ODS 124 - revisited for the 21C

[Stephen1966]

Here's a schematic of the Dumbleator I'm building as a side project. It isn't entirely finished but I'm down to the final component choices and a couple of little tweaks... one thing not shown, I will be using a 7025 instead of a 12AX7. This is really a topic for a different day but I thought I would share it because it includes a CF in the first stage.

Dumbleator 1.0.pdf

Good health!

Stephen

[Stephen1966]

Okay, following up on the last few posts, I think I may have misunderstood Tony's words of caution as referring to the CF...

Correct me if I'm wrong Tony, but you were talking about a regular gain stage when you were speaking of "too large a source impedance"? A problem with a solution in the form of the CF. You did say "a cathode follower prevents [this]", I should have read it a little more carefully. It doesn't suggest to me now that the CF in a single triode mixer would be in any way insufficient. And, as rootz pointed out, I have to have a CF capable of handling the output of V1b but given that CFs operate at much greater voltages compared to a regular gain stage, I don't think that would be a problem. The problem if there were to be one, might be arcing and oscillation between the grid and cathode but I'm confident a small grid stopper and break out resistor will prevent that.

Rootz, I am going to run up a schematic for your circuit in '6L6 crs post reverb masters CF #002 x #124' as this looks very promising. I take it 'U50' in that sim is the Master. I ran a calculation on the CF and with a cathode (series) resistance of 101.5k I got the unloaded voltage gain of 0.984 (-0.14dB) and the output impedance of 615 Ohmns which seem very respectable values - just slightly less than unity gain and very little output impedance. Doubling the cathode resistor has barely any effect on the latter but whether we could expect the same frequency response is another matter. Again, when I read what you wrote more closely, it seems that you also, were not suggesting that the CF would be insufficient to the task but rather if the benefits would be worth the trouble it takes of implementing them.

My apologies to you both... snatching five minutes here and there is not good for my concentration.

To answer that last point, yes. I think it is worth taking the trouble even if the benefits are minor compared to a circuit without a CF. Because the more I think about what I've seen and learned so far, the more I think MrD was not setting out to add tone, rather he wanted to release it. His careful construction methods, the obsessive attention to detail, seems all about removing the signal artifacts a poorly designed and executed circuit injects. Very low SNR, hi-fi like sound, rock solid mechanical connections, quality component choices. The amp shouldn't interfere with the pure musical expression, only amplify it... a million times In my career as a music engraver I have encountered this before in pages of poorly cast music. Many times sadly. It is basic signal theory but a good amp, like a good page of printed music, should be transparent. It should not interfere with the pure musical expression; it should not get in the way.

So a CF? Yes, absolutely. I'm very excited about this idea.

Rootz, if I may ask, what do you see as the benefits of the cathodyne splitter in your second simulation? Forgive me if I am missing something that should be obvious but in my uneducated understanding of it, the signal would have two origins in the input side of the circuit - that from the preamp source and then a second from the mixer plate, each out of phase with each other. Does this create some kind of positive feedback, like in a Trainwreck where the result is a gain factor, higher than the tube actually produces. Or, are we looking at more headroom? Some kind of DC coupling? What do you think about the possibility of runaway oscillation?

Once again, many many thanks, for your patience at the very least. It feels very close now.

Stephen

[rootz]

I’m on the phone right now, so a quick reply and I hope to not miss your points and questions.

First the benefits of the cf based tweed mixer Tony talks about. That point is true. Draw up a Thevenin circuit of the output of the mixer triodes and you can see how much of a load it represents and also see why you lose some signal in the mixing stage. Dumble’s solution with two CF’s in the 002 is likely to be cleaner and colour the signal less.

