Modded the S.E.X. for Cathode Feedback

[Dr. Toobz]

I have always had a love/hate thing with SET amps - love the tubes, 3-D sound, and warm mids, hate the poor damping factor and what I perceive to be a reactive, uneven frequency response with some loads. The Crack amp, with its OTL, cathode-following topography really intrigued me - perhaps the closest to my ideal of "perfect sound." Obviously, OTL's for speakers are tricky, so I've been thinking about modding the S.E.X. to use different tubes and/or feedback to lower the output impedance and give a more even response.

With CB's guidance, I did some minor rewiring of the S.E.X. amp to give the power triodes local cathode feedback (CFB). I basically undid the MQ irons from being wired as autoformers, ran the red primary wire to ground, desoldered the ground bus from each cathode, ran a jumper from the cathodes to the back terminal strips where the twisted pair for headphones meets up, and reversed the secondary wires (orange to ground, black to positive/headphone wiring). No bias resistor was used. The change is nothing short of amazing. The amp is cleaner sounding, more detailed, and much more balanced - all frequencies now sound in check with each other, whereas I had problems with tipped-up mids and treble previously. I'm hearing a bigger, deeper soundstage and little details previously buried in what I presume was harmonic distortion. Best of all, bass is better than anything I've heard to date - thunderous and musical, and a bit tighter than before. I estimate I lost about 3dB gain in the process, but no biggie.

I know NFB gets a bad rap, but this application of local cathode feedback should be an exception. It makes the S.E.X. sound more refined, and if you have a preamp, the loss of gain is easily made up for (I simply turned up the Quickie a bit to compensate). I suppose I can always "wet" the sound by turning up the Quickie and adding 2nd harmonic distortion to the signal, but the key here is the sound can be colored without affecting the damping factor, etc. like it would be if the power amp was responsible for the coloring. My Beyers (and even Grados) also sound much better, which leads me to believe it is less about the 120 ohm headphone jack resistors than people think.

Next up may be experiments with 6EM7 tubes and different OPT's for more power!

[Paul Birkeland]

I'm secretly hoping that PJ posts on this topic and offers further insights. I'd be somewhat curious about saturation when drawing DC current through the secondary of an ungapped transformer, as well as suitability of this modification with the Speco...

[Paul Joppa]

I have some thoughts but won't get to this until I'm back, late June. Thanks for sharing experiments; it builds the knowledge base for everyone!

[arsun]

Interesting modification. I will be watching this thread closely and wait for PJ's feedback . In the mean time, Dr. Toobz, can you post the instructions in a bit more detailed fashion including the terminal numbers a la Bottlehead manual, if you don't mind of course?  Thank you...

[Dr. Toobz]

Interesting modification. I will be watching this thread closely and wait for PJ's feedback . In the mean time, Dr. Toobz, can you post the instructions in a bit more detailed fashion including the terminal numbers a la Bottlehead manual, if you don't mind of course?  Thank you...

I'd rather wait to have others study this mod a little more empirically before publicly suggesting that people should do it (and providing detailed instructions on how). After all, it is an experiment still in progress, and may be found to be less-than-ideal (also, my taste in sound/amps appears to be somewhat different than many fellow BH's, meaning that NFB might make your amp sound worse to you). PM me if you want more detailed directions.

I think, for example, that the Specos in the stock S.E.X. would not be up to the task of handling much (if any) DC on the secondary winding and would therefore saturate easily. So, the fact that I hear no downsides in the bass department, for example, might reflect the fact that I have beefy MQ irons and plate chokes to make up for any saturation losses. CB also has suggested that if I go to a 6EM7, which has a large-triode plate resistance of ~750 ohms vs. ~2000 for the 6DN7, I might be in the vicinity of being able to use a 2.5k/8 ohm transformer to get more power (probably 4 watts or a little better). I may try this, too.

In my opinion, having a lower Z-out (and therefore, a less reactive response) helps bring a lot of bookshelf speakers into the offing even more so than increasing the wattage, esp. if you're like me and only listen at rather low volumes from a few feet away. A lot of commonly available monitors have nasty impedance curves and despite being listed as "nominal 8 ohms" will bottom out at 3 or 4 ohms. Then, there's phase issues, too, which make them even harder to drive at certain frequencies (seemingly, regions near the crossover points). In my experience, this leads to the speakers having weird bass and sounding too loud in the mids, esp. at low volumes. Thus, being able to have a solid one or two watts at most/all frequencies that won't "trace" the speaker's impedance curve as much as a no-NFB SET is more important to me than having a higher wattage output but less predicable frequency response, esp. at the extremes.  

