Further upgrades to an already fully upgraded SII-45?

Guest · 2244

0 Members and 1 Guest are viewing this topic.

Deke609

  • Guest
[Preface - Apologies if this results in a double or triple post on the forum. I tried to post this a few days ago and it disappeared, and posted it again yesterday but as of the time and date of this current posting, all posts after July 19 are gone]

I really like my SII-45 with the 3 BH upgrades (particularly the SR upgrade). With the EML 45Bs, VCap caps, and used as a power amp after the Beepre, I think it sounds superb. But I wonder if additional "upgrades" could make it even better. I have two ideas: (1) bigger output iron; and (2) incorporating elements of the Kaiju SR circuit. I'm not sure if either is feasible or a potential sonic benefit, so I am hoping for guidance from the experts.

Idea 1 - Bigger Chokes and Maybe Bigger OTs -- I think I've read posts from PJ to the effect that, within reason, bigger chokes are better - I hope I am not misstating PJ's position. And in a 2002 post on AA [http://db.audioasylum.com/mhtml/m.html?forum=tubediy&n=4050], PJ suggested a minimum ratio of 4H for every 1K of OT impedance, with a ratio of 8:1 being better, and 16:1 recommended for, quoting PJ, "an all-out assault on solid bass (i.e. the best parafeed)". I like the sound of that!  With the stock 4K OT, the 16:1 target results in a plate choke of 64H or higher. That's a bit of an odd value, but I am confident I could get some nice 75 or 100H chokes from Electra Print. BUT, that means big parafeed caps. In various threads PJ has suggested a midpoint target uF calculated as 2L/R*R where L is the plate choke inductance in Henries and R is the OT primary impedance in Kohms, and that anywhere between 1/2 and twice as big as the resulting value should work. I'd like to stick with the midpoint as a balance of bass and high frequency performance. So at 75H and 4Kohms, we get 9.375 uF, and 12.5 uF at 100H and 4K. This is potentially doable, but the cost of the VCaps to hit those values will likely be close to the price of new and bigger OTs.  100H and 6K OTs gives 5.5 uF, and 125H and 8K OTs gives 3.9 uF.


Questions for the experts:

(1) Other things being equal (including use of high quality parafeed caps), which if either would you predict might sound "better"? Option A - bigger choke, keep the 4K OT, and a big parafeed cap, or Option B - big choke and big OT, with smaller parafeed cap?

(2) My understanding is that going with a bigger OT requires a new operating point for the 45B - I'm not sure if going bigger with the choke does.  Does the SR circuit or any other part of the SII circuit prevent or limit changes to the operating point of the tubes?

Idea 2 -- incorporating elements of the Kaiju SR circuit. The SII SR upgrade circuit and the Kaiju SR circuit look pretty similar, but I've notices some differences. The Kaiju appears to shunt regulate both halves of the driver tubes, whereas in the SII with SR, only 1/2 is regulated. Further, one of the Kaiju C4S boards has paralleled cap and resistor in parallel with the LEDs, and the Kaiju has a zener string.

Question 3 for the experts: could any of the above Kaiju SR "extras" (or any others that I missed) be incorporated into the SII and potentially make a positive difference?

I should reiterate, in case it is relevant to the analysis/answer, that I am using my SII-45 as a power amp after the BeePre. If the stock SII's need to perform preamp duties is a limiter, I do not have that limitation.

Finally, I should add that I will soon have a Kaiju, so I have the "Just get a Kaiju" answer already covered.  But it is ultimately destined for a Jager setup that I hope to put together next year. The SII will remain my headphone amp for use with my Audeze LCD4 headphones which to my ears are almost speaker-like, but with incredible bandwidth, especially at the low end.

Many thanks in advance, Derek



Offline 2wo

  • Hero Member
  • *****
    • Posts: 1240
  • Test
Reply #1 on: July 23, 2019, 11:39:37 AM
I have an SR-45 with the MQ for Bottlehead iron which sounds great and I recently built a new 45 amp useing EXO45 at 4 or 4.5k, I don't recall off the top of my head and the Dowdy plate choke at I am guessing >80 H. With a 10uf PF cap and the bass is outstanding...John

John Scanlon


Deke609

  • Guest
Reply #2 on: July 23, 2019, 03:20:00 PM
Thanks John - with either 45 amp, did you by chance start with different iron and then upgrade to the MQ iron? Or go from a smaller choke to a bigger one? If yes, what were your impressions of the difference?


