Princeton Reverb circuit refinements

[tristanc]

The Princeton Reverb is a classic circuit. It’s fine as is.

But, my understanding is that these were built to as-cheap-as-possible-yet-passible-tone requirements and, thus, BoMs. They struck lucky and their popularity shows that. However, there are mods people do to make them better, and components added in the reissue to make them safer / more suitable for the modern era.

Inspired by Merlin's new book, given cost is no barrier to a DIY clone of a Princeton Reverb, what circuit refinements would you incorporate?

I don’t care about aesthetics / period correctness on the inside. I do care about safety / good engineering practice and, ultimately, the classic PR sound.

Given the constraint of using a ‘traditional’ turret board from a kit, I was thinking:

• MF resistors instead of CC
• bleeder resistor added
• reduce post-PI coupling caps to 0.022uF or 0.047uF
• add 470k stopper to PI
• adding stoppers to grid and screens on the 6V6s
• using a red LED for the tremolo oscillator bias
• add protection resistor on the speaker jack
• add variable bias (and balance?) control
• do my best to ground things appropriately, even if it means running short lengths of wire to connect the various star groupings, and connecting only to chassis at the input jack.

The PRRI schematic has a bunch of these things done already (diodes here and there for protection, fuses, e.g.). I don’t want to go overboard as that would complicate the pre-made turret board.

Have I gone overboard? Not far enough? Destroyed the mojo?

[tubeswell]

IMHO, the best mod you can do for a PR that won't affect the mojo of the original blackface amp vibe is to keep the entire circuit as-is, but swap out the filter cap can for separate filter caps at each HT rail supply node and change to a Merlin-style daisy-chained galactic ground return system. This will get rid of some/most of the hum in the noise floor.

Also, if your PT is such that the amp will run at higher voltages (i.e 430-450ish), you could also swap out the PR OT for a DR (6k6 25W) OT and go to a 12" speaker - this opens up the sound and ensures slightly improved power output. And of course, you will probably want to run JJ6V6S if your voltages are going to be that high.

Also recommend a couple of fail-safe protections, namely:
1) 1.5kV protection diodes in series with the rectifier tube plates
2) a 1k load resistor across the reverb transformer secondary and a 2-3kV 100-2000pF cap across the reverb transformer primary to help prevent the RT from failure.

[stratomaster]

Reverb HT moved to the unused node with the 18k feeding it reduced to 12k and the reverb driver resistor reduced to 1.5k will give you a nearly silent noise floor from the reverb circuit.

I like to use a large cathode cap (250μF) on the recovery for good measure like the standalone reverb units did.

I also like a 3.3μF cap on the 2nd stage cathode instead of messing with the coupling caps.

[waldner]

Rob Robinette has a nicely laid out version of the Princeton with his suggested mods here:


https://robrobinette.com/AA1164_Princeton_Reverb.htm#Suggested_Mods

[tristanc]

Rob Robinette has a nicely laid out version of the Princeton with his suggested mods here

Thanks - that's what started me on the journey. Also comparing the 'original' scheme with the reissue. There's a fine balance between copying it verbatim in a new build and making some alterations whereby you end up with a completely different amp.

This is my 2026 amp build challenge - taking my time and, hopefully, using up some of the components I have on hand. I like the idea of making a few small changes (a cap here and there, re-routing the grounding e.g.) and doing it within the existing turret board(s). ie, what's the biggest improvement you can do with the fewest components.

[burchyk]

I have the reverb driver power feed moved to the unused node (what stratomaster has also suggested) and can confirm it did drastically reduce hum.

One more from the Merlin's book that I liked is reducing V1a grid stoppers. Mine is effectively 10k (down from 34k?) on high input, which reduced hiss when the volume is up.

[tristanc]

Thanks - that's good to know. I'll also try the "ultrapath" thing Merlin talks about in his new book.

Attached is where I am currently. It's a bit difficult to see, but components marked are altered or additions. I've modified the power nodes and grounding - whether this will destroy the mojo... 

