Troubleshooting reverb circuit - solved

[Bombacaototal]

Thanks Erwin, seems like we are getting closer to tackle this. What do you mean by bad design filtering in switching power supply? Is it the layout of the power supply?

[ijedouglas]

So you think it could be the supply cap feeding the reverb? I actually use a cap can (high grade CE manufacturing) which feeds the tonestack, send/return and mixer circuits, and another f&t cap in the supply board for the reverb driver. Or do you mean the mains supply cap? I use a quad of the 100uF blue Sprague 350V

It may be either but in my case the caps were good and it appeared the interference was not via the power cord but rather some type of RF interference. I initially tried foil shielding inside the cab which worked great but later removed it and I just move the router when it bugs me

What happens if you plug the amp into the power conditioner (same as computer/desk etc)?

[Bombacaototal]

With the amp on the same power conditioner, there is barely an issue, surely much more mild and less noticeable. From what I have tried it did not cure completely, but I think it is because I have been focusing on listening to the issues.

[erwin_ve]

Thanks Erwin, seems like we are getting closer to tackle this. What do you mean by bad design filtering in switching power supply? Is it the layout of the power supply?

A copy paste: Disadvantages include greater complexity, the generation of high-amplitude, high-frequency energy that the low-pass filter must block to avoid electromagnetic interference (EMI), a ripple voltage at the switching frequency and the harmonic frequencies thereof.

Very low cost SMPSs may couple electrical switching noise back onto the mains power line, causing interference with devices connected to the same phase, such as A/V equipment. Non-power-factor-corrected SMPSs also cause harmonic distortion.

[Bombacaototal]

Thanks Erwin, so if I understand correctly, the power conditioner is coupling noise into the amp, and in theory, therefore, there is no issue with the amp, per se? Weird thing is that I did use a second amp in the exact same configuration with no issues

[sluckey]

Take the amp to a friend's house. Might be revealing.

[erwin_ve]

Thanks Erwin, so if I understand correctly, the power conditioner is coupling noise into the amp, and in theory, therefore, there is no issue with the amp, per se? Weird thing is that I did use a second amp in the exact same configuration with no issues

I agree with sluckey.
A lot of devices these days got a switch mode power supply: Laptop, phone charger etc. If one of them lacks the proper filtering for EMI/ switching noise, your amp could pick it up. Maybe not every amp but some amps are more prone to this(positioning, layout).

[Bombacaototal]

Thanks Erwin and Steve for the Input. Makes sense..and therefore given, from previous experiments, the main areas where the issue propagates were the RCA return jack and the reverb recovery circuit (even with the RCA send disconnected), do you believe there is any relationship between those parts of the circuit, or maybe their layout, or their grounds, and how this is coupling with the switch mode power supply.

Or in other words, if I want to try to mitigate this (assuming it is possible in any way or form) should I focus of maybe changing the layout, or the power supply, or the ground, or maybe too hard to tell, in which case I will carry on moving things and changing things, until I find a solution, assuming there is a solution.

I wanted to lend this amp to a friend who will take it on the road, and hence will have to deal with all types of venue supplies, and I think it may be unwise, unless I can mitigate this is some shape or form

[Richard1001]

Looking at the schematic at page 1, are you sure it is correct?
The reverb return signal is amplified by two 12ax7 stages before reaching the mixer, and amplified again in the mixer tube before reaching the master volume.

Also, i think the .005 cap on v5a should connect directly to the 470k resistor and master volume. I cant understand why one would draw a separate resistor between the output of v5a and the master volume.

[Stephen1966]

Asides from the several contributions made since my last post, and with the ground scheme sorted, I don't see any reason why you shouldn't send this amp out into the big wide world

Robben Ford talked about having to deal with dirty power supply on tour and there's a clip of SRV in the studio kicking up a whole barrow load of noise. I'm sure there's more. This is a Dumble after all! One of the most dynamic and touch sensitive amps on the planet. Load a power conditioner into the touring rig and you are good to go.

