Marshall MG hiss (knobs on 0, nothing plugged in)

[j_flanders]

I bought a used Marshall MG10CD... yeah, I know.
I guess it's one of those pieces of gear, like a DS1 or Metal Zone, that you need to have owned to understand all the hate.  <3)
I have found the "10" stands for: needs 10x less gain, 10x less treble, 10x less output and 10x less hiss.  :duh
I'm planning to mod it and turn it into a head with a Celestion fitted cab but I need to tackle some issues first.

Problem 1: hisssssssssssss
I can probably fix the gain and treble/voicing issues but I'm really struggling to figure out where the hiss (white noise, not hum) comes from. I see other MG owners complain about the exact same issue.
I understand that with lots of gain, there will be hiss but the problem is that even with nothing plugged in, there's quite a lot of hiss. With no gain other than coming from the power amp I don't understand where this hiss comes from.
At the levels I'm playing, turning up the gain to 10 only adds hum, and pretty much no extra hiss.

There's a pretty good analysis and nice schematic of this amp here: https://www.electrosmash.com/marshall-mg10
But unfortunately he made some mistakes when tracing, especially in the tone stack and Sallen Key sections, so the frequency plots are wrong.
R18 and R19 are 10k instead of the 100k in the schematic and he's missing R12 47k to ground right before R13

Link to the schematic in case the horizontal scroll bothers you:
https://www.electrosmash.com/images/tech/mg10/marshall-mg10-schematic-parts-big.jpg



Here's the only correct schematic I have found:
Link to the schematic in case the horizontal scroll bothers you:
http://oi63.tinypic.com/eqrqm1.jpg



With nothing plugged in, the preamp section is muted right before the tone stack, so turning gain and volume controls do not affect it.
If I ground the circuit right before the power amp (R22) the hiss is completely gone.
So, it must come from either the tone stack or those Sallen key filters.
If I bypass/jumper R15 and thereby disable that passive LP filter, the hiss is brighter but not louder so I think I can narrow it down to the tone stack or that first Sallen key filter.
Any suggestions on what to try next? Or any ideas on what could cause the hiss in that part of the circuit?

I have a Vox Pathfinder10 and you can honestly not tell if this amp is on or not when nothing is plugged in. Even with my ear against the grill cloth, one inch away from the speaker I still cannot tell if it's on. So it must be possible to have a dead quiet amp with nothing plugged in. And this amp is pretty bright as well and mutes/grounds the circuit in a similar spot.


Problem 2: OD channel volume knob mystery:
Another issue I can't seem to figure out is the following: on the OD channel, with the volume knob on 0 there is still some sound coming through. It's all treble frequencies, no bass.
When I ever so slightly turn the volume knob, like 1/10th of a mm, the bass frequencies kick in and the sound seems normal but it's actually already too loud by then.
I measured the resistance between wiper and ground to make sure there's no 'residual' resistance left when on '0' and it measures to about 1 Ohm.
One Ohm would give me 1/15000 of the maximum output (load of 10k + 5k pot). Maximum output of that opamp is about 2V because of the clipping leds in the NFB so output from the volume voltage divider would be 0.00013 volts. Can I hear that after the power amp amplified it? How would it explain the missing bass frequencies?

[Enzo]

Hiss comes from gain.  Your PA is quiet, and it is after the mute.  I'd have a new 4558 in that filter before the ink was dry on my computer screen.  A 4560 on your colorful drawing.

Sound at zero on the control?  Probably crosstalk.  The strong signal at the OD stage radiates around, and even though the volume control is down, the sensitive gain stages later in the amp pick up that radiated signal.

R18,19 are 10k on the factory drawing.

[j_flanders]

Thanks for the quick reply Enzo. It was middle of the night so I had to wait until now to try your suggestion.

I replaced the 4560 with a socket and tried a couple of different opamps in there. There was no difference in noise.
Your remark about crosstalk got me thinking that somehow perhaps the noise of the OD channel finds it way through even when nothing plugged in.

I lifted one leg of R46, the 10k resistor that is connected to the volume pot.
Surprise: even when breaking the connection between R46 and the volume pot there is sound coming through.
It sounds the same as when having the volume pot on 0.
I also lifted the negative leg of c10. Same result.
Obviously the volume pot no longer works and it seems the tone control is bypassed as well, at least I can't hear a difference when turning left or right.

Removing the Sallen Key opamp gives no sound, so it isn't finding its way directly to the power amp.

How do I figure out where or how 'it' finds its way?
Do I simply check for continuity between any of the components in the NFB and components after the tone stack? I think my DMM has 50 Ohms as a limit for a continuity beep. (if that matters)

Do I start lifting legs that are connected to the output pin?