And good point on the grid stopper. I'll likely get rid of the 2nd input (I've never used it when there) and re-purpose the hole for a 'boosted' input (a FET stage in front of the 1st stage). That means I can used a 10k.

[stratomaster]

Thanks - that's good to know. I'll also try the "ultrapath" thing Merlin talks about in his new book.

Attached is where I am currently. It's a bit difficult to see, but components marked are altered or additions. I've modified the power nodes and grounding - whether this will destroy the mojo... 

And good point on the grid stopper. I'll likely get rid of the 2nd input (I've never used it when there) and re-purpose the hole for a 'boosted' input (a FET stage in front of the 1st stage). That means I can used a 10k.

You'll likely find that the ripple is so low after moving the HT node for the reverb transformer over that the ultrapath is redundant with minimal to non-existent benefits.  There was no audible reduction in hum nor visible change on a scope with the ultrapath in or out with my PR.

On the reverb recovery a large cathode bypass cap (100+μF) is beneficial in reducing H-K hum, a log pot is useful for dialing in subtle reverb while maintaining access to Surf City, USA.

Bonus: If you study the same section of Merlin's book that covers the ultrapath you'll come across a frequency response chart of the spring reverb.  Use a plate resistor bypass cap on the recovery stage to aggressively roll off above the upper range.  This kills hiss while maintaining the full response of the reverb stage.  I think I used a 470pF, but I'd have to go back and look.

[tristanc]

You'll likely find that the ripple is so low after moving the HT node for the reverb transformer over that the ultrapath is redundant with minimal to non-existent benefits.  There was no audible reduction in hum nor visible change on a scope with the ultrapath in or out with my PR.

Thanks - good to hear that you’ve tried both.

On the reverb recovery a large cathode bypass cap (100+μF) is beneficial in reducing H-K hum, a log pot is useful for dialing in subtle reverb while maintaining access to Surf City, USA.

I’ve been wondering why LED biasing wouldn’t be appropriate for that stage?
And yes, a 100k log pot is on its way - surprisingly difficult to source.

Bonus: If you study the same section of Merlin's book that covers the ultrapath you'll come across a frequency response chart of the spring reverb.  Use a plate resistor bypass cap on the recovery stage to aggressively roll off above the upper range.  This kills hiss while maintaining the full response of the reverb stage.  I think I used a 470pF, but I'd have to go back and look.

Thanks for this - I see he mentions a 100pF grid to ground to kill oscillations, but also refers to hiss just before this. I don’t see any harm rolling off the top end here given its way above where you’d be wanting reverberations?

[pdf64]

...

Attached is where I am currently. It's a bit difficult to see, but components marked are altered or additions. I've modified the power nodes and grounding - whether this will destroy the mojo... 
...

Nevermind mojo, the bias supply won't work without any negative voltage; R37 needs moving to the AC side of D3 

[tristanc]

Nevermind mojo, the bias supply won't work without any negative voltage; R37 needs moving to the AC side of D3 

Thanks! Good catch!

So it seems re-doing the power nodes and grounding is likely to make the biggest impact. Smaller things like LED bias are, well, small and nice-to-haves.

Oh, one question I have (being really picky) is where the reverb transformer secondary should be grounded. I've assumed it's acting like a coupling cap so the ground should go to the following (recovery) stage node.

The fun challenge for me is making the changes within the confines of the existing turret boards / chassis. I think I'll move much of the tone stack to the pots, and maybe use some tag strips next to the valve sockets (using existing mounting screw holes) to free up space. I'd like to avoid drilling holes for turrets or in the chassis if I can avoid it.

[stratomaster]

I’ve been wondering why LED biasing wouldn’t be appropriate for that stage?

I haven't put pen to paper, but I think you'd end up in a different spot. The LED results in a fixed cathode voltage regardless of current through the triode, whereas the cathode voltage "falls out" from the cathode current and resistor value. I think the former is useful for non-audio stages but may alter the behavior of the stage noticably on an audio stage.