FWIW, I really don't like single driver speakers, which wouldn't care as much about impedance, since many of them have quite flat impedance curves around 6 or 7 ohms, even in the bass frequencies. So, for me anyway, a tube amp with under 1 ohm Z-out and regular monitors seems to be the way to go - hence my curiosity about NFB. I guess being rather young and only having listened to soild-state for most of my thirty-some years makes me have certain expectations of what constitutes "good sound." The warm, non-fatiguing sound of tubes, coupled with powerful, tight low-end response, a flat output with no one frequency more prominent than another, and neutral-sounding speakers seems to be my ideal.

[Dr. Toobz]

A question for anybody that knows some formulas: how does one figure out a) how much NFB the cathode-feedback arrangement here would result in (in dB), and b) once that figure is known, how to use it to figure Z-out? I figure that the amp output as-is would be between 2 and 3 ohms, correct? 8000/8 (OPT) =1000; 2200 (plate resistance)/1000=2.2 ohms. Now, the transformer itself has both DC resistance and AC impedance; let's say it adds 1/2 an ohm, which is what I've seen quoted in other posts, which would mean a total of 2.7 ohms. So, what would this look like with the arrangement I've been playing around with (with CB's help)? Obviously, unlike DC resistance, AC impedance is also affected by phase, which complicates things, but I'm curious as to a rough figure....

I know that SS amps typically are around .1 ohm or less, with many push-pull tube amps that use global NFB (like Dynacos) coming in somewhere between there and an ohm. Most SET's I've seen are 2-4 ohms, and some cheap pentode P-P amps (like those S-5 kits that use 11BM8 tubes) are even higher, like 5 or 6 ohms.

In sum, wouldn't an amp with a 3 ohm Z-out do something funny with an unbalanced speaker (let's say one that has an impedance of 3 ohms at 100Hz and 30 ohms at 5000Hz)? The voltage drop from a no-load condition would be higher at 100Hz than 5000Hz, meaning that the 100Hz note would show a reduced sound pressure level in comparison to the 5000Hz note, correct?

EDIT: I found some equations by Googling: according to the formulas on this page (http://www.veiset.net/tech/cathode_FB.html), I calculate about 2.6dB cathode feedback and a final output impedance of 1.34 ohms, assuming mu=15.4, rp=2000, OPT is 8k/8ohm. So, the 3dB guess I made was close! A 6EM7, assuming mu=5.4 and rp=750R, would result in a Zo of .63 ohm. This formula doesn't seem to take into account any impedance added by the winding, so it's a low-end estimate.

[Dr. Toobz]

I should mention that the ungapped irons do saturate very easily, from my testing. So, something like MQ Robin Hoods would be needed, or perhaps some Edcors. I think I'll just stick with the BH1's and pitch the NFB for now, though it was really interesting to hear the difference that a minor local feedback loop made. I think NFB gets a bad rap due to global feedback and too much of it, especially in "sand" amps, but a little CFB under a triode actually sounds pretty nice.

[Fred_P]

This CFB thing sounds promising, may I hope for any more updates?

I also consider the iron upgrade, but would wait for any further CFB results and recommendations before making a decision on that.
It is not urgent at all, just wouldn't want to leave this topic fade away.

[Dr. Toobz]

The CFB sounded good, but I'm not so keen on dropping $200 right now on another set of MQ irons (Robin Hoods). Maybe Edcor's (cheaper) offerings would work, but I doubt their stuff is anywhere near as good as Mike's at MagnaQuest. That would be their XSE series, I believe, and you'd probably want to stick with an 8k to 8 ohm model to keep similar power and distortion specs to the current S.E.X. amp. (FWIW, the Robin Hoods have 5k primaries, so maybe the power would be a little higher, but the increased distortion from the lower impedance load would be somewhat offset by the negative feedback). The Edcors can handle something like 30mA of DC, if I remember correctly, but are likely to have a lower effective inductance than MQ's stuff. If I can find somebody here to buy my BH1's, that would offset some of the cost of the new gapped transformers, and then I would probably make the cathode feedback permanent. Then I could give you some more info on this.....

You could certainly try a feedback loop with the Specos, but I doubt it would work very well. The BH1's themselves saturated very easily from the (small) DC current being pulled through the secondary, since one should recall that the bottom of the secondary is connected to ground. The primary sees no DC, since the parafeed caps block it, so this leads to an unbalanced situation. That's essentially why I reverted back to the stock, grounded-cathode arrangement with no NFB for now. And I'm not a fan of global NFB...