Check for AC on components after the disconnected volume pot and work my way back?

I already check with a strong backlight and a magnifying glass for solder or trace bridges on the pcb side and on the component side.

[Enzo]

Crosstalk is radiated for the most part, it isn't about continuity.  Imagine they are playing loud music next door, and you can hear it in your apartment.  You won't find an open window between apartments, it just radiates through the walls.

Your pot on zero, or lifting the resistor to it won't affect crosstalk because they are not in its path.  The stuff before the pot/resistor are somewhere sending out signal, and a later stage in the amp, perhaps that very filter IC, is picking it up.

Is this a real problem?  I mean when you are on clean and PLAYING, does this dirt channel bleed interfere?  Or it is just a matter of you notice it and wish it were gone?

Fix one thing at a time.

Your hiss, so a new IC in the filter changes nothing.  Fine, now we know wwithout further gnashing of teeth.  But with that socket empty is the hiss still there or is it now gone.  A new IC may change nothing, but an IC is still an amplifying element, so it is a clue if the noise comes from before or after that filter.  OH wait, you already said no sound.

How to isolate it?  A scope or signal tracer is good, step along the circuit.  Does the contour control affect the tone of the hiss?  COnnect a clip wire to ground and use it to ground the signal path anywhere that has no DC on it.  Does grounding R18 kill hiss?  R13?  R10?

If you are loathe to shunt circuit directly to ground, add a cap, like a 0.1uf or a 0.047uf in series with your ground wire.  That will shunt signal to ground without creating a DC path.

Crosstalk?  You can use similar tests to locate where the signal is re-entering the signal path.

[j_flanders]

Crosstalk is radiated for the most part, it isn't about continuity.

Ok, I stopped lifting components after those two as I was afraid this was going to be the case and that I'd do more harm than good, eventually lifting traces rather than legs.

Is this a real problem?  I mean when you are on clean and PLAYING, does this dirt channel bleed interfere?

No, it's not a real problem. It does not bleed through. I just thought it might be related to the hiss which I find rather problematic.

But with that socket empty is the hiss still there or is it now gone.

There's a tiny amount of hiss left but it's negligible and normal.
No sound, so the crosstalk happens in the preamp section, not in the power amp section.

Does the contour control affect the tone of the hiss?

It does not affect the hiss. It does not affect the tone of the radiated sound (which is very trebly, lacking all bass)

How to isolate it?  A scope or signal tracer is good, step along the circuit. 
COnnect a clip wire to ground and use it to ground the signal path anywhere that has no DC on it.  Does grounding R18 kill hiss?  R13?  R10?
If you are loathe to shunt circuit directly to ground, add a cap, like a 0.1uf or a 0.047uf in series with your ground wire.  That will shunt signal to ground without creating a DC path.
Crosstalk?  You can use similar tests to locate where the signal is re-entering the signal path.

Thanks, I'm going to try that tomorrow.

In the mean time I traced the PCB to see if there could perhaps be a mistake, but as expected there wasn't any. It might still show where the radiation could be occuring I guess.
I'll post it below, it's rather big.

Another thing I noticed is that the component leads were bent and then soldered. In many places those snipped of and bent leads are right on top of and touching the neighbouring traces. For all I know they might even have protruded, though so far the continuity tester says no.
This seems like a good way to radiate? Is it worth the effort to snip them all off, a lot shorter? Here is a picture to show what I mean:


I guess the real problem is that this amp is too loud. It's actually loud enough with the volume on 0, it's just that it's all treble then. The tiniest amount above that and it gets too loud but sounds normal. I just need way more range on the volume knob.

Can you suggest some solution for this. I could go ahead and insert some voltage divider here or there but isn't that going to mess up input or output impedances of certain sections, or affecting the frequency response of the tonestack or those filters.

I already tried a lot of things:
1) Double r46 to 20k
2) Lower the 5k volume pot value by putting a 1k or 2k resistor parallel to it
3) Replace the clipping LEDS (1.8Vf) by 1N4148 (0,7Vf) and because of the lower clipping threshold I raised R8 to 5K and replace R4 with a 5k pot to have less (treble) boost.
4)Connect a cabinet to the headphones output. The current limiting resistor is probably rated for 16Ohm headphones rather than 8Ohm speakers but they must have taken 8Ohm HP into a account. It hasn't burnt through. It was too muffled sounding so I changed c17 to a NP 6uF cap.

But I would still like to have a simple voltage divider somewhere that does not affect the voicing.

Thanks for your time and your suggestions Enzo!

[j_flanders]

PCB tracing. (scroll to the right for the preamp section)

[phatt]

Try lifting one end of R7 which looks like it reads 820 Ohms.
That should turn down the gain a fair bit without changing the tone too much.