More importantly, however, is I think the shunting of h-K leakage hum relies on the reactance of the bypass capacitor being significantly lower than the plate-cathode impedance.  That is, the path to ground through the capacitor is more attractive to the mains frequency cathode input is more attractive than the one to the plate.  This demands the presence of a capacitor.

So while you can get the two biasing arrangements to look similar in terms of load lines, the specific behavior we're looking to exploit needs a capacitor in place. If I'm wrong on this point, I'd love a correction. 

Thanks for this - I see he mentions a 100pF grid to ground to kill oscillations, but also refers to hiss just before this. I don’t see any harm rolling off the top end here given its way above where you’d be wanting reverberations?

There was a thread on TDPRI that I can't find at the moment where they discussed that a capacitor bypassing the recovery stage grid leak doesn't behave like a simple low pass.  Instead it looks a lot more like the frequency response plot of a guitar pickup with a tone control circuit due to the presence of the inductance of the reverb tank output transducer.  Conventional wisdom is that this cap shunts hiss and high frequency to ground, but in that thread simulations showed it behaved as a treble peaking circuit before the roll off.  It's for this reason that I prefer to do the treble shaping with a plate bypass cap.  You can even add a cap across the outer lugs of the Reverb pot for a steeper roll off of the hissy frequencies in conjunction with the plate bypass cap--though I haven't tried this yet. The reverb circuit noise is plenty low on a PR with just a large bypass cap and the cleaner HT node.

[tristanc]

I haven't put pen to paper, but I think you'd end up in a different spot. The LED results in a fixed cathode voltage regardless of current through the triode, whereas the cathode voltage "falls out" from the cathode current and resistor value. I think the former is useful for non-audio stages but may alter the behavior of the stage noticably on an audio stage.

Thanks! I've learned something.

There was a thread on TDPRI that I can't find at the moment where they discussed that a capacitor bypassing the recovery stage grid leak doesn't behave like a simple low pass.  Instead it looks a lot more like the frequency response plot of a guitar pickup with a tone control circuit due to the presence of the inductance of the reverb tank output transducer.

And again - learning something new. Thanks.

[tristanc]

I'm passing over the schematic now stripping out additions that may over complicate. So far:

- removing the ultrapath cap given I'm re-doing the power nodes & grounding
- abandoned OT protection diodes (don't think I'd have room, and have never installed them...)

Where I could do with some input: should I bother with a balance pot for the power tubes?

It's a cheap part which means I don't need to pay loads for 'matched' tubes. Albeit only likely to happen maybe twice in the lifetime of this amp. Though it should be an easy addition to the turret board (famous last words).

I've put all my workings here https://github.com/tristancollins/PrincetonReverbUpdate. There's a branch called "mods" with these circuit modifications: https://github.com/tristancollins/PrincetonReverbUpdate/tree/mods.

Latest schematic here: https://github.com/tristancollins/PrincetonReverbUpdate/blob/mods/schematic/PrincetonReverbUpdate.pdf

I'll do my best to borrow Rob Robinette's layout DIYLC files and update them for this schematic. Maybe even a spice model. We'll see.

[stratomaster]

R45 is in the wrong spot. It needs to go directly to the grid. The way you have it now is a voltage divider with the 1M.

I'd also recommend wiring the bias supply closer to what is recommended by Merlin in his site.  With 25μF as you've got it, it's under filtered enough to be audible.  Instead of wiring the pot at a variable resistor, wire it as a voltage divider and take the bias off the wiper with a 100k safety resistor and a 2nd cap to ground off the wiper or the bias input lug on the Intensity pot.

[tristanc]

R45 is in the wrong spot. It needs to go directly to the grid. The way you have it now is a voltage divider with the 1M.

Thanks - another good catch!

I'd also recommend wiring the bias supply closer to what is recommended by Merlin in his site.  With 25μF as you've got it, it's under filtered enough to be audible.  Instead of wiring the pot at a variable resistor, wire it as a voltage divider and take the bias off the wiper with a 100k safety resistor and a 2nd cap to ground off the wiper or the bias input lug on the Intensity pot.