The frequency response stays pretty much intact with 150k mixer resistors. But there is a trade off between loading down the mixer triodes with low value mixer resistors with better noise figures and frequency response and slightly less THD with larger mixer resistors. There is no free lunch... I think Tony’s point is to use a CF at both of the mixer triode’s outputs. Yes, that would give you a 4 triode reverb! A side note for the frequency response of the mixers. I think the Miller effect from the voltage divider in the dry signal had a much more prone influence in that respect. Giving the mixer triodes some local negative feedback will lower the output impedance a bit, making it easier to drive the load the mixer resistors and output impedance of the other triode present. You could even make the dry mixer triode unity gain just by setting the feedback resistors right. That would be much the same in some ways as a CF. I went with a divider and moderate feedback because that was how it was done in the Wonderland. And I really, really liked the cleans on that one, but missed the OD. So I have a wonderland with OD essentially, not a 124 or 94 with reverb.

But really, I never put so much thought in my designs without building something. I still have no clue how my design with just a CF on the dry side will sound. It will work, it simulates well, it seems to preserve the frequency response, it has loads of headroom so it won’t clip, there are no large series resistors so keeps noise from those out. But you’d have to experiment to be sure it sounds the way you like it.

The cathodyne was a brain wave to give you zero frequency depended loading on v1b, have unity gain and provide enough headroom to not clip when v1b puts out maximum signal. Iirc that maximum signal is somewhere around 200V peak-peak for v1b. So in any way the cathodyne has to carefully biased to meet the headroom needed. In short: I was trying to make the basic idea a bit better, not to add some new functions like positive feedback.

Long post again. If I were you, I’d start with drawing up a chassis that gives you the possibilities to experiment with more than one idea. Don’t expect to get it right the first time. I took me more than one chassis to get an amp I really like and bond with. And still I plan on doing a #60 just out of curiosity.

[talbany]

Steven

So I have a wonderland with OD essentially, not a 124 or 94 with reverb.

So in this case Dumble used the C.F (tweed) in a clean only amp and not one with OD..IMO It's much more difficult (or trickier) to implement a parallel type reverb into a cascading style preamp with overdrive and still preserve the balance between the two clean and O.D signals as well as try to retain the (theoretical) frequency response from the original ODS design was intended during the mixing process, yet? at the same time Dumble didn't use the tweed in his ODS's with reverb? Perhaps for some reason he didn't like the sound or there was something about a plate driven mixer that he preferred

But really, I never put so much thought in my designs without building something. I still have no clue how my design with just a CF on the dry side will sound. It will work, it simulates well, it seems to preserve the frequency response, it has loads of headroom so it won’t clip, there are no large series resistors so keeps noise from those out. But you’d have to experiment to be sure it sounds the way you like it.

IMO this is how it's done..Put your Ideas down run the sim's and if everything operates correctly and looks good on paper,then the only real way to see how it sounds. The only real way to know for sure is to build the thing and tweak if necessary

This is the main advantage to going with a proven/tested design! you know it already works and you also have the benefit of knowing it's already been put under some level of scrutiny by some members ears and to them have some level of comfort of knowing it's probably not going to sound all that bad..So if rootz says his design sounds good and meets your criteria then there you go a tested Idea!

If I were you, I’d start with drawing up a chassis that gives you the possibilities to experiment with more than one idea. Don’t expect to get it right the first time. I took me more than one chassis to get an amp I really like and bond with. And still I plan on doing a #60 just out of curiosity.

And at some point I would like to do a 102 with reverb using the tweed who knows exactly how it will sound.
Sorry if I made things more confusing? Just thought I would point out a few of the "Pros" of the mixer since Steven asked about it a few posts back!

Tony

[Stephen1966]

I’m on the phone right now, so a quick reply and I hope to not miss your points and questions.

First the benefits of the cf based tweed mixer Tony talks about. That point is true. Draw up a Thevenin circuit of the output of the mixer triodes and you can see how much of a load it represents and also see why you lose some signal in the mixing stage. Dumble’s solution with two CF’s in the 002 is likely to be cleaner and colour the signal less.

