Tried your idea , reduced .001 cap down to 220pf, it screwed up by bias setting and fried one output transformer. I’m going back to all original parts. A little rinsing at 10k square wave isn’t worth it. My sinewave was good. This experiment will cost me a new-transformer and a complaining wife.. My original parts and results are good enough for me I’m not made of money
Thanks for what you offered,but….
Thanks for what you offered,but….
Lmitchell posted this in his other Citation II thread today, but probably meant to post it here:
In any case, don't short C17,C18,C19,C20 or C16,C22,C26,or C13,C21 either.
C26 appears to be mis-wired to the output tube plate on the schematic.
and his post #8, above here:
I don't see any suggested mods above that could have damaged the OT. But shorting out the C17,C18 or C19,C20 bias isolating caps would cause runaway tube current. R63,R64 or R65,R66 15 Ohm, 1 Watt cathode resistors should have fused in that case still. So I would recommend checking/testing those resistors. Maybe change those to 1/2 Watt metal film.
In any case, don't short C17,C18,C19,C20 or C16,C22,C26,or C13,C21 either.
C26 appears to be mis-wired to the output tube plate on the schematic.
Last edited:
I don't understand what he's referring to or how he managed to blow an output transformer. Baudoin0 suggested a very standard low pass filter across the input stage's plate resistor, probably about a 40-50kHz cutoff. I suggested restoring the zobels at the outputs and adjusting the phase-lead cap upward a bit. None of these things should have caused much of an issue, unless the amp went into sustained oscillation and fried an output. I guess that's possible but it seems unlikely.
One thing I just noticed about the Antec MP-60W66 OT. Their data curve shows BIG losses at 20KHz. If the Amp was sitting there for some extended time with a high level 20KHz test signal on it, the OT probably -would- get rather hot: 8 or 9 dB loss!
https://www.antekinc.com/content/MP-60W66.pdf
https://www.antekinc.com/content/MP-60W66.pdf
In Post # 3 . . .
I see 3 "Push" negative feedback loops, and 3 "Pull" negative feedback loops, a cathode to cathode driver negative feedback loop, and a global negative feedback loop.
That is 8 negative feedback loops.
I hope that is enough negative feedback loops per channel.
I see 3 "Push" negative feedback loops, and 3 "Pull" negative feedback loops, a cathode to cathode driver negative feedback loop, and a global negative feedback loop.
That is 8 negative feedback loops.
I hope that is enough negative feedback loops per channel.
Hegemen himself said the UL loops were a mistake. And the local plate to grid loops on the 12BY7 drivers just waste gain from the outer loops. So there are 4 loops you can take out.
I have accumulated some parts to build a "modified" Citation II. All the plate to grid loops are coming out. I suppose a triode output stage would be the natural alternative to UL, but instead, I have a Toroidy OT, 6.6K, 10% CFB. 10 % CFB Fdbk goes to some JJ KT77 output tube cathodes, and attenuated CFB goes back to the driver cathodes (crossed). So driver and output grid 1's remain high Z. Fixed output screen voltage (so still gives 10% UL). Driver tubes changed to 6197 video tubes. (more linear than 12BY7, don't need any local loops, and got a huge box of them for $0.35 each, and they are near matched, right out of the box.) The Toroidy OT is near flat up to 50 KHz. One problem though is that one of the Toroidy CFB windings gives crazy effects when shorted on a freq. scan, while the other CFB winding looks as expected. I'm winding a second CFB layer on the toroid outside (a full layer) currently, to parallel with the errant CFB winding. (113 turns) I might use a 6F12P tube for the input and splitter. Super 6197 curves below:
Various TV Sweep tubes will get tested as outputs too, just a screen V change. (Is it still a Citation II? The "Mongrel II"? )
I have accumulated some parts to build a "modified" Citation II. All the plate to grid loops are coming out. I suppose a triode output stage would be the natural alternative to UL, but instead, I have a Toroidy OT, 6.6K, 10% CFB. 10 % CFB Fdbk goes to some JJ KT77 output tube cathodes, and attenuated CFB goes back to the driver cathodes (crossed). So driver and output grid 1's remain high Z. Fixed output screen voltage (so still gives 10% UL). Driver tubes changed to 6197 video tubes. (more linear than 12BY7, don't need any local loops, and got a huge box of them for $0.35 each, and they are near matched, right out of the box.) The Toroidy OT is near flat up to 50 KHz. One problem though is that one of the Toroidy CFB windings gives crazy effects when shorted on a freq. scan, while the other CFB winding looks as expected. I'm winding a second CFB layer on the toroid outside (a full layer) currently, to parallel with the errant CFB winding. (113 turns) I might use a 6F12P tube for the input and splitter. Super 6197 curves below:
Various TV Sweep tubes will get tested as outputs too, just a screen V change. (Is it still a Citation II? The "Mongrel II"? )
Last edited:
Well, I'm sorry you had a meltdown. It's possible, as smokingamp suggests in the other thread, that a sustained 20kHz square wave at high power overheated the output transformer. I don't know how much power you were inducing but generally you test for square wave response at 1 watt ouput. 10kHz testing is usually sufficient to tell you what you need to know in terms of HF stability, and it's a little less demanding on the amp.
