Reverb Wet/Dry mixing

[hkc]

I'm working on a reverb amp design and I'm wondering how much amplification to give to the reverb return signal before mixing it back in with the dry.

It seems like most reverb amps have a single 12ax7 amplification stage on the recovery side so they're getting something like 70x gain.
Accutronics says that one can expect a 1-5mV signal from the tank so a single gain stage will not be giving us more than about 350mV of wet signal.

On the dry side I'm expecting more like 5V after the first preamp stage so it looks like I will have a wet/dry signal difference of about -24dB.

So my question is, is that enough reverb? Is the -24dB wet signal big enough to make a difference or should I really be adding another gain stage to the recovery. With another gain stage I can get rough equality between wet and dry. Another option might be to cut the gain on the first preamp stage which would reduce the difference but wouldn't really bring the wet and dry into parity.

In my design I have one more high gain triode available that I can either use to boost the reverb or add crunch to the dry side. I expect I shall try both but I was hoping someone could offer an opinion based on experience. So far I'm just going on Spice data.

Hank

[Chuck H]

If you look at how others have done it you'll see that the wet/dry mix isn't usually simply coupled. The dry signal is heavily padded at the mix and then the mixed signals are typically sent to an additional amplification stage to reamplify the mix. Put another way, you do need two gain stages to amplify the signal from a reverb tank if you want a strong reverb component at the end of the preamp. You won't typically need all that the two gain stages have to offer, but you do need more than just one can give.

[Mick Bailey]

My usual fallback is to take a look how Fender does it - maybe take a look the AA769 twin reverb schematic. Depends on what you have for a gain structure in your design and how much reverb you want. A high-gain rock-oriented amp will have a different requirement to a clean 'surf' amp, or the sound from a 6G15 standalone reverb, where the reverb will go to very nearly 100% wet.

[guitician]

I had to use a voltage divider to bring down that 5v signal so it mixed more like 50/50 wet/dry with the 350mv reverb recovery stage.

[Chuck H]

I had to use a voltage divider to bring down that 5v signal so it mixed more like 50/50 wet/dry with the 350mv reverb recovery stage.

Sooo... At 50/50 with the reverb recovery stage (at only 350mV, so two 350mV signals in parallel) the "mixed" signal would certainly need another gain stage to bump it prior to the PI (or, if single ended, the power tube grid).

I'm not questioning you guitarician. Just using your post as spring board

This is my point above. I've seen a few circuits where a novice designer (Nothing wrong with that BTW!) have tried putting the reverb recovery stage directly into the "other" PI input (non inverted side) as a late mix. Usually they report that it works! Don't believe it. Yes it "works" but I'm pretty sure not as well as hoped. A strappin' PI circuit may have a gain of 35 (35 x 350mV = 12.5V). So if that same strappin' PI is swinging, say, 100V, you're pretty far from a 50/50 mix. You'll never get a "surf" tone out of that. I say "why bother" if you're not going to do it up There are a lot of other variations too that people have tried. Ignore them! Look to what manufacturers have used as reverb circuits!!! Nothing new under the sun. Seek examples from those that have made it work already.

JM2C

[hkc]

OK. So I've been looking at schematics trying to understand how Fender did it and what I seem to find is that the reverb mix on Fender amps is very dependent on the volume setting. At low volume you can get a lot more wet into the mix but at high volume the reverb is going to be overwhelmed by the dry signal. They seem to have improved the situation somewhat in later designs like the Blues Junior by using op amps and no transformer in the reverb circuit. The op amp recovery amp has a gain of 111. I just played a blues junior the other day and it definitely didn't seem to have anything close to surf'n reverb.

Chucko's comment that the dry signal was "padded" mystified me a bit. I assume that you mean it was attenuated. Actually it seems that the only attenuation is the channel volume control. The idea of feeding the wet signal into the non-inverting side of a long tailed pair phase inverter is interesting but a little misguided. (I understand, that was the point) The person who did that probably thought that they could simply use the PI as a mixer, trying to get double duty out of one tube. Unfortunately that seems likely to screw up your Class AB operations. If I had the extra double triode to play with I think I'd just use one of the units as a second gain stage for the reverb and put the other into the mixed path for crunch.

But this is totally speculation. The amp is on the bench 3/4 done so one of these days I will be able to plug it in and find out how it really works. ;^)

[Chuck H]

Well it sounds like you have the chops to figure the circuit for yourself!!! My implications weren't meant to discourage though. For whatever you think of the vintage Fender design, it certainly does get surfin' reverb! The hitch is that once the amp is clipping the reverb will be a garbled mess Enter the Blues Junior... A more modern amp will take measures to mix the reverb after any preamp distortion to avoid the garble. They do this because this is the way modern amplifiers are used!!! Whatever the final mix is I can't say (because I haven't looked). But my point is that the old BF amps and the new HR and Blues amps have different goals based on different users.

If you look at a typical BF type reverb you'll see that via the 3.3M or 4.7M resistor and the load of 220k results in a voltage division of roughly 20:1!!! The recovery stage uses a 50k or 100K pot (Which doesn't actually attenuate much. Less than 1.5dB) and then goes through a 470K, sharing the same load, to join with the dry signal. Having only been attenuated at about 2.2:1, or, one tenth as much as the dry signal at the mix circuit output. Then the mixed signal is sent to a re amplifying stage to pump both up for the phase inverter input. So the dry signal is attenuated about /20 and the reverb is just about cut in half. Meaning that the mix stage IS attenuating the dry signal about ten times more than the reverb signal to balance them before re amping.

