None of the options are very expensive fortunately... 🙂 Within the likely usage, the 6AU6/6136 is a fine option. Not that I'd want to try something else...LOL
Douglas
Douglas
I'm trying to get back to this project, but my recuperation has been lengthy. I'm okay, but have been experiencing some minor complications. The good news is that I officially retired last week, so I should be able to start clearing space to get back to work soon. I'm looking forward to getting started on the physical phase of this project. If the amp ends up performing really badly then I'll switch to something more conventional.
Hopefully soon...
Hopefully soon...
@rongon Welcome to the land of the retiree. I've been enjoying my amp building and restoration since retiring. Always hoped I would have time and energy for this, at least for a few years. I'm working up a project to make an SP-10 clone. A fellow on another site worked up some PC boards for the passive components and for a power supply filter set. He gave me a set of the boards to do an amp and I'm getting ready to cut an aluminum top plate and start drilling and punching. I'm going to make mine with 12V tubes. I have a good Stancor power transformer which has a 12.6V winding but no 6.3V and I decided to try a 12SN7 as the input/PI tube. We'll see how it goes. With summer coming and yard work looming I'm not certain how soon it will be running.
John
John
Great to hear that you're working on a similar project. I hope it turns out really well.
I can tell you that in LTspice, this RCA SP10 circuit looks like it performs really well. I understand that simulation and real life are often two different things, but my experience has been that if a circuit simulates really well it will come out working at least reasonably well in real life -- although perhaps not as well as in the sim. Simulation can help predict what circuits are likely to perform well and which are likely to perform less well, but it can't predict exactly what kind of performance you'll get in the end.
I have a steel chassis that was given to me. I'd rather not have a steel chassis, but that's what I have. It was drilled for 9-pin mini sockets for the output tubes (EL84s) and a single 9-pin mini socket for a cathodyne preamp stage/phase splitter for each channel. I've re-punched the output tube holes to fit octal sockets (I'm planning to use 6AV5GTA) and I'll need to punch the chassis for another two 9-pin mini sockets (I'll actually use 7-pin mini tubes, probably 6AU6A or 6CF6) and one octal socket for a 6SN7 voltage amp stage (one triode per channel), with a ZVN0545A MOSFET used as a split load phase inverter to save space.
The B+ will be about +325V 125mA per channel.
The output transformers will be 5k:VC, something salvaged from amps I have lying around (maybe from a Dyna ST70, maybe from an old harman-kardon receiver that used 7355 outputs). LTspice predicts distortion under 0.02% at 1W out into 8 ohms and 22W max output at 1% THD. That's probably overly optimistic, but it does look promising.
I could also use 6JC5 outputs for max 15W output, but I think I'd like to try the 6AV5GTA tubes. I have a bunch of them.
It's slow going. Every time I do any work on the thing, I get tired and have to take a rest. Hopefully I can peck away at it over the summer.
Good luck with your build, and please keep us posted!
I can tell you that in LTspice, this RCA SP10 circuit looks like it performs really well. I understand that simulation and real life are often two different things, but my experience has been that if a circuit simulates really well it will come out working at least reasonably well in real life -- although perhaps not as well as in the sim. Simulation can help predict what circuits are likely to perform well and which are likely to perform less well, but it can't predict exactly what kind of performance you'll get in the end.
I have a steel chassis that was given to me. I'd rather not have a steel chassis, but that's what I have. It was drilled for 9-pin mini sockets for the output tubes (EL84s) and a single 9-pin mini socket for a cathodyne preamp stage/phase splitter for each channel. I've re-punched the output tube holes to fit octal sockets (I'm planning to use 6AV5GTA) and I'll need to punch the chassis for another two 9-pin mini sockets (I'll actually use 7-pin mini tubes, probably 6AU6A or 6CF6) and one octal socket for a 6SN7 voltage amp stage (one triode per channel), with a ZVN0545A MOSFET used as a split load phase inverter to save space.
The B+ will be about +325V 125mA per channel.
The output transformers will be 5k:VC, something salvaged from amps I have lying around (maybe from a Dyna ST70, maybe from an old harman-kardon receiver that used 7355 outputs). LTspice predicts distortion under 0.02% at 1W out into 8 ohms and 22W max output at 1% THD. That's probably overly optimistic, but it does look promising.