It's also possible the the capacitor change caused the transformers to respond badly and created some oscillation, which would screw up the biasing, since the output tubes would be working at peak power trying to accommodate it.
At any rate, you're probably justified in leaving well enough alone.
It's also possible the the capacitor change caused the transformers to respond badly and created some oscillation, which would screw up the biasing, since the output tubes would be working at peak power trying to accommodate it.
At any rate, you're probably justified in leaving well enough alone.
Have you checked C26 to make sure it is NOT connected to the V4 output tube plate? ( erroneously shown connected on the schematic )
Also, I would recommend the Toroidy TTG-EL34PP 6.6K 60 Watt OT if you are ordering an OT replacement. 50KHz BW instead of 10KHz, same price, except for more shipping cost. (no affiliation, just have one here that I've tested)
Also, I would recommend the Toroidy TTG-EL34PP 6.6K 60 Watt OT if you are ordering an OT replacement. 50KHz BW instead of 10KHz, same price, except for more shipping cost. (no affiliation, just have one here that I've tested)
smoking-amp,
Thanks for your post about your modifications!
I consider 10% Screen UL to be about as close to Pentode / Beam Power mode as you can get.
I need to think about the similarities of 10% cathode feedback, versus 10% Screen UL; I can not quite wrap my head around that right now.
You said:
"Hegemen himself said the UL loops were a mistake. And the local plate to grid loops on the 12BY7 drivers just waste gain from the outer loops. So there are 4 loops you can take out."
That reminds me of what I just learned today:
Elon Musk has Reduced the Mileage Range Specifications of his electric cars.
As to the old inflated specifications; Trust but Verify.
It is interesting what specification changes are admitted after the fact of the initial sales.
When it comes to how many negative feedback loops should be used, I generally refer to Albert Einstein:
"You should make things as simple as possible, but no simpler".
I had the good fortune to use and support the Sony-Tektronix 370B curve tracer.
Thanks again, and have fun!
Thanks for your post about your modifications!
I consider 10% Screen UL to be about as close to Pentode / Beam Power mode as you can get.
I need to think about the similarities of 10% cathode feedback, versus 10% Screen UL; I can not quite wrap my head around that right now.
You said:
"Hegemen himself said the UL loops were a mistake. And the local plate to grid loops on the 12BY7 drivers just waste gain from the outer loops. So there are 4 loops you can take out."
That reminds me of what I just learned today:
Elon Musk has Reduced the Mileage Range Specifications of his electric cars.
As to the old inflated specifications; Trust but Verify.
It is interesting what specification changes are admitted after the fact of the initial sales.
When it comes to how many negative feedback loops should be used, I generally refer to Albert Einstein:
"You should make things as simple as possible, but no simpler".
I had the good fortune to use and support the Sony-Tektronix 370B curve tracer.
Thanks again, and have fun!
re: 6A3sUMMER
Check the ug2g1 ratio for the output tube being used. Typically around 8 for many audio tubes, but 11.5 for the KT77 or EL34, around 3 to 4 for TV Sweeps. That is the ratio of gm or sensitivity between g1 and g2.