The frequency response stays pretty much intact with 150k mixer resistors. But there is a trade off between loading down the mixer triodes with low value mixer resistors with better noise figures and frequency response and slightly less THD with larger mixer resistors. There is no free lunch... I think Tony’s point is to use a CF at both of the mixer triode’s outputs. Yes, that would give you a 4 triode reverb! A side note for the frequency response of the mixers. I think the Miller effect from the voltage divider in the dry signal had a much more prone influence in that respect. Giving the mixer triodes some local negative feedback will lower the output impedance a bit, making it easier to drive the load the mixer resistors and output impedance of the other triode present. You could even make the dry mixer triode unity gain just by setting the feedback resistors right. That would be much the same in some ways as a CF. I went with a divider and moderate feedback because that was how it was done in the Wonderland. And I really, really liked the cleans on that one, but missed the OD. So I have a wonderland with OD essentially, not a 124 or 94 with reverb.

But really, I never put so much thought in my designs without building something. I still have no clue how my design with just a CF on the dry side will sound. It will work, it simulates well, it seems to preserve the frequency response, it has loads of headroom so it won’t clip, there are no large series resistors so keeps noise from those out. But you’d have to experiment to be sure it sounds the way you like it.

The cathodyne was a brain wave to give you zero frequency depended loading on v1b, have unity gain and provide enough headroom to not clip when v1b puts out maximum signal. Iirc that maximum signal is somewhere around 200V peak-peak for v1b. So in any way the cathodyne has to carefully biased to meet the headroom needed. In short: I was trying to make the basic idea a bit better, not to add some new functions like positive feedback.

Long post again. If I were you, I’d start with drawing up a chassis that gives you the possibilities to experiment with more than one idea. Don’t expect to get it right the first time. I took me more than one chassis to get an amp I really like and bond with. And still I plan on doing a #60 just out of curiosity.

Rootz, this is a real gift. Thank You. You are a true gentleman and a scholar.

[rootz]

I agree it is difficult to design a frequency independent reverb, in any amp. That is why I put the effort in making sure the loading by the mixer stages and reverb input is much like a master volume. The basic idea: present the same (kind of) load at the output of the od relay and it will sound close to a non reverb amp.

And Tony, you are not confusing at all. The goal is to implement a reverb that does not get in the way of the expected ods sound. Every idea that eventually gets us closer to accomplish that is helpful.

So onto a new idea with cf mixer stages. I ditched the pre reverb driver and used that triode at the end of the reverb signal line to form a cf. See what it starts to look like? Only the reverb send pot taxes the high end a tiny bit. At most around 1dB and only around 5 to 20 kHz if my math is correct. The input of the dry mixer will not present any load other that the 1 Meg resistor preceding it.

If you don’t want the tiny bit of loading on the high end, you could split the parallel reverb driver into a pre driver with 100% feedback (unity gain) like the #60 and a single triode driver. I’d rather take the extra current capability of a parallel dual triode as a driver though. You do not need extra voltage gain from a pre driver as you’ve got more than enough signal coming out of the preamp already.

Now about building several ideas into one chassis. You could do that by splitting things up I think. Make room for 6 preamp tubes. Just a single row like you’d normally make. Design a board with the reverb section between the preamp and the PI components. Or maybe you could build an ods board regular style and add a separate board to the right of that for all the reverb circuitry? Never tried that, but that would make testing with different designs a relative breeze. The reverb uses two high voltage nodes at most. I’d suggest to use a dual can cap for that. The reverb driver supply can be used with or without a cap. I opted for none in the wonderland, but use one in my current amps. This cap can be placed on the PSU board while still keeping the size of that board under control. 22uF/500V caps are small these days.

[Stephen1966]

Steven

So I have a wonderland with OD essentially, not a 124 or 94 with reverb.

