I've got some 6EW6, 6J11 and 6BN11 tubes around. Let me see if I can find them. I have heard that IF tubes can be microphonic, so some selecting may be required.
If you are looking for really good small triodes, I found that Westinghouse (Japan) 6JC6A tubes are amazing in triode. (but not so hot in pentode), while USA made 6JC6As are great in pentode, but not so hot in triode. Mu around 60.
If you are looking for really good small triodes, I found that Westinghouse (Japan) 6JC6A tubes are amazing in triode. (but not so hot in pentode), while USA made 6JC6As are great in pentode, but not so hot in triode. Mu around 60.
Last edited:
Thanks again very much. I did buy a supply of Admiral branded 6JC6 (not A, octagon etching, RCA?) on your recommendation back in the Tube From Hell era, but haven't even tried any yet. All of these last gen IF amplifier pentodes are very compelling, and right now a bargain for DIYers. Nobody used them for guitar amps, so all are from the Valve Era and not modern (lesser) copies, but still available in quantity for smallish bucks. Might be a good choice for the OP and others interested in the circuit. What would you suggest for a target cathode current (as a pentode)?
Iain's amplifier project required 12 pin compactrons, even a pair of 6AX3 rectifiers (12 pin versions of 6AX4), so we're restricted by fashion. Just ask David Bowie. https://www.itishifi.com/hifi/diy-compactron-set-part-2
All good fortune,
Chris
Iain's amplifier project required 12 pin compactrons, even a pair of 6AX3 rectifiers (12 pin versions of 6AX4), so we're restricted by fashion. Just ask David Bowie. https://www.itishifi.com/hifi/diy-compactron-set-part-2
All good fortune,
Chris
This is great, thank you both!
Under the influence of all those posts with plate curves of obscure (to me, at least) radio and TV receiving tubes, I've been collecting various types when I see them available for cheap. I got a bunch of shiny new 6AC7s for 35 cents each, also some Sylvania 6JC6A (triode mu = 60? I need to try that!), 12HL7 (which I tried as triodes in a 'spud' headphone amp and they sound excellent), 6JC5, 6CF6 (I couldn't find 6CB6A for cheap), also 6HB6/6HA6 and 6GE5.
I've thrown together an LTspice model of the SP10 circuit. I can upload it if you'd like to play around with it. Let me know if you'd like me to do that. It looks like the circuit will make only about 9W max into the chosen OPT model (Hammond 1608). I'm sure a better OPT would allow it to make another watt or two.
I can see that the 6AU6s are running in starvation mode, with combined plate+screen dissipation of only about 75mW, with plate current of only 0.32mA. The sim says THD will be kind of high, at about 0.09% at 1W output into 8 ohms (2.828V rms).
From plugging different (virtual) tubes in and out of the circuit, along with reducing the value of R14 (screen load to the 6AU6s) to 470k (about half of the original 1M) and R15, R11 (plate loads on 6AU6s) to 150k (again, about half the original value), it appears I can increase the plate current through the 6AU6s to about 0.64mA. The THD and max output power stay exactly the same in the simulation.
However, if I swap in a 6SN7 and change values of R2, R3. R4, R5 to bias the 6SN7, the THD can be reduced to half of what it is with 12AU7, with no changes to the 6AU6 driver and 6V6 output stages. It looks like the THD at the output is dominated by the 12AU7 voltage amp-phase splitter stages.
The biggest improvement comes from simply goosing the B+ higher and replacing the 6V6s with an output pentode that can take the higher voltages. Simply leaving the circuit as-is except changing the B+ to +430VDC increases the plate current through the 6AU6s to 0.43mA. That. and putting in 6L6s for the outputs, reduces THD at 1W out to 0.06%, and max power out into 8 ohms is increased to a whopping 15.6W. I think the 6AU6 drivers are running out of steam before the output stage clips. It would take some redesigning of the driver stage to get this thing working at higher voltages.
This is what the virtual scope probe sees at the grid of one of the 6L6GC outputs at the slightest onset of clipping at the speaker output. It looks like the 6AU6 drivers can only swing about 16V peak cleanly, no? A more muscular driver pentode would be needed here, I think. The output from the 12AU7 phase splitter looks fine while this is happening. I think the 6AU6 driver stage is the limiting factor.
More later... Gotta go to work.
Under the influence of all those posts with plate curves of obscure (to me, at least) radio and TV receiving tubes, I've been collecting various types when I see them available for cheap. I got a bunch of shiny new 6AC7s for 35 cents each, also some Sylvania 6JC6A (triode mu = 60? I need to try that!), 12HL7 (which I tried as triodes in a 'spud' headphone amp and they sound excellent), 6JC5, 6CF6 (I couldn't find 6CB6A for cheap), also 6HB6/6HA6 and 6GE5.
