marshall 4210 jcm 800 help!

[alchemy]

I just acquired an 83 jccm 800 4210 1x12 combo, unworking, but for a song. Replaced the filter caps and it worked but sounded like crap. Replaced all electrolytics and did an el 34 swap (bias resistor and feedback wire),replaced a bunch of other caps that had drifted off value, matched the PI resistors and caps, checked out every diode and transistor, swapped a couple wires (plate and grid from v3 that were backwards), removed a resister soldered to the undrside of the board going between a cap and a transistor (cant remember which right now)and still it sounds like crap. So here's what it's doing: It has the common channel bleed that these amps get,(on both channels) it has a very high pitched shrill out of tune sound. Plucking single notes it is less but still there, while playing chords sound just awful. I am beginning to think it is a parasitic osscilation, but having never ran across one before I'm not sure. I tried moving around all the wires i could while it was being played and noted no differences. Checked all of the grounds with a meter and found no high resistance. I feel like it is somehow related to the gain circuit, as it gets much worse as the gain is turned up (maybe it's just amplifying it more) and if i touch the gain knob I start picking up the radio. Gain pot is also scratchy. All tube have been swapped, and i also resoldered all of the pot connections. So I guess you could say I'm at a loss....Any help steering me in the right direction? Oh yeah, here are all of the measurements I took. I'm leaving out the heater voltages etc. A couple things looked off to me, but I am still learning so I'm not sure....

Edit: heater voltage is 6.7-6.8v, not too high i don't think (wall voltage fluctuates but is steadily 2-3 volts over 120v. Also, I removed the feedback wire and the channel bleed went away, maybe almost completely. The clean channel now sounds pretty good, but the boost channel still has some weird "out of tune" qualities, and slight quite distortion underneath the notes played. This is subtle on higher notes, but on say an open E or A it is pretty bad. Just measured the choke as well. 112 ohms. Looks close to the specs on replacement chokes i've seen, at least for resistance. Is this the proper way to test a choke?

Any and all help would be most appreciated!!!!
V1
1-197v
2-0
3-1.45
6-169
7-0
8-1.5
v2
1-141
2-0
3-1
6-295
7-141
8-143
v3
1-196
2-0
3-1.5
6-155
7-0
8-.9
v4
1-454
2-0
3-3.4
6-169
7-0
8-1.25
v5
1-254
2-30
3-51
6-245
7-32
8-51
v6
3-466
4-460
5--44
6-465
v7
3-467
4-459
5--44
6-462

[g1]

Those amps had problems with the switching circuits, sometimes weird problems, maybe like this. I think that resistor to the transistor may have been a mod. to deal with the channel switching issues. Can you put it back in and try?
Otherwise you may have to try to hardwire around the switching scheme and see if it still has problems in the gain channel.

[Enzo]

My first thought is that the signal path is interrupted, and the tinny sound you are getting is just the remaining crosstalk through the amp.

I would apply a steady signal to the inoput and trace it through the amp, looking for a place where the signal drops when it ought no to.

[alchemy]

Well, i may have found the problem, though i could only play it for a minute so i'll have to check it more thoroughly later. Last night I cleanned up all of the chassis grounds, and it seems to have gotten rid of most of the ugly subnotes. I had previously checked the grounds with a meter, but apparently that doesn't always work!!
As for tracing the signal, i have never done it, but recently aquired piles of test equipment so i do have the capability, assuming i can figure out how to set it up. I'm assuming that i would need to put the amp on a dummy load (which i dont have) and then put either a signal generator on the input (what frequency?) or just play guitar into it and then either listen with a signal tracer or hook it up to a scope. I have both a tracer and a good scope, but i'm not sure how to set up the scope for this so i think the tracer is the way to go. From what i can tell i just clip the ground to the chassis and then touch the probe anywhere i want in the circuit and listen for a change/abnormal noises/etc...does it sound like i'm on the right track? any and all help with testing procedures would be most welcome. I am trying to learn this stuff by reading books mostly, but sometimes a little human instruction goes a long way...cheers!!!

