of the bass.
So my winning combination was the 4.7uF cap and 3.3uF cap in parallel. From the measurements, I know that diminishing returns in the bass response start at around 6.8uF, so I will be looking to find a parallel set of caps with similar values that add to ~6.8uF, perhaps 3.3uF + 3.3uF or 3.9uF + 3.9uF. Obviously this will make the size restriction more stringent, but I think the performance is worth it. It's pretty incredible the gains that can be made in this topology by tuning the parafeed caps.
I did briefly drop the 0.22uF cap in parallel with my 0.47uF coupling cap to see if similar gains could be had there. A few back-to-back comparisons was not enough to say definitively, and I was pretty fatigued at that point, so I'll revisit it.
-
Beautiful response!
-
Cool!
-
Were the caps all the same brand and type?
-
Yessir, all Solen 630V metallized polypropylene, purchased a set for the purpose of this experiment before committing to more expensive caps.
-
Also probably worth mentioning - on this amplifier, I ultimately abandoned the EL83 pentode CCS on the 45 in favor of a second cascode MOSFET CCS with ground-isolated heat sinks on the exterior of the chassis.
After crunching the numbers, I was only going to get something like 30dB PSRR from the EL83, so the performance hit was just too much.
So sorry PB! Didn't mean to waste your time, but thank you for helping me understand the particulars of a pentode CCS.
-
This is fascinating stuff. Thanks for sharing your measurements and impressions of different cap values and combinations.
Does anyone know or have theory-/experience-based speculations about whether larger cap values take longer to break in than smaller ones? Just curious about whether some part of Keenan's initial preference for a small cap over a big one, and subsequent liking of two smaller caps in parallel might be explained by this.
Keenan: were all the caps new, or had they had equal hours of use, before you tested them?
cheers and many thanks, Derek
-
30dB of PSRR isn't so bad if your power supply ripple is already low.
-
I was only going to get something like 30dB PSRR from the EL83, so the performance hit was just too much.
How much did you calculate that you'd need?
30dB PSRR means 3% of power supply ripple is left, so a power supply with 2mV of ripple (noisy) becomes 60uV (inaudible).
-
How much did you calculate that you'd need?
30dB PSRR means 3% of power supply ripple is left, so a power supply with 2mV of ripple (noisy) becomes 60uV (inaudible).
In my simulations, I think I was going to have around 6mV of ripple on the 45 B+, I just felt more comfortable with the FET option knowing it was going to be nuked into the noise floor. Maybe some element of "fear of the unknown" as well. I had a custom transformer wound for this amp, so it was one less heater winding too since it would have needed elevation.
Don't make me regret it, PB! I have a quad of nice Valvo EL83 on hand, so maybe I will give it a try in a future amp.
And Derek - all of the caps were brand new and unused before the listening tests, other than the bit they were in the amp for the FR sweeps.
-
I just wanted to thank PB and PJ for their input on the parafeed topology, this amplifier was completed in April.
Below is an FFT at 350mV into 300 ohms. Not sure why I didn't measure at 1mW into 300ohms at the time, but this is what I have. I think this is what I determined to be listening volume with my headphones.
Also the FR plot, I ended up using two 3.3uF Rike Audio S-Cap-2 aluminum foil PIO caps in parallel, after my experiments with the Solen caps, so 6.6uF total on the output.
I put some globe 45 tubes in this amplifier as well, they are very nice sounding tubes, I got them for a bargain on eBay since they were untested. One had a filament to grid short, I whacked the tube a few times and broke the short, they test over 100% emission on my Jackson 648-R.
-
Looks good!
-
Thanks! I'm very happy with it, I appreciate the help, every amp build is a learning experience. I'm working on something truly insane now....
It is a three-stage all DHT class A2 hybrid amplifier. Gain stage is a CCS loaded 841 (high mu cousin of the 801A) with filament bias, cap coupled to the second stage which is a transistor source follower buffer, which is direct coupled to the grid of the output stage, a fixed bias 801A with an A2 bias point (420Va, -20Vg, 40mA Ia) into a 11K:8ohm OPT. The source follower drives the grid of the 801A far into A2 territory, the grid swings from around +50V to -100V for roughly 6W into 8ohms out of the 801A! I just finished the prototype, sounds darn good too.
Anyway, thanks again.
-
I thought it was a #45 amp?
-
Oh yes the parafeed amplifier is a #45, this is a totally new design, sorry for the confusion. I have the DIY bug, I probably should take a break.
-
The 801 rewards a lot of effort! I was wondering how those FR graphs could be from such a high impedance transformer.
-
Wow, that looks incredible! It is an A1 design with a choke input supply?
Yeah, not getting the same FR results with the 11K:8ohm traffo HA! Bass extension is great but I do have some HF rolloff, -3dB at 16.5kHz, 20kHz is at -4.4dB. Below is the FR at 1W into 8ohms.
Even with the HF rolloff, it sounds very good and the bass definition is fantastic.
-
No, that's A2 biased (about 0V on the grid). It is a choke input supply since the 801A doesn't need a whole ton of voltage to make a decent amount of power. I ran it at 325V/60mA.
-
Very nice, I bet it sounds great. The 801A certainly does reward a lot of work, didn't realize how finicky a tube it is until I committed to using it. Glad I did though, I have learned a ton working through the...peculiarities of an A2 biased amplifier. My transistor buffer stage will hide inside the chassis, no one has to know :) I plan to trick some tube purists at future meets, certainly no one would put a transistor in the signal path of an all DHT amplifier? Pure evil.
-
The grid current demands of the 801A were modest enough in my design that I just used a 6S4 cathode follower. It did require a negative voltage rail for the driver though.
-
I was going to use a cathode follower at first, I had the 6BX7 in mind, but I had read of good results in A2 designs using source followers, so I decided to give it a try and I am happy with the results so far. Going to try a few different chips and see how it affects performance. I'll probably try a cathode follower A2 design in the future if the mood strikes again, but this is a ton of work.
-
I'm curious PB what primary impedance you used with 0Vg bias point?
-
3K.
-
I see thanks, I assume some feedback was used? I’m afraid I may have to live with the HF rolloff in my design, perhaps I should have looked into using feedback myself, but I suppose -3dB at 16.5kHz isn’t horrible.
-
Yes, I used feedback between the plate of the 801 and the plate of the first stage driver tube.
-
Got it, thanks PB. My OPT can also be wired for 6.5K:8ohm rather than 11K:8ohm which I'm sure will help my HF rolloff. I bypassed my 1K grid stoppers along the signal path with a 100ohm resistor and took FR measurements, no change, so I think it is safe to say it is due to leakage inductance / stray capacitance of the OPT. I think I am going to look into using a 6.5K impedance, altering my bias point, and adding NFB. Simply changing the primary impedance to 6.5K and my output impedance will be far too high, damping factor less than 2 with 8ohm speakers. NFB is new to me, we'll have to see if it will fit in with the rest of my design.
-
Going off of what I said above, I think I will give this bias point a try, 350V 50mA -9Vg. Below is a load line with 6.5K primary, this is at 87.5% of max plate dissipation.
I guess it is time to learn about NFB, been putting it off, hope it will gel here.
-
This thread has an approach that is far less intimidating than others:
https://www.diyaudio.com/forums/tubes-valves/356952-output-tube-plate-driver-cathode-feedback.html (https://www.diyaudio.com/forums/tubes-valves/356952-output-tube-plate-driver-cathode-feedback.html)
Remember that you need lots and lots of gain so that when you add the feedback network you end up with enough leftover.
-
Awesome, thanks for the link, I'll do some homework. The measured mu of my CCS loaded 841 gain stage is 29, so it can swing a hefty amount of voltage into the grid of the 801A. With the new bias point, will need about 100Vpp to drive to full output, so I should have an extra 50Vpp of headroom with a standard DAC input of 5.6Vpp, so hopefully that will be enough.
