Mesa Subway Blues some hum/buzz 100Hz, out of ideas

[iefes]

Hi everybody, I've finally found a solution for the hum/buzz problem. I installed a small sub-PCB which converts the AC heater voltage to DC and feeds it to the three preamp-tubes. The hum is completely gone now. The power tubes are still heated via AC.

DC is produced by rectifying the 6.3VAC with four 1N4007 Diodes and smoothing the voltage with four 4700uF caps two of which are referenced to Ground. Works like a charm :-) I don't know why my amp was humming more than others though. Maybe any parts around V2 got microphonic, e.g. that capacitor in the smaller Epoxy-block. However, now it is solved.

But I have another issue still occurring that I had encountered before: With the Volume-pot fully ccw there's still some significant amount of signal passing through the amp and out of the speakers. It is very muddy and lacks a lot of treble, but still it's actually pretty loud. I tried swapping the pot for a new one but this didn't help. Also I tied the grid of V1b to Ground and STILL there is signal coming out of the speakers. That's really strange. Could the signal be passing through to the second tube using another way?
When the anode decoupling cap of V1b is tied to Ground there is no signal passing anymore and the hiss that the amp has is gone as well. So it seems that the signal is somehow transferred from V1a to V1b. I tried different tubes in that position so it shouldn't be a bad tube.

Any ideas regarding these issues would be really great. What else can I test to clarify what is causing the issue?

Thanks heaps!

PS: I don't understand why this post didn't make it to the end of the thread, sorry for this.

[Chuck H]

Congrats on the DC filament success. I can't say if my old amp had the hum issue. If it was a small amount I may have chosen to ignore it.?. re: Bleeding signal with the volume down...

If it's not a tube (you tried others in that socket), it's not a pot (you tried others), it's not a decoupling filter (those are new) and it's not the ground for the volume control (you grounded V1B grid without joy) the only other (logical) possibility is that the cathodes of V1A and V1B share a questionable ground. You can try lifting the grounded end of either cathode circuit and temporarily grounding it elsewhere just to see if there is any change. Otherwise I would suspect that one of the other eliminated issues above is not actually eliminated.

You mentioned that you replaced the filter capacitors but did you install them in the same year that they were purchased or do you know that the caps you bought were fresh? You tried another pot and another tube but was pot of the same make and batch? Was the tube? These things happen. Could be NONE of the pots in that batch actually turn down all the way or that NONE of the tubes in that batch don't bleed between the elements. Did you ground the V1B anode (after the decoupling cap I assume) in the same location as the volume pot ground? And if not did you test the volume pot ground node against the other ground location for resistance or voltage potential? You need to know your meters "null" reading at the time of the test to do this. Just touch the probes together and note that the result isn't zero. Do this whenever measuring for small voltage or ohms. Subtract that reading from your test measure.

Hope this helps.

[iefes]

If it's not a tube (you tried others in that socket), it's not a pot (you tried others), it's not a decoupling filter (those are new) and it's not the ground for the volume control (you grounded V1B grid without joy) the only other (logical) possibility is that the cathodes of V1A and V1B share a questionable ground. You can try lifting the grounded end of either cathode circuit and temporarily grounding it elsewhere just to see if there is any change. Otherwise I would suspect that one of the other eliminated issues above is not actually eliminated.

You mentioned that you replaced the filter capacitors but did you install them in the same year that they were purchased or do you know that the caps you bought were fresh? You tried another pot and another tube but was pot of the same make and batch? Was the tube? These things happen. Could be NONE of the pots in that batch actually turn down all the way or that NONE of the tubes in that batch don't bleed between the elements. Did you ground the V1B anode (after the decoupling cap I assume) in the same location as the volume pot ground? And if not did you test the volume pot ground node against the other ground location for resistance or voltage potential? You need to know your meters "null" reading at the time of the test to do this. Just touch the probes together and note that the result isn't zero. Do this whenever measuring for small voltage or ohms. Subtract that reading from your test measure.

Hope this helps.

Chuck, thanks for your thorough answer. What do you mean by "decoupling filter"? The filter capacitors of the B+ supply? If yes, you're right they are new and fresh and have been replaced before I replaced the pots. I did not have the issue with the old pots installed which were very scratchy, which in turn is why I replaced them. The new pots potentially come from the same batch but the signal passes through to the second stage even with the grid of V1b grounded, so I don't think the pots are to blame here. The tubes I tried were both new and old ones of different brands and they didn't change the issues I'm observing.
I'm not 100% percent sure if I grounded the grid at the same spot as the anode coupling capacitor but I measured the resistance of the different points that should be grounded to a common ground point and got a reading of <1 Ohm all of the times. So there shouldn't be any signal passing.

