jcm800 2204 copy: very harsh distortion

[vanslycken]

I copied the preamp circuit of a 2204 JCM800, except for R8 (voltage drop resistor for V1a and b). (The power amp is more or less a standard 4 tube pp amp.) The distortion sounded terrible, very harsh. Apparently the third stage was driven too hard.
I can't make out from the schem what the B+ is on these amps. It was about 400V on mine, I've lowered it to 350V but the sound was more or less the same.

Now, I've solved the problem but I'm wondering how come this circuit didn't work in my setup.

The way I solved it was by removing C2 (100p cap across V1a) and making C1 33µ (grid resistor for V1a) as V1a could handle some more amplification. Then with a 22n/1M to V1b (no volume pot, just fixed resistor). V1b overdrives mildly. Then I changed R11 to a 25k pot, so at maximum only about 1/20th of the signal gets fed to V2a. Turned V2b to another gain stage (100k/820R). EQ after V2a, after the Eq a voltage divider (220k/100k) so about 1/3 of the signal goes to V2b. Sounds good now, but it's not quite the same circuit.

My main question remains: why didn't the "stock" circuit work here? Master volume was down to "civilized" levels so I doubt it has anything to do with the PA because that was still in normal operation (not overdriven).

[txstrat]

Well, did you make all changes to the stock circuit at the same time?
Do you still know - if you have checked - what was the change that tamed the circuit?
Did the harsh-distorted sound change when the vol pot was full open? I'm asking, cause I found this in one of my builds too. When I turned the pot full open the harshness was completely gone - same when I turned it down to 6 or less. I suspect the pot but haven't changed it yet.

[vanslycken]

I have tried the volume full open and everything inbetween. The harshness went away when I changed R10/R11 for a 1M potmeter and turned that down until I got a pleasant sounding overdrive.
That overdrive wasn't distorting enough so that's why I added the extra gain stage. I put the eq before that last gain stage because I thougth it couldn't do any harm to load the 3d gain stage down a little - and I pot another 1M pot afer the eq. That way I tried to find the best sound.

[txstrat]

I assume you're looking for a sort of vintage circuit with master volume. Did you have a look at the JCM 800 lead 100w (2203)?
http://www.schematicheaven.com/marsh..._100w_2203.pdf
It's basically a standard vintage Marshall circuit with a preamp and a master volume plus two different input sensivities.
The low input skips tube V1A while high adds V1A with a preamp vol pot. I've copied this circuit twice and with a bit of tweaking the coupling caps I found a great "sort of" Marshall sound. I highly recommend this circuit for Marshall fans who like preamp distortion and a master vol for the convenience of being not too loud.

[vanslycken]

Sorry for the late reply (I have been busy).
It is this circuit that I built and i used the high input.
If you have cloned this circuit, can you tell me what your B+ voltage was? I think in my amp it might be a bit too high.

[txstrat]

I think in my amp it might be a bit too high.

No, I don't think so. My circuit is running at 423v (no load) and some 390 volts with tubes plugged in. Never had a problem and the amp is almost 17 years old. What I'm trying to say is, I don't think it's the voltage. Have you doublechecked all the values of caps and resistors and tried a different pot?
That's what I would do.
The voltages might vary a couple volts (I'd say 20v - 40v more or less wouldn't make a noticeable difference) so I think it's not the missing voltage drop resistor.
Also cold solder joints can result in harsh distortion.
I would assume C2 was there to prevent the amp from oscillating. No reason for harsh distortion, quite the opposite.

[vanslycken]

Thanks for the reply, that rules out one of my concerns.
Perhaps it's the fact that I'm trying to get preamp distortion at low volume, I suppose the real JCM800 sound is more power amp distortion...

[txstrat]

I wouldn't necessarily say so.
The low input is sort of a bassman copy. The high input adds a 4th gain stage in front of that. With the preamp volume you dial in the amount of drive you want. Open the preamp vol pot and you get lots of preamp distortion (not metal, though) that you can expierence at bedroom levels with the master volume turned down. This circuit is sort of made for this.
I think there must be something else that's not right with your circuit.
When I encounter problems like this I follow every lead again and again and doublecheck all the solder joints too.

[vanslycken]

Well, I have started from scratch again, rebuilt that circuit, same thing happens.

[txstrat]

Have you tried other tubes and a different speaker as well? - that's one thing.
2nd: As I mentioned before the amp might be oscillating ultrasonically (that's why C2 was there) - you could put c2 back in and try shielded wires to the grids of the first stages to prevent picking up something.
Could you post a layout of your amp innards?

[vanslycken]

I want to use this amp with my own speaker cabinet so I haven't tried other speakers. I've used a couple of tubes but all I have here are sovtek 12ax7wa. C2 was there in the original build but removing it did very little.
Even at very low master volumes it distorts quite ugly.
Now, meanwhile I've installed a grid leak resistor of 470k after my drive pot (which is still 25k but I've omitted the 470k before it), it distorts quite nicely now.

As for the layout, it's all air soldering for now, I try to make things sound right before I solder them to turret boards.

[vanslycken]

just a little head's up on my head :-)

I finally got the old oscilloscope out and measured around. Turned out that the ugly sounding distortion was the PI (long tailed pair with 12AX7) distorting hard. I hadn't used the feedback circuit as used in an original Marshall, so an input signal of 5Vpp was the absolute maximum before overdrive.

Because I've made some other changes making an effective feedback loop impossible, I've used a cathodyne PI which handles the large input signals very well. Problem solved!