Mixing 2 Channels for Reverb

[Dagan8]

Hello Everyone.

I am working on a design for a donor amp that I am gutting. It is basically a AB763 50W amp with some mods and tweaks. I am just working on the schematic and I need some help on the reverb input area. I am trying to wire it so that I will have reverb on both channels. I think I need the 2 X 220K mixing resistors but I'm not sure if that is the best way. Will it create some bad side effects like channel bleading? I really like the original Black Face reverb but I'm not too concerned if I loose a bit of reverb strength since a rarely use it above "3" on the reverb knob. Any ideas? Please keep in mind that this schematic is a work in progress and there is likely mistakes in other areas.

Cheers
Seth

[Chuck H]

Your idea looks OK. One thing that stands out for me is that with the high series resistance at the signal coupling and the relatively lower impedance where the two channels are coupled to feed the reverb there will some filter effect (for better or worse) via each channels coupling cap. You could reduce this effect some with the modification I've shown below.

I would actually couple the two channels differently to avoid the high series resistance. Look at the 6g16 Vibroverb schem. Feeding the reverb driver from the plates of the two channels and feeding the mixer stage from the split plate loads would significantly reduce channel interaction. Further, since there is no trem circuit that loads the signal, you don't actually need any gain from the mixer stage. You could make it a "virtual earth" mixer. This could allow you to reduce series resistance (less HISSSSS) and channel interaction even more.

[pdf64]

Another method is to link the plates of the 2nd stage of each channel, see
AB763 etc Channel Link Mod picture by pdf64 - Photobucket
Pete.

[Dagan8]

Hello Chuck

One thing that stands out for me is that with the high series resistance at the signal coupling and the relatively lower impedance where the two channels are coupled to feed the reverb there will some filter effect (for better or worse) via each channels coupling cap. You could reduce this effect some with the modification I've shown below.

Ok. I think I see what you did here. You added the separate coupling via the extra and relocated 500pF caps. So this is a simple solution but it still could allow some channel interaction. Correct?

I would actually couple the two channels differently to avoid the high series resistance. Look at the 6g16 Vibroverb schem. Feeding the reverb driver from the plates of the two channels and feeding the mixer stage from the split plate loads would significantly reduce channel interaction.

Hmmm. I have never really studied the 6G16 circuit before. I modified my schematic with what I think is the configuration you are revering to and attached it here. Is this what you mean?

Further, since there is no trem circuit that loads the signal, you don't actually need any gain from the mixer stage

In the Fender Custom Vibrolux Reverb schematic I think I see this attempt to reduce the gain of the stage after the mixing. They removed the cathode bypass cap and add some series resistance via a 1M which I assume could still add some hissss but probably less than the original 3.3M. I added this mod to the attached schematic above as well.

You could make it a "virtual earth" mixer. This could allow you to reduce series resistance (less HISSSSS) and channel interaction even more.

I'm assuming that you are referring to a completely different way of mixing the channels and reverb return here?

[Chuck H]

Don't know how well sharing the plate load like that will work. But yeah, that's the general idea. I did some more redrawing. I kept the two channels sperate as per your original schem and included the VE mixer stage. It should work well but it's a more complicated circuit and I've never used it so it may require tweaking (what doesn't.

Pete's idea is also a good one. I thought about that too but wasn't sure of a good way to implement it. Anyway, some of my thinking out loud is drawn below.

[Dagan8]

Hi Chuck

Thank you for taking so much time to help me with this.

Don't know how well sharing the plate load like that will work. But yeah, that's the general idea.

I see your concern here. Although i did just copy the circuit from the Fender Custom Vibrolux Reverb schematic on their website so it should work in theory. Looks like they use a smaller total resistance because they are shared? I could easily just use two separate split plate loads like in your modified drawing though.

I also notice that the Fender Custom Vibrolux Reverb does away with the 220K mixing resistors into the reverb driver like in my original idea and your latest version. Do you see any major drawbacks in removing these?

included the VE mixer stage. It should work well but it's a more complicated circuit

I will have to look at this more and decide if I want to add the complexity. I could always just stick with the low gain stage with the 1M series resistance if I chicken out.