So 10% CFB with a KT88 (mu 8) would be equivalent to an 80% UL. (cathode 8 times more sensitive than g2) KT77 becomes a 115% UL. Or one can look at this in terms of an equivalent triode. 100% UL gives a triode with mu 8 for the KT88 or mu 11.5 for the KT77. With 80% UL equiv., the KT88 becomes (u = 8)/0.8 = mu effective 10 or KT77 becomes (u=11.5)/1.15 = mu effective 10. Looks like it's just 1/CFB for mu. 50% CFB gives mu 2.
Hegemen didn't like UL because it runs out of Vg2 when you need it most for current drive. (unless going +Vg1) Same thing for triodes. (unless going +Vg1) So the tubes loose their grunt when you need it most. Pentode keeps it's full Vg2 and similarly 10% CFB keeps 90% of its Vg2. Maximum current capability is preserved when you need it.
Even more, the linearity of CFB is better than UL or triode. For UL and triode the NFdbk is being applied to a 3/2 power electrode pair (g2 or plate and k), while drive is applied to a more square law electrode pair (g1 and K).
CFB NFdbk is applied to the same electrode pair as the drive. (g1 and k) So the square law characteristic cancels out fully there. So called local Schade resistive N Fdbk (R from plate to grid1) works similarly to CFB except in the current domain (using a current driver).
The problem with CFB is finding a good CFB OT that doesn't cost $$$$$$.
For the minimum # of NFdbk loops, I see a local loop around just the output tubes (No xfmr included) as a big positive, since that can linearize the most distorting tubes and lower Zout, without the xfmr phase problems. (a driver stage can be included for more loop gain, but trickier stability)
Then a global loop does not need so much gain to finalize the linearity and Zout job. Job easy now. Poorer or cheaper OTs can be accommodated too.
Local NFdbks for phase splitting, whatever, are fine if they make things easier and stay local.
Interactive, nested loops, however, (like the Citation II !!) can be a design nightmare or a tweakers paradise, depending on your view and schedule.
CFB does break the rules here, since getting good CFB results from the OT requires extra special winding talent. The CFB windings need to be closely coupled to the secondary, with matched coupling, to get good accuracy and phase response. But if you find a winder who can -actually- DO it for a good price, you are flying above the clouds.
Another way would be to use a LOW % UL type OT with the secondary wound between and around the UL sections for close coupling to the output. Actual UL use (to the screen grids) is then optional, but could be optimized for TV Sweeps (with a DC drop). The UL taps would then get used for attenuated NFdbks to the driver tubes (cathodes or grids).
Check the ug2g1 ratio for the output tube being used. Typically around 8 for many audio tubes, but 11.5 for the KT77 or EL34, around 3 to 4 for TV Sweeps. That is the ratio of gm or sensitivity between g1 and g2.
So 10% CFB with a KT88 (mu 8) would be equivalent to an 80% UL. (cathode 8 times more sensitive than g2) KT77 becomes a 115% UL. Or one can look at this in terms of an equivalent triode. 100% UL gives a triode with mu 8 for the KT88 or mu 11.5 for the KT77. With 80% UL equiv., the KT88 becomes (u = 8)/0.8 = mu effective 10 or KT77 becomes (u=11.5)/1.15 = mu effective 10. Looks like it's just 1/CFB for mu. 50% CFB gives mu 2.
Hegemen didn't like UL because it runs out of Vg2 when you need it most for current drive. (unless going +Vg1) Same thing for triodes. (unless going +Vg1) So the tubes loose their grunt when you need it most. Pentode keeps it's full Vg2 and similarly 10% CFB keeps 90% of its Vg2. Maximum current capability is preserved when you need it.
Even more, the linearity of CFB is better than UL or triode. For UL and triode the NFdbk is being applied to a 3/2 power electrode pair (g2 or plate and k), while drive is applied to a more square law electrode pair (g1 and K).
CFB NFdbk is applied to the same electrode pair as the drive. (g1 and k) So the square law characteristic cancels out fully there. So called local Schade resistive N Fdbk (R from plate to grid1) works similarly to CFB except in the current domain (using a current driver).
The problem with CFB is finding a good CFB OT that doesn't cost $$$$$$.