So in this case Dumble used the C.F (tweed) in a clean only amp and not one with OD..IMO It's much more difficult (or trickier) to implement a parallel type reverb into a cascading style preamp with overdrive and still preserve the balance between the two clean and O.D signals as well as try to retain the (theoretical) frequency response from the original ODS design was intended during the mixing process, yet? at the same time Dumble didn't use the tweed in his ODS's with reverb? Perhaps for some reason he didn't like the sound or there was something about a plate driven mixer that he preferred

But really, I never put so much thought in my designs without building something. I still have no clue how my design with just a CF on the dry side will sound. It will work, it simulates well, it seems to preserve the frequency response, it has loads of headroom so it won’t clip, there are no large series resistors so keeps noise from those out. But you’d have to experiment to be sure it sounds the way you like it.

IMO this is how it's done..Put your Ideas down run the sim's and if everything operates correctly and looks good on paper,then the only real way to see how it sounds. The only real way to know for sure is to build the thing and tweak if necessary

This is the main advantage to going with a proven/tested design! you know it already works and you also have the benefit of knowing it's already been put under some level of scrutiny by some members ears and to them have some level of comfort of knowing it's probably not going to sound all that bad..So if rootz says his design sounds good and meets your criteria then there you go a tested Idea!

If I were you, I’d start with drawing up a chassis that gives you the possibilities to experiment with more than one idea. Don’t expect to get it right the first time. I took me more than one chassis to get an amp I really like and bond with. And still I plan on doing a #60 just out of curiosity.

And at some point I would like to do a 102 with reverb using the tweed who knows exactly how it will sound.
Sorry if I made things more confusing? Just thought I would point out a few of the "Pros" of the mixer since Steven asked about it a few posts back!

Tony

Tony, thank you for clarifying... I can see now why the field of real amp builders and designers like you is so small. The learning curve is mind-blowing (and tube blowing apparently) I'm going to follow rootz's advice. I'll get my head around the maths it's probably the easiest part, but you are both right - sooner or later, I'm going to need to fire up that iron. It's easy enough to follow a schematic, but knowing how to fix it when things don't go according to plan is the trickiest part. Is this why all the old amp builders have beards and translucent vampire skin, I wonder

[Stephen1966]

...
Now about building several ideas into one chassis. You could do that by splitting things up I think. Make room for 6 preamp tubes. Just a single row like you’d normally make. Design a board with the reverb section between the preamp and the PI components. Or maybe you could build an ods board regular style and add a separate board to the right of that for all the reverb circuitry? Never tried that, but that would make testing with different designs a relative breeze. The reverb uses two high voltage nodes at most. I’d suggest to use a dual can cap for that. The reverb driver supply can be used with or without a cap. I opted for none in the wonderland, but use one in my current amps. This cap can be placed on the PSU board while still keeping the size of that board under control. 22uF/500V caps are small these days.

Great idea! A prototyping chassis. I can do that first. I've already made a start... there seems plenty of room between the preamp and PI sections.

[rootz]

Thank you Stephen!

I forgot to mention the new schematic has an extra switch. That is to switch between reverb input from the od relay or from the output of the dry mixer follower. So to find out if there is a better way to separate the reverb from the preamp and completely eliminate any loading on the high end by the reverb input. It works both ways. I haven’t really looked into feeding back part of the mixed signal through the 220k mixer resistor to the reverb input. Personally I’d ditch the new switch and just tie the reverb input to the output of the od relay. I wouldn’t be bothered with 1dB of high end loss at the maximum. One step extra off axis from your speaker probably has much more effect on the high end.

[rootz]

Stephen, I’d ditch the relay to bypass the reverb. I think that idea is not very useful in a live situation. When you want to switch of the reverb, just shunt the reverb recovery input to grond. That is easier and might save you a lot of long wiring with capacitive shielded cable. Better for your high end.