Cheers
Seth

[Chuck H]

With the two plates shared there's obviously no need for channel mixing resistors. And the split plate load does much of the wet/dry balancing for you. Also, it's easy to control gain without the need to pad the signal with a voltage divider. The split plate is the divider and you can change the resistor ratio to suit your preferences for gain. I do think you still need mixing resistors to couple the reverb and dry signal to the summing amp. With the VE mixer these resistors could be comparably small. And the VE would be less crowded if you were only summing the reverb and the one signal from the shared plate. I did use the VE as a summing sage for a reverb once. I liked it. The need thing about the VE is that it's input impedance is so low that there is almost no interaction between the input signals. That's why, if you look at my latest drawing, I changed the 470k mixing resistor prom the reverb pot to 100k. Then after posting I realized that it doesn't even need to be there.

[Dagan8]

Ok. Chuck. Very cool.

Here is the schematic with the shared split plate load and the VE mixer stage together. I think......

Hopefully I interpreted it correctly.

Cheers
Seth

[Chuck H]

Yupper. Ha, ha. I just drew up the same thing (with a couple of value differences). It's starting to look like an original design instead of a hybridized circuit. If you wanted to keep simplifying, another thing you could probably do is share the cathodes for both channels first stages, then both channels second stages and re-share the reverb recovery and summing stage cathodes. Use 820R or 1k resistors. Of course you could leave them seperate too. It'll make tweaking easier later if you want to try different value cathode bypass caps.

I would probably use a 47k to couple the reverb to the summing stage. It'll add reverb depth.

I would change the coupling cap from the shared plate to the summing stage to .047uf. I just don't see the point in .33uf. and besides, the VE will work better if the input rolls off more lows than the output. Otherwise the REALLY low frequencies aren't attenuated.

[Dagan8]

Hi Chuck

Perfect! I think I will leave it as is for a starting point and not share anymore cathode or plates. I will finish adding all of the edits along with finishing the rest of the circuit and post in a new thread for final inspection by all of the wonderfully helpful people on this forum.

Thanks again for all of your help. Time to go out enjoy the sun while it lasts.

Cheers
Seth

[Chuck H]

Time to go out enjoy the sun while it lasts.

Ah, in Canada I see. I'm on Whidbey Is. Wa.

Yeah, yesterday I almost called some authority to tell them it must be the end of the world. A huge orange ball was shining in the sky!!! Then I remembered it was the sun. Been awhile since we saw it here too.

[Chuck H]

OK... A couple of things I noticed about the schem...

The presence control is set up wrong. The cap should be on top of the pot. The point of this incarnation of a presence control is to keep DC off the pot. Now, some may argue that with the cap in series that no DC can pass "through" the pot so it doesn't matter. But, just put the cap ahead of the pot.

There's a filter between the plate and grid of the second triode of the non BF channel. The schem shows a pair of 20M resistors with a .047uf cap in between. 40M is roughly the same conductivity of plywood!?! What is this circuit doing and why is it included? Did you copy this preamp from a known design? I honestly doubt you would be able to tell if that circuit were there or not.

Last... You have the reverb driver cathode resistor value plugged in but no HV indicated. This circuit value will be chosen the same way you choose a power tube cathode resistor. Because that's what it is. In this circuit the reverb driver acts as a power tube to provide watts that drive the reverb tank transducers like a normal power tube would drive a speaker. In this application it's important to recognize the driver tubes limits and also be sure you're driving the reverb efficiently. Point is, if that's an arbitrary value plugged in from another amps schem, it may not be the ideal value for your amp and your reverb driver tube.

[Brigham]

Your working are really very impressive thanks for sharing your work.

[Dagan8]

Hi Chuck

Thanks for looking!

The presence control is set up wrong. The cap should be on top of the pot. The point of this incarnation of a presence control is to keep DC off the pot. Now, some may argue that with the cap in series that no DC can pass "through" the pot so it doesn't matter. But, just put the cap ahead of the pot.

