AB763 Fender Alternate Reverb Arrangement

[Matador]

The stock 60's Fender Reverb topology has the reverb signal being injected in parallel with the 'dry' signal across a 3M3 || 10pF LPF network that is placed in series with the dry signal. This arrangement works since the dry signal path and the reverb signal path both undergo 360 degrees of phase shift thus can be summed across this network.

It's also pretty noisy: shorting across the 3M3 || 10pF network causes the background hiss to drop considerably (and yes, you lose the reverb), which got me to thinking: if I were building a single channel version of the circuit, the normal and vibrato channel are mixed together at the input to the PI through 220K mix resistors. If there is no dry channel, one just shorts one of the 220K resistors to ground. But what if we were to use it to mix the reverb in with the dry signal?

Schem is here: https://robrobinette.com/images/Guitar/Deluxe_Models/AB763_Deluxe_Reverb_Annotated_Schematic.jpg

One thing I notice is that injecting the reverb at the 220k mix resistor places the reverb signal out of phase with the dry signal, causing cancellation. However since we pass through the reverb transformer, there's an opportunity to add 180 degrees more phase shift by reversing the leads of the reverb transformer, and which point we align back with the polarity of the vibrato channel, and in theory this should work and be far less noisy.

Has anyone tried this? I've attached the 'stock' topology and my 'proposed' topology.

 

[CJ]

There is no phase with the reverb signal , the springs turn it into one big blob.

 

[Matador]

I only ask because I input a 400Hz sine wave, and scoped the phase at various points, and the reverb is definitely out of phase with respect to where the normal and vibrato channels are mixed.

But are you saying in reality, it can be mixed in 'anywhere' along the signal chain where the gain staging makes sense?

 

[CJ]

You might notice a difference if you flip the phase with the transformer on a steady signal of 400 hz, but when you play a guitar signal I don't think it will make much difference, try and see.

Since those springs have a delay time, your guitar signal phase will come out of the pan at different angles depending on what note you hit.

But I get wanting to avoid the 3.3 meg / 10 pf network. Almost all of your signal is going through that 10 of cap as Xc is only 40 k ohms compared to 3 meg.

Dang it I had the best sounding amp, it was an epiphone tube amp chassis with a diy circuit, I wanted to add reverb so j spliced something in there, the problem was that when you turned the reverb knob up, the gain went up, but there was a point there where the overdrive and the verb sounded awesome. So what do I do? Tear it out to try something different and I lose the magic tone and I fic not bother to write a schemo so the tonf is gone forever.

 

[merlin]

If there is no dry channel, one just shorts one of the 220K resistors to ground. But what if we were to use it to mix the reverb in with the dry signal?

You can do that. It's not really different from the stock arrangement except you have an extra tube stage in the dry path before mixing, which is kinda putting the cart before the horse. The reason Fender had to use that noisy 3M3 resistor is that the wet signal is very small compared to the dry, so he had to knock the dry signal level down a lot. If you put the extra gain stage in the wet recovery path instead, you can use smaller mix resistors and hence less noise. Some amps even use the two grids of the phase inverter for mixing wet and dry.

As CJ said, phase is irrelevant. Sinewave testing is misleading because signals appear coincident when they're actually being delayed.

 

[Matador]

The reason Fender had to use that noisy 3M3 resistor is that the wet signal is very small compared to the dry, so he had to knock the dry signal level down a lot.

That's interesting: for the dry signal path, the 3M3 goes into the 220K of the next grid (V4), which is a 94% reduction in signal level (ignoring the 10pF). However the reverb pot leads into a 470K attenuator that works against the same 220K follow-on grid resistance, which is a 68% reduction. Both signals are then brought back up via the gain in V4.

It seems to me one could mix right at V4 with 100K's into the 220K, so long as the gain is reduced from the dry signal *and* the reverb signal, which could be done by splitting the plate resistors. Whether or not it would be measurably less noisy I guess needs experimentation.

 

[merlin]

one could mix right at V4 with 100K's into the 220K, so long as the gain is reduced from the dry signal *and* the reverb signal, which could be done by splitting the plate resistors.

I think you would only want to split the plate resistor of the dry signal, but yes, that ought to be a lower noise solution

 

[CJ]

Did you see the Swart circuit a while back?

 

[Matador]

Did you see the Swart circuit a while back?

This one?

 

[CJ]

yess sir, very nice sounding reverb,

 

[Matador]

I wonder what the purpose is of the 0.022uF cap leading from the reverb level pot into the 100K mix resistor? The pot is already isolated from DC from the coupling cap of the driving stage.

V1B is also missing a DC ground reference: doesn't this make the operating point of the stage fluctuate as V1B ages?

 

[musipol]

Looks to be referenced thru the vol pot, isn't it?

 

[merlin]

I wonder what the purpose is of the 0.022uF cap leading from the reverb level pot into the 100K mix resistor? The pot is already isolated from DC from the coupling cap of the driving stage.

The 0.022uF cap is indeed redundant. Possibly a relic of an earlier circuit iteration.

V1B is also missing a DC ground reference: doesn't this make the operating point of the stage fluctuate as V1B ages?

As Musipol said, ground ref is through the volume pot.

 

[Matador]

Here's what I'm going to try: it's basically the vibrato channel of the AB763 with the trem stripped out of the preamp. The reverb drive is take from the second gain stage plate (called J9/TUBE3B on the schem), and is mixed along with the dry signal into the grid of the third gain stage. The second gain stage has a split plate resistor so I can adjust the dry/wet mix leading into the final preamp stage (although I could also just tailor the values of R15, R16, and R17 to accomplish something similar). In fact, I would probably substitute 100K pots for R15 and R16 and adjust until the dry/wet balance is good when the reverb control is maxed.