[alchemy]

perhaps someone could tell me how to post sound clips and it would be easier than just trying to describe the sound...

[Enzo]

If my schemario is correct, and there is no guarantee of that, you would be looking for a gross loss of signal between two points. Signal tracer is fine, scope is fine. A steady signal like froma generator is easy to trace, but I often just use a music signal from the shop stereo. Generator frequency? Doesn;t matter, whatever you are comfortable listening to. 1kHz is a common test signal, but I find it shrill and very tiring. even 400Hz I find tiring. I usually use 100Hz unless I have reason to go otherwise. I can listen to 100Hz for a long time.

A dummy load removes the ear fatigue, but a speaker is OK. If the problem us before the FX loop, just plug something into the FX return to silence the power amp.

[alchemy]

cool thanks for the help. the signal tracer i have has an external speaker, no earphones, so i was thinking that the sound coming out the amp speaker might drown it out. that's why i was thinking of a dummy load, but i'll give it a shot. to make sure i'm clear, i should hook the ground up from the tracer to the chassis and then probe the input circuit all the way through, correct? is there anything that i specifically DO NOT want to probe with it hooked to ground? sorry for all the questions, still learning, and up until now i've always just used a meter, a chopstick, and my eyes and ears, but this one's really got me stumped!
on a side note, are there any good books dealing specifically with tube amps that explain how to set up a scope and troubleshoot?

oh yeah, one last thing: is it normal to have 30+volts on the grids of the phase inverter? everything i read specifies 0 volts for preamp tubes and bias voltage for power tubes but does not specify what the PI should be. I know the PI is not a preamp tube, but it is cathode biased so shouldn't it be 0volts, or am i missing something?

thanks for the help. it is much appreciated...

[Enzo]

Your signal tracer is just a little amplifier for listening. It should have a capacitor at its input blicking any DC voltage from the circuit under test. SO you should be able top probe anywhere you want.

What is bias? It is the voltage difference between grid and cathode, not grid and ground. That is why in power tubes we can either ground the cathode and put a -V on the grid, or we can put the grid at ground, and make the cathode +V. Either way there is V difference between grid and cathode, and in both cases the grid is more negative than the cathode.

In most preamp circuits, how is the tube biased? Cathode bias, meaning some +V will appear on the cathode while the grid is at DC ground potential. How does that +V get there? BY way of voltage drop across the cathode resistor. The voltage drop is caused by the current through the resistor and tube. Ohm's Law.

A typical 12AX7 triode will be conducting maybe 1ma, more or less. A typical cathode resistor is 1.5k ohms, 1500 ohms. Assuming 1ma of current, that means V = IxR = .001 x 1500 = 1 volt. SO in a typical preamp gain stage we se about +1v on the cathode and zero on the grid for a grid bias of -1v.

Now look at your PI stage. First, the two triodes share the same cathode resistance, so the currents of the two are added up. And that means the voltage dropped across that common resistance will be twice what it would be for one triode. I see from cathode to ground: 470 ohm, 22k, and 4.7k, and we will ignore the 100k from the NFB in parallel with that 4.7k. Adds up to 27170 ohms. ABout 27k.

Assuming about 1ma per triode, we then have 2ma flowing through the 27k. 27000 x .002 = about 54v. My rough guess then is about +54v on the cathodes of the PI. How close am I? Depending on many factors, it could be 60v or 40v, or whatever. But let's call it 54 for now.

Now look how the grids are terminated. There are two grids, with a 330k and a 120k grid resistor respectively. They are not connected to ground, they are connected to the bottom of that 470 ohm resistor. Remember we have about 2ma flowing through that resistor, so V = 470 x .002 = 0.94v. The bottom of the 470 ohm resistor is about 0.94v less positive than the cathodes. Let's call it 1v. Easier to write. That point is then +53v, right?