-
29 ain't gonna do it! You'll want to move to a pentode up font to give you a lot more wiggle room.
The first schematic in the link I sent you shows a pentode driver that will have an amplification factor over 1000x. That sounds like a lot, but 1000x is only 60dB ;)
-
Nothing is ever easy, is it? Okay well no point in speculating on what I can do without the knowledge of how/why to do it, let me do some reading and come up with a plan :) bummed I wouldn't be able to use my 841 driver, guess I am going back to the drawing board.
-
So rather than going straight to a total input stage overall PB, I decided to try something else first, NFB from the secondary of the OPT to the cathode of the driver stage, the 841. Picture I found attached.
I used 500ohm on the 841 cathode bypassed with 10uF film cap + 50ohms to ground (R2 in the diagram) . This got me my 500Va / 9mA / -5Vg bias point and LF extension down to 20Hz. I tried bypassing with an electrolytic first and it sounded horrible, the film cap sounds great. I then altered the value of the feedback resistor (R1) from the OPT secondary. I tried 2K, 1K, and 500ohms. Obviously I am sacrificing gain from the driver stage for other benefits. Here are the results:
No NFB
Output impedance: 3.95ohm
HF -3dB: 16.5kHz
THD at 1W: 0.78%
Power output: 5.85W
R1 = 2K
Output impedance: 3.47ohm
HF -3dB 18kHz
THD at 1W: 0.71%
Power output: 5.85W (still enough gain to drive the 801A to clipping)
R1 = 1K
Output impedance: 2.50ohm
HF -3dB 20kHz
THD at 1W: 0.67%
Power output: 5.3W
R1 = 500ohm
Output impedance: 1.02ohm
20kHz at -2.3dB
THD at 1W: 0.61%
Power output: 4.3W
What do you think? Other than the decreased gain, not sure if there are other issues caused by using feedback resistors as low as 1K or 500ohms, the 1K results seem to be a nice compromise to me and gets me a 1:3 damping ratio with 8ohm speakers.
-
Putting feedback around the output transformer will make the amplifier at best conditionally stable, and at worst it could oscillate. I just had a home brew amp someone brought in my workshop the other day that oscillated at 150kHz just sitting with a speaker hooked up. This is not a desirable outcome!
We would tend to recommend the shortest loop you can possibly use to get the job done. Cathode feedback is a super short loop, but that requires a very special output transformer and doesn't work so well with parallel feed. Output tube plate to driver plate is a very short loop, and it's what we used on the Seductor (and what I have used on several personal projects). Output tube plate to driver cathode is a bit longer of a loop but far easier to implement, and that's where I would recommend starting. Output transformer secondary to first stage cathode is the longest and most problematic loop.
Still, your experiments are a decent demonstration of why the 841 has to go in favor of a pentode. There are directly heated pentodes...
-
Point taken, PB, I have not ruled out the pentode! Just wanted to see what could be done here. The amp appears to be stable, at least as it sits here in my garage, no oscillations on my scope and it sounds quite good. But I will look into the pentode further with plate to cathode feedback. I appreciate your $0.02.
-
I am going to mock something up with the EF37A in LTSpice, seems a popular choice as a pentode input with NFB.
-
I took your pentode suggestion seriously PB, I have been doing lots of reading and I am working it up now.
I was looking over some of the work Gary Pimm did on CCS loaded pentode stages. Looks like he came up with two different ways, the first with a high value resistor in parallel with the CCS, the other with the resistance to ground (picture attached). I am looking at using the EF37A or 6J7G on the input, I mocked up a model using plate to cathode feedback with Pimm's "resistance to ground" method. I need to take a closer look at the "paralell resistor" method, I am doing something wrong in LTSpice, since there is very little current across the resistor, there is no Vdrop and I am getting full B+ on the tube plate. I am sure the answer is obvious but my brain is getting tired.
In my sim with 250K to ground connected to the 6J7G plate, the open loop gain of the amp is roughly 59dB and the closed loop gain is roughly 44dB, so about 15dB of NFB. This leaves me with a mu of about 30 from the driver stage (open loop mu ~165), plenty of swing to drive the 801A to full output. With this setup, I would rewire my OPT to 6.5K:8ohm and a 370V / 50mA / -11 Vg bias point on the 801A.
I penciled in 50pF on the phase compensation cap, but will need to dial in the value. Overall I think this looks pretty good! I know it isn't accurate, but in LTSpice I am getting 0.25% THD at 1W, 0.56% THD at 3W. If that is anywhere close to real life results, that is an incredible improvement.
Will check my work a few more times and dial in part values, have some tubes on the way to give this a try.
-
I would suggest bypassing R6 and rerunning the sim. Try 1000uF.
As far as plate voltage goes, you need to set the CCS to deliver adequate current for the plate load resistor plus the plate current of the tube itself, so that can be a little tricky.
-
I'll give the bypass cap a shot, PB. Yes, setting the CCS has been a bit challenging! A little bit of guess and check to get the correct plate current. I'll report back when I have some real world measurements.
-
If you bypass the bias resistor, you can turn it into a pot in your build and just adjust it to get the plate voltage to sit where you want it to.
Alternatively, you don't have to use a CCS at all.
-
That's a good idea! Let me think on it.
Yeah, a 75K-100K resistor on the plate would get the job done. I would lose the very low output impedance from the mu output of the CCS though, but in terms of HF rolloff, unlikely it would make a big different with the source follower buffer. I'll probably give both methods a shot and see what measures / sounds better. The resistor load certainly is less of a bother...
-
Yes, you don't need the mu output and the fet buffer, that's a little overkill.
-
Yeah, overkill for sure. Distortion is significantly better in the sims with the CCS load however, 0.25% THD with CCS at 1W, 0.42% THD with the resistor load at 1W. I'll give them both a go and see what the real world has to say about it.
-
When you go to the plate curves and try to draw out the DC load line for the pentode driver, you'll see the plate voltage problem pretty clearly. This is made worse by a fair amount of your bias voltage being relatively fixed because of how the feedback is setup. By going to more of a resistive plate load, you can stabilize the DC operating point a little bit. The extra 0.2% THD is probably worth the trade.
-
I see...very small changes in cathode voltage will result in drastic changes in plate voltage and bias point. Is this what you are referring to, PB? It does seem a good tradeoff for the sake of simplicity. As I have continued to try different operating points, balancing the CCS current, plate voltage, cathode voltage, while maintaining the appropriate amount of NFB is...difficult. They are interdependent, getting the operating point dialed in with the CCS will be quite a hassle. I had hoped to leave an electrolytic out of the signal path by leaving the cathode unbypassed, I have had good success with this in the past with CCS loaded triodes, but I suppose that may not be an option with a resistively loaded pentode.
-
I see...very small changes in cathode voltage will result in drastic changes in plate voltage and bias point. Is this what you are referring to, PB? It does seem a good tradeoff for the sake of simplicity.
Yes, we ran into this issue with our Eros phono kit, which uses a CCS loaded pentode with a resistive load from plate to ground. PJ cooked up a servo circuit to deal with this issue.
-
Interesting, I think I will save myself the headache and just go with a resistive load, thanks for the warning. If I could pick your brain yet again PB, is there a practical method to determine to appropriate phase compensation capacitance in parallel with the feedback resistor? I have been told by some that the value isn't critical, just throw something ballpark on there and check for oscillations / square wave ringing. I have read others that will put a scope probe on the input and output and increase the frequency on a signal generator until they find 180 degree phase shift.
-
You may not need one at all! You could look at 10kHz square waves and add/subtract capacitance to examine the shape of the waveform.
-
Good point, I had wondered if I might not. Okay, I'll give that a go and keep some capacitors handy just in case. As a non-engineer guy picking this stuff up, finding good resources on specific topics can be tough, its a big help to be able to bounce things off people, so thanks again.
