Marshall JTM45 Problem

[steppenwolf]

Hi Guys!

If you are measuting 900 ohms from one leg and 70 ohms from the other then there is something wrong. Also these two readings should add up to equal the total resistance between the two legs (was this 1600 ohms?)

No, it was around 90 Ohms to 70 Ohms, so that should be fine! Primary Raa reads 7600 Ohms, should have 8k, so it should be allright!

That does look strange, some crossover notch, but also some breaking thru at the very peaks. Maybe a layout problem, it does kinda look like positive feedback, but if you swapped the wires and it squeals then definitely not at the OT. You could try to move some of the power tube grid wires or PS plate wires, feedback wire from speaker, even back as far as the tone stack output....keep going back and reposition wires and see if anything changes. Or try a 12at7 and see if it gets better or worse, or load the secondary down 1/2 and see what happens to it, or disconnect the 47 pF, see it it clears up.....try and find something that affects it.

I tried to move around some wires, put out the 47pF Cap, but that doesn't really change a thing.
Then I tried to disconnect the NF and putted in a alligator clip wire. Then I took the wire and moved it around the chassis, and that changed the sine considerably...That brings me to the idea, that it could be some canceling out somewhere.

The wires from the PI to the grids of the KT66 are inside a shielded cable.
Could that be a problem. They are really close and parallel in there and they are out of phase...Moreover there a are pretty high voltages passing through there. That could explain, why it occurs only on higher frequencies, as those are more prone to cancelling out I think. It starts to develop that ugly notch just as the sine begins to cut off maybe because distortion causes overtones...
What do you guys think about that?

I'm thinking power supply...
Have you tried to monitor the B+ DC voltage while increasing the volume? How much does it sag?
Did you check the rectifier tube?
Are you getting full wave rectification i.e. ripple freqency at B+ should be 120Hz. Does it look on the scope like a nice sawtooth wave?
If your rectifier is bad it could be that you're not getting enough DC current out of power supply once the power tubes start pulling more than bias current.

I tried to monitor the B+. It is around 435V at idle and about 390V a full throttle, it gradually decreases with the volume going up...

Thanks, Stefan

 

[dabo]

Check the output tube sockets for burns and anything that looks wrong. I've had a few quirky problems from arcing depositing metal from the tube pins on the sockets.

 

[Junction]

[quote author="steppenwolf"]Hi Guys!
No, it was around 90 Ohms to 70 Ohms, so that should be fine! Primary Raa reads 7600 Ohms, should have 8k, so it should be allright![/quote]

No I don't think that does sound right, I would expect maybe only up to 10% variation ... 90ohms on one half to the center tap and 70 ohms on the other half could indicate a shorted turn or two in the primary.

I think you really need to isolate whether it is your OT or not by swapping with a known good one. You could do the same check on a good one to verify the normal measured resistance before ripping it out to replace yours.

Michael

 

[plexi]

You got it to change by moving the feedback wire around....Is there a position for that wire (feedback connected) where the notch goes away completely? Like up in the air away from everything...It may be worthwhile to unsheild the 2 grid wires and lay them under the board up against the chassis and see if that makes any difference. Sheilded wire is not really needed at all at that location because the signal level is very high and hum/interference is never usually a problem. Capacitive coupling between the 2 grids is not neccesarily what you want.

Keep at it, you are getting closer to the real problem.

I would expect Mercury to have a tighter DCR between the 2 sides, but as the transformer gets physically bigger, the difference is more because the mean length of turns is greater for the outside primary half than the inside primary for a given turns ratio. If they used a smallish guage wire, (like original) then even more DCR difference. Plus it is 8K instead of 4K so more turns for required primary inductance.

You can inject signal to the secondary and see if you see the same AC SIGNAL swing on EACH primary half. Regardless of DCR, these should be very very close.

 

[steppenwolf]

Hi Guys!

I just replaced the OT to a new JMP 50 Watts that lays around and has never been used... The notch DIDN'T disappear...It's still there, nothing changed...
And I must say that I am pretty happy about that as the MM tranny was very expensive;-)I just send an email to Paul at MM if this difference in DCR is OK. Let's see, if he can tell me more about, usually there are very friendly there!

I feel as if I'm getting near the problem, I'll try to unshield those wires and lay them out on the chassis, let's see if that helps...

The NF wire does change the sine, the notch persists but it changes...

I really hope I'll get this fixed as I love the JTM45 sound very much and I have it at reach!

Have a nice day and thank you so much for support!

Stefan
EDIT:
Paul @ MM said my readings are allright
I put a sine into the secondary and scoped the primary with two probes, the signals are spot on, same amplitude (phase reversed) at all frequencies...

