Solid-state Vox Pacemaker with 120Hz buzz

[Rhodesplyr]

I've finished basic repairs on a 1966 solid-state Thomas/Vox Pacemaker--replacement of all electrolytics and input stage transistors. The amp works and sounds good, all functions work, and all voltages at idle are right where they should be. The problem is that there's an objectionable amount of constant 120Hz background buzz.

The buzz is unaffected by volume control position, so it doesn't seem to be coming from the input stage; however, it is modulated by the Tremolo circuit as you turn up tremolo depth. Otherwise, as I said above, the buzz is constant.

I installed 3-wire grounding on the amp. I've checked for bad ground connections, but can't find any.

Any diagnostic suggestions? Can installing a 3-wire grounded power cord ever *cause* something like this?

I feel like I'm missing something obvious, but it's been a l-o-n-g day...

[Enzo]

Grounding issues cause 60Hz hum. 120Hz hum is from power supply ripple. It maye be ripple directly on a power supply, or it could be ground return from some stage is sharing conductor with the ground return of some power supply element, like a cap.Won't be the mains cord. Well unless you moved some other power supply ground to make way for it.

RG is much more familiar with these than I am.

[woodyc]

I doubt this will help but... I fixed a Kensington a while ago with the same symptom. In that case, someone had previously done a cap job and then screwed the board back in a little too well. A trace directly under one screw got a hairline crack. It was a trace carrying ground to a supply cap.

[Rhodesplyr]

A little more refinement in my observations:

The noise first appears at the driver transistor stage, the transistor directly connected to the output stage driver transformer. The noise isn't on the supply to the transformer, but it is at the other side, at the collector of the driver transistor. C, B, and E of this transistor are at the correct voltages. E is fully bypassed to ground, and there's no unusual noise at E or B.

It's as though the power supply ripple is getting picked up by the driver transformer, though, of course, it could be coming from either side of this transformer. Time to take a closer look at the secondaries, the output stage, and the NFB path that surrounds the driver and outputs.

Ripple on the first power supply capacitor (2,700uF) is ~0.3V, peak to peak. Diodes are new ultrafast/soft-recovery types.

[R.G.]

I tried to respond to your PM, but your storage is full and you can't get any more messages.

The grounding inside Thomas Vox amps appears to be crazy, but is actually a fairly sophisticated local-star arrangement. It's hard to tell, but that's what it is.

Enzo is correct: 60Hz is hum pickup, 120Hz is full-wave rectifier ripple or pickup. I'm assuming you checked this on the scope to find the actual frequency. Plus two points to you for recognizing the difference!

If you recapped, there is one point that you absolutely have to check. This amp has a full-wave center tapped rectifier section. The center tap from the transformer must go to the negative side of the first filter cap, nowhere else. And the negative side of the first filter cap must either be the ground point, or must have one wire from it to the amp's master ground point. If this is not true, the current pulses from the rectifiers charging the cap contaminate the entire ground system with pulse voltages from the charging pulses. This is one place where moving a single wire's end position can make the difference between silent and buzzy.

"Buzz" typically means you're hearing pulse waveforms as opposed to real hum. We hear pulses at a low (i.e. 60 or 120Hz) as a low frequency hum, but with a buzzy quality. This is what classic fluorescent light buzz is.

As the other guys noted, there are other things to look at as well. Cracked traces, or a break in one of those #@$#* black grounding wires can do it to you too. But the rectifier buzz is subtle and hard to find otherwise until you know about it.

No, a three-wire conversion does not cause hum, nor ground buzz, unless the safety ground in your shop is high impedance or otherwise noisy itself. Buildings can have ground problems too.

[Rhodesplyr]

R.G.,

You are a genius. Thank you!!

I installed a new first stage filter cap under the chassis, near the old one, and, since it was convenient, I intercepted the ground wire to the PCB for the capacitor's ground return. I should have known better than to do this because I understand the idea of having a dedicated ground return for the first filter stage to keep charging pulses in a closed loop. In fact, I've done it in things I've built and have recommended the practice to others! Duh. I guess I just wasn't thinking when I did that. Or, rather, the ground system in this amp, as you say, *appears* to be so haphazard that I didn't think it mattered very much.

It's amazing that three inches of wire could contaminate the ground so severely. Is it the case that this problem is more pronounced in solid-state supplies that are operating at lower voltage and higher current?

As implied in my post, I grounded the first stage filter cap directly to the PT center-tap and used a separate ground wire to the PCB. The buzz disappeared.

The only noise remaining is a little transistor "rush." Does Q4, the driver transistor, tend to get noisy? I find the most random/broadband noise at its collector. If it needs to be replaced, what would be a good sub?

[Enzo]

NO, not only solid state. I had an AMpeg something or other, GU-10 maybe?, that hummed like crazy 120Hz. FOund a B+ ground wire running to a terminal strip instead of right to the cap/CT ground. A 3" wire from that proper point over to the terminal strip. Move the ground wire over to the real ground point - other end of that 3" wire, et voila. No more hum. Obviously the amp had hummed that way since it was made.