-
Alright PB, I have some preliminary pentode results...
Just to recap, in the original design with the 841 and no NFB, I used a 420Va / 40mA Ia / -21Vg bias point for the 801A with an 11K:8ohm output. My FR stated to roll off around 6-7kHz and my -3dB point was 16.5kHz, my output impedance was 4ohms.
For this new NFB pentode setup, I figured I could get the output impedance down to a reasonable level even if I rewired my OPT for 6.5K:8ohm. I rebiased the 801A to 360Va / 50mA Ia / -10Vg with the 6.5K primary.
The pentode I used is the Mullard EF37A. I biased it with a 400V B+, 200K plate load, 200V on the plate, 100V on g2, 3.2V on the cathode. NFB was taken from the plate of the 801A to the cathode of the EF37A (can't say exactly how much, will have to run some sims to get an idea of the open loop gain). Ended up with a mu of 20ish with the NFB added.
Output Z: 2.8ohm
THD @ 1W: 1.2%
Power output: 4.7W into 8ohms.
Frequency response below, vastly improved from the build with 841 driver and no NFB. Doesn't truly start to roll off until around 16.5kHz, down 2.5dB at 20kHz.
Going to play with the bias and load of the EF37A to see if I can squeeze out a little more gain. Wondering if I should try the same with my 11K:8ohm OPT wiring and 420Va / 40mA Ia bias point. I was expecting more than 4.7W with this new arrangement, will have to give it some thought, must have made a mental error somewhere.
-
It may help to draw out and post the schematic.
-
I figured it out PB, with my OPT wired for 6.5K:8ohm, the copper losses go up from 0.5dB to 0.8dB. I am getting around 572Vpp into a 6.5K load, with the copper losses I should be getting a little over 5W out, and that is what I am seeing. The 801A is clipping asymmetrically though at the positive peaks with my 360V 50mA bias point, might drop it down to 350V 50mA and see if I can get a more symmetrical clipping, my OPT are made for 50mA on the primary, so can't crank up the current, that would put me at 17.5W plate dissipation, so not a bad place to be for longevity.
Thanks for suggesting the pentode NFB route PB, someone else had suggested it too early on in my project, I poo poo'd it at the time, "NFB in my amp? Never!" I won't be saying that ever again. I lugged this monstrosity into my house today (it is built on a piece of plywood, picture below). Gave it a listen with my Snell J/II that I recently restored. This amp sounds INCREDIBLE, I mean, really it is something else, even compared to my 45 parafeed amp and my 6A5G SET, which are no slouches IMO. The detail, bass definition, eerie level of imaging...
I was listening with headphones today with a 8.2ohm resistor paralled on the output, at a pretty high listening volume for headphones. I left the volume where it was, unplugged my headphones and measured an FFT. The distortion was 0.023%. So super low THD with headphone listening, around 1.2% THD at 1W into 8ohms.
Going to further optimize the NFB tomorrow and maximize it while still driving the 801A to clipping. These EF37A tubes are pretty cool.
-
Nice! You are not the only person I know looking for the perfect amp to go with 8" Snell 2 ways. These are my current A2 amps that get the most use. The feedback loop I use is from the output tube plate to the plate of the driver pentode.
-
WOW! That is something, what tube is that and how much power? Cathode follower is a 6BL7 or 6BX7?
I made some changes to the EF37A plate load this morning and optimized the bias point / NFB. I used two 1K trimmers stacked on the cathode and bisected by the feedback resistor so I could adjust the bias point and feedback loop. I maximized the NFB while keeping the 801A just below clipping at max signal input (2Vrms) and kept the EF37A at a 220Va / 96Vg2 bias point.
Now getting an output impedance of 0.38ohms! And the FR has improved further, I wonder if I am hitting transformer limitations at this point given how low the output impedance is.
-
Yeah, you're running up against transformer issues there. The good news is that there are tons of 5K transformer options that aren't so limited.
The amps in the photo are 100TH with a 6BL7 as a directly coupled cathode follower to drive the grid of the 100TH. The extra half of the 6BL7 regulates the B+ for the cathode follower, which is something PJ suggested when he was visiting last year. The input tube is an EL41, which was about the only pentode I could find that didn't care too much about high plate voltage and had the middle of the road transconductance that made it work properly in the circuit.
-
Funny to think I am running up against transformer issues, these are from a reputable builder, although I suspect the designer may have sacrificed some HF extension for a high primary inductance, 100H on a 6.5K primary, these can be rewired for 11K and 23K as well, I thought I would be using them at 11K. The thought of replacing them isn't super appealing, will have to do some soul searching and decide if 17kHz is good enough, certainly beyond my hearing ability. I'll think about it, but the sound I am getting right now is sensational, the bass is jaw dropping...worried if I replace them I could end up with something less satisfying, "the devil you know is better than the devil you don't."
The 100TH...that one is out there, never even heard of it! Very cool looking tube though. I imagine when you have been building tube amps as long as you have PB, you have to get more and more esoteric to stay occupied. I also imagine they keep you nice and warm in the cold winter months :)
-
When you make a transformer that is either 6K or 12K, depending on how you wire/load it, that is already a bit of a compromise. I think this is more doable with a 5K/2.5K, as it's easier to land right in the middle and make a fully optimized 3.75K:6 transformer and ratio it accordingly. It's still a compromise though!
You are using the big Lundahls right? I have been pleasantly surprised by the performance of the 5K One-Electron output transformers, especially up top. They aren't insanely expensive and are plenty big to get all the power out of an 801 that you could possibly get. The datasheet says you'd be -1dB at 30kHz with this iron, but the last amp I made with them was -1dB at 42kHz, which was rather surprising. Said amp was -1dB at 22Hz, so about the same down low as your current transformers.
-
Interesting, always compromises in this hobby. Well thank you for the recommendation, I'll check out those One Electron transformers, -1dB at 42kHz, that's pretty impressive. That's right, I am using the Lundahls, these are the LL9202. I've used Sowter as well, tried reaching out to Magnequest recently for a potential project I am cooking up but never heard back. I've liked the sound of the Lundahls so I have stuck with them, but always open to trying something else. Thought about trying Electraprint as well, they seem to have a good reputation in the DIY world.
-
If budget allows, Sowter and Monolith are really solid options. For a ~10W amp, the prices aren't completely insane either.
-
Thanks, I'll give the OPT changeout some thought, I have an order pending with Sowter right now actually...it might not mean much considering I am only three scratch builds into this thing, but this might be the best amplifier I have ever heard...no kidding, it's that good. So perhaps it is worth squeezing out some more performance with a wider bandwidth OPT, assuming the LL9202 are not a large part of why it sounds so great. I'll sleep on it! For a week.
I think the most I have ever heard of someone getting out of an A2 801A is 8W, you didn't mean to imply you could get 10W out, did you PB?
-
I got 8.5W. Perhaps with Sowters you could go over 9W.
-
Wow that would be interesting. I am in the process of reworking my 6A5G amplifier, transplanting into a new chassis. It uses 3.3K 60mA OPT, maybe I will throw them in this prototype and see how it works out. Only thing is I would have to rethink the bias supply, it uses a LND150 CCS and a 56K resistor to ground to set the bias. If I just used a resistor to ground instead, the bias would be set by the gate-to-source voltage of the FET buffer, something like -4 to -5V. Wouldn't be adjustable though, have to look into it more deeply and think it through...
-
Just be sure your loadline still makes sense in terms of giving you the power you are expecting. I ran the 801A at 300V/60mA (the 60mA was measured as plate+grid current) and got about 8.5W.
-
For sure, would have to alter the B+, bias supply, etc. to give this a try, but shouldn't be too bad I don't think.