I hope my reference to the differences between vintage and modern ideologies didn't get lost here. The old designers simply didn't have any idea someone would eve turn the amps up that loud

[Enzo]

They seem to have improved the situation somewhat in later designs like the Blues Junior by using op amps and no transformer in the reverb circuit.

I don't think the nature of the reverb circuit is what makes the difference. Low impedance pans are driven by the transformer, high impedance pans are not. Solid state reverb drives inevitably use the high impedance. little transformers cost money. They can design the reverb to be as strong or as wimpy as they like with most any system. The op amp gain can be whatever we want, and 12AX7 stages can also be designed for a range of gain.

the Blues Junior... op amp recovery amp has a gain of 111.

How did you determine that? If I look at the Blues Jr schematic, I see the signal level leaving the reverb pan as 191mv, while the signal leaving the op amp is only 63mv. That looks like a gain of 1/3 to me.

Further ;looking at the BJ schematic, I see the output of the preamp as 1.36v at TP10. That them goes into a (roughly)1/10 voltage divider at the master control. meaning at 1.36v in, we get about 0.14v max at the master volume control. That is 140mv, which is mixed with the potentially 63mv signal from the reverb.

The input side takes a 27mv signal and runs it up to 388mv to drive the pan, but that doesn't determine the level in the mix.

And note the reverb drive signal is tapped off the wiper of the master volume control, so the drive signal in the BJ is dependent on the volume setting too. But that is desirable, if we left the reverb drive level at whatever max the signal has, then as we turn the channel volume down, the reverb would stay at strong levels, so you'd wind up with reverb ten times louder than the dry at low volume.

http://bmamps.com/Schematics/fender/..._sch_rev_d.pdf

Compare the 65 Twin Reverb
http://bmamps.com/Schematics/fender/..._reverb_sm.pdf

The reverb there takes 1.9mv from the pan and boosts it to 110mv out the reverb recovery tube. Which is mixed with the dry at the grid of V4 pin 7.

Chucko's comment that the dry signal was "padded" mystified me a bit. I assume that you mean it was attenuated. Actually it seems that the only attenuation is the channel volume control.

Look at the Twin drawing. The output of the preamp is at TP6, about 3v. But when it gets to pin 7 of the V4, it is down to 120mv. R31 and R32 form a roughly 1/15 voltage divider. There is the padding/attenuation. They brought the level of the preamp output down to the level of the reverb output and combined them. THEN V4b boosts the signal back up to 4v. And that is why the FX channel has that extra gain stage, V4b.

[Merlinb]

How did you determine that? If I look at the Blues Jr schematic, I see the signal level leaving the reverb pan as 191mv, while the signal leaving the op amp is only 63mv. That looks like a gain of 1/3 to me.

That doesn't add up. A non-inverting opamp can't have a gain of less than unity. There's something funny about those measured voltages. In principle the gain should be 1+ 330k/2k = 166...

[J M Fahey]

Agree.

I see the signal level leaving the reverb pan as 191mv, while the signal leaving the op amp is only 63mv. That looks like a gain of 1/3 to me.

That's most certainly a typo.
No way a magnetic reverb pan can put out that much voltage.
Add the proper dot to turn it into 1.9mV and we can talk.

FWIW 63mV/166=0.38mV at the output transducer which is somewhat on the pessimistic side, I guess they are playing it very safe now.

Big problem with specifying reverb tank levels is that frequency response is *horrible*, and full of huge peaks and dips (much worse than a speaker) so tou might have a +20dB peak at 1000 Hz and a -20dB dip at 1040Hz , no kidding, so measuring ba tank with a steady state signal is a hit or miss job, useful only to check whether it works or not.

In fact, they *should* be fed pink noise, like a speaker and for the same reasons, if you wish band limited to guitar frequencies, for any meaningful number to be achieved.

Did Leo Fender use pink noise?
Don't think he even was into the concept, but most certainly used a guitar frequencies source: a guy playing a Strat or a Tele and Leo adjusting levels by ear until he liked it.
State of the art.

FWIW back in 1969 one of my first succesful commercial products was building this SS reverb circuit (which could also be added to tube amps) straight from Popular Electronics pages, including the PCB design (PCBs made with my Mom's nail enamel, of course, and drilled with a hand cranked drill)
Notice the same gain structure ad Fender's and most others: a low power Class A driver (Op Amps were practically an exotic Lab device way back then) , in this case straight coupled to the drive coil, a high gain recovery stage, and 2 mixing resistors for dry and wet signals:


For the full fascinating Popular Electronics article:
http://www.swtpc.com/mholley/Popular..._1968_pg41.jpg

From way back then, I still remember (must have read that article 1000 times) something like "reverb tank attenuation is around 40dB" or thereabouts.

I added reverb to *tons* of amps, first my friends and then the word spread out, the limiting factor was getting enough tanks, very hard fo find.
Oh well.

FWIW now I make my own , go figure.

[Chuck H]

...the limiting factor was getting enough tanks, very hard fo find...now I make my own...

My hero

[g1]

That doesn't add up. A non-inverting opamp can't have a gain of less than unity. There's something funny about those measured voltages. In principle the gain should be 1+ 330k/2k = 166...

The BluesJr.III has numbers more in line with that.

[Enzo]

OK, I was just looking at the print and not thinking too deeply, sorry.

[Mick Bailey]

The limiting factor was getting enough tanks, very hard fo find......now I make my own...

Michael Faraday made and insulated his own wire. Big deal. He should have made reverb tanks then maybe people would have heard of him and he would have been famous. Oh - hang on....

Anyhow, there should be an ISO unit called the Fahey.

[Enzo]

And a slap-back echo unit called the Yehaf?