12AU7 -> small PT as SE OT -> spring reverb?

[lonote]

Try this —

https://web.archive.org/web/20180921032 ... rb_driver/

Fully agree this circuit provides exemplary reverb. The only downside is sourcing suitable NOS tubes. Get a sleeve of five of whatever you choose so spares are available twenty years from now.

Thanks for that, I always forget about that resource.

Those tubes look reasonably priced at the moment.

[Bergheim]

I don't have any pentodes at hand at the minute, but plotting a few loadlines, it seems the 12au7 (which I do have in stock) performs about the same current-wise as an EF80 with equal supply voltages and anode resistor. One third of the output impedance compared to the pentode, so a bit more high end roll of. Might actually not be a bad thing though, IMO the Lamington was quite bright, especially at higher settings. Only one way to find out

[lonote]

+1 on the transformerless design. I copied one from John Polstra's Channel Road Amplification page utilizing a 6GH8A as driver/recovery for use in my 807 amp. To my ears this is the best sounding reverb I've ever heard. Notes are not swamped or muddied by reverb but rather the reverb follows the notes and decay. Sounds a lot more like a good room than typical spring reverb and I used the Mod 3-spring medium decay short tank 8EB2C1B. This circuit is not nearly as gainy as a trad transformer type design, but it's also less prone to noise. Highly recommended...

Fully agree this circuit provides exemplary reverb. The only downside is sourcing suitable NOS tubes. Get a sleeve of five of whatever you choose so spares are available twenty years from now.

With these recommendations, I am thinking I will try this design. Did you both just use a typical bypassed triode circuit for the recovery stage?

Asking as I have always found the recovery stage in the Fender-style circuits to be the nosiest part of the reverb circuit. Wondering if there were any specific tweaks to include?

[didit]

I used the triode paired within, in conventionally familiar form so yea. Details are lost, at least for the moment. It was breadboard level. When I get my workshop back in order I might find the scrap remains. Plan basing it as the reverb sub-circuit for my flavour of a Dumble-esque Princeton/Deluxe combo.

Wondering if that noise was actually inherent to the signal from the transducer vs an artifact of the recovery amp.

..

[Bergheim]

The output signal from the reverb tank is tiny, not very much higher than the noise floor, so noise is kinda part of the nature of the beast. Good lead dress and shielded signal cable all the way from the tank to the recovery stage input grid will help.
For the recovery stage itself, maybe try parallel triodes for better S/N ratio than the regular triode.

[jabguit]

With these recommendations, I am thinking I will try this design. Did you both just use a typical bypassed triode circuit for the recovery stage?

Asking as I have always found the recovery stage in the Fender-style circuits to be the nosiest part of the reverb circuit. Wondering if there were any specific tweaks to include?

I used the triode section of the 6GH8A for recovery. 10K Ra, 560Rk bypassed with 4.7u. Noise is less than typical Fender circuit, but gain is less so that would explain the lower noise.

cheers,

[Helmholtz]

... or a pentode for better S/N ratio than the regular triode.

Acc. to Merlin pentodes tend to have a higher noise figure than triodes due to additional current partition noise between plate and screen.

[Bergheim]

... or a pentode for better S/N ratio than the regular triode.

Acc. to Merlin pentodes tend to have a higher noise figure than triodes due to additional current partition noise between plate and screen.

Whaat, I figured the much higher signal gain outweighed the higher noise level? Anyway, removed the pentode suggestion from my earlier post just in case.

[Helmholtz]

Whaat, I figured the much higher signal gain outweighed the higher noise level? Anyway, removed the pentode suggestion from my earlier post just in case.

Comparable noise figures of active components are specified as equivalent input noise (EIN), so actual gain doesn't change S/N.

[Bergheim]

It's been a little while, but here it is. Please have a look and let me know what you think

Bergheim Rødampen 2025-04-21.png

I left the 12AU7 reverb driver cathode unbypassed to increase the output impedance to 8.6k (up from 5.1k bypassed), functioning better as a constant current source. Not that big of a difference sonically, as 8.6k is still kinda low for a 600 ohm tank, but more importantly the unbypassed cathode resistor increased input headroom just enough to avoid clipping the grid.
In regards to current drive, I found that driving the reverb tank with the specified 3.1mA (0.0058W) RMS simply isn't enough to enter surf territory. No slapback effect, and actually a pretty dry character to the reverberation in general. Also, it leaves a very weak output signal to recover from the return side, wrecking the S/N ratio and greatly amplifying the oscillation risk. Followingly, I biased the 12AU7 driver to 8.7mA, and now it swings up to 14mA (0.12W) RMS into 600 ohms @ 1kHz. No visible signs of clipping or transducer saturation on the oscilliscope, and the reverb sounds lush and clear as ever. The THD in the driver stage is off the charts, but spring reverb is so messy that it's impossible to notice.

