Princeton Reverb circuit refinements

[stratomaster]

Yes. I run a 1M input resistor and a 1.2k bias resistor. Its a much smoother overall sound to my ear, but I don't have an A/B comparison.

[shooter]

a 'side kink' that moves up and down the wave with volume setting

those can be "transients", an AC signal riding on your fundamental Frequency
does your scope have a FFT/spectrometer mode?
IF so;
look for a spike that comes n goes, usually it's a harmonic of the fundamental you're using for testing

[johnfromcyrene]

Yes. I run a 1M input resistor and a 1.2k bias resistor. Its a much smoother overall sound to my ear, but I don't have an A/B comparison.

do you have a pi grid stop resistor?

[tristanc]

Can you temporarily solder a 3.3k resistor in parallel with the 1.5k cathode resistor on your cathodyne?

There's a thread on TDPRI where the role of this resistor was investigated thoroughly. The conclusion was that the 5E3 value of 1k gave a much better PI balance than the 1.5k used on the 1164.

Thanks - I think I've just found that thread. In the spirit of trying to alter the bias point, I quickly swapped in a 12AT7. Messes up the tremolo but the distortion / fuzz is still there. I'll have a think about what that could mean. Next time I've the chassis out I'll clip in the 3.3k and report back.

a 'side kink' that moves up and down the wave with volume setting

those can be "transients", an AC signal riding on your fundamental Frequency
does your scope have a FFT/spectrometer mode?
IF so; look for a spike that comes n goes, usually it's a harmonic of the fundamental you're using for testing

The FFTs are in the PDF along with some harmonic values. I've pulled the graph out (attached).

I also spotted another thread on here from a few years ago. Still wading through it, but it was suggested issues with the OT could explain things like I'm hearing. I may have a spare OT so temporarily hooking that up may be another option to try.

[shooter]

looks like there's a low freq oscillation riding on your fundamental or something is breaking down with signal load.  since it's affected by volume, it's probably creeping in before the PA section.

[tristanc]

looks like there's a low freq oscillation riding on your fundamental or something is breaking down with signal load.  since it's affected by volume, it's probably creeping in before the PA section.

Interesting - I'll check the cap can, wiring around it and do a bit of chopsticking. Could the oscillation be cancelled out by the push-pull output until it starts to break up, when the feedback collapses and the rogue signal appears?

I'm using the standard JJ cap, and would hope 40uF would be enough filtering. But another thing I could do would be to clip in another big cap to see what happens.

Thanks

[tristanc]

Some progress. I pulled out the chassis to get a 4 channel scope on it. In doing so I didn't bother connecting the reverb tank. "Hmmm, that's strange, why's the trace changed from yesterday..."

Then I realised that probing the reverb driver grid(s) could make the waveform change. A lot of playing about made me realise without the reverb tank connected the distorted tone was great. Plug it in and I get this high 'hash' behind the notes. I also could see some ringing on the reverb send.

Great - a line of inquiry. So I added some capacitance from the grid to ground. That seemed to help a bit. Doing some googling suggested adding some stoppers - so I put a 1k from the PCB to the 1st grid, then bridged to the 2nd grid with a 470ohm.

That 'seemed' to partially resolve it. However, putting everything back in the cabinet and listening just now the overtone is still there... Darn it.

The traces captured suggested I've had an impact but perhaps not enough? Channel 1 is input, 2 is speaker output (resistive load), 3 is reverb out, 4 is reverb return. State 2 is just as the ringing appears and clipping about to be seen. 3 is clipping, 4 is amp up full. A very different picture to the previous traces, so something has changed.

This post was useful: https://www.tdpri.com/threads/silverface-princeton-reverb-reverb-related-distortion.1171963/

He seemed to sort it by moving the 10pF cap a little bit.

[stratomaster]

Recent Fender amps use a 10k grid stopper at the socket of the reverb driver.  You can try this.

I've also seen 560pF caps from plate to cathode on certain reverb driver tubes. 

An overlooked place for tweaking the hash out of Fender reverb is the recovery triode plate resistor.  A fairly aggressive 470pF cap can be used without losing much of the useful reverb information due to the mid-focused and bandwidth limited nature of the circuit. 

You can also temporarily install a dwell pot to see if attenuating the input to the reverb driver kills the hash.  If so, then warming the bias or slightly raising the plate voltage (or a little of both) may be in order.