I have to admit PB, I am not seeing how you got 8.5W out of the amp with a 300Va / 60mA bias point and a 3K load, I believe it, but something is missing. I've attached a load line, I am seeing at max something like 360Vpp into 3K, limited at the positive peaks, that's about 5.5W by my math. What am I missing here?
-
I went back over my notes and it looks like I was more like 330V/60mA.
-
Hmmm still don't see 8W at that load line, but maybe it is the real world difference between the datasheet and live implementation? You aren't the only person I have heard that from, I know of another builder who used a nearly identical bias point, 300V 60mA, and got 8W output. Maybe there is some magic in that bias point.
-
Also the curves are hand drawn off a "bogey" valve, so some variation isn't surprising.
-
Fair point, well over the next week or so, going to see if I can throw these 3.3K 60mA OPT in the prototype and see what kind of sound / power output we get!
-
PB - I tried out my 3.3K transformers this morning.
I nixed my bias supply and just used 520K to ground, not ideal since the grid voltage isn't really adjustable, it is fixed by the gate-to-source voltage of my FET buffer, so -3.1V on the grid.
I was able to adjust my bias point to some degree using the trimpot on the Maida regulator I am using. I dialed it in to a 335Va / 57mA / -3.1Vg bias point.
With this, I am getting 156V to 524V peak-to-peak swing, with the 801A hard clipping beyond 524V on the positive peaks. This results in 16Vpp into 8ohms, about 4W out. This is pretty darn consistent with the datasheet if I draw out the load line.
Ideally, I would get to a true 0V bias point and 320V with a 3K (rather than 3.3K) primary impedance, it is doable, have to put +3V on the gate of my FET. At that bias, if the datasheet is to be believed - and so far has been consistent with my measurements - even assuming no copper losses, that makes 5.5W into a 3K load. Even if you did something crazy and biased at +10V on the grid at 280V on the plate, that is 400Vpp into 3K, makes a 6.4W.
So, I am just not seeing how it is physically possible to get 8W out of an 801A A2 amp at any 0V bias point. Am I missing something here?
I added additional NFB given the reduced gain needed from the driver stage, got the output impedance down to 2ohms. FR of the LL1620 below, think I am hitting transformer limitations on the high end again.
Thanks for your thoughts.
-
Here is my set of load lines, as I had considered 3K and 5K iron for this project. This is swinging 520 to 70V, so 161V RMS. A 3K:8 is about a 19.5:1 step-down, so that gets you 8.25V at the speaker posts, which is just over 8.5W.
-
I see, thanks for the pic.
But the amp is going to hard clip asymmetrically on the positive peaks, is that not typically considered when quoting an amplifier's power output? This is exactly what I see in my measurements, the positive signal is chopped off above 524V whereas the amp can continue to swing downward until the plate hits saturation at 70V or so.
-
The 8.5W was at 10% THD I believe. You'd have to make a plot of THD vs. power to determine where you'd say hard clipping starts. The asymmetrical clipping will be pretty brief IMO in the grand scheme of things.
-
Gotcha PB, I see now where I was making the mental disconnect. Well regardless, this is basically a full A2 amp in this setup, going to give it some listening time, then decide how it compares to the LL9202 setup with the 6.5K primary. Thanks for letting me pick your brain, yet again.
-
I made an error in a previous measurement PB, my 0.38ohm output impedance wasn't reproducible, so something was off. I was more consistently getting 2.8ohms output impedance with a 220K plate resistor on the EF37A. I wanted to improve this, but using a bypass cap on the cathode of the EF37A degraded the sound quite significantly. I decided to try the CCS setup.
I simulated in LTSpice to get ballpark values, then went for it. I used 1.56mA out of a cascode CCS with a 470K resistor to ground. I figured maybe if I used a larger value without going crazy, I might not have DC bias instability issues but could still get a damping factor of 3 out of the amp. EF37A bias point was Va 240 / Vg2 95V / Vk 2.77V. Leaving the cathode degenerated, I was still able to get a open loop mu of 125 out of the EF37A. I returned to my 6.5K LL9202, I think the sound is better than the full A2 setup with the 3.3K LL1620.
THD @ 1W 0.44%
Output Z 2.44ohm
-1.67dB @ 20kHz
Bandwidth improved as well. I monitored the plate and cathode voltages of the EF37A with music playing at varying volumes, did several shut-downs and start-ups as well - the bias does not drift, it finds its way right back to the same point every time.
I think I'm gonna stick with this approach and improve up on it!
FR and unclipped full output below, this is a 5W amplifier :)
-
Still working on this tricky amplifier, I was finding that adding capacitance to stabilize the feedback loop was hurting the HF response and there is none left to spare. I pulled the trigger on a pair of 5K:8 60mA custom OPT from Sowter UK, no more of this bandwidth issues from the Lundahl LL9202, I'll have to repurpose them. Specc'd the transformers for 60mA in case I want to go for a 0Vg bias point, something like 330V 55mA, either than or 360V 50mA.
-
I kind of feel like I'm corrupting you...
I've had Sowter make plenty of iron for me over the last couple of years. Spending the dough on their stuff lets you focus a lot more on the circuit itself rather than worrying about what the iron is doing.
-
No worries on the corruption :) I have an order pending with Sowter for another project, in truth I was already considering the switch after all the frustration with the Lundahl transformers. I am very committed to this build, so I think it will be worth it, could always resell the Lundahls. Good thing to be able to focus on the circuit, having some issues with ringing on square waves that I am working through. Maybe the issue will disappear with the Sowter OPT in place, wouldn't that be nice.
-
You could also experiment with a zobel across the secondary. Heck, you're paying Sowter for a custom wind, Brian might be able to tell you the optimal values for R and C.
-
Good idea, I'm wondering if it is an artifact of the Lundahls specifically. I checked square waves on the pentode and source follower with the 801A out of the circuit (no feedback), no ringing whatsoever. It is present on the primary and secondary, near the audio band based on the period of the ringing, roughly 50uS, so around 20kHz, don't want to shunt 20kHz to ground.
Anyway, I'm probably going to shelve this beast until the Sowter OPT get here, probably not worth troubleshooting issues that might go away with a more optimal iron.
-
Hey PB (and PJ if you happen to stop by :)) ) - still working on my 801A A2 amplifier, still waiting for the Sowter OPT, they are a bit hampered with COVID so I think building is slow, should have them soon.
Wanted to show you something pretty cool and also ask if you have any input on something else. I have been hanging around diyAudio and discussing ideas for these high-gain CCS fed pentodes. A certain someone who maintains a certain DIY tube blog suggested a fancier g2 supply for my pentode driver rather than the 1Meg resistor from B+ I was using before. The idea is to use a local feedback mechanism with the DC voltage fed to g2 via FET, with the voltage derived from a divider from the mu output of the CCS. Seems similar ideas are floating around a few places, but the one that inspired the suggestion to me was done by JC Morrison on his transconductance gain block here:
http://www.labjc.com/?p=5317
In this arrangement, any drift in the plate voltage is communicated down to the screen, I put together some Spice models, built out the circuit on a protoboard and gave it a shot, schematic below.
What I found is this setup provides VERY stable anode / screen voltages. At maximum output into a 8ohm dummy load, sine waves as well as real music, I never see a voltage drift more than 0.5V on either, pretty nice, so this is going in the final design. With the exact setup I have shown below, I am getting 0.14% THD at 1W out of the amplifier.
This design is 95% of the way there, just a matter of dialing in the feedback / output impedance / FR / harmonic profile when the Sowter stuff gets here. The only niggling thing left is some somewhat ugly overshoot / ringing I am seeing on square waves, shown below as well, the output from the EF37A. The amplifier is stable, perhaps it is a non-issue or will resolve with the Sowter OPT, still need to do some more troubleshooting, but the Lundahls seem quite prone to this type of ringing, we will see.