Details aside, your problems are not necessarily with the amp, but with the environment it sits in. As long as we check all the bases here you will have done your bit. The world can take care of itself

[Bombacaototal]

Many thanks for the encouragement Stephen! Yeah, sounds good, I cannot thank everyone enough for the help and contribution. It is good to make sure all bases are covered and at least my amps is sound, even if slightly more prone to coupling interference from switch mode power supply, and a power conditioner should be a must in anyone's rig. Fingers crossed it will go smoothly with him for the test drive

[Stephen1966]

Many thanks for the encouragement Stephen! Yeah, sounds good, I cannot thank everyone enough for the help and contribution. It is good to make sure all bases are covered and at least my amps is sound, even if slightly more prone to coupling interference from switch mode power supply, and a power conditioner should be a must in anyone's rig. Fingers crossed it will go smoothly with him for the test drive

No worries, mate!

[Stephen1966]

Okay, back on the reverb circuit. The RETURN pot is still giving me hum, however, I made some changes and the situation seems to be somewhat improved.

First, I took the reverb transformer ground lead and ran it over to the ground point (GND 3) shared with its filter caps in the PSU. Not a noticeable difference, if any at all, but it is now isolated away from the clean stages at the opposite end of the chassis so I am leaving it where it is now.

Second, I reduced the mixing resistors on either end. I had a 470k on the input to the RVB and a 330k on the output. Originally, they were both 470k. I put 220k in both sides. With the cap on the Return pot disconnected, I got a lot of hum still, but it intensified the reverb to public toilet level. This added intensity meant that I could dial the RETURN back, get a decent intensity of the effect and reduce the hum.

Third, by replacing the .015 cap on the RETURN with a .01 I lost a little bit of intensity but it took the hum way down and now, at stage level, it's playable, and playing, it's hardly noticeable but I still can't crank up the RETURN all the way or the hum becomes intrusive again.

If I turn down the RETURN all the way, the amp is dead quiet with only a slight noise at an otherwise very loud level for Master and Volume. The hum from the reverb is not affected by the volume setting, but naturally, turning up the Master amplifies it. I get the hum at the same level whether the volume is at zero, halfway or higher. Which means I can tame it further by driving the front end harder but dialing back on the Master.

There is really only one more thing to try now, which is to isolate the RCA jacks or rather, the SEND jack because this is a 9AB3C1B with the C indicating the input is insulated and the output is grounded. Testing for continuity between the tank jacks and its casing, I found the case had continuity with its input jack shielding whereas its output jack was isolated from the casing. So if it's necessary to ground the input side to serve as a shield, isolating the jack may not help. What's more, it's an Accutronics which doesn't have the greatest reputation. That aside, this hum is present whether the tank is connected or not, so if it is a ground loop it could be forming because I have two ground lines coming off different sides of the reverb board leading the GND1 via the casing of the can cap (that could be source of the hum - can caps are notoriously noisy) and then a grid leak resistor grounded to the chassis via the RETURN RCA jack. The can cap casing itself serves as the intermediary for the reverb board's grounds, it's isolated from the chassis and has the single line off to GND 1. The can cap seems to have no effect at all on the clean preamp channels which are also branched off GND 1 with it. The hum only becomes obvious when turning up the reverb Return pot.

In any case, there's been some improvement with the mods I made here, and it's possibly something I will have to live with now, which isn't so bad. But experimenting with isolating the RCA jacks, and maybe rethinking the ground scheme a little bit more, perhaps even a better tank as well. It is getting easier to see where any remaining problems might be now. Considering where I started with it!

Stephen

[Stephen1966]

Tried a couple more things today. Thinking about the RCA jacks on the chassis, I don't think it would make any difference to the hum; the hum is present whether the tank is connected or not; the SEND>IN jack is coupling with the chassis to provide a shield with the case of the tank and the OUT jack is isolated in the tank from its case. Both the OUT and IN meet ground in the chassis but they are effectively different branches which don't connect up in the tank. So no ground loop there. Looking at the grounds in the rest of my reverb circuit though I noticed one mistake I had made, which was to ground the reverb board ground branches to the case of the can cap, and then on, to GND 1. I've separated these, so the can cap case is now grounded to GND 3 along with the other filters that feed the preamp stages and the reverb grounds are now on a separate lug on a mounting screw of the can cap. The case of the can is now isolated from the reverb and preamp sections.