Yes, I usually follow his bias supply recommendations - the bias board (original and in the kit) has limited turrets as is. Perhaps this one will need some drilling. Or maybe a new board.

Having put the idea in my head - the extra step to actually just lay out a PCB that fits the mounting holes wouldn't be much more effort....

[stratomaster]

Yes, I usually follow his bias supply recommendations - the bias board (original and in the kit) has limited turrets as is. Perhaps this one will need some drilling. Or maybe a new board.

No need. Just a little outside the box thinking. Hang the 2nd cap off the input to the Intensity pot.

[tristanc]

So after many hours trying to clever and shoehorning the circuit changes on to the existing turret board I gave up and laid out a PCB:

https://github.com/tristancollins/PrincetonReverbUpdate/tree/fab

They should arrive next week.

I have gone a bit overboard and used multiple ground plane zones to emulate multiple star grounding points. Just tried to keep things logical and compact - and all on one board.

Attached are an updated schematic, a layout (borrowing Rob Robinette's source file) and a quick render of the board. This should be fairly quick to assemble. All the sources are on GitHub in KiCAD 10. I'll be updating things as the build progresses (I'm expecting something to go wrong with the board - I did it very quickly) so at the end should be a verified PCB should others wish to use it.

[ac427v]

After spending many many hours troubleshooting hum on my build, I replaced the reverb recovery cathode bypass cap with a 220u. C9 on your schematic. Dead silent! (Thanks Stratomaster). Now I do that mod in all reverb circuits. The part is dirt cheap. Just add it to your order "in case"    ;-)

[stratomaster]

You left off the screen node filter cap in your schematic. You'll need that.

Edit:
On closer inspection I see you're adding several board mounted filter caps and splitting the load with a can cap.  You do have screen and reservoir filtering, just in an unusual way.

You'll want to run the numbers again on the power rating (and voltage rating for that matter) on your bleeder.  It's dissipating nearly 2 watts. Replace it with either a 3W or a 330k 1W.  Check the voltage rating of any resistor you plan to use on HT. You'd be surprised how many commonly used ones aren't actually up to the task.

[tristanc]

Thanks - funnily enough I’ve been going through ratings just now. Most power rail ones are 3W, and the kit I’ve used uses mostly 1W elsewhere.

But worth doing a check and I’ll also update the schematic with ratings.

And yes, I’m making use of the cap can as a ‘mega cap’ as the returns of A, B and A’ (power valve, screens and LFO) all return to its shared cathode anyway. Rather than replace it.

[tristanc]

After spending many many hours troubleshooting hum on my build, I replaced the reverb recovery cathode bypass cap with a 220u. C9 on your schematic. Dead silent! (Thanks Stratomaster). Now I do that mod in all reverb circuits. The part is dirt cheap. Just add it to your order "in case"    ;-)

I’ve a 100u in my stash of random parts ready and waiting!

I think there will be a few components tacked in for the 1st power on, to get a rough picture of the voltages. I was targeting the values on the original Fender schematic, but I’m also conscious the reissue (which people seem to like) varies a bit from those. As I have it drawn, I end up with a 1st stage B+ of 275 (E). The PI and 3rd stage + reverb recover at 290 (D) and the reverb driver at 340.

[tristanc]

It's coming together nicely. Populated the PCB in a little over an hour. Next will be doing the flying leads from PCB -> off board components. I've mostly tightened up the chassis parts. This will likely be a "once it's in, it's in" build...

[tristanc]

It came together nicely in the end. A few hiccups here and there mostly due to impatience, but it's working and sounding pretty good.