I could also use 6JC5 outputs for max 15W output, but I think I'd like to try the 6AV5GTA tubes. I have a bunch of them.
It's slow going. Every time I do any work on the thing, I get tired and have to take a rest. Hopefully I can peck away at it over the summer.
Good luck with your build, and please keep us posted!
The 6AV5 would be an interesting option. I have those and some 12AV5's that I could try in mine. I'm planning to use a higher primary impedance output in keeping with the original that called for a 10k transformer. I have some decent 8k options, Heyboer and O-Netics, some UTC LS-63's which can be wired for 10k and some Stancor A-8072's at 7.6k and pair of Heath/Stancor outputs from UA-2's that are 10k. Not sure which way to go yet for the OPT's. I can punch and drill my top plate and leave the transformer selection to the last minute before the plate has to be painted or coated. Probably go with an 8k option although I've thought about doing some testing to see how they behave open loop since these won't be in the feedback path. Not quite sure how to go about it and what criterion to use for the choice.
12AV5 would be cool too. I have a few of those too.
The original design used 6V6GT outputs, which want a 10k:VC OPT. If you switch to different output tubes, you'll want to match the OPT primary impedance to those particular tubes.
However, since the OPT is outside the feedback loop, and the feedback should lower the output impedance from the output tubes, I think this circuit can work into a wider variety of primary load impedances than a pentode or ultralinear output stage. I think it will react like a triode output stage -- higher Zpri will yield lower THD and better damping but lower max output power, while a lower Zpri will produce higher THD and worse damping but higher max output power.
I figure since 6AV5GTA as triodes work well into a 5k plate-to-plate load, that's what I'll shoot for in this design.
Wow, you have a pair of LS63s? I had a pair of those, and I regret selling them off a few years ago. They have a 6k p-p primary too, so you could try the amp with those and compare 6k to 10k by changing just a couple of connections. I used to use a pair of 300Bs connected to the 6k primary load. It was very clean and had really nice bass. About 13W of output power.
I also have a pair of O-Netics 8k OPTs, which I was planning to use for a push-pull GU50 pentode monster. But I'm reconsidering that.
But right now I need to take my fitness walk, or the doctors will be mad at me...
The original design used 6V6GT outputs, which want a 10k:VC OPT. If you switch to different output tubes, you'll want to match the OPT primary impedance to those particular tubes.
However, since the OPT is outside the feedback loop, and the feedback should lower the output impedance from the output tubes, I think this circuit can work into a wider variety of primary load impedances than a pentode or ultralinear output stage. I think it will react like a triode output stage -- higher Zpri will yield lower THD and better damping but lower max output power, while a lower Zpri will produce higher THD and worse damping but higher max output power.
I figure since 6AV5GTA as triodes work well into a 5k plate-to-plate load, that's what I'll shoot for in this design.
Wow, you have a pair of LS63s? I had a pair of those, and I regret selling them off a few years ago. They have a 6k p-p primary too, so you could try the amp with those and compare 6k to 10k by changing just a couple of connections. I used to use a pair of 300Bs connected to the 6k primary load. It was very clean and had really nice bass. About 13W of output power.
I also have a pair of O-Netics 8k OPTs, which I was planning to use for a push-pull GU50 pentode monster. But I'm reconsidering that.
But right now I need to take my fitness walk, or the doctors will be mad at me...
I'm off to the SP-10 races. Got my top plate cut to size and all the larger holes drilled and punched yesterday. A few small holes to go for socket mounting screws and the PC boards. Now to get onto the PC board stuffing. I drilled it for the Heyboer 8k OPT's that I have on hand. Don't have any experience with them but they are reputed to be decent to good OPT's. If they don't work out somehow I'll substitute something else. I have a few choices. The UTC LS-63's are too big for my layout but there are others. I wanted to keep the chassis to manageable dimensions. I'm at 16"W x 14"D. With a wood wrap base it will be slightly larger.
That sounds great nerdorama!
I've seen a lot of good reviews of both the O-Netics and Heyboer OPTs.
I think I'm settling on this circuit, below. I have all the iron for it, and I have a chassis that's big enough to fit it all. I'll have to strip that chassis (it was a PP 2A3 thing that developed a mysterious problem that I was never able to fix...) and then I'll need to punch a few new holes for tube sockets. Nothing major, and the power supply stuff is almost wired already.