The two grid resistors connect from that +53v point. No current flows through those grid resistors, so according to Ohm's Law, there will be no voltage dropped across them. There is +53 at the bottom of the 470, and it should be there on the grids too.

So +54 on the cathodes, and +53 on the grids, for a net bias of -1v. Pretty typical.

But wait. You measured +30 or something on the grids. Not +53.

It is your meter. It is loading the circuit. Nothing is wrong with the meter other than how we used it. We are used to thinking our meters are uninvolved observers, but they are not, they have resistance themselves, and often it doesn;t matter, but in this case it does.

I don;t know your meter, but let us assume it has a 1meg internal resistance. If I conect it between ground and the left grid, we wind up with a circuit from the +53v point through the 330k resistor, and then through the 1 meg of meter to ground. So the meter and the 330k form a voltage divider. Total resistance would be 1.33meg or 1330k, however you like to think. The red probe of the meter then will see 1meg/1.33meg x 53v, when we solve the voltage divider. That means about 39-40v. So in this experiment, the meter would read 40v instead of 53. The other grid should read a bit higher since th grid resistor is smaller.

The trick here is to measure the voltage across the 330k resistor itself, and of course the same way on the 120k. Now instead of forming a voltage divider, the meter is simply parallel the resistor. SInce no current flows through the resistor, there will be no voltage drop other than leakages. SO you take the reading, and calculate the grid voltage based upon voltage drop from that 53v point. Not much I bet. And that means you really have close to +53v on those grids.

All of which is to say, I bet your grid voltages there are fine.

[alchemy]

damn. thank you. that finally makes perfect sense. i'll have to recheck the grid as you desribed, but one thing i do know is that we are looking at different schematics. mine has two 10k's after the 470k, and the 330k and 120k on the grids are both 1M. i'll do the calculations and remeasure it correctly later on, but i think you are right, that it is ok. i just wanted to know why that voltage was there, and you explained it perfectly. thank you.

[alchemy]

so i did the math and came up with 40v on the grid (not taking into account the meter since i don't know what it is.). I'm getting 30 and 32 on the grids so that seems about right.
checked through the circuit with the tracer. the only strange thing that i noticed was it seemed i was getting gain and distortion on the normal channel when i shouldn't have been. i played the amp for a while and here's what i noticed: while on the normal channel if i turn the gain knob on the boost channel it seems to effect the low end of the normal channel. to the point that about 3/4 of the way up it vibrating everything in the house, sort of like runaway bass.
on the boost channel, there doesnt seem to be any bleedthrough from the normal channel. however, if i run the master all the way up, the volume rather low, and turn the gain knob, the harsh unpleasant out of tune distortion is very apparent. also, when touching the gain knob it will pick up the radio, as well as pop and crackle as i wiggle it back and forth (seems to be in time with connection to the chassis being broken and remade.)
Is any of this helping to make it clearer?
The problem seems to be with the gain circuit but i'm getting confused about what might be causing it. Can diodes and transistors go bad in other ways that do not cause them to read bad on a meter?
is

[bluesfreak]

Sounds to me like your gain poti in the boost channel losses ground and thus it crackles and picks up radio. Sometimes in these amps the BC184 which are uswd for switching do fail and do cause strange symptoms as they might not fail complete but partially, also the 33µF cathode caps might worth a replacement, me personally I do like 25µF there as a replacement but your mileage may vary...

[alchemy]

yeah i've been going over the schematic again and again and the only places i can see for the signal to bleed through is at those transistors that switch the channels. i had pulled them and checked them and they tested ok, but maybe it's possible that they failed in a way that wont show up with a meter. the thing that seems wierd to me though, is that it seems like where they are in the circuit that if they were allowing some of the signal from one channel through to the other that it would be the whole signal and that it would be evident on all of the controls, not just the gain control. should i just try replacing them?

is the pot grounded through it's case to the amp chassis? doesn't look like it from the schematic or the pcb, but it's certainly acting that way.

which caps are you speaking of? i have some 330mf on my schematic on some of the cathodes in front of the switching transistors,but no 33mf. Actually, i can only find one 330 in the amp, the others have been changed to .1 and .22 (this is noted on the schematic, i think it came from schematic heaven and seems to match up pretty good). Thats quite a difference from 330 to .1. Why would they change it by that much, and why is one a big electrolytic(330)and the others (the smaller ones) are the lego caps?

thanks for the continued help folks.

