Marshall 75 Nasty Problem

[Dr Kaos]

I have for repair a Marshall 75 Reverb Model 5275.
The problem is of course an intermittent fault, it goes crack loudly thru the speaker even with all volumes off at random times.
I am sure it is a poweramp fault because it a sharp DC shift but I haven't yet seen it on the scope.
Wondering if anyone knows of this.

Dr Kaos

[DrGonz78]

A schematic is a great place to start. To assume an amps fault is common is always "our" first hope but not always a likely scenario. Actually, look at every problem first as if it is a new thing w/ only "your" own defined solution as the tech.

Okay I will post the schematic for us all...
http://www.drtube.com/schematics/marshall/5275.gif

I would first check the all the voltages you can including: 15v rails for the preamp, 24v rails, and 45v rails. Lets see the power amp by numbers. Lets not assume anything until we see it with our test equipment. Not knocking your gut instinct, but lets put it to the test. Also, check to see if we have any DC on the output at all. If DC on the output then don't use a speaker till we have removed that DC on the speaker output. Excuse my recommendations if you already knew this... Good Luck

[km6xz]

Can you give us a clue as to why this is a "nasty problem". It sounds like a typical random problem caused by the intense mechanical and electrical stress that an amp lives with.
Is the DC shift to one rail or some value less? How does the owner reproduce the problem, is it with playing it hard or when it is idling with no signal? Did it destroy the speaker and blow fuses when it shifts at the output to a rail. Is it a thump or actually DC offset. If it is a DC offset, you are lucky, the only possible causes are in the power supply and the rather simple power amplifier. If a thump, it could be anywhere in the signal path since it is capacitor coupled between stages.
You should refrain from doing shot-gun "fixes" like resoldering the whole board so you never know whether you actually solved the problem. Try to reproduce the conditions where the unit fails, level, heat and vibration so you see it occur while connected to your test instruments so even if the fault occurs briefly you will have some basic data from which to formulate a hypothesis. The correct diagnosis is one that explains all the measured data. If it does not explain it all, it is not the correct diagnosis. A trap many techs get into is "marrying" their first assumptions which usually accounts for the "dogs" that take hours or days to resolve. When it is resolved, the problem is usually one that was overlooked for hours because it did not fit the first assumptions. Take a set of basic measurements and turn it off, study the circuit and figure out how that set of fault data relate to a cause. If 15-20 minutes of analysis does not lead you to the fault cause, spending 3-4 hours more is not likely to improve that. Put it aside, work on something else and come back to it with a fresh mind. If I run into one that does not make sense, I take more data and leave it for the next day, usually in the morning the diagnosis becomes crystal clear, often accompanied with the sound of a slap to the forehead in how obvious it was. In my old shop, with lots of techs, I encouraged techs to discuss each other's diagnosis or lack of one. If serious effective measurements when the fault is happening does not result in a correct diagnosis I have found that the only causes of that is the tech did not understand the circuit or they formed an opinion too early and stubbornly held to it despite evidence. The 30 minute rule was in effect: "if you do not have a diagnosis that fits the evidence in 30 minutes, get a set of eyes on it by asking for assistance". The physical act of dis-assembly/re-assembly and correcting the identified defect ought to be the longest part of the total repair activity. If it is not, the circuit is not well understood or the tech is not approaching it with an evidence based point of view.
I am writing this about general tech work because the statement of "nasty problem" really stood out as if it was a particularly difficult or mysterious problem. It most likely is, like most defects, very simple and logical.

I would not use the speaker at all until you have identified and corrected the problem that is causing this. You can disable the speaker by simply inserting a 1/4" phone plug into the extension speaker jack.