Stay safe fellas! Hopefully 2021 is a better year.
-
What happens if you adjust the value of C2?
-
I changed C2 to 4.3uF without any change in the overshoot / ringing. I discovered the ringing goes away with the feedback loop disconnected. The outputs of the EF37A and source follower are clean, it originates at the output stage. The ringing is 45-50kHz when open loop, roughly 20kHz closed loop. Adding any feedback compensation capacitance starts to roll off the HF, so that isn't a viable solution, especially with it being so close to the audio band.
I wonder if this is just another issue with the LL9202 OPT and it will go away with the Sowter iron in place, maybe I should just wait it out and see.
Below is the 1kHz square of the 801A plate with the feedback loop disconnected.
-
I think hanging around and waiting for the Sowter iron might not be such a bad idea.
-
Agreed, these Lundahl OPT are so quirky, I think it's best to just wait it out. Not sure what I will do with them afterward, despite their oddities they are very nice sounding transformers. Thanks again, PB.
-
What do you think of this PB...
A friend of mine made a comment that got the gears turning on how to simplify my 801A A2 design.
Originally, I was using a source follower buffer to drive the 801A grid into A2 with V+/V- supplies such that it could swing from +60 to -70V. What I am experimenting with (in LTSpice) is eliminating the source follower, elevating the 801A filaments (~200V), and direct-coupling the 801A grid to the mu-output of the EF37A CCS. This does away with the V+/V- supplies, allowing the grid to swing positive/negative with respect to the filament, and eliminates a coupling capacitor. The output impedance of the DN2540 appears low enough to provide the grid current for A2 and does not seem to be adversely affecting the plate current of the EF37A. Could be different in the real world, I've attached my functional but not refined schematic below, 6J7 has been substituted for the EF37A for proof-of-concept purposes. Thanks for looking.
-
I would rather see that 200V be a negative rail sitting under the driver so it doesn't have to deal with the craziness of the output stage.
-
Hey PB - didn't see your reply for some reason.
Yeah, the more I have looked into this, I think that while it helps with some issues, it creates a host of other ones. For instance, the bias of the output tube would be highly dependent on variances between different input tubes, changing each EF37A driver would require a manual adjustment. Also, as the EF37A comes up to bias, the plate voltages initially swings up to 250-270V or so before settling down to 200V, meaning the 801A grid will momentarily be at +70V, which will likely blow my plate fuse. Even if I increase the rating of the fuse, seems to me putting that stress on the 801A with each startup is not good design. It also increases my B+ requirements, pushing up against the limit of 600V components...
I may just go back to the old setup! Why didn't you tell me the 801A is such a pain, PB?! (Just kidding, of course).
In your 801A A2 amplifier, what kind of harmonic spectrum were you seeing? So far, with the NFB what I have measured as been a higher H3 than H2, not your typical SET dominant H2 spectrum.
-
I didn't capture anything that specific. The future owner of the amp was very eager to get it!
-
I can see why, it's a real looker :) no worries, just thought I'd ask.
-
Hey PB - Sowter iron arriving in the next few days. I've abandoned my idea above on elevating the 801A filaments, back to using a negative bias supply. Quick question for you - on your 801A A2 build, did you use a dedicated transformer for your negative bias supply, or did you run it off the 801A B+ secondary? Right now, I am using Pete Millett's shunt-regulated 300B bias supply for my source follower V- (adapted for my use). It's a bit unwieldy though since it has a dedicated toroid, trying to slim things down a bit.
-
I got lucky and the power transformer I used had a separate 100V winding that I used to generate the B- supply.
-
Darn! That would be perfect...perhaps I'll look around, see if I can find something suitable. Thanks for the $0.02.
-
Actually, quick follow up question for you PB if you don't mind: say I found a suitable mains transformer with an appropriate B+ winding for the 801A and a ~100V winding for the source follower V-. In that case, it would be appealing to run the source follower V+ off the 801A B+ winding. The concern here is the stress that could potentially be placed on the 801A grids if driven excessively into clipping in A2, as the V+ would supply any voltage asked of the driver.
Do you think that is a good reason to limit the source follower V+ supply, such that it does not run the risk of excessively high positive grid voltages? My initial though was to limit the peak-to-peak voltage swing of the pentode driver by simply applying the appropriate amount of feedback such that the remaining gain is ONLY enough to just clip the 801A.
Just to summarize, do you think I could use a 300V supply on the source follower drain and not worry about stressing the 801A grids by limiting the gain of the drive via feedback, versus using a lower voltage V+ supply to protect the grids, but making the power supply larger and more complicated since I would need another lower voltage transformer for the V+ supply...
Hope that makes sense, as always, I appreciate your time :)
-
You could certainly use an avalanching device to add some protection there, though I've never been able to flog one of these amps hard enough to run into that issue.
-
I hear ya, thanks, I'll think it over some more. I doubt it would ever be an issue in real life use, just one of those hypothetical type of things. Maybe I don't use the maximum amount of NFB and I have some extra gain on the driver and my cat paws the volume knob and melts my 801A grids LOL who knows.
-
Got the Sowter iron, 5K:8ohm 60mA. Designed the PCBs for the EF37A g2 supply, source follower, and my 0V 801A bias supply. I don't know if I mentioned it, but what I am trying to do is use a 6.3VDC regulator on the EF37A heaters, then drop that across an adjustable voltage divider to make the gate of the source follower FET a few volts positive, such that I get a perfect 0V bias point. We'll see how it goes, need to round up all of these parts, wait for a warm spell, then throw together a single channel and measure.
-
That's one of the tough parts of using the 801 IMO, you often will need to be able to move the bias to either side of 0V.
-
Yeah it is one of the final challengers I am facing, that and the driver g2 supply. Should be able to adjust from around +1.5V to -3.5V on the grid, we’ll see how stiff the regulator is, hopefully I get it right the first time around.
-
Hope you and everyone at Bottlehead had a nice holiday, PB.
Wanted to show you what I came up with as a solution for the 0V bias point problem, I think this is a good approach. I got rid of the source follower FET! Switched to a cathode follower, a 6BX7 with the plate current set by a current sink. The negative grid voltage is generated by a LND150 CCS and a 56K resistor to ground. Since the CCS is adjustable, the cathode voltage can be adjusted to either side of 0V by setting the grid voltage appropriately.
I think a single section of a 6BX7 will get the job done, but could always use two and parallel both triodes if necessary.
-
You can just use an adjustable bias pot to put either 0V or some amount of negative voltage under R5 to set the bias. I wouldn't adjust the current on the cathode follower, just run it reasonably hard.
Being a regulator type tube, you could use the other half of the BX7 as a series reg for your cathode follower and screen supply for the first stage.
-
Yeah just using a pot would simplify things wouldn't it. That -100V supply will be shunt regulated so should work fine. The 6BX7 current would be fixed by the current sink, was going to run it at around 90% plate dissipation, 325V plate-to-cathode at 28mA.
Thanks for the idea, the 325V B+ is regulated as well, but could definitely think of creative ways to use the second 6BX7 triode if I use two tubes.
Have you ever used cold cathode voltage regulator tubes like the VR105, PB? I was going to experiment with using one on the screens of the pentode drivers. The feedback circuit that is there now is a bit complex and I have seen some oscillation issues, but was recommended to avoid bias drift since the screen will track changes in plate voltage, although it hasn't been a big issue for me so far.
-
How much screen current are we even talking about? The LR8 is often recommended for low current apps like small pentode screens.
-
Less than 0.5mA on the EF37A, not much!
-
PB - Is this what you had in mind when using the other half of the 6BX7 as a series reg?
My 325 B+ is regulated, however it might be beneficial to drop some voltage across the other triode to run the cathode follower at a higher plate current. Going over the datasheet, I see each triode has a 10W dissipation, but both together have a disappointing 12W, meaning using a single tube for both channels really hamstrings how far each cathode follower can be pushed with each at a 6W dissipation, only 18mA with a 325V B+.