[Enzo]

I was looking at the 1987 or 1988 version from my Marshall MAster book. Identical to this: http://www.schematicheaven.com/marsh...h_50w_2205.pdf

But the discussion still applies, roll with the changes.


The pot that crackles and hums when you touch it? Tighten the nut. See if the rear cover is coming loose.

The only place something can bleed through? How about the air? Two circuits only inches apart can radiate one into the other. It is called crosstalk. On a poorly designed mixer, you could put a hot signal into channel 1 and bring up the fader on channel 2 and hear some of channel 1 there. Crosstalk.

On the clean channel it is easy to pick out the distorted boost channel sound, but when on the boost channel, how would you know if a little clean were leaking into it? It would disappear into the dirt.

[Ronsonic]

It has been awhile since I worked on one of these so please excuse any haziness. Channel bleed is a known issue for them. If you google the more common head number 2210 bleed problem, you'll get a lot of hits. I seem to recall modding the early version to the later 1988 spec to solve a lot of that. Also ISTR that it behaved differently with a footswitch than without. Korg tech support may even have info on this problem it was so infamous.

Common problems were in the voltage supply to the switching circuit, the cap and rectifiers would go bad and inject some hideous noise, warbling and other artifacts that modulated with the signal. And yes that "IC" transistor array was as subject to failure as any other solid state device buried in the guts of a screaming tube amp.

I'll email myself at the shop and when I get in monday I'll see if I left any notes or a trail of bread crumbs.

[alchemy]

Enzo:
"The only place something can bleed through? How about the air? Two circuits only inches apart can radiate one into the other. It is called crosstalk. On a poorly designed mixer, you could put a hot signal into channel 1 and bring up the fader on channel 2 and hear some of channel 1 there. Crosstalk."

Sorry i guess i left that out in the last post. I meant the only place i can see in the schematic. I'm well aware of parasitics and initially (and still somewhat) thought that was the problem. I tried moving wires all over the place as best i could without cutting the rubber bands that hold the bundles together, and heard absolutely no change. I'm trying to take it one step at a time and rule things out as I'm still on the far side of the learning curve with this much circuitry. I have some sort of gut instinct that it has to do with the solid state switching set-up, at least partially. But if that does not remedy the problem the next step will be to cut the rubber bands and completely redo the lead dress, get the grid wires short and on different planes, etc, etc...as for not hearing the clean bleed through to the dirty channel i expected that, i was just surprised that the tone knobs had no effect on either channel.
That schematic is the later version. The one i have is here: http://www.schematicheaven.com/marshall.htm and scroll down to jcm 800 4210 reverb
Ronsonic:
This is not the IC version but the earlier version with only transistors, but I would appreciate any notes you have on these beasts. I'll check out the 2210. It seems from what i can tell that these were common problems, but most searching i've done just leads to threads that were A:inconclusive, B:folks started hacking the amp apart, C:completely rewired it PTP,or mostly D: "i ripped out the diodes and transistors and it sounds like shit. marshall channel switchers suck. i sold it and spent $$$$ on an "all tube" amp and put a diode filled pedal in front of it." I'm all about older PTP "all tube" amps (that's really all i've ever worked on), but I feel like this has a freakin sweet little beast of an amp hiding inside waiting to come out.
edit: i think the only 2210's are later versions as well, but perhaps the same tips will apply, though i seem to remember lots of folks replacing the IC chip which i don't have...

Thanks alot for the input guys, and please keep it coming. I am learning an immense amount through this troubleshooting process. That's the only reason I have yet to get frusterated...