For my SII-45, I can see two potential sources of "better": higher quality iron (better construction and materials) and bigger iron. I'm just not sure which, if either, to give more weight.



cheers and thanks, Derek



Online Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5751
Reply #3 on: July 23, 2019, 05:10:03 PM
Output transformer - keep the stock 4K, it is already overkill for a 45 in terms of bass handling capabillity. You might consider a Magnequest EXO-45/46 if you can get one with an all-nickel core, but it's a tradeoff - the 5K impedance reduces tube distortion and the nickel core reduces transformer distortion, but the physically smaller core and nickel laminations both reduce deep-bass power capability and the 5K impedance reduces available power full-range. (These effects would be very subtle.)

The Magnequest BCP-15 comes in two versions; the 50 henry, 40mA version would fit and is a small increase. We do not at present offer the Kaiju plate choke separately, though it has a 60-henry tap. If you go for a custom choke, the largest that fits easily would be a "1.25-inch stack of EI-100 laminations in a horizontal channel frame," the dimensions of the Kaiju choke.

Operating point - you are already operating at the maximum plate-cathode voltage, so you can't set up "optimally" for a higher impedance. Using a higher impedance just gives less distortion and less power.

SR driver - it is identical to the Kaiju circuit (each dual triode uses one triode to regulate the voltage and the other triode to amplify the signal). the Zeners in Kaiju protect the rulator from the higher voltage and are not needed in a Stereomour which has a lower power supply voltage. Both have RC stabilizing networks. I suppose you could re-wire it for the 5670 which has less gain but is more linear than the 12AT7 - but when we were developing the 5670 circuit, we found that distortion measurements were identical in the circuit - the current-source plate load reduces distortion to the point that the output tube dominates.

Hope that helps!

Paul Joppa


Online Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5751
Reply #4 on: July 23, 2019, 05:18:43 PM
I should add that the BCP-15 has exposed terminals with full power supply voltage, which is a serious safety hazard if top-mounted, unless you can hermetically seal and insulate them. Dust and oil aerosols can get into even the tiniest gaps, and can be conductive. I'm serious!

Paul Joppa


Deke609

  • Guest
Reply #5 on: July 23, 2019, 05:26:10 PM
Many, many thanks PJ.  Yes, that helps and greatly simplifies things.


{edit: ignore previous version of this post - I was mixing together two different formulas}  It's been a long day!

« Last Edit: July 23, 2019, 05:29:39 PM by Deke609 »



Offline Paul Birkeland

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 19319
Reply #6 on: July 24, 2019, 10:22:38 AM
Other things being equal (including use of high quality parafeed caps), which if either would you predict might sound "better"? Option A - bigger choke, keep the 4K OT, and a big parafeed cap, or Option B - big choke and big OT, with smaller parafeed cap?
I would not expect you to notice much of a difference between 40H and 80H.  If the MQ OPT isn't available, I'm sure Sowter could make you a pair of 4K parallel feed output transformers with some nickel in the stack (they offer a 2.5K and a 5K already).  This is what I would do if you have to do something different. 
My understanding is that going with a bigger OT requires a new operating point for the 45B - I'm not sure if going bigger with the choke does.  Does the SR circuit or any other part of the SII circuit prevent or limit changes to the operating point of the tubes?
A bigger choke will have higher DCR and will drop the plate voltage, so you'll have to raise the B+ or accept lower output power.

The Kaiju appears to shunt regulate both halves of the driver tubes, whereas in the SII with SR, only 1/2 is regulated. Further, one of the Kaiju C4S boards has paralleled cap and resistor in parallel with the LEDs, and the Kaiju has a zener string.
The Kaiju and the upgraded Stereomour have very similar driver circuits, but with different tubes.  Your first statement above is a bit confusing to read, but in both the Kaiju and the upgraded Stereomour II, there is a hybrid shunt regulator for each driver stage.  There's no difference in "how much" is regulated between the two amps.  The Kaiju has the zener string because the raw B+ voltage is higher, and we left the soft start components in that were part of the Paramount driver circuit.


Question 3 for the experts: could any of the above Kaiju SR "extras" (or any others that I missed) be incorporated into the SII and potentially make a positive difference?
The plate chokes would make a positive difference, but it's pretty small when you're already running the #45 mods.  For a stock 2A3 Stereomour, the difference would be more significant.