Maybe I could set up a series reg with the other section and drop some voltage for a hot bias out of the knee of the curves on the cathode follower triode? My other thought was to use two 6BX7 and parallel the two cathode follower triodes, both running at 325V 18mA (~6W a piece).
Here is the article that discusses the circuit below: https://electronicspost.com/series-triode-voltage-regulator/
As usual, thanks for your tube genius.
-
18mA is plenty. You don't need to drop a ton of voltage across the regulator half, and I'd suspect you could do 9W/3W and get good results.
-
Gotcha, thanks again. I'll think it over, either using a single 6BX7 with one triode per channel at 325V / 18mA at 6W dissipation, or use two 6BX7 with the series reg at something like 9W/3W. The series reg is definitely cooler, if that counts for anything :)
-
I was reading through some old posts in this thread (gotten sort of long, hasn't it?). Something caught my eye that probably didn't even register before: is one of your amplifiers based on the 3C24, PB? I recently picked up a six pack of these tubes, along with a pair of HK54 (50W big brother), all pretty cheap as far as old triodes go. I am getting my hands on a digital curve tracer very soon, thought I would trace them to find ideal operating points and think about building an amp with them some day. Found some of those Eimac HR2 heat dissipating caps too, would need some sort of shield so the HV isn't exposed, I am thinking polycarbonate from a local plastics shop with holes for ventilation. Working on this 801A design has been extremely educational, doesn't seem a far reach to build a 3C24 A2 design with the knowledge gained, maybe late 2021 or 2022.
The HK54 curves are sort of ridiculous, need to blow them up on a tracer. Seems like all roads lead to transmitting tubes in this hobby.
-
I ended up just using glass and extrusions.
-
Sweet idea PB, Hmmmm
-
https://www.isel-us.com/enclosure-extrusion-profiles (https://www.isel-us.com/enclosure-extrusion-profiles)
-
Very cool, I'll keep that in mind. The cherry plates on those tubes look fantastic.
-
Here is what I decided to do with the 6BX7, going to drop the FET current sink and use the other 6BX7 triode as the sink. Since the bottom triode is only dropping 100V, can keep the total dissipation of the 6BX7 under 12W and push the top triode cathode follower to 20mA plate current, perhaps a little more. Will come somewhat close to the heater-cathode limitations of the bottom triode (100V), will probably drop the bottom supply down to 90V which should still be fine for full grid swing of the 801A and will give a little more dropout headroom for the negative supply shunt reg. One less PCB inside the chassis too and gets the heat out on top!
-
So the reason I don't do that (there is a guy in Japan that does though!) is that the BX7 takes ages to warm up and the output tube does not, so the grid is just flapping in the wind. This is why I tend to use a cathode choke or a resistive load under the DC cathode follower. When you first fire up the amp, the output tube would see the full negative bias supply and have its own time to heat up before conducting.
-
Very good point, I figured there would be some issues with startup conditions but hadn't gotten into it yet. Let me take another look at a choke or resistive load. Another option - I have some time delay PCBs, relay switches after ~30s, can handle up to 500V. I could put them between the B+ and OPT to allow the 6BX7 to come up before B+ is applied to the 801A.
-
Would it be overkill to use one 6BX7 per channel with a resistive load and parallel the two triodes?
-
I don't think that would be a problem. The two triodes in parallel will let you use a lower resistor value, but will be somewhat demanding on that B- supply.
-
Yeah it would be demanding, always a compromise to be made. I was planning on a TL431 shunt regulated B- supply, could probably spec it for 30-40mA. I'm trying hard to use the 6BX7 since I have a bunch of them in pairs, but if it came to a point where I was using a single 6BX7 for both channels, I think it would make sense to switch to the 6AK4 or 6CK4 to take advantage of the higher single triode plate dissipation. Also, maybe it is total tube amplifier voodoo, but in my experience I have found equivalent single-triodes to have better soundstage and imaging than equivalent dual triodes (6J5 vs 6SN7, for example), although I have not found a objectively measurable way to explain that (crosstalk, distortion, etc.). Not sure if the same applies to cathode followers, haven't tried it yet.
I really like the second triode as a current sink if I can keep the 801A happy on startup, might really go with the time delay idea.
-
Check out the 12GN7.
-
Hey PB - I'll check out the 12GN7, thanks. I'm still brainstorming some ideas on the 6BX7 cathode follower feeding 801A startup biasing issue.
What about this - placing a high-value resistor between the 801A grid and negative supply. At startup before the 6BX7 is biased, the 801A will see the full negative supply and be placed at cutoff, then the grid will rise to its 0V bias point as the 6BX7 come up to bias.
Any reason you can see that wouldn't work? My thinking is this would also protect the 801A in the event of a failure of the 6BX7, although the 801A will be fused as well. I might be making a mental error though, I am experiencing post night shift brain.
-
The problem with the high value resistor is that just a whisper of grid current will shift the bias substantially. You can place a transistor between B+ and the primary of the OPT and have it sense current draw on the top 6BX7 in order to turn on and allow the output stage to work, but I favor a chunky resistor under the cathode follower as the simple/reliable solution.
-
I gotcha PB, thanks for the input. I think I am going to go with the resistor load and parallel triode sections. I crunched the numbers, a 6BX7 as a current sink with a 160ohm cathode resistor only represents about a 2.6K AC load, not great! Compare that to a 6K resistor on a single triode or a 3K resistor on parallel triode 6BX7 cathode follower.
Check out these curves PB - I recently purchased a curve tracer. The first set are both triodes traced and overlaid on eachother. With a single triode resistively-loaded cathode follower, operating point would be around 330V, 15mA, -35Vg1.
Compare to the second set of curves, this is both triodes in parallel, operating point would be 330V, 30mA, -34Vg1. The linearity is much, much better!
So, think I will use the 3K resistor on the parallel cathodes of the two sections running at 15mA a piece, just have to spec my negative supply for it.
-
Fancy.
-
Random question for you PB - what did you think of the sound of your 801A A2 design? Was it worth the effort?
-
Yes, the current owner is quite happy with it.
-
Thanks, still plugging away at mine, I am hoping the payoff is worth it, it has been a lot of work with much more to be done.
-
Here is my recent iteration PB, haven't prototyped this yet, but I feel pretty confident this is close to the final design.
I dropped the CCS load on the pentode and the screen DC feedback. What I found was I was getting an H3 dominant THD spectrum and the sound suffered for it. Instead, what I am going to try is using a high-value resistive load with a VR105 glow tube regulating the screen at 105V. Inspired by some other DIYers, going to use 1.5V battery on the EF37A g1 and buffer the feedback loop with a PMOS on the EF37A cathode.
This gets a ton of sand out of the gain stage (less opportunity for oscillations), simplifies it significantly, gives me an H2 dominant THD spectrum and I believe the gain will be enough to get a good Zout. I think I've pushed the resistor load as well as the 6BX7 grid leak (max 2.2Meg) to their limits while maintaining a good operating point on the EF37A to maximize the gain.
I feel pretty good about this, I will pray to the oscillation gods before I prototype it and take some real-world measurements.
-
Let us know how the breadboarding goes!
-
Got another random question for you PB, hope you don't mind.
Have you ever thought about loading a 6080/6AS7G cathode follower with an output transformer? Think a Bottlehead Crack with an OPT load. I recognize finding a transformer that is right for the job would be challenging as it would need to be gapped for 80-100mA at a low turns ratio, could do 1:1 but if you are going to the trouble, perhaps 2:1 or 4:1 to drop the output impedance. These are like, interstage and line output turns ratios, which typically don't come gapped for 100mA. Also would likely need to put a resistor between the primary and ground to bias up the cathode.