[Dr. Toobz]

Here's a question for somebody who might know: in a cathode feedback arrangement, how would you figure out how much DC current the cathode, which is tied to ground through the secondary winding, is drawing? This will affect whatever transformer I end up getting, which will need to be gapped accordingly. For the S.E.X. amp, would a 30mA DC rating be sufficient?

[JC]

Probably the quickest way in this instance would be to look at the characteristic curves for that section of the tube.  IIRC, the plate Voltage in the SEX is ~345, with a bias Voltage of -17 on the grid.

Another approach would be to calculate the current based on the Voltage drop across the plate choke, if you know the DC resistance of the choke.

Either way, the current flowing through the plate comes from the cathode.

[Paul Joppa]

Nominal current is 28mA for 6DN7, 34mA for 6EM7.

[Dr. Toobz]

Nominal current is 28mA for 6DN7, 34mA for 6EM7.

Thanks for an exact number. I pulled out the tube curves from an old GE sheet I found, and I couldn't get that exact. (Line was somewhere between 20 and 40ma; I had figured around 30mA).

No wonder why the un-gapped BH1's saturated! :-) FWIW, I did place an order for some Edcors - the cheap ones first (XSE10-8-8k) and will report back regarding my findings (in about 4 or 5 weeks). I'll be curious to see how good these irons are with feedback. Hey, maybe these would be decent upgrades from the current Specos in a future S.E.X. II amp? I've read nothing but good things about them, esp. below 5W RMS (which makes them perfect for the S.E.X.), and it would seem that at such wattages, the usable frequency range is from about 14Hz-40kHz. The specs only have them as -1dB at 70Hz and 18kHz, but I'm sure that's at full power (10W).

EDIT: I should mention, the Edcors clocked in at $17 and some change, per transformer. Only about $12 more than a Speco - so I'll really be curious as to how these perform.

[Dr. Toobz]

Well, the Edcors came, and I like how they sound - no bass saturation to be heard. Interestingly, though, I have yet another observation about the amp and how components change the sound:

I had taken out the MQ upgrades and put back the stock 30H chokes in with the Edcors, and then, put the 50H MQ chokes in instead. Oddly enough, I've found that while the larger choke value does increase the stability of the low bass frequencies (by about an octave, I'd say), it also changes the soundstage and makes things sound a lot more "punchy" and dynamic. And I'm not sure I like that. The 30H chokes yeild a very smooth, laid-back sound with less bass, but a much more natural midrange and treble. The soundstage is also more intimate. I'm interpreting this to again be the result of second harmonic distortion. If I'm not mistaken, the bigger plate choke should lower distortion, as it will have a higher DCR and also a greater inductance, and can "hold" more energy for when the tube requires it (such as on a loud bass note or transient). The smaller plate choke, therefore, should have a somewhat higher distortion than the larger value, as it can't react as well to what the tube "needs" at certain times. I'm thinking this must be mostly even-ordered, as it's very pleasing and natural-sounding, vs. the harsher, brassy quality of 3rd and 5th harmonics. When using the Quickie with headphones, I noticed the same effect - overloading the tubes caused an increase in 2H, which sounded very "rich" and pleasing, and resulted in a closer soundstage. Euphonic disortion, anyone?

I may leave the circuit as-is with the 30H chokes and simply use a subwoofer to handle bass frequencies. The Heresys don't have much bass below 55Hz anyway, and I much prefer the Crack to any transformer-coupled headphone amp I've ever heard....

The CFB does result in a bit drier treble - very vivid and clear - but I like that, as it lets the warmer mids punch through better.

[Dr. Toobz]

In case anybody is interested, I've arrived at the final formula for my amp. It sounds great, especially over my Klipsch speakers, which seem to like the lower Z-out.

In short, I:

-Used 30H stock chokes to load the power tubes,
-Went with 2.2uF Auricaps
-Used XSE10 8k:8 ohm OPT's, wired as standard tranformers
-Took the hot leg of the secondaries back to the cathodes, which are not grounded and not using a bias resistor
-Used C4S boards to load the Va circuits - the CCS sounds a little less warm, but much smoother and has better bass and soundstage
-Used an HLMP-6000 LED in series with a 499 ohm resistor for each cathode on the Va halves (in place of the stock 1.27k resistors). This cleaned up the bass.

My end product is not quite a BH stock amp, but still has that BH "house sound" and suits my need for lower output impedance and a bit drier treble. The fun part has been making the stock amp into something of my own - now, I plan on building a power amp from scratch....