Done! Moved

There's a filter between the plate and grid of the second triode of the non BF channel. The schem shows a pair of 20M resistors with a .047uf cap in between. 40M is roughly the same conductivity of plywood!?! What is this circuit doing and why is it included? Did you copy this preamp from a known design? I honestly doubt you would be able to tell if that circuit were there or not.

I'm really curious too. It is a known D-sign. Looks like a local feedback loop but with very very little feedback. I will take it in and out to see if I actually hear a difference.

Last... You have the reverb driver cathode resistor value plugged in but no HV indicated. This circuit value will be chosen the same way you choose a power tube cathode resistor. Because that's what it is. In this circuit the reverb driver acts as a power tube to provide watts that drive the reverb tank transducers like a normal power tube would drive a speaker. In this application it's important to recognize the driver tubes limits and also be sure you're driving the reverb efficiently. Point is, if that's an arbitrary value plugged in from another amps schem, it may not be the ideal value for your amp and your reverb driver tube.

Yes. I am planning on tweaking the dropping resistors in the power supply to give me the HV I need. I will tweak the cathode resistor it needed.


Chuck. I have drawn up a couple versions showing the different mixing design variations and I am planning on providing some extra eyelets to allow some experimentation. I was looking at the options some more and I came up with this idea attached. It looks way too simple so I am likely defying the laws physics somewhere. It uses the shared split plate load. The 2 reverb sends come from the plates and the mixed channel output comes from the split load and enters the PI after the "Mixing" stage. The reverb output enters before the "Mixing" stage just like the common BF variants. I'm not sure if I will have a feedback issue?

Cheers

[Chuck H]

That looks pretty good. The series reverb amplification is sort of like something Bruce Zinky did with the Prosonic amps. With this incarnation you could jostle the 220k mix resistor values for reverb balance.

I must admit that I'm partial to the VE mixer because it's novel and effective at the same time. But IMHE it is a pain to layout on an eyelet board.

BTW. Good call on including extra eyelets for tweaking. I've been in that place where I wished I had included them but didn't.

I think your NFB loop is very carefully condidered. At first the ratio looked low. But then I saw that you have it coupled from the two ohm secondary. So I'm guessing you considered ratio and tap based on something like the Super Reverb? Looks just fine.

The reverb coupling looked a little confounded in that there is already a 100k load via the pot and then another 220k load added. Then I considered... You don't ever want to rely on a pots wiper for a grid load, and there's not much gain to be had in this circuit above the 220k value. So this seems fine. Since the 470k resistor is post voltage divider it doesn't actually attenuate much, but does have a low pass filter/grid stopper effect. And a grid stopper on the second stage of two cascated triodes seems like a good idea too. So the 470k serves that purpose. If the reverb seems dull you could decrease that resistor value as a starting point in the tweaking.

Your bright switch value is going to sound a little subtle. But I don't think that's a bad thing. Higher values do have that instant gratification effect but are usually too much of a punch in the face to be diversely useful. On BF type circuits I like to hardwire a 47pf "bright" cap and then use a 100pf on the switch. The bright switch a fun circuit for tweaking.

I know Fender used 500pf and 1000pf PI input caps a lot. But there's a reason not to follow that lead in a hybrid design. If your cloning it's fine. The only way to get the same sound as the original amp is to make your clone the same. But for a new design there may be some things to consider. Like the fact that you don't know what the amp sounds like yet. The 1000pf value has a little rolloff at low E. Not a big deal, unless... If you plan on overdriving the amp and the PI will be clipping, the 1000pf value might show it's failing. When Fender made the BF amps they didn't expect people to intentionally drive them into distortion, so the 1000pf value gave them what they wanted for clean reproduction. But when a LTP PI clips the input impedance drops. That means the knee frequency of the input cap rises. Since the 1000pf value already cuts into the audible range a little this effect is more pronounced as the PI starts to clip. If you plan to drive this amp into distortion I would use a higher value PI input coupling cap value. If the amp seems boomy or farty you can start trimming the bypass caps in the preamp. Triimming at the PI input cap isn't the best course for overdriven amps. You don't want to pass low frequencies you can't use though. I've gone as small as 2200pf, but I usually use a value of 4700pf or .01uf. Just something to consider.