Paul "PB" Birkeland

Bottlehead Grunt & The Repro Man


Deke609

  • Guest
Reply #7 on: July 24, 2019, 10:52:14 AM
  Your first statement above is a bit confusing to read, but in both the Kaiju and the upgraded Stereomour II, there is a hybrid shunt regulator for each driver stage.


I must be reading the schematics wrong, b/c when I compare the SII SR schematics with the Kaiju schematic, I see one TL431 board feeding the cathode of 1/2 of an 12AT7 in the SII, and for the Kaiju I see two TL431 circuits, one for each half of a 5670. One appears to provide 4V bias. As far I can see, in the SII, the 12At7 is biased by a cathode resistor. 


Anyway, that's where I was coming from. But, as you and PJ have made clear that the SR circuits in the two amps are functionally identical, it doesn't really matter.


cheers, Derek



Offline Paul Birkeland

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 19319
Reply #8 on: July 24, 2019, 11:17:55 AM
Yes, that's a difference in how the bias is obtained for the Kaiju vs. the Stereomour.  The Stereomour doesn't benefit from adjustable cathode bias voltage and really needs the full mu of the 5670, while the Stereomour has less stringent requirements for the operation of the 12AT7, so the lower gain from just a cathode resistor isn't a problem, and some deviation in plate voltage isn't so critical when you need less voltage swing from the driver.

Paul "PB" Birkeland

Bottlehead Grunt & The Repro Man


Deke609

  • Guest
Reply #9 on: July 24, 2019, 11:21:39 AM
Thanks PB. That clears up things for me.



Deke609

  • Guest
Reply #10 on: November 11, 2019, 10:38:24 AM
@PB or @PJ - I'm doing some planning for my SII-45B rebuild (using the higher power EML 45B tubes). Jac, by way of Roger Modjeski, reports good results with the 45B's with B+ of 495V, plate voltage and grid bias at about 410V and -85V respectively, with 9.2K OT primary impedance: 3.6W output with about -29dB 2nd harmonic distortion. I'd like to try something similar, but with 9.7K primaries (the LL1679 OTs look to have lots of power handling to spare, so no worries about saturation/loss of bass). 

But I also want to implement the Kaiju 5670 shunt reg/driver circuit. The Kaiju has raw B+ of approx 450V, but I'll be using closer to 500V (by way of a different PT). So the question I am hoping you can help me with is: To deal with extra raw B+ voltage, can I just tweak the zener string so that it shunts 200V instead of 150V (assuming that this is its purpose)?

(And, yes, I do realize that all of this would be avoided if I got the new monoblocks and converted them to run the EML 45Bs - and I would have already ordered them had I not already paid for all this new iron - the announcement of the new amps came after I ordered the Lundahl chokes and OTs).

many thanks in advance, Derek



Online Paul Joppa

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 5751
Reply #11 on: November 11, 2019, 12:04:47 PM
The Zener string in the Kaiju is there to protect the transistors in the plate load C4S from the startup transient voltages. It's a tricky balance among the voltages and currents and the heat dissipation of the transistors. Without doing a complete analysis, I am unwilling to say what changes may or may not work.

I don't recommend using the 5670 at such a high output level, either - it would use up most of the overload margin.

Paul Joppa


Deke609

  • Guest
Reply #12 on: November 11, 2019, 12:24:13 PM
Many thanks PJ. I just knew it couldn't be that easy!  ;D  I'll "hit the books" some more on CCS loading and shunt regs to see if I can find something promising.

An interesting alternative to the 5670 ciruit might be the EML 30B driver tube - similar gain to the 5670 (32 versus the 5670's 35) - and Jac reports that it is very linear.  It's otherwise a completely different species of tube, but choke loaded, it could be made to work with the higher B+.  That might be something to play with as I try to figure out an analogue to Kaiju's 5670 circuit that would work with the higher B+.

cheers and thanks, Derek



Offline Paul Birkeland

  • Global Moderator
  • Hero Member
  • *****
    • Posts: 19319
Reply #13 on: November 11, 2019, 01:20:59 PM
with 9.2K OT primary impedance: 3.6W output with about -29dB 2nd harmonic distortion. I'd like to try something similar, but with 9.7K primaries (the LL1679 OTs look to have lots of power handling to spare, so no worries about saturation/loss of bass). 