Last minute changes:

• removed reverb TX protections - didn't want to spend the time to get tag strips installed.
• removed the FET boost - added way too much noise, but could have been a wiring issue. Will revisit later.
• removed OT protection & didn't wire in extension jack - will add a smaller resistor at some point - the one I have is too big to fit nicely.
• didn't use shielded runs from input -> 1st stage grid or volume pot to 2nd stage grid

Non-obvious (to me) issues:

• my grounding scheme means the reverb tank needed modifying - the input side of the tank was left floating. All I had to do was solder the little jumper pads provided for this purpose. I just didn't realise without the help of Michael at Modulus Amplification... Obvious in hindsight.
• I can still hear some 'thumping' at higher trem settings, and that's with the diode added to the intensity pot. I'll add an audio example soon.
• there is quite a bit of hiss
• there's quite a bit of mains hum (there as soon as I flick the switch, not coming through the speaker, more mechanical I think?)


I've no other PR example to compare against, only my own amp builds and they have been very quiet with regards to hum and hiss - and they tend to be high gain.

[edit] Actually, it's not that bad at all. With fresh ears today, and a less critical state-of-mind, it's fine.

The one thing that does stick out is at high volume settings the onset of what I presume is blocking distortion on lower notes. I reduced the coupling caps and added the stopper to the PI, but there's still an ugly buzz behind the note.

Voltages:

HT: 450
Screens: 440
LFO: 440
Reverb driver: 346
Reverb recovery, PI, Stg 3: 289
Stages 1&2: 262

Biased to 40mA (41.7V at the grids), so a little cold. Will up this to 45mA to get closer to 70%.

[tristanc]

So I've used the amp for a jam session in a pub. It worked well and sounded great.

I did 2 things prior:

- biased to ~70%
- replaced the wire runs to the 1st and 2nd stages with shielded cable (I was being lazy 1st time round)

The latter definitely reduced noise - I can't tell if it's on now. In the pub, with the volume on 10, it was still silent. I'm happy with that. Only the tiniest amount of hiss with the reverb up full too.

I'd consider the PCB tested and validated. If anyone wants one of the remaining boards, just let me know.

[dogburn]

Great build - looks very professional. I've been building a modified AA1164 as well, and have done a number of the same mods, though I'm doing a simplified version with 1-tube reverb and no tremolo (very close to what Lectroid recently built). I definitely picked up some useful info from this thread.

[johnfromcyrene]

When I moved the reverb to the unused node I got a problem with the tremolo. It wont run sometimes and then will randomly start up again. Not sure why. In the original location I didn't have the problem. Anyone else experience this?

[sluckey]

When I moved the reverb to the unused node I got a problem with the tremolo. It wont run sometimes and then will randomly start up again. Not sure why.

Tell us exactly how you moved the reverb to the unused node. Sounds like you may have also moved the tremolo too. The tremolo circuit needs to stay on node B.

[johnfromcyrene]

I left the tremolo on node B. I moved the wire only for the reverb to the new node. I thought maybe it was the tube so I switched it and it seemed to work ok for a while but now its back to randomly going on and off. When its on it usually stays on though.

[sluckey]

Replace the tremolo tube cathode RC with a red LED as shown on the latest Selmer Zodiac schematic I posted. The tremolo signal will get much stronger and prevent dropouts.

[johnfromcyrene]

thats the 3.3k and 25uf connected to pin 3?

[johnfromcyrene]

Is there a way to fix oscillation in the princeton reverb without putting in grid stop resistors on the power tubes. I tried it and it worked but I like the amp better without them. Perhaps a snubber cap like they did on the bassman. not sure if its the same.

[johnfromcyrene]

Replace the tremolo tube cathode RC with a red LED as shown on the latest Selmer Zodiac schematic I posted. The tremolo signal will get much stronger and prevent dropouts.

I had upgraded the power and output transformer to the deluxe in my princeton. Would that make the tremolo not work right?

[pdf64]

Is there a way to fix oscillation in the princeton reverb without putting in grid stop resistors on the power tubes. I tried it and it worked but I like the amp better without them. Perhaps a snubber cap like they did on the bassman. not sure if its the same.

Did you identify where the positive feedback loop started, eg which circuit stages affected the oscillation, what control settings affected it?