I think last thing I can do is to lift the cathode caps and resistors and connect them to another ground-point as you suggested. I will do this later tonight hopefully. Otherwise I can't really see a reason why the amp would behave like this now but didn't do it in the first place when I got it a few weeks back. It shouldn't be an issue with the grounding-scheme as I didn't change it. Probably I'll lift the PCB again and see if there's something on the back of it that looks strange.

Thanks so far. Any more ideas are very welcome!

[iefes]

Okay, I had a glance under the PCB but it looks nice from both sides. I haven't checked the different ground-connections but measured all different ground-points against the main chassis ground and they're all the same between 0.6 and 0.5 Ohms, if I can trust my DMM. I've also tried to connect those ground points to other ground-points using jumper cables.

Another thing that I observed today: I put in another tube from my drawer, a Groove Tubes GT12AX7R. With this tube there was almost no crosstalk (with the volume control at 0), I could hear the notes coming out of the speaker very quietly but not to compare with the bedroom-levels I got with other tubes (one of them was new!). I can't believe that it's just a matter of the tubes. Otherwise this would have been observed more frequently. I read about crosstalk in 12ax7's but not as loud as I experience with my amp. So, it seems different tubes somehow affect the issue but I don't really think they're the reason for it. What do you guys think? Does this observation help to come up with other ideas of what this could be caused by? Thanks heaps!

Edit: Chuck, in the schematic that shows your modification you did to your amp years ago there's a 10uF filter cap for the "C"-voltage (which powers V1 and V2). I've used a new 33uF just like what was in there, but decreased the dropping resistors to a total of 25k (yielding roughly 250V; you used 20k). Did you put the smaller cap in there for a specific reason? Could it be that the signal can somehow pass over the HV line and that the cap value plays any role? I'm trying to think of every possibility.

[Chuck H]

There are two things the smaller cap does. One is that it recharges faster ("fills up" faster after a drain because it's not large). Probably not significant for a single, early preamp tube that probably never taxes the supply. The second thing it does is less obvious. Low LF frequencies aren't bypassed by the smaller cap and would remain on the HV rail. These frequencies should suffer some cancellation due to local NFB since the rail shares one node for each two series, inverting gain stages. The end result is that the smaller cap sounds and "feels" a little tighter. Some might wish to describe this as a simple LF cut with the same result as any high pass filter, but it DOES feel a bit different to the player. All things being relative there are a number of circuit specifics when doing an LF cut this way that are not the same as an 'in line with the signal chain' HP filter. But I haven't done enough measuring and testing to qualify this or offer specifics. I just tried it based on a suggestion and liked the way it sounded.

[iefes]

Thank you Chuck. I quickly tried a 15uF instead of the 33uF in that position using jumper cables, but it didn't improve the crosstalk issue. Just wanted to be sure.

But I'll switch to a 10 or 15uF eventually when I have sorted out the crosstalk.

Thanks!

[iefes]

I haven't solved the issue yet, but I have some updates:

I was at a befriended amp tech who had a look at the amp. We tried to decouple the HV path between the two stages using a series resistor (4k7) and a 10uF cap to ground on either end of the resistor. Didn't help at all. I tried to lower the C+-voltages down to almost stock voltages (168V), which didn't help. Then I figured I could try to remove the anode resistor on V1b to see if the crosstalk still happens, and yes, it does. So there's still the crosstalk with V1b disabled. I reconnected the anode resistor and lifted the cathode resistor and bypass cap, but the crosstalk remained. Then I lifted the leg of the coupling cap (47n) behind V1b that is going to the anode. Like this the amp is silent, there's no crosstalk anymore but of course no signal passing at all. But this tells us that the signal is bleed to the anode of V1b somehow.
I cleaned the PCB around V1 throughly using acetone, cotton buds and a toothbrush. I also removed the tube socket temporarily to clean the PCB underneath it. This all did not help. I suspect there's some microscopic crack in the PCB that somehow forms a connection between the two anodes. Or the whole PCB conducts around these areas. I can't see any other explanation at this point. Of course I measured the resistance between the two anodes but my meter doesn't read any electrical connection between the two, i.e. infinite resistance.

When I worked on the amp with my friend one strange thing happened: We tried several things using jumper cables. One of the things was to connect lug 3 of the treble pot to lug 2 of the volume-pot. This shunts the tone stack and the Anode of V1a to ground with regard to AC, when the Volume-Pot is fully ccw. And it worked. We were able to quiet the amp by turning down the volume. Okay, then we remove the jumpers to have the amp stock again. And suddenly the volume control worked as it should even without the jumpers We were confused, tried switching the amp off and on again but still, the amp behaved exactly as it should and we didn't know what happened. We left the amp for a while (1 hour or so) and when we came back to check again, the crosstalk was back and we weren't able to reproduce the behaviour we had seen before using the jumper cables. Since then I wasn't able to reproduce any normal behaviour of the volume-pot and that's why I tried all the things mentioned above.

What do you guys think? Should I rip out the whole PCB and rebuild the thing on Turretboard or similar?

Thanks!