[tristanc]

Recent Fender amps use a 10k grid stopper at the socket of the reverb driver.  You can try this.

I've also seen 560pF caps from plate to cathode on certain reverb driver tubes. 

An overlooked place for tweaking the hash out of Fender reverb is the recovery triode plate resistor.  A fairly aggressive 470pF cap can be used without losing much of the useful reverb information due to the mid-focused and bandwidth limited nature of the circuit. 

You can also temporarily install a dwell pot to see if attenuating the input to the reverb driver kills the hash.  If so, then warming the bias or slightly raising the plate voltage (or a little of both) may be in order.

Thanks - this is really useful and gives me a few things to try.

With fresh ears this morning I gave another listen. Still there as before. However, unplugging the tank (something I hadn't done before yesterday) removes the annoying tone almost entirely. So it gives me a sole area to focus on.

Annoyingly, I had planned on putting a 100pF 2kV cap across the reverb transformer primary (I assume the equivalent of plate to cathode?) and a 1k across the secondary following Merlin's book. But got lazy when the layout would get messy and left them off... So:

1) I'll try plate to cathode with the cap I have.
2) add in the 1k at the reverb send

The driver is running at ~343V B+, with 6.2V at the cathode - so 4.2mA going through both triodes. A quick look at the tube curves shows this is pretty cold (and/or a weak JJ tube from the looks of it). Maybe too cold?

3) clip in more capacitance across the recovery plate resistor
4) clip in another resistor across the 1.5k Rk? Get it down to 1k perhaps?
5) Increase the stopper(s) to 10k (and the 'bridge' stopper too, if appropriate)

Will report back.

[EDIT] thinking about it and looking at the later 019320 schematic - could this all be due to the driver being biased too cold? c.v. the ~ 3.2V on the later version, 6 is close to cut off. If I'm feeding it a hot signal that could be causing something to 'leak' back in to the clean signal line? Hence hearing the rattiness when the reverb is connected but turned to a minimum.

[stratomaster]

How hot of a signal? If your RMS AC voltage at the driver tube grid is ≈70% of the cathode voltage you could have grid clamping and that would affect the dry signal. 

The part that has me scratching my head is that disconnecting the tank eliminates the problem. This would imply it's a problem that is fed through the tank and mixed back into the signal through the recovery stage OR that the effect of eliminating the load of the tank changes the overall plate load in a favorable way to reduce grid clamping.  I'm leaning towards the latter.

[tristanc]

How hot of a signal?

I'll try and take a reading tomorrow.

The part that has me scratching my head is that disconnecting the tank eliminates the problem. This would imply it's a problem that is fed through the tank and mixed back into the signal through the recovery stage OR that the effect of eliminating the load of the tank changes the overall plate load in a favorable way to reduce grid clamping.  I'm leaning towards the latter.

Agreed - that is odd. Unless it's parasitic coupling to a close-by wire? Or something silly like having the reverb transformer oriented the 'wrong' way round? Or wires reversed?

I'm wondering what order I do the experiments in. I've limited time tomorrow to do some play testing, figuring I can always scope things properly later on when I have more time to investigate and tidy up any bodges.

1) clip in another 1.5k across the 1.5k Rk to get closer to the 680r value I see on other schematics
2) clip in plate to cathode 100p (or higher - 560p)
3) clip in the 1k at the reverb send (protection - likely no difference to this case unless it's impacting plate loading when unplugged. Could be red herring.)
4) clip in more capacitance across the recovery plate resistor
5) Increase the stopper(s) to 10k (and the 'bridge' stopper too, if appropriate. A bigger job - one for later)

I could go one by one until it's, hopefully, resolved, or it could be a sum of a few of them. Who knows...

[stratomaster]

I think the most useful test isn't listed. Add a Dwell control temporarily to see if you can induce/cure the effect by attenuating the input to the driver.

Also, I don't understand why you have a "bridge" grid stopper.

[tristanc]

Good shout - I’d need my iron out for that. Perhaps I can do the opposite - add capacitance to let more signal in and see if the issue gets worse?

The 2 stoppers thing was from a google - there being 2 potential gain loops that could pick up interference, so belt and braces approach was to add a 2nd resistor. Overkill maybe, but also removed the length of wire from one grid to another. I can remove on the tidy up.

[tristanc]

I'm about to give up... Just to make sure I'm not chasing a figment, attached is a sound sample. Should I just settle for it?