This didn't make an enormous difference but it is further isolating the PSU grounds from the preamp grounds so worth doing in any case.

I also replaced the 150k grid leak resistor on the recovery tube. The AB763 uses a 220k in this position and normally the grid leak of a preamp tube is 1Meg. I didn't alter the value because coming after the RETURN from the tank, it could be performing a grid biasing function though the ground sides of grid leak and the cathodes of the recovery tube are separated by the width of the chassis. Grid leak grounded on the RETURN RCA jack and cathodes on the front of the chassis, attached to the lug on a mounting screw of the can cap.

Again, no discernible difference. I also played around with a 200pF cap across the grid leak and from the RETURN pin of the jack to the chassis, two Fender fixes that Rob Robinette discussed, but again, no real difference.

I think I have this as good as I can get it now, unless I can find one of the square getter tubes that Tony mentioned but having swapped all the tubes around, that could be an expensive disappointment if it didn't work.

[Stephen1966]

If you have been following my roller-coaster ride with this circuit, you will recall that I had an enormous problem locating the source of the hum. When troubleshooting it's always a good idea to reduce the problem down into managable chunks.

To recap then, I tried: rolling tubes; replacing the 270pF input cap and the 1.2M feedback resistor with #60 correct NOS parts; increasing the dry signal by dropping the mixer resistors, ultimately, to 220k (down from 470k); increasing the voltage to the RVB transformer and then; a breakthrough, putting a cap on the return pot to GND like in the Wonderland. That last had the most effect and I thought, why not, I can live with it, as long as I never turn up the Return pot all the way, because when I did, it was hum city.

Then, came my experiments with a passive (mixer) loop. That https://ampgarage.com/forum/viewtopic.p ... 15#p442815 was instructive and led me to the conclusion that I needed an active loop in its place. I toyed with the idea of a miniature tube like in the Tube Town version (but with better parts) but that led to the realisation that even a miniature tube draws as much current as a 12AX7 and with 10 tubes already, the power transformer would probably melt down if I tried added an eleventh!

This led to the realisation that a three-tube reverb, could be a two-tube reverb, and thanks to rootz, I found a two-tube design that could work in the #60 reverb's place: https://ampgarage.com/forum/viewtopic.p ... 15#p442815 This would free up one of the tube sockets and allow me to install a Dumbleator in the chassis. That's where I am heading now. The new board is built and I'm just waiting on parts before I pull out the #060 board.

The #60 reverb now works though. The problem is solved. It was not the circuit. To get there though, I went back to the only place where the hum becomes a problem (the Return pot) and concentrated on that.

With scope probes on a heater pin and on the grid of the recovery tube, Volume, Master and reverb Return, all the way down, this is what I saw:

SDS00001.jpg

There it was! The source. The AC heaters. This is what Blencowe calls Rectifier Induced Heater Hum and you will notice that the AC signal (yellow trace) is clipping; it isn't a nice symmetrical sine wave like you would expect with the AC coming direct of the secondary of the power transformer. The clipping as I understand it, comes from the rectification of the HT secondaries, because the heater and HT coils interact with one another. I would speculate this is normal behaviour and a trade off of the solid state rectification in the HT supply. Perhaps the clipping is softer in a tube driven rectified circuit but I would expect it there as well. When you look closely at these traces you will see the blue trace and rectifer hashes on the trailing edge of the AC peak align almost perfectly - producing a 50Hz correlation. This wasn't the signal I saw in Blencowe's oscillogram though. His, is almost a steady 0v signal off the grid with rectifier hashes. So that was with the Return pot, all the way down. When I turned it all the way up, this is what I got:

SDS00003.jpg

Now we have something like a second order set of higher harmonics at 250Hz, superimposed over the 50Hz ripple present at the beginning. A holy crap of a signal. The signal that kept me up nights.