What I would do however:
1) test with reverb wired in or not to compare the effects of high end loading. Just hard wired.
2) make a setup for a dual reverb return pot that switches with the od relay. So a return pot for clean and one for od. How? Replace the 100k return pot with two 250k ones. Connect both to the 3n3 cap. Wire the wipers to a SPDT relay. Add a 1 Meg grid leak at the relay output. Wire the relay in parallel to the od relay, so it switches at the same time. Voila, you now have dual return knobs and can set the reverb level for clean and od independently. A very useful feature imho, which I will incorporate in my next amps.

[Stephen1966]

Stephen, I’d ditch the relay to bypass the reverb. I think that idea is not very useful in a live situation. When you want to switch of the reverb, just shunt the reverb recovery input to grond. That is easier and might save you a lot of long wiring with capacitive shielded cable. Better for your high end.

What I would do however:
1) test with reverb wired in or not to compare the effects of high end loading. Just hard wired.
2) make a setup for a dual reverb return pot that switches with the od relay. So a return pot for clean and one for od. How? Replace the 100k return pot with two 250k ones. Connect both to the 3n3 cap. Wire the wipers to a SPDT relay. Add a 1 Meg grid leak at the relay output. Wire the relay in parallel to the od relay, so it switches at the same time. Voila, you now have dual return knobs and can set the reverb level for clean and od independently. A very useful feature imho, which I will incorporate in my next amps.

Thanks, this seems a great practical solution. My one concern with previous designs was of introducing even the smallest of filters between the OD/Clean relay and the Master (a 220k resistor and 22n cap in your latest design) when the reverb is out of the loop. But you make a good point, the high end cost is negligible and I would add, it probably wouldn't be noticed in a live situation, if that net is always there. This is where going between a hard wired in and out of the reverb would be indicative of the true cost. This solution really brings a second level control to the OD, right? I'm having trouble visualizing the switching, though. If the 250k pots are wired in parallel off the 3n3 cap I can't see how a SPDT switch makes any difference. If the wipers are attached to a single pin of relay it's going to be an on/off switch. But... Is there a connection to the wet OD signal before the 100k OD level pot to one of these 250k pots? I would still like to switch the reverb independently of the OD, I don't want to be forced to activate the reverb when I only want the OD. So a separate, independent SPDT switch shunting the reverb recovery input to ground is still required? Your use of the singular indicated only the reverb 100k pot but perhaps you meant the OD 100k return (Level) pot as well?

Help

[Stephen1966]

This is the only way it makes sense to me...

Dual pot sketch (1).pdf

[Stephen1966]

Sorry, I'm calling the OD RATIO pot the Level pot - I meant RATIO (at the output of the OD = B100k)

[rootz]

Yeah, that is in the direction of what I meant, but not quite correct yet.

Follow my corrected schematic as attached and add the pots and relay like in my drawing. Maybe I’ll have some time left tonight to update my schematics in a more clear way. For now this will have to do.

[Stephen1966]

Here is the plan of the chassis with 3 extra tube sockets for the reverb, all inline.

Dumble chassis - 4+6 tubes.pdf

This is an off the shelf large chassis from Tube Town. The document is layered so the SSS layout can be found by opening all the layers up. By default it only shows the markup information.

I've evenly spaced the preamp sockets whereas the SSS skews them for tag strips and extra components around the sockets. I would probably use tag strips as well while prototyping the different reverb circuits. The number of components in the reverb is quite light so the board (85mm x 70mm) might be made smaller, allowing the PI to be shifted right, away from the power section. The clean/overdrive board and the PI board are about the size of a regular main board, side by side. These also, might be reduced but are probably fine as they are. Because of the transformer mountings, the standoffs for the boards would need to be reconsidered but shouldn't present a problem once the transformer locations are determined. The rectangle at the bottom left corner is the space required of the filtered and fused IEC module I will be using. We might squeeze the PT a little further left but I would like to leave a little room for the rectifier and relay filter boards attached to the sides.

A work in progress... but a step in the right direction, I think.

Good health!

Stephen