[Dr Kaos]

Can you give us a clue as to why this is a "nasty problem". It sounds like a typical random problem caused by the intense mechanical and electrical stress that an amp lives with.
Is the DC shift to one rail or some value less? How does the owner reproduce the problem, is it with playing it hard or when it is idling with no signal? Did it destroy the speaker and blow fuses when it shifts at the output to a rail. Is it a thump or actually DC offset. If it is a DC offset, you are lucky, the only possible causes are in the power supply and the rather simple power amplifier. If a thump, it could be anywhere in the signal path since it is capacitor coupled between stages.
You should refrain from doing shot-gun "fixes" like resoldering the whole board so you never know whether you actually solved the problem. Try to reproduce the conditions where the unit fails, level, heat and vibration so you see it occur while connected to your test instruments so even if the fault occurs briefly you will have some basic data from which to formulate a hypothesis. The correct diagnosis is one that explains all the measured data. If it does not explain it all, it is not the correct diagnosis. A trap many techs get into is "marrying" their first assumptions which usually accounts for the "dogs" that take hours or days to resolve. When it is resolved, the problem is usually one that was overlooked for hours because it did not fit the first assumptions. Take a set of basic measurements and turn it off, study the circuit and figure out how that set of fault data relate to a cause. If 15-20 minutes of analysis does not lead you to the fault cause, spending 3-4 hours more is not likely to improve that. Put it aside, work on something else and come back to it with a fresh mind. If I run into one that does not make sense, I take more data and leave it for the next day, usually in the morning the diagnosis becomes crystal clear, often accompanied with the sound of a slap to the forehead in how obvious it was. In my old shop, with lots of techs, I encouraged techs to discuss each other's diagnosis or lack of one. If serious effective measurements when the fault is happening does not result in a correct diagnosis I have found that the only causes of that is the tech did not understand the circuit or they formed an opinion too early and stubbornly held to it despite evidence. The 30 minute rule was in effect: "if you do not have a diagnosis that fits the evidence in 30 minutes, get a set of eyes on it by asking for assistance". The physical act of dis-assembly/re-assembly and correcting the identified defect ought to be the longest part of the total repair activity. If it is not, the circuit is not well understood or the tech is not approaching it with an evidence based point of view.
I am writing this about general tech work because the statement of "nasty problem" really stood out as if it was a particularly difficult or mysterious problem. It most likely is, like most defects, very simple and logical.

I would not use the speaker at all until you have identified and corrected the problem that is causing this. You can disable the speaker by simply inserting a 1/4" phone plug into the extension speaker jack.

Thank you for the replies and thank you Dr Gonz for posting the schematic.
I think it is a nasty problem because I cannot readily diagnose the problem with my test equipment because it is only present momentarily at random intervals. If it were steady state I would have it tracked down quick.

So far I have replaced the two very tired looking zeners with 1 watt types spaced above the board.
By inserting a jack into the return socket I have narrowed it down to IC5a and the poweramp.
The voltage at pin1 of IC5 is 0, the voltage at the poweramp output is 0 and not drifting, it just goes bang every so often and the last time it did it I believe i heard the supply rail hum making me think it swung to the rail.
I replaced C42 the Miller capacitor as I have seen them cause this problem before in other amps.
Even if I leave a scope connected while I do something else I still have to be looking when it happens (no capture facility).
Ultimately I will shotgun the small transistors, I have already hit hem all with heat and freeze.

Dr Kaos

[DrGonz78]

Just got to say to measure pins 4 and 8 on IC5 as with all of those IC's 1,3,4,5 are LM1458 chips. Look at the pinout for that chip... We want to know if the preamp 15v rails are intact. Look closer on the schematic to see if the 15v rails are intact as well as all other higher voltage rails. Could be on the preamp side or the output side...? Have we narrowed it down yet for real? Does the amp have a power amp in jack? The return must be what you are referring to and does this nasty problem go away on plugging into the return jack? Just changing parts because that may have caused problems before in this type of amp really is not troubleshooting. Actually it might just be shooting for more trouble... Well it might have fixed the problem, but it didn't. Time to measure the voltage on the chips.