Again, this isn't what I was expecting at all, even if Rectifier Induced Heater Hum was the culprit, this second set of traces didn't make any sense at all. Stepping away, I considered brute force methods of removing the hum. I couldn't convert the AC heaters to DC because that would pull even more current (I'm already at the maximum I am comfortable with), I considered elevating the heaters, to take the hum out of the audible range but that would mess with the operating point of my preamp tubes and might have been more trouble than it was worth. Blencowe did suggest an alternative though, with a 500R humdinger pot, in this case, in place of the artificial CT I had with two fixed resistors. This was surprising, how it altered the sound of the hum, but the hum was still very audible. By placing probes on the opposite legs of the heaters though, using a 15 turn 500R trimpot, this is what I was able to do with the 180 degree phases of the AC signal:

SDS00010.jpg

I inverted one side as well to make sure they were properly balanced. Still the bloody hum though, so I grabbed a chopstick and started poking around. A funny thing happened when I poked the heater pin of V3 - the amp went dead quiet.

You can get me drunk and tie me naked to a lamppost right now! It was a cold solder joint.

Fixing that (mostly) cured the problem. Next then, I was curious to see what would happen to the signal if I removed the cap from the Return pot I had placed there earlier. The original #60 didn't have this so different PS scheme and all, I was hopeful the cap wasn't needed because a .01uF cap is hardly a burden on the wet reverb signal, whereas a .1uF cap would kill it altogether, but small as it is, it must be muting the signal. This led to the two following discoveries:

Here are the traces (blue and yellow heater AC) and red, is the grid with the Return pot on full:

SDS00014.jpg

This is without the cap, and you will see that that there is a corresponding rectifier hash again on the trailing edge of the AC peak corresponding to 50Hz hum coming off the grid. This time, because I had fixed the cold solder joint, not nearly so bad, but still present and better than anything I had tried before. This hum (sans cap) is not something you would start a topic on TAG with. It would be accepted and life would move on. But, being who I am, I thought I would put the cap back and see what happened to the signal off the grid then. And here's the signal with the cap:

SDS00015.jpg

The signal is all but flat and both these last traces were with the Return pot up full. So, I am speculating that the cap can go even smaller than .01uF. It is barely, barely noticeable at .01uF and once you start to play it isn't noticeable at all if you aren't listening for it. We might go down as low as 750pF or 500pF with this cap, but I would still add it into the circuit, because without, it just isn't as good.

The irony (sorry my American friends) the irony is that I will be removing this circuit from the amp in the coming days. I really have to get a recording of it before I do, because it is a beautiful reverb, deep and transparent, it doesn't appear to suck the tone at all and in clean mode you can really milk the notes for all their worth. Dick Dale lives! And more, overdriven notes are clearly delineated; it doesn't get mushy or muddy at all. I want to try a couple of other things with the circuit before I rip it out. Such as, reducing the 1.2M LNFB resistor in the Schade network on V4. I don't really understand the Schade network principles and reading around, many would argue it should be called by other names, in this instance at least. My understanding is that is more effective with pentodes. At one point, as I was thrashing the thing in OD I thought I heard clipping in the reverb so reducing this resistor's value might help to keep the headroom it displays at all other times. I'm not certain, because I only heard it the once, couldn't repeat it, and could be mistaken. Even so, it's easy enough to drop the resistance with another resistor tagged in, in parallel. I'm going to take great care with this circuit board, I don't want to try anything that is non-reversible or destructive with it, because it works very nicely now and it might come back another day.

So here's my offer to everyone who has ever looked at the #60 and thought what the hell is going on with the reverb. We have a working example here and if you can think of any other mods I might try before it goes, this is your chance.

That, and if anyone could teach me how to solder, that would be much appreciated.

DipTrace Schematic - SKYLINER reverb 2.6.pdf

Stephen

[Edit: and yeah! A BIG thanks to everyone who helped me get there.]