[R.G.]

You are a genius. Thank you!!

You're welcome, but it's not genius, just places I've stumbled before.

It's amazing that three inches of wire could contaminate the ground so severely.
Is it the case that this problem is more pronounced in solid-state supplies that are operating at lower voltage and higher current?

It's hard to believe until you get tripped by it. Many commercial hifi amps have the same problem to various degrees, because it is so subtle to fall into. Solid state rectifiers make this worse, and bigger capacitors make it worse.

The only noise remaining is a little transistor "rush." Does Q4, the driver transistor, tend to get noisy? I find the most random/broadband noise at its collector. If it needs to be replaced, what would be a good sub?

Could be. The driver is not a likely suspect, but could be. The suggested replacement was 2N2219, but you can use 2N5320, 2N5321, or 2SC3503. This sub *may* make the thing less stable.

It is also possible that the resistor/cap on the collector to ground may be in bad shape and not controlling the high frequency rolloff of the stage. Check those out, or just replace them.

[Rhodesplyr]

R.G.,

I've isolated the noise at the output to the output stage. It's around 20mV peak-to-peak at the speaker terminals and sounds a bit like the roar of a distant highway, but more constant. Of course, I'm listening to it in a very quiet room.

Is this in the normal range of background noise for an amp from this era with germanium output transistors?

I replaced the 470 Ohm resistors, R22 and R24, with KOA Speer 2W carbon films, but I didn't replace the 3.3 or 0.33 Ohm resistors. Do you do this in your Pacemaker restorations? I could put in wirewounds. Or, to ask my question in another way, if they are carbon composition, do these low-value transistor output stage resistors tend to be noise sources or not?

I read 4mV across each of the 0.33 Ohm resistors at idle, translating to ~12mA bias current. And I read 0.12V across each of the 3.3 Ohm resistors.

As you can tell, I'm still on my learning curve in terms of noise reduction in Jurassic solid-state amps, and I appreciate the benefit of your experience. With these things, it's hard for me to tell if it's working as well as can be expected or not.

[R.G.]

I've isolated the noise at the output to the output stage. It's around 20mV peak-to-peak at the speaker terminals and sounds a bit like the roar of a distant highway, but more constant. Of course, I'm listening to it in a very quiet room.

That description makes me think "1/F noise". There is a noise mechanism which gets larger as frequency goes down. It is often a characteristic of contaminated contacts, or contaminated junctions.

Is this in the normal range of background noise for an amp from this era with germanium output transistors?

I don't know for sure. I do know that germaniums had poorer sealing/environmental protection, so they often had contaminated surface leakage paths and that gave them problems with 1/F noise. I don't think it was normal for the amp when it was new.

I replaced the 470 Ohm resistors, R22 and R24, with KOA Speer 2W carbon films, but I didn't replace the 3.3 or 0.33 Ohm resistors. Do you do this in your Pacemaker restorations? I could put in wirewounds. Or, to ask my question in another way, if they are carbon composition, do these low-value transistor output stage resistors tend to be noise sources or not?

I don't think it's the resistors. Try this: remove the transformer primary lead from the driver transistor collector temporarily. Make sure it's not shorted to anything, then power the amp and listen/measure noise. If it's the same, it's generated in the output stage, and my first suspicion would be the two output devices. If it's essentially gone, the noise comes from the driver or before.

As you can tell, I'm still on my learning curve in terms of noise reduction in Jurassic solid-state amps, and I appreciate the benefit of your experience. With these things, it's hard for me to tell if it's working as well as can be expected or not.

Me too. You mentioned my "Pacemaker restorations". I have a Pacemaker, and one day I'll get the time to tear it down and work it over. That'll be my first Pacemaker restoration.

[Rhodesplyr]

R.G.,

Thanks for the diagnostic suggestions. I now think that the remaining noise is coming from transistor Q3, the one modulated by the tremolo LFO signal. One clue is that this background noise itself is modulated by the tremolo. I arrived this conclusion by grounding the signal via a 10uF capacitor. If I do this at the base of Q3, the noise disappears. It also disappears if I connect the capacitor to the collector of Q3. (I'd already replaced R15 with a 1Meg metal-film resistor to eliminate that as a noise source.) All these tests are with the volume control turned all the way down.

So, the question now is: what transistor to use to replace Q3? This is Thomas's 86-5050-2, with the yellow dot. I'm cognizant of Enzo's points about how one shouldn't over-think transistor substitution. In this situation, I'm not sure how critical gain is since it's this stage's gain that's being modulated to produce the tremolo effect.

In your document on Thomas transistors, you don't list a standard sub for this Thomas number. I know that in my Thomas/Vox Continental Organ, manufactured around the same year, transistors with a single yellow dot are 2N2924 (used as master oscillators on the divider boards and marked A1461), but the transistors in the Continental are in a TO-98 package, the kind with a small depression on top vs. the round button shape in the Pacemaker. (I know the package makes no difference in terms of function, but what TO package type are these small black plastic buttons?)

On datasheets, the 2N2924 is rated hfe 150-300 @ 2mA , so any number of NPN transistors should fall into that range.