 

[pucho812]

Check the rectifier.... :wink:

 

[steppenwolf]

I have a diode rectifier on a tube socket that I could drop in just for a test, will try that. Hope I will not damage anything due to the expected higher voltages with a solid state rectifier...But my electrolytics are rated @ 500V, they should handle the power!

Have a nice day!
Stefan

 

[steppenwolf]

Hi everybody!

Took me some time to update, but work was hell the last week...

I swapped the rectifier tubes, nothing changed, I unshielded the wires going to the power tubes 5,6k grid stoppers, and the notch is still there...

I really don't get it, I changed the wiring, pretty Marshall standard but nothing...

As the problem seems to be somewhere behind the PI, and I excluded the OT by changing it and the power tubes by changing them, too, I don't have any possibility left, anybody an idea?

The JTM45 is definitely not rocket science, but it drives me mad right now!

Have a nice day!
Stefan

EDIT:

I just monitored the BIAS voltage before the 5,6k and had a reading of around -50V at both tubes...That should be ok
When turning up the volume knob with a signal applied, the voltage slightly changes to around -52 or so on both, and stays pretty stable even at higher volumes, but than the notch appears with the volume getting higher and the BIAS suddenly increased parallel to the appearence of the notch to about -65V. But I changed the volume from 6 to 7...
On full it measures around -73V one one tube, the other one measures -61V :?:
The voltages stay similar till the notch appears and than increase dramatically but not equal on both sides...

Is this normal???

 

[CJ]

sounds like you are starting to lose your bias as the pwr increases.
check all the grid stoppers, sometimes, if a tube has shorted grid to plate, the resistors can get burnt out of value.

do the plates glow red at top volume?

 

[AnalogPackrat]

You know, I just solved a very similar problem on my silverface Deluxe Reverb. I did most of the blackface mods but it would get spitty/farty at moderately high volume (with single coils), especially low notes. Ugly. My recently modded Vibro Champ was kicking its ass tone-wise.

I ran it into a resistive load and scoped it as I cranked the volume. Sure enough, beyond a certain point I got nasty coring--what looks like severe crossover notching. I couldn't get it to go away.

Well, tonight I pulled the chassis again to look for problems. Stared at the schematic again. Decided to reverse one of the changes in the PI--namely changing the 1M resistors from grids to cathode (actually to between the 470R and, in your case, 10k connected to the cathodes of the PI triodes). Being lazy, I just tacked another parallel 1M on each of the existing 1M. Problem solved. Tone heaven. Try it. That JTM-45 schemo looks an awful lot like my Deluxe Reverb.

<edit><edit again--too much beer>

A P

 

[steppenwolf]

Thanks for your advice!I'll try that after work!

Do you know why this happens? I mean, many of my pals have the same schematic but don't have those problems... :?:

 

[plexi]

Steppenwolf,

I'm still here but I have nothing for you......I hope you can get this fixed, the bias will increase negative when you drive the power tube grids, that is normal.

Some crossover notch is normal too,

What isn't normal is the breaking through at the peaks of the waveform as your scope picture shows....what happens when you drive it into cliping hard? Does it square up ok?

Did you ever try EL34's and run 8 ohm load on 16 ohm tap?

 

[steppenwolf]

Hi!

I tried to parallel two 2,2MOhms to the 1mOhm in the PI, didn't change anything...

.I hope you can get this fixed, the bias will increase negative when you drive the power tube grids, that is normal.

I thought that, too, but it increases unsymetrical, meaning that I have a reading of -65V on one and -55V on the other...I tried differnet tubes, even KT66 types, but it is still there...Is this normal?

What isn't normal is the breaking through at the peaks of the waveform as your scope picture shows....what happens when you drive it into cliping hard? Does it square up ok?

Sorry, what exactly do you mean by breaking through?
When I drive it hard, it is squared out, but a the top and the bottom, it isn't a line, it's a fat greenisch line for lack of better words, I thought that would be caused by the hum sitting on top...

Did you ever try EL34's and run 8 ohm load on 16 ohm tap?

I have a MM OT, it is really hard to change impedance as you have to resolder and reattach different cables...
I have 500Ohms screen resistors, it could be unstable with EL34?

 

[AnalogPackrat]

I guess your problem isn't the same as mine. I think I was getting blocking distortion at the PI. Sounds like yours is HF oscillation (a.k.a. breaking through) which is what makes the line on your scope trace fuzzy. Zoom in and you'll see a HF waverform superimposed on your actual shape.