Something like this, gyrator loaded, direct coupled with an OPT load with a theoretical 1:1 OPT. Ever seen anything like that?
-
Are you talking about doing this for a headphone amp or for A2 drive?
I have thought about using a small push-pull output transformer in place of the center tapped choke to do an Altec 1570 variant, and possibly using the secondary as part of a feedback network of some sort.
-
Interesting amplifier, would love to see it. I was thinking more along the lines of a headphone amp. I have seen the idea discussed somewhat for speaker amplifiers, with a high-gain input tube, but obeying Vhk could become a problem at high outputs.
With the 6AS7 OTL headphone amplifiers like the Crack being so popular, I'm surprised I haven't found anyone trying this, but could be due to the oddball transformer that would be needed, probably a custom job. And who knows if it would really sound better than a simpler, cheaper output cap.
-
When you start your design assuming that you'll use an output transformer, you'd then really free up your choice of tubes to drive said output transformer, and there are many candidates that are far more linear than a 6080. Edcor makes some small 1.7K output transformers that would work for such an experiment.
If you aren't using an output transformer, then you want a tube with a good amount of transconductance that can also pass a pretty heavy amount of DC current.
-
Good point PB, I'll take a look at those Edcors. The EC360 might be an interesting choice, I'll see what's out there. I think this could make for a cool design, going to build on the idea over time and see what I can come up with, thanks for the $0.02.
-
Hey PB - going back to when you were using the 3C24, did you find an efficient way to degas them? I am attempting to trace their curves, have five different samples, all appear gassy and will not conduct on my pulse tracer.
I have read anything from running the filaments for five hours to running the plates red for 10-20 minutes since the tantalum plates are self-gettering, although for the latter I would concerned about arcing given the gas...
I have a pair of HK54 which I was able to trace A2 curves on no problem.
-
I degas them by throwing them away and using a different one when they are gassy. That's the big downside to the 3C24.
-
Yeah seems to be an issue, thought one out of five would pan out, bummer. Seems many sellers don't have the means to test them, so buying is a crap shoot. Alright well I guess I will chuck these and keep an eye out for a tested pair, thanks.
-
It's worth it to buy NOS in box with the 3C24. The ones with black plastic bases seem to be the best bet.
-
Sounds good, I was able to find a tested pair, black base like you suggested, thanks.
Can I ask your thoughts on a full A2 3C24 output vs. a partial A1 + A2? The below schematic seems to be the most popular design choice with the 0V grid being the positive swing boundary. Attached a 3.3K load line, wonder if it is worth adding a low voltage negative supply to get the additional 10-15V of negative grid swing? Thought about choke-loading the cathode follower for the negative swing, but might cause feedback issues.
-
I've built one or two of those. They sound decent enough but they don't make a lot of power. This is what I ended up doing instead, and it makes about twice the power with no GNFB.
-
Thanks for the schematic PB, choke loading the cathode follower with plate-to-plate local feedback, nice. What is the purpose of the 2.5V bipolar supply? I'll keep this in mind, think I would like to do something similar i.e. choke loading the cathode follower with local feedback and avoid a negative supply.
-
That was there as an experiment for tweaking the cathode follower bias to dial in the plate current perfectly. I didn't end up liking it so the power supply it there but that's no longer reflected in the schematic.
-
Got it. Well thanks for sharing, appreciate the input as usual, will report back when all of these ideas become real world things!
-
I got my functioning 3C24 PB. Traced the curves too, all looks well. Plate resistance around the bias point I am thinking is around 9.4K.
I might breadboard this with an E80F and 6AH4 with a negative supply, could be low voltage since the 3C24 grid only needs to swing down to around -25V. I have a spare Lundahl mains transformer, would only need one of the four 6.3V 3.1A windings, was thinking I would place the other three in series and use a voltage doubler for the negative supply, should get me something like -50VDC.
Did you really leave the EF86 cathode degenerated in your design, still had enough gain?
Also, just wanted to say, I know I am talking about my own designs on this thread (which might be renamed to "L0rdGwyn Pesters PB"), but I am a super BH ambassador, probably responsible for 10-15 BHC sales ;) people often ask me how to get into tube DIY. I tell them to read Valve Amplifiers and build a BHC. Excited about the YouTube channel announcement, will follow along.
-
What you're messing around with is something PJ and I have been working with for a while, and to be completely honest there was a good 6 month period where it was me pestering PJ and seriously trying his patience, so this thread feels rather appropriate.
One thing PJ told me was that the unbypassed cathode resistor on the first stage pentode would improve performance of the amp. While this is absolutely true, I eventually made a really gnarly spreadsheet on the computer (that started out as a pentode operating point performance predictor from PJ) and this was able to show what bypassed vs. unbypassed cathode resistors looked like. The performance differences between the two were small enough that I don't worry as much about squeezing out that aspect as much as I did before.
-
The unbypassed driver cathode resistor does not provide very much feedback - around 6dB or so, I think.
-
Well I do appreciate the feedback (no pun intended), thanks gentlemen. In one of my 801A A2 prototypes, I left the cathode of an EF37A unbypassed, which significantly improved the sound, however the internal feedback was too much and I did not have enough gain remaining for the loop, at least in that particular iteration. In a DIY thread PB linked, which I am now following quite closely, a user implemented a PMOS transistor on the cathode in output plate to input cathode local feedback, buffering the feedback loop with very impressive results. This is what I am in the process of trying, using a resistive load. If it works out, might make it in a 3C24 prototype as well :D
-
I've also split the cathode resistor into two resistors in series and bypassed one of them to split the difference a bit.
-
Here is my latest A2 transmitter craziness PB. A few months back when I was thinking about the 3C24, I picked up a pair of HK54. Price was $65 for a NIB pair, good deal in retrospect, was just curious but didn't know if I'd ever use them. Well now more serious about the idea as I am learning more about this topology, might even try them out in place of the 3C24 if I can find more pairs.
I traced A1+A2 and grid current curves on my tracer, came up with a 425V 105mA bias point using a 3K load, plate resistance is about 9K. I put together an LTSpice DHT model using Dmitry Nizh's modeling tool and made a simulation. Using 6AU6 on the input, get an open loop gain of around 220 from the stage. Given that the HK54 will draw a whopping 50-60mA of grid current at peak swing, thought a cathode follower was probably impractical (bias + grid current would make for a monster power supply), so a FET buffer instead. The gate would be biased positive using a voltage divider from the regulated B+, then a bias pot to dial in the HK54. For the output transformer, I subbed in the Lundahl LL1623 120mA model, 23H 164DCR primary, rated for 25W at 30Hz.
With this setup, I can apply about 22dB of feedback and get a damping factor of about 3.5:1, could improve that by changing the pentode to something like the 6EW6. This makes for close to a 15W amplifier, not bad for a 450V B+.
-
60mA isn't impractical for a cathode follower. Try high gm tubes like the 6BX7, 6BL7, 6080, 6AS7, 5998, 12GN7, etc. With your 40V of bias, what's the standing grid current?
What I really like about what you've cooked up there is there's very little swing above the 0V grid voltage line, so that could make for a far simpler design overall without sacrificing much.
-
I thought the exact same thing on the negative grid swing, sacrifice a little power for a more simple power supply. I mocked it up both ways with that in mind. With the negative grid boundary being 0V, I rebiased to 375V 120mA. This brings the quiescent grid voltage to around 48V.
At idle, the grid will draw about 8mA. When I was mocking it with a cathode follower, I was looking at using the 6AH4, but might be better to go with 6BX7 or 6BL7 and parallel the sections, like I am doing in my 801A amplifier, which draws far less grid current. At one point I had thought to use the EC360, similar to a 6AS7G but single triode, but with the bias current + peak grid current, the B+ supply would be drawing something like 500mA. Going with a 6AH4 / 6BX7 / 6BL7, etc., the current draw would be much more manageable. As it so happens, I have quite a few 6BX7 and 6BL7 handy.