Amp at 7/10, bit of reverb & trem, high output tele bridge pickup. A low E plucked with guitar volume up full, then down a bit to clean up, then back up again. Then some high notes with IMD. Some power chords. Recorded via Boss Waza TAE. Sound in the room is a bit more muffled and the rattiness more prominent at living room conversational volume.

Today I tried:
- a 12au7 in the reverb driver slot,
- changed the bias of the 12at7 (two 1.5k in //, 5.1V on cathode, 320V on plate),
- grounded out the reverb entirely,
- removed the tube,
- added in a cap to the reverb transformer primary and 1k to the 2ndary.

No difference. Which makes me think the reverb may not be the source.

Oh, I'm hitting the driver with 11.8Vrms.

I also tried various caps in places (plate resistors, e.g.), increasing existing cap values, and basically experimented seeing if I could filter off the tone I'm hearing. No joy.

I took the opportunity to solder in a 10k grid resistor to the reverb driver and the 3rd stage grid. Latest schematic attached.

Should I just call it done and play it?

[Adrien]

Hi guys, been following this thread with interest, a lot of great info here, thanks to everyone who contributed!  I have recently completed a similar project to update my PR.  I originally built it with the Hoffman board and layout probably 10 years ago. I had some issues with a horrible blocking distortion from the reverb driver at higher volumes along with a fair bit of hum.  I also wanted to tidy up some of my old sloppy wiring and less-than-ideal grounding, so decided to do some updates as well with some of the mods in this thread. 

The Hoffman layout was a bit easier to modify in terms of power supply and grounding, I added a couple extra power nodes with axial caps along the board, moved the reverb driver power to node C and split the grounding as discussed here among other things.  Was a bit tricky in spots but managed to do it just using existing turrets. 

End result is great, MUCH better noise floor overall, the reverb is silent, and the only hum is mechanical from the PT.  As-built schematic is attached, it ended up very similar to what you have Tristan. This also fixed the horrible distortion I originally had which was much worse than what's in your recording, it sounds like what was described in the TDPRI thread you linked (https://www.tdpri.com/threads/silverface-princeton-reverb-reverb-related-distortion.1171963/).

That said, after all these mods I still have a buzzy distortion when cranked, the exact same as your most recent sound sample. The general consensus I've seen is that PRs are not great when cranked, and truly fixing the issue kinda turns it into a different amp...so I think I'll just leave it.  Happy to play it clean and crank my 6v6 Plexi instead  .

Just my 2 cents.  Big thanks to you for documenting your project so thoroughly, it has been a big help to me and I'm sure many others 

[dogburn]

tristanc - I don't think you need to give up yet. It might help if you did an audio file where you play it loud with the umpleasant distortion and then the same with the reverb tank unplugged, which you said removed it. Then we could really be able to understand what you are talking about. Also, play some chords, and tell us what guitar you are using and what pickup. I recently built a PR with 1-tube reverb and no trem - when I put that back together I will test it some more and compare to your sound files - when first testing that amp, the distortion wasn't very pleasing either, but it may have improved with some of the tweaks I did.

And on another note, I built a Princeton 6G2 a couple of years ago (my first tube amp build), and I didn't like the distortion on that either, but apparently I had biased it too low (calculating based on 12W for a 6v6GT instead of 14W), and when I upped it some, it sounded a lot better. Cold bias can produce unpleasing distortion for sure, so you might want to tweak yours a bit higher to test out.

[Adrien]

And on another note, I built a Princeton 6G2 a couple of years ago (my first tube amp build), and I didn't like the distortion on that either, but apparently I had biased it too low (calculating based on 12W for a 6v6GT instead of 14W), and when I upped it some, it sounded a lot better. Cold bias can produce unpleasing distortion for sure, so you might want to tweak yours a bit higher to test out.

This is a good point, I'll try this too.  One downside I can foresee with a hotter bias is it might weaken the trem somewhat.

[stratomaster]

Oh, I'm hitting the driver with 11.8Vrms.

This is wild. This number should be safely below the cathode voltage if you want to keep the driver stage happy.

The fact that you are getting an issue without the tube in place though tells me that you're hearing 2 different problems, though. And your most recent observations seem to contradict your prior reports.  I think some is the cathodyne rattyness (a fixed bias arrangement or the "Stokes mod" may be worth a try) and the rest is grid clamping from slamming the driver stage.