For the minimum # of NFdbk loops, I see a local loop around just the output tubes (No xfmr included) as a big positive, since that can linearize the most distorting tubes and lower Zout, without the xfmr phase problems. (a driver stage can be included for more loop gain, but trickier stability)
Then a global loop does not need so much gain to finalize the linearity and Zout job. Job easy now. Poorer or cheaper OTs can be accommodated too.
Local NFdbks for phase splitting, whatever, are fine if they make things easier and stay local.
Interactive, nested loops, however, (like the Citation II !!) can be a design nightmare or a tweakers paradise, depending on your view and schedule.
CFB does break the rules here, since getting good CFB results from the OT requires extra special winding talent. The CFB windings need to be closely coupled to the secondary, with matched coupling, to get good accuracy and phase response. But if you find a winder who can -actually- DO it for a good price, you are flying above the clouds.
Another way would be to use a LOW % UL type OT with the secondary wound between and around the UL sections for close coupling to the output. Actual UL use (to the screen grids) is then optional, but could be optimized for TV Sweeps (with a DC drop). The UL taps would then get used for attenuated NFdbks to the driver tubes (cathodes or grids).
Last edited:
The old Tek 576 curve tracer is near ideal for converting to vacuum tube curve tracing, since it is a dumb and straight-forward design. (no CPU) I just added a pot in the Fdbk loop for the step generator to change the gain, and put HV transistors in the step amplifier circuit to handle 250V. Then a boost power supply for the step amplifier got added, which can be switched in or out. Some caps and diodes around the step amplifier got upgraded for the higher step V too.
Can do up to 170V total for steps now, instead of the stock 20V.
Last edited:
The Tek museum recently de-assessed two of these for cheap ($100 IIRC, or something else dirt cheap). Only kicker was no shipping - ya hadda pick up at the museum. Still crying.
All good fortune,
Chris
ps: I've been there; cool museum, just west of PDX
All good fortune,
Chris
ps: I've been there; cool museum, just west of PDX
To smoking -amp…
could you provide a schematic of your test setup?
Are the Toroidy OT, 6.6K transformer that much better than the Antek? Is it worth the upgrade? At what cost?
Can the 6197 tubes be replace the 12by7a tubes? Would the feed back need to be eliminated or altered?
I am a little leery of changing C12 cap.
Thanks for the feedback, obviously you know about these tubes than me.
i’m trying to learn From my mistake.
could you provide a schematic of your test setup?
Are the Toroidy OT, 6.6K transformer that much better than the Antek? Is it worth the upgrade? At what cost?
Can the 6197 tubes be replace the 12by7a tubes? Would the feed back need to be eliminated or altered?
I am a little leery of changing C12 cap.
Thanks for the feedback, obviously you know about these tubes than me.
i’m trying to learn From my mistake.
The test setup I used was simple, but requires a sine wave generator with a freq. sweep feature for a nice scope display. I have an old Heathkit IG-1275 Lin/Log Sweep Generator. SS , 50 Ohm output, and works good. You connect the signal output to the OT primary ( use 1/2 or full primary) thru a series resistor. Use max signal gen. amplitude. The resistor emulates the output impedance of a tube stage. I used a 500 Ohm resistor that was handy. Not real critical, but higher R there will allow distributed winding capacitance to show up clearly in the results. 50 Ohm straight from the generator hides the capacitance effects. (I think some OT adverts use this!) I expect winding capacitance to be a major issue with low cost toroidal OTs.
Then connect the scope input to the full secondary 16 Ohm or the 8 Ohm section, if the signal is enough to see on the scope. Then set the sweep generator to sweep thru 10 Hz up to 100 KHz. The sweep generator usually will have another output connector with the sweep ramp signal on it. Connect that to the scope's Horizontal input jack. (not all scopes will have this) And adjust the scope Horiz. gain and positioning for a screen width sweep. You will then see the AC signal passing thru the OT displayed vertically versus freq. horizontally (will likely need some input sensitivity to see the reduced secondary voltage) I also put the direct signal gen. V on the 2nd trace of the scope (T conn.) to check for constant amplitude signal coming from the signal gen. You can manually adjust the HF end signal sweep freq. setting to see where the amplitude is 1st dropping off.
If you don't have a sweep generator, you can just sweep the signal gen. freq. manually to see where response drops off on the scope.