Put together a schematic below, still makes for a 10.5W amplifier without a negative supply and avoiding the A1/A2 transition point, the curves get much more cramped beyond. Would either need to run the B+ supply through a voltage divider or make a separate low voltage supply to bias up the CF grid. With each section of the 6BX7 biased at 10mA, idle current draw on the B+ supply would be around 300mA, then around 400mA at peak grid swing.
Another advantage of this approach - voltage swing around the 375V bias point is more linear, should make for a good first watt.
-
I have a pretty strong proclivity toward amps that look like that design.
You could split the 2.5K cathode load into a Hammond 154E and a resistor in series as an experiment.
-
Ahh yes, something to try in breadboard, will keep that in mind. Could potentially get the full negative swing after all, so long as it doesn't anger the feedback gods.
Well I'm excited about this, altogether not an overly complex design without the negative supply. Need to decide if I want to change to a higher gain pentode for more than 22dB of feedback for a better damping ratio. Might try different amounts of feedback and tune by ear. The 3K 120mA 23H Lundahl is probably the best off-the-shelf transformer for the job, but will see if there are others out there.
-
I haven't found a single cathode choke in the feedback loop to be a problem so far. Our old Seductor amp used plate to plate local feedback and we did run into some instability depending on the value of the cathode bypass cap at the output, which was a bit of a surprise.
A damping factor of 3.5 will allow the amp to work in place of something like a 300B/2A3 amp without the bass character changing so much.
The One-Electron UBT-3 is a decent transformer for not a ton of money, but the maximum recommend primary current is 110mA.
The PRC-1 and our Kaiju OT's could be used in parallel feed well in that circuit as well.
-
That's good to know, gives me a lot of confidence using the choke then, would be great to get the 15W without a negative supply, I'll give it a shot. Thanks for the recommendations, let me think it over the transformer situation, think this would be next in the DIY queue after the 801A amp, which is probably two months or so from completion.
-
Hey PB - at long last, I have some preliminary results on my 801A A2 build. Since the circuit was going to need a breadboard anyway, I decided to put together a reusable prototyping board, I combined the two projects into one.
I finished the board a few days ago and finished a single channel of the amplifier. The plate-to-cathode feedback is buffered by a PMOS on the cathode of the EF37A. I had some heater regulators from another project on hand, so I put DC on the EF37A heaters then ran one end through a bias pot to apply grid bias to the pentode and dial in the operating point. I used a VR105 glow tube to regulate the EF37A screen.
In this first iteration, this comes out to around 11dB of feedback. Output Z is 1.7ohm, which is great, hit one target on the first go around. Clipping starts beyond 6.3W into 8ohms.
THD is a bit higher than expected, 1W into 8ohm spectrum below, which I attribute to the open-loop gain of the EF37A being much lower than simulated (I had thought I would be able to apply around 15dB of feedback), so that is what I will be investigating next. The harmonic spectrum is favorable, so even as is I feel confident the sound will be pleasing. Ultra low distortion wasn't necessarily a very high priority, we'll see how it sounds once I finish the other channel.
Edit: I should note that going from 1W to 6W, the distortion *only* climbs to 2.5%.
-
I've never heard of the 801a's before, you have very much peaked my interest as i use early 01a's in my preamp which are also tungsten thoriated DHT's, and fed with Coleman regulators. Any chance of sharing your design?
P.S. That is one of the cleanest proto boards i have ever seen :)
-
Thanks, mcandmar, don't mind sharing the design, but I have to warn that it is a bit of a doozy, complicated, at least by my standards. Both the output transformers and mains transformer are custom jobs, although you could probably make it work with off-the-shelf stuff, but likely more transformers needed, especially for the 801A filaments, I was able to get everything wound on a single core with internal shielding. The 801A is a tricky tube to use given its high plate resistance, necessitating either a high impedance OPT or use of NFB, I went with the latter after getting subpar results with the former. I've attached a pretty accurate amplifier schematic below.
-
I like it!
-
Thanks, PB. Here's a technical question maybe you can help me out with: as I said I'm not quite getting the gain I expected in the pentode stage, I'd like to squeeze out a little bit more and apply more feedback. One way to do that would be to increase the grid resistor for the 6BX7. For autobias, the data sheet lists 2Meg max, I have a little less than 1Meg in my design since it is using sort of a mixed bias.
Think I can push that to 2Meg?
-
Those test sockets are sweet. Nice flat frequency response !
-
Do you have a little bypass cap on the screen? For the paralleled cathode followers, 1M seems like a reasonable grid leak choice.
-
Yeah I thought 1Meg was a good choice, I will leave it as is. At the moment I do not have a bypass cap on the screen, I have seen it done both ways, got the idea from one of Pete Millett's amplifiers who left it unbypassed. I tried it bypassed with 0.1uF and without, but did not see a measurable difference in gain, distortion, FR, etc.
-
Those test sockets are sweet. Nice flat frequency response !
And thanks, Thermioniclife! Sorry didn't see your comment until just now. The sockets are very nifty, sort of make the whole thing work. I have some noval 9-pin boards as well that can be swapped in if the design calls for it, will use adapters for any oddball tube bases if need be.
-
Your prototype board is way too neat, I'm sure that's your problem ;)...John
-
HA! That has to be it, no doubt.
I finished the second channel, the heater regulator I was also using as a bias supply was injectin HF noise into the EF37A grid, so I had to make a switch, nice and quiet now. Maybe it was the tube swap, but the feedback and measurements improved in the final circuit, but HF extension, lower distortion and output impedance.
Output Z: 1.53ohm
Power: 6.2W into 8ohm
I had a listen with headphones, then lugged it into my stereo, the sound is very nice, will do some more tinkering of course but it won't be long before it is boxed.
-
There are lots of 100-ish watt transmitting triodes, now you can go for 40W!
-
Oh man, scary to think about, but something tells me I'll mess with the 100W transmitters some day.
My next victim is the HK54, 15W single-ended, pretty good for 92dB/W speakers :D not sure when it will come to fruition, but have a nice protoboard to mess around with it now.
-
My next victim is the hk54 HK254
-
HA! Good one PB, I saw those while shopping for HK54, 100W with a 37.5W filament, woo boy! I'm making my way there, 25W now, 50W next, then maybe the big fellas. Getting ahead of myself but thinking of using EL34 to drive the HK54 grid.
-
The EL34 isn't a bad choice. Maximum cathode current is plenty high and high voltage isn't particularly bothersome.
-
I thought so too, took some measurements at a potential operating point, I found gm and linearity were better in the EL34 than say, parallel 6BX7 at high voltage, 440V B+. Shouldn't have a problem with the 60mA grid current at peak swing, I have a pair of NOS EL34 burning a hole in my pocket too.
With a similar feedback circuit to the 801A design, the challenge is the gain. For a 3K load to get 15W output with a 3:1 damping ratio with the HK54, will need something like 25dB of feedback. I grabbed a couple of high-gain pentodes to try out, 6EW6 and 6AH6, the 6EW6 being the higher of the two. With a 5-pack of NOS 6EW6, what I found was a high degree of variation in gm / the curves from sample to sample, even from the same exact batch, BUT they can be made to be matched pretty well by adjusting the screen voltage, so I would need an adjustable screen supply to match the feedback in both channels.
So we'll see, it's going to take some trial and error.
-
I finished my 801A amp PB, here are some pics.
-
Looks great! How does it sound?
-
Really good, best of the amplifiers I have made so far, the real standout is the bass, unlocks a new level of potential from my Snells! Great clarity and detail too, the speakers disappear.
Thanks for your $0.02 along the way. First got the idea for an 801A build over a year ago now, figuring out how to make it happen was a serious learning experience.