It's certainly your money to spend, but 10K single ended transformers are not particularly easy to make and the good ones I have bought have been stupidly expensive (over $1000/pair).  The 40H of the 1679 will get you down to about 40Hz, and things will fall apart in a hurry below that.  If they are making you a special parallel feed version and you're using it with a 100+H plate choke, that should be somewhat better, but 3.6W with 2nd @ -29dB should be easily doable with a 5K output transformer and a 300B at a very light duty operating point.  This is often referred to as a "Japanese" operating point, as it seems to pop up more over there in the DIY community.  You could run a 300B with 500V of B+ and a 5K transformer and get about 7W at similar THD.  I would not be surprised if the MonAmour with the JJ2A3-40 running on the 60mA position would do about 5W at those conditions; perhaps I'll have time to check on that this week.

An interesting alternative to the 5670 ciruit might be the EML 30B driver tube
The EML30A ought to be loaded with a plate choke having about 150H of inductance, which is painfully expensive/hard to find.  If you had a ton of B+, you could probably get away with a high voltage current source to load a 30A with 7-10mA and no shunt regulated driver stage.

The direction you're going will be something that is completely unrelated to a 45 amp and won't work with any tubes other than the oddball EML.  This isn't to discourage your experimentation, but just some cautionary advice that it's way easier to backwards in iron quality from the Kaiju than it is to go the other way!  In particular if you run a 10K/40H output transformer and something like the Lundahl recommended for the EML driver tubes, it would not surprise me to have you come back here asking why there isn't much bass  ;D

Paul "PB" Birkeland

Bottlehead Grunt & The Repro Man


Deke609

  • Guest
Reply #14 on: November 11, 2019, 02:22:56 PM
Thanks PB.

The 40H of the 1679 will get you down to about 40Hz, and things will fall apart in a hurry below that.  If they are making you a special parallel feed version and you're using it with a 100+H plate choke, that should be somewhat better, but 3.6W with 2nd @ -29dB should be easily doable with a 5K output transformer and a 300B at a very light duty operating point.
[snip] ... This isn't to discourage your experimentation, but just some cautionary advice that it's way easier to backwards in iron quality from the Kaiju than it is to go the other way!  In particular if you run a 10K/40H output transformer and something like the Lundahl recommended for the EML driver tubes, it would not surprise me to have you come back here asking why there isn't much bass  ;D

Yeah, even I *know* (well, sort of - it's more a fuzzy sense than true understanding) that 40H would be a step in the wrong direction! The 40H figure is for the SE (big gap) version. The PP (tiny gap) version is 150H. There is no datasheet for the unofficial "PPZ" (zero gap, parafeed) version that I ordered, but Jac states that going to zero gap gives you roughly double the inductance of the PP version - so approaching 300H.

I'll be using them with 70H plate chokes.  So I'm hoping bass will be pretty good.

And 9.7K primaries is just one of things I want to try. Since the OTs can be configured for 2K6, 4K5 and 9K7, I've got lots of room to play.  The first thing I might try is just swapping in the OT's configured for 4K5 - pretty close to the existing 4K.  And then add the 70H (150 DCR) choke with additional dropping resistor if necessary to stay close to the original operating parameters.  I have no idea how the OTs will sound, but Lundahl's amorphous core versions get pretty good reviews and are reported to sound similar to mumetal/permalloy - we'll see.

Quote
The EML30A ought to be loaded with a plate choke having about 150H of inductance, which is painfully expensive/hard to find.  If you had a ton of B+, you could probably get away with a high voltage current source to load a 30A with 7-10mA and no shunt regulated driver stage.

Eesh. Yeah, 150H might be a tall order. I was hoping to be able to use the SII 40H choke. But Lundahl does make 165H @ 25mA choke. And it's not super pricey.  But a HV CCS would be great, particularly if I can put it together myself from a schematic. I was just reading about John Broskie's idea for a shunt reg board that can put out 600V at 30mA load - but I don;t understand enough (yet) to know if this could be adapted to my purposes.

I'll stick with the 12AT7 to start with. Maybe it can be tweaked?  Any potential benefit to emplying the Kaiju's TL431 cathode bias scheme for the 12AT7?

Quote
The direction you're going will be something that is completely unrelated to a 45 amp and won't work with any tubes other than the oddball EML. 

Yup. But I'll disagree with you in one regard: to my ears, the EML 45B captures the 45 sound, but with better punch and clarity. I've got some pretty decent NOS 45 pairs that sound pretty darn good. But after the EML 45Bs got about 200+ hours on use, they beat out (to my ears) every other pair of 45s  I have.  The EMLs are the only ones I want to listen to now. 

cheers and thanks, Derek