New project: Marshall MG15CDR

[Roly]

{oh you are in good form. :dbtu:}

[bluesky6]

Well I have done a mod a Fender Princeton 112+ that did a really good job. It involves of course changing the pot to an audio taper, but that did not really get it done.

I have a Fender FM65R which is a direct descendent of that Princeton and what I did was to build a "master volume" control/attenuator between the pre-out and the power in: put a 100K resistor in series with a 1.5K (after some experimentation), pre-out to one end of the 100K, ground to one end of the 1.5K and power in to the junction of the 2 resistors. This was very effective and didn't need me to pull the amp apart. And it remains virgin to my prying hands to this day...

So I looked at the input of the amp and tried to increase resistance of the input signal just before the first op amp. This would R1 and R2 on the Fender amp and looking at the schematic for your MG15CDR it looks very similar of course. R1 is 10k and you could raise that to 20-30k and/or R2 (1meg) can be brought down to 750ohms/500ohms. The idea is that at the input of the amp R1 is resisting the guitar signal just a bit more. R2 then lets more of the signal bleed into the ground, thus reducing the signal a bit more. Both of these approaches have made more differences than just changing out the volume pot from reverse log to log.

In your case I get that you just want to increase resistance using higher value pot. However, my example is to show that there are other ways to get similar results.

Thanks. This is a good idea. Or perhaps just change the gain of the first input stage. I think the RC networks on the input also act as a high-pass/low-pass filter (if I understand an analysis of the Vox Pathfinder 15 circuit correctly), so any change in resistor value will have to be accompanied by capacitor changes or they will affect the freq response of the amp (to Phil's point). I'll experiment.

[phatt]

I had another quick glance at the Princeton as well as the MG15 and my immediate thought for MG15,,,(assuming I'm looking at correct schematic?

*Ed Whoopsy I meant C8 Not C10*
C10 is way too small at 10pF go much bigger. try 1nF or even higher to get rid of diode hash. some call this harshness sand in your dirt channel.

C4 10pF, go bigger try 200pF

For better volume balance between channels raise values of R8 560R or R49 10k. you will have to try a few values to get it to happen right unless you know your R/C maths well.
Even when one does the maths you often end up still having to swap a few values anyway. The maths may well be spot on but sadly our hearing can vary greatly and may not add up the same way,, winky. 

After that first part is a tone setup followed by basic cab sim idea which I'd have to simulate along with the fist part to see how well it works which I'll do when time permits.

@ Dr gonz,
Not sure of those values you mentioned I'll assume you meant
500 *K*  Ohms not 500 Ohms?
1 Meg down to 500k will be fine and will help pull off a little edge but not much.

Also the two input circuits you mentioned are very different. The Princeton input has a high pass filter built into the very first stage which Fender use in a lot of there models.
the Mg input is Not the same input.
just thought I'd mention it,
Phil.

[DrGonz78]

Yeah I meant 500-750k ohms good eye there Phatt. I see what you mean about the circuit inputs being different and how what works on one might not do the same on the other. I will bust out the MG15 and try a few mods to try here as some experimentation and to help me learn this stuff better. Thanks for the good info!  :dbtu:

[bluesky6]

I had another quick glance at the Princeton as well as the MG15 and my immediate thought for MG15,,,(assuming I'm looking at correct schematic?

*Ed Whoopsy I meant C8 Not C10*
C10 is way too small at 10pF go much bigger. try 1nF or even higher to get rid of diode hash. some call this harshness sand in your dirt channel.
....

I thought that was a typo but thought you were referring to C1

....

Also the two input circuits you mentioned are very different. The Princeton input has a high pass filter built into the very first stage which Fender use in a lot of there models.
....

Is that the circuit around U1A on the Princeton schematic? I noticed that this same (values may vary) design is used in the clean input stages in the FM65R, Fender Reverb aka 15R and Champion 110. My op-amp theory is too rusty/limited to help me understand the "dual" input implementation that Fender is using here.

[phatt]

Yeah I meant 500-750k ohms good eye there Phatt. I see what you mean about the circuit inputs being different and how what works on one might not do the same on the other. I will bust out the MG15 and try a few mods to try here as some experimentation and to help me learn this stuff better. Thanks for the good info!  :dbtu:

Well easy to miss type it,, and look at me i just made a stuff up also on C8  :lmao:

I've spent some time tonight simulating the circuit and it maybe a tough one to fix without major alterations.

The tone stack is looking into that cab sim bit (low pass filter) and my guess is that it maybe loading the signal and limiting the useability of the tone controls.
By the time the signal heads off to reverb and power amp node the output is flat up to 10khZ before it dives. and that would make it very harsh and scratchy through a small speaker,, well any speaker really.

Phil.

[phatt]

I had another quick glance at the Princeton as well as the MG15 and my immediate thought for MG15,,,(assuming I'm looking at correct schematic?

*Ed Whoopsy I meant C8 Not C10*
C10 is way too small at 10pF go much bigger. try 1nF or even higher to get rid of diode hash. some call this harshness sand in your dirt channel.
....

I thought that was a typo but thought you were referring to C1

....

Also the two input circuits you mentioned are very different. The Princeton input has a high pass filter built into the very first stage which Fender use in a lot of there models.
....

Is that the circuit around U1A on the Princeton schematic? I noticed that this same (values may vary) design is used in the clean input stages in the FM65R, Fender Reverb aka 15R and Champion 110. My op-amp theory is too rusty/limited to help me understand the "dual" input implementation that Fender is using here.