Here's my proposed circuit:
I've seen a lot of good reviews of both the O-Netics and Heyboer OPTs.
I think I'm settling on this circuit, below. I have all the iron for it, and I have a chassis that's big enough to fit it all. I'll have to strip that chassis (it was a PP 2A3 thing that developed a mysterious problem that I was never able to fix...) and then I'll need to punch a few new holes for tube sockets. Nothing major, and the power supply stuff is almost wired already.
Here's my proposed circuit:
- Low B+ (330V) with 150mA per channel power transformers means I can run 6L6GA or 6P3S outputs low and hot, mostly in class A operation. This reduces the 3rd harmonic distortion, leaving 2nd harmonic dominating until nearing clipping. I figure that should 'sound good'.
- The 6L6 outputs dissipate 16.7W at idle, which is well within the limit of 19W or 20W for 6L6GA, 6L6GB or 6P3S (not the -E version). But that's beyond what I would ask of 6V6GTA, and those are expensive now, so I don't want to burn them up unnecessarily.
- Simulation says I should get 18W per channel at 1% THD.
- At 1W out, THD should be really low (0.015%) with 3rd harmonic distortion 10dB below the 2nd harmonic.
- I don't have any 6CB6 so I'm going to use 6CF6 instead. I have some 6AU6A in stock (same pinout), in case the 6CF6s don't work out.
Last edited:
rongon...what LTP are you speaking of?
You could be psychic, as getting rid of the concertina/voltage amp 12AU7 in favor of *SOMETHING in a LTP rigging is one of the things I would try. It is driving a pair of 6AU6's so it is not going to take any heroics. That is however not stopping me from wanting to build it with a 5687... 🙂
Say also, try something out of George's favourite family of tubes: horizontal output/sweeps. 31JS6C ought to do. Type 27GB5 is another that won't break the bank.
Douglas
You could be psychic, as getting rid of the concertina/voltage amp 12AU7 in favor of *SOMETHING in a LTP rigging is one of the things I would try. It is driving a pair of 6AU6's so it is not going to take any heroics. That is however not stopping me from wanting to build it with a 5687... 🙂
Say also, try something out of George's favourite family of tubes: horizontal output/sweeps. 31JS6C ought to do. Type 27GB5 is another that won't break the bank.
Douglas
I'm sorry, I was calling the 6AU6 differential pair a 'long tailed pair' -- which is incorrect. My bad.
I'm still planning to use a 6FQ7 concertina for the input stage and phase splitter.
An LTP first stage would be a great idea. You don't need more gain, so a pair of low-mu triodes should work fine. 5687 at low plate current would probably work well. Maybe 6DJ8?
I decided against an LTP phase splitter first stage because then the harmonic distortion spectrum would have a dominant H3, and also because of the added complication of getting a CCS into the tail. I wanted to try to get a dominant H2, hence the concertina. The distortion profile seems to be dominated by that first stage. Change the 6FQ7 to a 12AU7 and the THD goes up significantly.
My question was whether the 6AU6 diff pair needs grid stoppers. Does it?
I'm still planning to use a 6FQ7 concertina for the input stage and phase splitter.
An LTP first stage would be a great idea. You don't need more gain, so a pair of low-mu triodes should work fine. 5687 at low plate current would probably work well. Maybe 6DJ8?
I decided against an LTP phase splitter first stage because then the harmonic distortion spectrum would have a dominant H3, and also because of the added complication of getting a CCS into the tail. I wanted to try to get a dominant H2, hence the concertina. The distortion profile seems to be dominated by that first stage. Change the 6FQ7 to a 12AU7 and the THD goes up significantly.
My question was whether the 6AU6 diff pair needs grid stoppers. Does it?
There's no hard and fast rule regarding the inclusion of grid stoppers. You could try it without grid stoppers to see if there's any parasitic oscillations. You could include the stoppers as a matter of course since they won't do any harm if not required for damping out parasitic resonances.
My usual practice is to include 1K grid stoppers and 470R screen stoppers for pentode gain stages. so far, no probs. It's the power finals where you really need stoppers since some types (807) love to make snivets when going into plate current cutoff. Adding 1K5 screen stoppers put an end to that.