I've thrown together an LTspice model of the SP10 circuit. I can upload it if you'd like to play around with it. Let me know if you'd like me to do that. It looks like the circuit will make only about 9W max into the chosen OPT model (Hammond 1608). I'm sure a better OPT would allow it to make another watt or two.
I can see that the 6AU6s are running in starvation mode, with combined plate+screen dissipation of only about 75mW, with plate current of only 0.32mA. The sim says THD will be kind of high, at about 0.09% at 1W output into 8 ohms (2.828V rms).
From plugging different (virtual) tubes in and out of the circuit, along with reducing the value of R14 (screen load to the 6AU6s) to 470k (about half of the original 1M) and R15, R11 (plate loads on 6AU6s) to 150k (again, about half the original value), it appears I can increase the plate current through the 6AU6s to about 0.64mA. The THD and max output power stay exactly the same in the simulation.
However, if I swap in a 6SN7 and change values of R2, R3. R4, R5 to bias the 6SN7, the THD can be reduced to half of what it is with 12AU7, with no changes to the 6AU6 driver and 6V6 output stages. It looks like the THD at the output is dominated by the 12AU7 voltage amp-phase splitter stages.
The biggest improvement comes from simply goosing the B+ higher and replacing the 6V6s with an output pentode that can take the higher voltages. Simply leaving the circuit as-is except changing the B+ to +430VDC increases the plate current through the 6AU6s to 0.43mA. That. and putting in 6L6s for the outputs, reduces THD at 1W out to 0.06%, and max power out into 8 ohms is increased to a whopping 15.6W. I think the 6AU6 drivers are running out of steam before the output stage clips. It would take some redesigning of the driver stage to get this thing working at higher voltages.
This is what the virtual scope probe sees at the grid of one of the 6L6GC outputs at the slightest onset of clipping at the speaker output. It looks like the 6AU6 drivers can only swing about 16V peak cleanly, no? A more muscular driver pentode would be needed here, I think. The output from the 12AU7 phase splitter looks fine while this is happening. I think the 6AU6 driver stage is the limiting factor.
More later... Gotta go to work.
Rongon,
Does the 6AU6 circuit have a missing part?
Put a bypass capacitor from the 6AU6 screens to ground. Then add the bypass cap to the scematic.
Or, if you want to connect the bypass caps to the cathodes instead of ground, you have to split the connection between the screens, use 2 caps, then use 2 separate resistors to connect the screens to B+.
You will probably get more voltage swing from the 6AU6 plates with the bypass cap(s).
Un-bypassed screens do not work well for RF; and they do not work well for Audio.
I hope that makes the amp closer to being finished.
Does the 6AU6 circuit have a missing part?
Put a bypass capacitor from the 6AU6 screens to ground. Then add the bypass cap to the scematic.
Or, if you want to connect the bypass caps to the cathodes instead of ground, you have to split the connection between the screens, use 2 caps, then use 2 separate resistors to connect the screens to B+.
You will probably get more voltage swing from the 6AU6 plates with the bypass cap(s).
Un-bypassed screens do not work well for RF; and they do not work well for Audio.
I hope that makes the amp closer to being finished.
I found those 6EW6 .... tubes, so hopefully can get them traced this afternoon. The GE 6AU6 datasheet shows them operating between 5 an 10 mA idle. Something must be wrong with the SP-10 schematic.
Hmmm, my saved copy of the SP-10 has a simpler bias network for the 6AU6s. Plain vanilla.
On closer inspection, it's just drawn differently. Same thing.
Sounds like a good idea.
Hmmm, my saved copy of the SP-10 has a simpler bias network for the 6AU6s. Plain vanilla.
On closer inspection, it's just drawn differently. Same thing.
Yes, that's the one. The only difference between that Sams Photofact schematic and the one in the back of the RCA RC-19 manual is the power takeoff 5-pin socket to the lower right (that's absent from the RC-19 schematic).
In simulation, putting a 1uF cap from the 6AU6 screens to ground gave a very little bit more gain and increased THD a bit. It was not a big change. I'll revisit that.
The weird things I see in the SP-10 circuit are:
I was thinking (which is always dangerous)...
In simulation, putting a 1uF cap from the 6AU6 screens to ground gave a very little bit more gain and increased THD a bit. It was not a big change. I'll revisit that.
The weird things I see in the SP-10 circuit are:
- 6AU6 screens fed from a common 1M ohm resistor (R14), unbypassed (no capacitor). That value of R14 is really high, limiting the screen voltage to only 65V.