The frustrating part here is that my oscilloscope seems to be relatively useless in tracking this noise. I suppose it's too low in amplitude at this point to be distinguishable on the trace.

[R.G.]

I now think that the remaining noise is coming from transistor Q3, the one modulated by the tremolo LFO signal. One clue is that this background noise itself is modulated by the tremolo. I arrived this conclusion by grounding the signal via a 10uF capacitor. If I do this at the base of Q3, the noise disappears. It also disappears if I connect the capacitor to the collector of Q3. (I'd already replaced R15 with a 1Meg metal-film resistor to eliminate that as a noise source.) All these tests are with the volume control turned all the way down.

Good debug strategy.

So, the question now is: what transistor to use to replace Q3? This is Thomas's 86-5050-2, with the yellow dot. I'm cognizant of Enzo's points about how one shouldn't over-think transistor substitution. In this situation, I'm not sure how critical gain is since it's this stage's gain that's being modulated to produce the tremolo effect.

I had a look at the schematic. I think most medium-high gain, low-noise NPNs would work. I have a saying I humorously refer to as Keen's Second Law: When in doubt, whip in a 2N5088.

I would measure the DC voltages you're getting on the transistor now, and compare them to the voltages on the schematic. If they're close, use that as a semi-yardstick. Assemble several prospective devices, (and include a 2N5088 for me... ) and try them. If the DC isn't too far off, the signal operations should be close.

A couple of things come to mind. One is that the circuit around that transistor makes it prone to reverse base-emitter breakdown on power off. This degrades transistors a little at a time, so it could be that any modern device you stick in there will be quieter. You can prevent this in the future with a signal diode (1N4148 works) connected anode to the emitter of Q3 and cathode to base. It's normally reverse biased and has no effect, but it completely stops the reverse breakover problem.

The other is that a shotgun approach sometimes works. If replacing Q3 doesn't help, it's not out of the question to replace every R and C around it. It's cheap and fast. On one Beatle I fixed with monster hiss, the problem turned out to be a noisy metal film resistor. Sometimes strange things go wrong.

The frustrating part here is that my oscilloscope seems to be relatively useless in tracking this noise. I suppose it's too low in amplitude at this point to be distinguishable on the trace.

That is unfortunately just the way it is. If you hunt noise gremlins often, you need to build yourself a noise amplifier. This can be a low noise opamp circuit with a gain of maybe 10x and 100x selectable, and a high pass cutoff down at about 120Hz to keep the worse hum out of it. This amplifies the noise up to where you can see it. You can see the noise amplifier's noise if you short its input and mentally subtract that out. But it complicates things if you don't do this much.

[Rhodesplyr]

R.G.,

We have success. I went in and replaced Q3 and shotgunned all contiguous resistors including R38, R39, R40, and R16 with Vishay/Dale metal films. I'd previously replaced R15. Other than a little ordinary background power supply hum, it's now surprisingly quiet.

Some amps lend themselves to diagnosis by replacing one part at a time, but, in this amp, as you know, you have to desolder wires to work on the PCB and resolder them to hear the results. Thus, I thought a shotgun approach was the efficient way to proceed.

I ended up using a 2N2924 to replace Q3 for a number of reasons. I tested Q3, and it showed a gain of ~120 at 1mA (old-style Heathkit tester). When I tested the 2N2924s I have--modern ones bought in the last year or two from Mouser--they set up in the tester very much like the transistor I removed. These 2N2924s have a gain of around 200 at 1mA. When I set up a 2N5088 to test, the control settings on the tester had to be very different to get the same test conditions, and gain was much higher (>500). That suggested to me that I might have the lowest chance of ending up with a problematic replacement with a 2N2924, even if its pedigree is less respectable than that of the 2N5088. Also, from working on Thomas/Vox organs, I know that these modern 2N2924s work just fine in place of the yellow-dot transistors in those units.

Interestingly, the PCB layout neatly fits the 2N2924 lead pinout, both vintage and modern, but the original button-shaped Q3 had to have its leads cross to fit the layout. I apologize for not testing a 2N5088, but, as I said above, the way this amp is built, experimentation with different parts would be a nightmare, and I have other repairs I need to get on with.

I think you'd originally advised me to replace Q1 and Q2 in this amp as a matter of course. I might add Q3 to that list. I also shotgunned many of the resistors around Q1 and Q2 with low-noise metal films the first time I went in. I didn't have to replace any of the film capacitors.

I ended up with around 5,000uF at the first filter stage. One might add more, as I believe you suggest on your website.

A 0.02uF capacitor across the On/Off switch eliminates a loud pop at turn-off, which I'm sure the speaker appreciates.

Vox Pacemaker: Done.

P.S. Two final details:

The 2N2924 as Q3 biases at idle right at the voltages shown on the schematic.

And did you think that power supply sag was exclusive to tube-rectified amps? Not so. If you play the Pacemaker loudly, you can watch the pilot lamp dim in time with your playing, suggesting that Thomas didn't over-spend on the power transformer.