Someone tested an Antek a few years ago here, when they 1st came out, and only got around 7 Khz high end. I see Antek says they now have improved OTs. But if you check the Antek 6.6K 60 Watt specifications page: https://www.antekinc.com/content/MP-60W66.pdf
You will see it has dropped off significantly by 10 KHz. Improved?
The Toroidy TTG EL34PP 6.6K OT I have here looks good up to 50KHz. Their data sheet quotes that figure too. They are in Poland I think. About 4 Zl to the dollar last I checked. I got mine when the distributor tme.eu carried Toroidy products, and their shipping was well discounted. Now I think you have to order directly from Toroidy and the shipping is $$. And the Toroidy checkout didn't even have an option for shipping to the US. I had to email them recently to set up an order for a new product offering.
The Citation II has long been idolized for super OTs. I was a little surprised that the Antek worked at all. But the circuitry is actually what you would expect for using low cost OTs. Most of the NFdbk before the OT, and reduced N Fdbk thru the OT.
I would check that C26 connection on the schematic you gave. It is putting plate voltage back into the input. Can't be correct. It could be what finished off the OT.
The 6197 tube has a different pin-out than the 12BY7, and different heater voltage configuration. 6.3V only. Most changes will require some tweaking of the N Fdbks.
Then connect the scope input to the full secondary 16 Ohm or the 8 Ohm section, if the signal is enough to see on the scope. Then set the sweep generator to sweep thru 10 Hz up to 100 KHz. The sweep generator usually will have another output connector with the sweep ramp signal on it. Connect that to the scope's Horizontal input jack. (not all scopes will have this) And adjust the scope Horiz. gain and positioning for a screen width sweep. You will then see the AC signal passing thru the OT displayed vertically versus freq. horizontally (will likely need some input sensitivity to see the reduced secondary voltage) I also put the direct signal gen. V on the 2nd trace of the scope (T conn.) to check for constant amplitude signal coming from the signal gen. You can manually adjust the HF end signal sweep freq. setting to see where the amplitude is 1st dropping off.
If you don't have a sweep generator, you can just sweep the signal gen. freq. manually to see where response drops off on the scope.
Someone tested an Antek a few years ago here, when they 1st came out, and only got around 7 Khz high end. I see Antek says they now have improved OTs. But if you check the Antek 6.6K 60 Watt specifications page: https://www.antekinc.com/content/MP-60W66.pdf
You will see it has dropped off significantly by 10 KHz. Improved?
The Toroidy TTG EL34PP 6.6K OT I have here looks good up to 50KHz. Their data sheet quotes that figure too. They are in Poland I think. About 4 Zl to the dollar last I checked. I got mine when the distributor tme.eu carried Toroidy products, and their shipping was well discounted. Now I think you have to order directly from Toroidy and the shipping is $$. And the Toroidy checkout didn't even have an option for shipping to the US. I had to email them recently to set up an order for a new product offering.
The Citation II has long been idolized for super OTs. I was a little surprised that the Antek worked at all. But the circuitry is actually what you would expect for using low cost OTs. Most of the NFdbk before the OT, and reduced N Fdbk thru the OT.
I would check that C26 connection on the schematic you gave. It is putting plate voltage back into the input. Can't be correct. It could be what finished off the OT.
The 6197 tube has a different pin-out than the 12BY7, and different heater voltage configuration. 6.3V only. Most changes will require some tweaking of the N Fdbks.
Last edited:
Thanks for the response.l’ll look into the Toroidy TTG EL34PP 6.6K OT transformer.
Maybe later this year if I get the money to cover the cost.
Is. It worth changing out the small tubes in a new design.
Maybe later this year if I get the money to cover the cost.
Is. It worth changing out the small tubes in a new design.
When 'shorting' (discharging) large capacitors, it's best to have some resistance in the grounding path, say a 10 ohm 20 watt WW resistor in the shorting lead to limit excessive discharge current. Some capacitors can be damaged by directly shorting them, especially high capacitance ones. After discharging the cap, check with a voltmeter to be sure it's drained, in the unlikely case that the resistor should become open circuit, or the resistor was not in circuit long enough to completely discharge the capacitor. Save the sparks for tesla coils!