If the hiss is still evident then you can always raise the value of C4,, it's 10pF on the schematic but I'd bet a 1nF cap there will kill the hiss and all the trashy hi freq that is often in these amps.
If it's too dark then try 470pF. You can leave the original 10p cap on the board and solder the new cap on the track side to test.

If you want more bass then make C3 larger,, try 1uF. again you can leave the original on the pcb and solder on the track side.
Phil.

[j_flanders]

Thanks for the suggestions Phatt. I tried some of those previously. Slightly revoicing the amp towards less gain and less treble won't be a big problem I think. But I will keep them in mind.

For now I've given up on the hiss. Maybe attenuating some high frequencies in the revoicing process will cure some of that.

At the moment the volume control is my biggest issue.

To track down the 'crosstalk issue' which is part of that problem I did and discovered the following:
1)Recorded some guitar playing into a looper and hooked it up to the guitar input of the amp.
2)Set my DMM to VAC and checked every point along the signal path.
Results:
I see fluctuating voltage (corresponding to the playing looper) up until the leg of volume pots (both clean and OD channel).
I see 0 Vac on the wipers because the volume knobs are on zero and thus wipers are connected to ground.
After the wipers all the points along the signal path show 0 ac voltage until I arrive at R32, the first resistor of the Zobel network, right before the speaker. That is the first point in the signal path where I see a small ac voltage again and it corresponds to my playing looper.
I did skip some caps when testing because I was following the path at the component side and the legs of those box caps aren't accessible there.
I removed opamp2 (in the filter section) from its socket. There is still sound coming from the speaker but it's quieter (even though I saw 0 volts ac on out1 and out2 when the opamp was in circuit)
I have no idea what's going on here!

Now for the volume pot problem it gets even weirder.
One of the things I tried in getting more usable range from the volume control was to insert a 100k pot in series right before the volume pot (after R46 10k).
I realise this changes the load for the opamp and will probably also affect the contour control coming right after that.
Now for the weird thing:
Previously I had this trebly sound with the volume knob on 0 and a contour control that didn't affect it when turned.
Now I experience the same thing but with the volme control on 1 or 2.
So I'm back where I started: having to turn up the volume control to get a 'normal' sound, which is then too loud.
I also noticed the sound gets bassier the more I turn up the volume pot. And the trebblier the more I turn it down, until I get to 0 and then there's only treble left...

So, please help me out here. Where do I insert a voltage divider that gives me more range on my volume knob?
I could place it after the tone stack but it would interfere with the input resistors of those Sallen key Filters.
I could place it right before the power amp but there it woud interfere with the input resistors of the power amp.

Is a buffered volume control the only way here? What else can I do?

[j_flanders]

In case anyone wants to tinker with a simulation of the circuit, here's a link for an online app (http://www.falstad.com)

Preamp section frequency response (no option for adding clipping diodes, or log type pots):
https://tinyurl.com/ybvl5aod

Power amp section (no idea how to include the speaker (inductor?) in the current feedback loop. I just used an 8 Ohm resistor.)
BTW, the spec sheet says typical gain of around 26, but this seems to have way more. Anyway, I don't understand a lot of what they did there and haven't found any other examples of using this power amp in inverting config.
https://tinyurl.com/ycu5denz

You need to have the Java plugin active in your browser.

Some screenshots below.

[phatt]

Re the Volume problem;
I assume the clean volume has no issue?
The clean already has substantial gain so when OD is engaged the gain then goes off the scale.
WHY,, because R8 is way too small.
The gain of that stage is likely  over 800 (I think) and you already have 10 times gain at input so the outcome is what you have noted,, stupid.

If you bridged R8 it would likely squeal or just not work.
There is a limit to how small the series input resistor can be with inverting opamps, at some point it all falls over, R8 needs to be at least 1k.

I'd try 5k then stage gain would then be,  500k/5k = 100

Oh that hiss you are chasing is likely Partly due to this stage being close to unstable (and what Enzo noted above) :dbtu:.

I'd add;
Bad pcb layout causing cross talk.
Pushing inverting stage to extreme causing a leak back through a ground path that is not well designed.

AND** R34 OR1 is not part of the Zobel network ,, that is the CS resistor (Current Sensing Resistor)
Which is used for Current Feedback,, a common trick to fake a Valve sound in SS amps. But it can cause issues like this.

I can't be sure of the power stage as the schematic is unclear in that section. So I don't know how that is working but R34 is most likely CS.

If the slight darkening of tone at higher gain is still present after reducing the gain by raising R8 then look at C8 220pF, making that a smaller value will help.

Gain for opamps help here;
https://en.wikibooks.org/wiki/Electronics/Electronics_Formulas/Op_Amp_Configurations

Cheers Phil.