- Huge value 330k ohm plate load resistors for the 6AU6s (R11, R15). Do they need to be 330k? How about 220k? 150k? What would that do?
- Input stage voltage amplifier 12AU7 (V1) has an enormous 220k ohm plate load resistor, and a large value Rk of 6.8k, for a plate current of less than 1mA. That's practically in starvation mode. The THD from that stage alone is up near 0.1% at 1W out into an 8 ohm load.
I was thinking (which is always dangerous)...
- Improve the screen supply to the pentode push-pull driver. Perhaps as 6A3sUMMER suggested, split R14 into a screen R for each pentode, with screen bypass capacitors from g2 to k, the way it's supposed to be done.
- Replace 6AU6s with 6AC7 (2x the gm) or perhaps there's a better pentode to use that won't require extensive modifications.
- Replace 12AU7 with 6FQ7 or 6SN7 and modify the parts values to get both the AF AMP and PHASE INV stages to operate with about 4mA plate current and 100V plate voltage.
- Increase B+ to over 400VDC and use bigger output tubes like EL34, 6L6GC, etc. for more power.
- Perhaps make a stabilized or regulated screen supply for the output tubes and use horizontal deflection beam tubes rated for 15W or more plate dissipation.
Low G2 voltage and low Ia/high Z load, isn't that an old trick to maximize gain in small signal pentodes? I believe I've seen it used in old guitar amps with pentode input stages.
If that's what they were up to, then time to ditch these 6AU6s for something more robust. 6JC6? 2x the gm.
We need a driver that can handle most any output tubes. This Amp should be the starting point for other creative designs.
Stu Hegeman ( HK Citation II ) used 12BY7 video amp tubes for some knockout punch. They are getting a little scarce apparently. 6197 is an improved and cheap 12BY7A equivalent. ( OK, I'm biased, got 200 of them ) Or 6HZ8, gives you the whole front end.
Then there is the 12HL7, the Tube from Hell. Another 2X gm and great triode curves besides.
I looked for the RCA SP-20 schematic, but no luck. Apparently the same tubes as the SP-10, but with a quad of 6V6 outputs. Could that wimpy starvation 6AU6 circuit still work?
Last edited:
A pentode-wired 6V6 (or two) is probably a very easy load for the driver stage compared to pretty much anything else, I guess it could work? I seem to remember that Futterman used a current starved 6AU6 as input stage in his original OTL amp too.
I'm glad I've been collecting these last-of-the-tube-era pentodes.
I have about 10 6JC6A.
I also have about 10 12HL7.
I could get a couple sleeves of 6197, since they're still cheap.
I have some Russian 6J9P pentodes too (I think those are equivalent to 6688, maybe E180F).
I also have some Russian 6J51P, which are supposed to be equivalent to EF184 (6EJ7).
The problem for me is that I don't know how to replicate that feedback network with the 330k resistors from cathode to grid of each driver pentode. I can calculate values for output plate to driver cathode FB resistors, but I would like to keep any DC blocking capacitors out of the FB loop.
I have about 10 6JC6A.
I also have about 10 12HL7.
I could get a couple sleeves of 6197, since they're still cheap.
I have some Russian 6J9P pentodes too (I think those are equivalent to 6688, maybe E180F).
I also have some Russian 6J51P, which are supposed to be equivalent to EF184 (6EJ7).
The problem for me is that I don't know how to replicate that feedback network with the 330k resistors from cathode to grid of each driver pentode. I can calculate values for output plate to driver cathode FB resistors, but I would like to keep any DC blocking capacitors out of the FB loop.
Apparently the service manual for RCA SP-20 was posted on AudioKarma, here:
https://audiokarma.org/forums/index...-for-rca-rs147-or-sp-20.932590/#post-14178656
I can't seem to log in there, but maybe someone else can?
I'll try logging in later...
I could see that current-starved pentode trick working for input stages with very light loading (like into the grid of a cathode follower), but as a push-pull driver stage? I guess it could work in the RCA SP-10 because a 6V6 is an easy to drive output pentode. Would an EL84 be an even easier load? Maybe I should look into that...
I wonder if putting source followers after the 6AU6 plates would improve matters.
I don't have any relevant numbers in my head but I think it's safe to assume that the input capacitance here is much lower than what we see in in output stages with bigger tubes wired for UL or triode mode. Still, it's probably a good idea to use driver tubes with higher gm and run them at more normal plate currents.
I'm sure that could be made to work, but it adds more tubes (er, Mosfets). Just using a bigger driver with a lower plate resistor should work for driving most anything.