It's probably a lead dress issue. Can you post pics of the wiring around the power tubes and PI? You could also try the bandaid of putting small caps from power tube grids to ground right on the socket. Start with 100pF and increase until the oscillation stops. This will start to be noticable once the caps get above maybe 500-1000pF, so you're better off fixing the lead dress.

Good luck!
A P

 

[steppenwolf]

I guess your problem isn't the same as mine. I think I was getting blocking distortion at the PI. Sounds like yours is HF oscillation (a.k.a. breaking through) which is what makes the line on your scope trace fuzzy. Zoom in and you'll see a HF waverform superimposed on your actual shape.

Guess so, too...When I zoom in, I can't see any superimposed waveform but if I zoom out, I can see the sine "dancing" at around 100 HZ, zooming in again, this "dancing" becomes the fat greenish lines...

If I put one cap from one grid to ground, the notch gets much more present... :?:

 

[AnalogPackrat]

You used 100pF, right? Both grids?

When I say zoom in I mean use delayed sweep if your scope has it. If you just try to mess with the main timebase you'll lose sync on the main waveform. What kind of scope are you using?

A P

 

[steppenwolf]

You used 100pF, right? Both grids?

I tried on both ends, but it didn't really change anything, the waveform gets cut a little, but the notch is still there, and also the fat lines, even if I use 330pF...

I'm using a Hameg HM412, it has delay option....

 

[CJ]

you should be able to dial out the notch with the bias control.
to get rid of the notch, lower the bias, so you are drifting into A/B land.
If that does not work, we can run it class A to make sure it isn't something else. Things might get a little hot, but no biggy we just need a few seconds to verify...

use a sine wave, not guitar, when scoping.
and put the scope at different places.
check the screen
check the grids.
check the wave form across the speaker.
be careful not to blow up the scope, you really need a 100:1 probe for marshall work.


funny, when i bias my amp for best sound by ear, i look at the results on the scope, and its notched. so a thousand theories vs one experiment...

:roll:

the bias difference is tube mis match at the high end.
you are starting to get grid current flowing in the opposite direction.

 

[AnalogPackrat]

Have you used the delayed sweep on your Hameg before? If you don't have the manual, you should be able to find one on the web somewhere. That will let you get a closer look at the fuzzy part of the peaks of your main output waveform.

So it sounds like you have two problems. One is what looks like crossover distortion (but may well be blocking distortion at the PI like I had) and the other is this oscillation thing.

Looking at the crossover notch thing, I would still try temporarily knocking down those 1Ms to half or even 1/3. Paralleling 2M2 only took you down to 666k. Just do a temporary job and tack a 1M over each existing 1M. Or even a 470k over each 1M. We're just trying to figure out where the source of the problem is.

On the oscillation front...can you post a photo or two of the wiring around the power tubes and PI? Can you try JJs KT66 instead of the Chinese ones you're using?

Just for educational purposes I see a few interesting differences between the Fender and Marshall PI & power sections. First, the Marshall takes both the power tube plate and screen AFTER the choke where Fender takes the plate before the choke and the screen after. And there's that extra 1k2 2W in series with both screens on the Marshall. It was also around the mid-60s when Fender started putting 1k5 grid stoppers on the power tubes. Interesting, no?

A P

 

[plexi]

Fat greenish line when clipped is the power supply ripple modulated on the sine. Ok, normal, no problem.

Crossover notch at zero crossings (or close to zero crossing) also normal, as CJ says, you can adjust this out with bias, but don't leave it too hot (class a) for too long, just long enough to test and see that notch go away, then reset at an appropriate bias voltage (class ab)

Now the breaking through as I see in your picture of the waveform, this is what is not normal....maybe not a full oscillation like we sometimes see with unstable op amp circuits, but on the verge of going full blown. It does look like a lead dress layout issue....you did say that it changed when you moved some wires around, keep looking in this area...maybe try a different PI tube like 12AT7 or 12AU7 to see if it is a problem that is caused by loading on the PI when the KT66 grids draw current. Or move the feedback wire far away from anything else (power tube grids/PI plates)

You have the Mercury 8000 ohm primary transformer...if you load the 16 ohm secondary with 8 ohms resistive load it will reflect 4000 to the primary side. This is the more "normal" bassman/marshall impedance for 6L6/EL34, etc....try it with both KT66 and EL34 and 6L6, whatever you have, just see it and how it changes what you are looking at.

The notch will always be there to some degree, just be sure we are talking about the same "notch" crossover distortion that can be adjusted out with the bias control.....the "notch" at the peak of your sine is what is puzzling.