Mesa MkV unusual problem

[Nyqusit5]

Hi,

I have a Mesa MkV in for repair.
The owner brought it back because the sound suddendly cut out while playing. Only a faint signal was audible on channel 2.

Following the signal flow, I noticed that it was stopped at V3a, and that the plate was at 0V, while the other end of the plate resistor was at 380V roughly. So I naturally decided to replace it thinking it was dead. After replacing it, I had the same behaviour. But simply touching both ends of this 100k resistor usually brings back the signal... almost. It only lasts for a brief period during which the tone seems more or less ok except on channel 2 where the tail end of notes is accompanied by a crackling noise. After some time, the relays start clicking randomly until the signal cuts out again (the switching jacks are fine and not the culprits).

The schematic is easily found on the web. I am not sure I can post it here (Mesa Copywright). And Mesa is unable to help me as, i quote "Although this is not a common problem, it does happen on occasion. The relays must be inspected/replaced."
Given the number of relays on the main board, this is just not a realistic repair procedure...

Any help, experience sharing or suggestion will be most welcome!

[nickb]

The schematic that I have is missing the all the relay coil / switch connections so that makes it tougher. Also the connections to the FET gates are not shown. ugh! Shame on you Mesa!

Relays chattering: suggests bad connections / bad switch(es). If it's all of them then maybe the power to them is a problem.

V3A: If the relay connected to the grid to not working properly then the grid could be floating and strange things will happen. Try putting 1 Meg from the grid to 0v as a diagnostic.

[Nyqusit5]

Yes, I don't have the switching circuit either...
I checked the grid connection but that's a good idea: I'll solder a 1M resistor just to make sure.

Thank you Nickb.

[bob p]

Mesa is notorious for publishing incomplete schematics.

i bought my Mk IV for a song because it was switching erratically and nobody knew how to fix it. instead of looking in the amp, i looked at the footswitch. i found an LED that had been stomped on enough times that the lead had broken off of the PCB, which interrupted the switching logic. Re-mounting the LED fixed the problem. (the amp had mounted the LEDs on long stalks, mounting the LED to the chassis at one end, and to the PCB at the other. I removed this arrangement, interposing some wire between the LED and the PCB to allow flexion within the footswitch chassis when it was under load {ie: "The Boot"}.)

my only point is to say that sometimes the problems aren't in the places that you'd think of first.

It does really surprise me that an amp as new as a Mk V would be failing so soon. This makes me inclined to look for a mechanical problem.

[Tom Phillips]

… the plate was at 0V, while the other end of the plate resistor was at 380V roughly. So I naturally decided to replace it thinking it was dead. After replacing it, I had the same behaviour.

The zero plate voltage is the big clue. I’d guess that some fault is still causing open circuit between the supply rail and the point where you measure zero plate voltage. The switching relays shouldn’t have any affect on the plate voltage no matter what condition they are in so let's forget them for now. I’d be looking for some kind of open connection in the circuit from the top of the plate resistor to the tube socket plate pin. It could a solder connection, PCB traces or maybe a bad via feed through on the board.

With the amp turned off and all caps discharged you could take resistance measurements to narrow down the problem area. Pull the tube and measure from the top of the tube socket. Measure resistance from the plate pin sleeve to the power supply rail (should read the plate resistor value). If you don’t get the proper reading then move on down that part of the circuit until you find where the fault is.

Let us know what you find.
Tom

[nickb]

...The switching relays shouldn't have any affect on the plate voltage no matter what condition they are in ...

Not so. If the contacts are bad on the grid side, the grid will be floating.

[g1]

But if the tube is not conducting, there should be the full B+ on the plate, not zero. I was thinking the same as Tom, sounds like a break somewhere between the plate resistor and the plate.

[Enzo]

I am confused. When you found 380v at one end of the plate resistor, and zero at the other, why did you not immediately replace the open resistor? And why replace the tube?

If touching the resistor makes sound come back, then likely the resistor has intermittent connections inside between the resistive part and the wire leads. And touching it moved it enough to make contact.

[g1]

Hmm, now I'm confused too. I had assumed the resistor was replaced, Enzo assumed the tube was replaced. Re-reading the original statement it seems it could be either:

Following the signal flow, I noticed that it was stopped at V3a, and that the plate was at 0V, while the other end of the plate resistor was at 380V roughly. So I naturally decided to replace it thinking it was dead. After replacing it, I had the same behaviour.