Last edited:
Thanks! Apparently the same exact driver!
Was there an SP-40 or 50 Amp? They must have upgraded that driver stage at some point. Maybe a warehouse backlog of 6AU6s and 6V6s had to be cleared out.
Was there an SP-40 or 50 Amp? They must have upgraded that driver stage at some point. Maybe a warehouse backlog of 6AU6s and 6V6s had to be cleared out.
I'd be happy to use a bigger driver.
Would that mean I could discard the resistor network in the cathodes of the driver stage pentodes? (R9, R10)
I'll have to try that in simulation to see if it looks like I can figure out parts values and retain the DC-coupled NFB.
Adding MOSFETs wouldn't be so bad, as long as I don't try to DC-couple them to the output tubes' grids. Retain the cap coupling. It's kind of a band-aid application, but it's a cheap 'n easy way of adding oomph to that wimpy driver stage. I'd be careful to use low Crss MOSFETs (like ZVN0545).
According to the RC20 manual, there's 23dB of NFB applied through R18, R19. Is that much NFB necessary? I wonder if backing that off to something like 16dB would ruin everything.
Last edited:
rongon,
The 6AU6 plate load is:
At DC, 330k , Rl
At AC, 330k Rl in parallel with 330k Rg . . . I am not a fan of that 1:1 ratio.
My preference. Some designers do it successfully.
I have never been a fan of starving tube current, just to get more gain.
And, for some circuits the driver plate and Rl impedance can not drive the capacitance of the next stage.
Some will try Triode wiring the outputs, and conclude that the high frequencies are rolled off . . . they are, but the driver is the cause.
So they conclude that [all] Triode wired is not good. A missed opportunity for them. Ha!
My preference. Some designers do Triode wired successfully, some do not.
Note: A little high frequency roll off, hardly ever hurt an Electric Guitar Amplifier.
But this is a Hi Fi thread.
The 6AU6 plate load is:
At DC, 330k , Rl
At AC, 330k Rl in parallel with 330k Rg . . . I am not a fan of that 1:1 ratio.
My preference. Some designers do it successfully.
I have never been a fan of starving tube current, just to get more gain.
And, for some circuits the driver plate and Rl impedance can not drive the capacitance of the next stage.
Some will try Triode wiring the outputs, and conclude that the high frequencies are rolled off . . . they are, but the driver is the cause.
So they conclude that [all] Triode wired is not good. A missed opportunity for them. Ha!
My preference. Some designers do Triode wired successfully, some do not.
Note: A little high frequency roll off, hardly ever hurt an Electric Guitar Amplifier.
But this is a Hi Fi thread.
Last edited:
Thinking about the resistor network, The 100K N Fdbk resistor from the output plate pulls the 1800 Ohm cathode resistor positive being DC coupled. The 330K and 1 Meg grid resistors just form a voltage divider from cathode to ground to get the grid voltage set somewhere negative of the cathode for correct cathode idle current. Of course the screen V has to be considered too.
Last edited:
Since that resistor ratio sets closed loop gain, it pretty much needs to be whatever it needs to be, and can only be scaled a limited distance towards smaller resistance values without imposing too much load on the output valves. RCA's choice to divide the resulting driver cathode voltage isn't really optimum, so a small positive DC voltage applied to driver grids (G1) seems worth exploring.
This combination of small positive DC on the grid and larger than normal cathode resistor was specified for the fancy high-Gm pentodes used for wideband / "baseband" amplifiers in their day (see 6688 or E180F), because it gives an extra strong dose of cathode resistor current regulation over lifetime. In our case, this effect is smaller (because the DC feedback contributes a fixed amount of DC current) but it's still some good negative feedback, rather than the DC positive feedback in the RCA original biasing.
WRT bypassing the common G2 supply, or not, this was usual practice in those days. Might have been looked on as similar to an unbypassed common cathode resistor in class A push-pull output stages, DTN Williamson etc.
All good fortune,
Chris
This combination of small positive DC on the grid and larger than normal cathode resistor was specified for the fancy high-Gm pentodes used for wideband / "baseband" amplifiers in their day (see 6688 or E180F), because it gives an extra strong dose of cathode resistor current regulation over lifetime. In our case, this effect is smaller (because the DC feedback contributes a fixed amount of DC current) but it's still some good negative feedback, rather than the DC positive feedback in the RCA original biasing.
WRT bypassing the common G2 supply, or not, this was usual practice in those days. Might have been looked on as similar to an unbypassed common cathode resistor in class A push-pull output stages, DTN Williamson etc.
All good fortune,
Chris