Yes indeed,, the first opamp has a hi pass filter built in which slopes everything below  ~100hZ

I'm running out of time tonight,, maybe tomorrow I'll do some more head scratching and simulating.
I have a horrible feeling the tone stack is not working well going direct into that cab circuit idea. xP
You might need a buffer between them.
Good night for now, Phil. 

[phatt]

OK some testing reveals;

Remove R7 and C7 (just lift one end for test)
Make C8 10 times bigger (Try 1nF)
Make R9 lower (try 22k)
R13/R14 can go bigger (try 330k) This will pull the cutoff way back down from 10khz to 3khz.
You can also make C48 small to wipe excess low end,, try it if the bottom is soggy.

As I've not got this amp I've no way to verify if this will work but simulation does give some clues as to the result.

IC3 (the power chip) obviously is connected with the channel switch and I can't simulate that chip but my guess is this alters the freq response of the power chip in OD mode.
Others here may know more about that section.

Hope it helps in the search for clues. :tu:
Phil.

Edit;
Oh forgot,, The yellow trace is the original circuit while green reflects the changes made.

[bluesky6]

I just managed to get back to the amp last night. Weekend was no-go because we had a whole day blackout in the neighborhood on Saturday (great to be living in the greatest country on earth...).

Swapped out the power rectifier caps. It helps with the sparkle/pop noise when shutting the amp down but not entirely. I may need to try the cap across primary on the PT next.

I also swapped out the volume pot for a 10K audio taper in lieu of the 5K and replaced R49 with a 22K. This immediately helped with usable volumes for playing in company.

After some staring at the schematics, it finally dawned on me that the speaker simulated out is really speaker simulated (unlike the Pathfinder 10) using U5A and U5B. If that is the case, a speaker swap may "interfere" with the intended design of the amp. In any case, a Celestion 8 Fifteen is on order and I'll try it out and report.

Next, I'll try out Phil's suggestions.

[Roly]

It helps with the sparkle/pop noise when shutting the amp down but not entirely.

There are two kinds of switch-off pop, a fairly sharp noise or crack at the instant the switch is thrown, or more of a bump/thump a second or three later.

The first is caused by the sparking of the on/off switch as it is opened, and can be treated by fitting a "snubber", a small resistance and capacitance, across the switch (or two across both contacts in the case of double pole switching); fitting a snubber across the mains transformer won't do much.

Cause
http://www.ssguitar.com/index.php?topic=1623.msg10391#msg10391
http://www.ssguitar.com/index.php?topic=1623.msg23552#msg23552

Cure
http://www.ssguitar.com/index.php?topic=2439.msg19970#msg19970

Circuit

Typical values would be between 47 and 220 ohms 1/2 watt for the resistor, and 0.01 to 0.1uF mains rated X2 type cap.

This is on the power mains so you must be certain the amp is isolated before you work on it (unplugged), and the new components secure and well insulated (e.g. heatshrink tubing) before you power it up again.

The second sound is the supply voltages collapsing on the amplifier circuits, and the only cure I know of is some form of speaker protector relay, HOWEVER this is very common and normally presents no risk to the amp or speakers.

[bluesky6]

The Celestion Eight 15 arrived today.

Pics to show differences compared to the stock Marshall speaker.

The Celestion has a larger cone surface area, shallower basket, larger diameter magnet, and a taller magnet "tower" (what's the right word for this?).

I used 7mm socket to undo the nuts and hand-tightened them on the new speaker. The latter was a somewhat tricky fit: if the bolts had been a little skewed, it would have been impossible to fit in the new speaker.

I'm going to do more work on the amp itself before putting everything together again and doing a full sound test.

[Roly]

You're lucky.   :dbtu:

The standard "Murphy" problem with a speaker change is that the holes don't align at all.

[J M Fahey]

Those speakers look very similar, don´t think sound difference will be dramatic.
The original, generic one already sounds very good.

[bluesky6]

The one thing I wanted to try out was to "improve" the OD channel. Like the Vox Pathfinder 10, I felt that the gain control was not able to provide a good range of crunch sounds. You basically went from silence to fuzz and fizz.

So the first thing I tried doing today was the same GE diode mod that I did with the Pathfinder: putting a GE diode in series with each of the overdrive LEDs. This did not help at all. I think because the Marshall used the diodes in the overdrive op-amp feedback circuit whereas the Pathfinder used the diodes in the signal path. Increasing the diode clipping voltage on the Pathfinder by adding in more diodes meant that the gain control had more play before you hit the fuzz.

Finally, I realized that it was effectively the gain of the op-amp that was the key factor in reducing the fuzz effect. Duh.

I changed R8, which is the input resistor to the op-amp and whose value contributes to the gain. The factory value is 560 ohms and I changed it to 5.6K.

The end result was a better range of crunch/overdrive sounds. Depending on the contour settings, I could go up to 12 o'clock on the gain before I hit the fuzz zone. And there was still sufficient play on the gain knob for heavy distortion and fuzz.

Pic shows the 3 changes that I made:
1. R49 for the volume on the OD channel
2. R8
3. Op-amp swap. I had a spare RC4559P so I ended up with that.

EDIT: I reviewed what I had done and realized that I'd most likely tested the original GE diode mod with the Contour set to 0. This would have meant fuzz all the way. Ugh! Looks like I'll have to tear the amp apart again next weekend.