The recovery circuit is two cascaded 12AX7 stages with a 50% resistive attenuator in between for a theoretical gain of ~1900, mixed 50/50 with the dry signal into the other half of the 12AU7. Enough reverb gain to push the power amp with the level maxed, which I think is pretty cool I actually had to tame the 12AU7 gain stage quite a bit to keep things together, so the 15k unbypassed cathode resistor you see is no error. Even with the 15k in, I felt the amp was too hot with humbuckers, so I set up the "Stupidly Wonderful Tone Control 3" from Yet Another Tonestack Calculator (https://tonestack.yuriturov.com/), with values modded to taste. With the control at 6, this circuit drops -6dB which fitted perfectly.

Will upload audio when I get around to record some.

[nuke]

Here's a circuit snippet from the Sears Silvertone 1484 and 1485 reverb circuit.

The reverb tank is piezo-electric, so there's no output transformer.

The tank is driven by a 6FQ7/6CG7 which has curves quite similar to 12AU7.

It uses two sections of the 6CG7 as a floating paraphase inverter. This supplies as much as 90vpp across the piezo. The left most section of 12AX7 supplies some gain into the drive, the second section is the recovery amplifier. Supply voltage to the 6CG7 is about 340v. The pot, unlike Fender, adjusts the drive into the tank, not the return gain.

This circuit could quite easily be tailored to a small push-pull transformer to drive an inductive reverb pan pretty easily with just some resistor changes and adapted to your 12AU7 if you like. A general purpose P-P output transformer for table radio would do the job. I'd shoot for 12k-15k plate to plate impedance on the transformer. Hammond makes several. Should do the trick quite nicely.

1484-reverb.jpg

[Bergheim]

Tanks for the input, Nuke.
Something along those lines was my initial idea, but in the end I went for a capacitor driven solution instead, as you can see from the discussion above.


A little more tweaking was done on the reverb circuit (I've updated the schematic above):

- The reverb driver has it's plate resistor connected to B+3, not B+2. Drawing error.

- After installing the reverb pan in the head enclosure, I got an "ultrasonic" (in the sense that the speaker couldn't reproduce it) oscillation at the reverb output with the reverb level above 2 o' clock. I was a little lazy so right off the bat I went for a 68k grid stopper and 180pF cap from grid to ground on the first recovery triode. I might have gotten away with a lower value resistor and/or cap to retain a little more high end, but it cured the oscillation and I'm pleased with the tone anyway so I'll leave it be for now.

- I concluded that the overall output level of the reverb was a bit over the top and somewhat uncontrollable at times. Certain notes seemed to resonate with the springs, creating a harmonic feedback that wouldn't stop until turning down the reverb level. Removed the cathode bypass cap and changed the 2.2k cathode resistor to 1.5k (to compensate a little for the removed cap) on the second reverb recovery stage. Lowered the maximum reverb output to a loud yet controllable level.

- Lastly, I wanted more drive into the reverb pan for a little more punch and surf. Swapped the 15k for a 12k plate resistor on the reverb driver, still with the 620R cathode resistor. A tiny bit less high end in the reverb tone due to the lower driver output impedance (7.5k down from 8.6k).

Some audio clips. Cheap Louder Strat with old strings. Fingers are noticeably rusty also
Close mic is an SM58 with the ball end removed, pointed at the outer edge of the dust cap. Room mic is a t.bone SC-400. Files are edited in Audacity with no post processing other than panning and adjusting overall levels.

shadows fbi.wav

Volume @ 2, Dwell @ 10 and Reverb Level @ 7

Volume 50% Dwell 100% Reverb level 50%.wav

Overdrive riffing, Volume @ 5, Dwell @ 10 and Reverb Level @ 5.

Volume 50% Dwell 100% Reverb level 100%.wav

Overdrive riffing Volume @ 5, Dwell @ 10 and Reverb Level @ 10.

[nuke]

Which tank did you wind up using? The schematic shows a 4AB31C with 8-ohm input.

[lonote]

First, thanks for updating with results & revised drawings. I find lots of dead-end threads frustrating and it is nice to know how things turned out.

I think the clips sound really good, a result I would be pretty happy with.

I would encourage you to fool around with the send/return settings & play the dwell against the output & explore those changes. I have added a dwell to my last few amps & you can get any number of usable settings between the two. So far I have used pretty much stock Fender circuits, so these alternates are pretty intriguing.

Your drawing lists the tank # which denotes an 8 Ohm input. Is that what you used in the end? (first post mentioned a 150 Ohm tank) EDIT: Nuke beat me to it..

[Bergheim]

Which tank did you wind up using? The schematic shows a 4AB31C with 8-ohm input.

Thanks! Corrected - I copied the wrong part number. It is a 4EB2C1B, 600 ohms input and 2.25k output.