So which was replaced, the tube or the plate resistor?

[Enzo]

yes, I guess one could read that either way. When he said "it" was dead, I thought tube. People call tubes dead, but rarely resistors. But could be.

Very hard for a tube (12AX7 preamp tube or similar) to bring its plate to zero volts through a resistor to B+. There is usually no path to ground in the tube. The cathode has a 1.5k or something and the grid typically higher. A plate to heater short is unlikely, but even there, probably not zero volts. And if the tube could do that, then the resistor ought to be getting very hot.l

[Nyqusit5]

Ok, maybe I wasn't so clear.

Yes, there ALWAYS is 380V on the B+ rail side of the plate resistor (no open circuit here), and yes, I measure 0V on the plate side when the amp won't produce a sound. I have already replaced this resistor twice and the behaviour is still the same: I know it doesn't make sense because if the tube wasn't conducting, I should read 380V on the plate side of the resistor anyway, but that is the way it is. (I have also tried other tubes in there just in case).

Unfortunately, the contacts are not intermittent as I have hardwired both ends of the resistor to eliminate this possibility. Furthermore, touching one end (either end) of the resistor with a probe doesn't bring back the plate voltage, but touching BOTH ends does.
Yes, this is very confuusing...

And the relays are still clicking (maybe a second problem, no related)...

[Enzo]

OK then, is the resistor getting warm or hot? What resistance is there to ground at that plate pin?

Before I knew you had replaced the resistor, when I was talking about loss of contact on the wires, I meant inside the resistor, not at the solder joint.

Touching both probes to the resistor returns plate voltage? You are then using two meters? A meter has internal resistance, so it completes a circuit. If you parallel an open resistor with it, it will take the resistors place. If you have 380 at one end and zero at the other, then I would expect a meter across the resistor to read 380v.

If probing across the resistor makes plate voltage return, what voltage is on the plate at that time?

You have replaced the resistor more than once? OK, was the original one open? And the new ones you replaced, did they measure intact or did they now measure open? If the plate end is shorted to ground, perhaps it is burning the resistors open. I don't know what circuit you have, but for example a 100k resistor with 380v across it will try to dissipate a watt and a half. So if they are 1/4 or 1/2 watt resistors, they could fail.

You tried other tubes? OK, but what does it do with no tube in the socket? If it behaves the same, then we know it is not tube related and there is something about the socket circuit.

[Nyqusit5]

The resistor is not getting warm. I am using 1W resistors and they all measure correctly in or out of the circuit and before, while and after being installed. I understood your point about losing contact Enzo, but I was also trying to reply to the other posts at the same time. Either way, there is no loss of conatct at any point that I can find (inside or outside the resistor)

Touching both ends of the resistor when it is in place measures 380V when no signal is getting through. When it does (after a few seconds, sometimes more, sometimes less), it measure about 180V: a normal voltage drop for a plate resistor. The signal still comes through even while not touching both or one end with a probe, for a few moements, then it stops again and voltage drop accross the resistor becomes 380V again at that point.

The schematic shows no resistor possibly grounding the 100K at the plate end, and I can't see any on the circuit (which is very difficult to trace). Measuring a resistance at the plate end gives about 550K which is coherent (100k+the two 220K accross the first filter caps).

Thanks to all of you for your suggestions!

[Mick Bailey]

If you're getting 380v across the resistor then one end has to be at 0v if your supply is 380v, and it will get hot. I wouldn't bother with AC conditions at this stage, you need to get the DC correct.

The only connections to that resistor are;
a) The HT
b) Coupling cap.
c) Tube plate connection

The HT is shared, so you can check this on another tube - V6a for instance.

You've changed the resistor. If it appears to have 380v across it and isn't getting warm, then I suggest something may be wrong with your measurement technique or your meter. Ohms law dictates that the resistor has to dissipate 1.44 watts given the stated conditions, and you'd notice this with a 1w resistor.

The coupling cap needs to be eliminated from the equation. I would lift it temporarily at the tube end and re-check the DC on your tube. You need to be getting a stable voltage on the plate. Maybe there's something bizarre going on where the coupling cap is getting discharged with your meter.

[Nyqusit5]

Yes, my next move was going to be the coupling cap too.
As far as measuring technique is concerned, I have used two diffrerent multimeters and the procedure is somewhat basic so I don't think there is any room for error here.
Thanks for your suggestions