Ghost notes after filter cap change.

[greengriff]

I recently acquired a 1977 Marshall 100w Master Volume Combo. Everything was original. The bias supply caps and one of the filter can caps tested high on my meter, but the amp sounded just fine. As a precaution I thought I'd replace all the filter cans and the bad bias supply caps. I did this and now the amp has clearly audible ghosting on mid to high register single notes, although it still sounds as good as before. Previously the amp had barely audible ghost notes, hardly any worse than the 50w that I also have. Is it likely that one (or more) of the brand new caps is bad). I tested them all before fitting with a cheap capacitance meter and all were 50uf or thereabouts. I have no way to test them under high voltage conditions.

If there is a bad cap is there an easier way for me to identify it than to swap a cap, play the amp, swap another cap, play the amp and so on? I just don't have the time to spare at the moment. Any help would be gratefully received.

[Helmholtz]

Correct polarity (positive lead connecting to ground) with bias caps?

[greengriff]

That's a good question! I didn't realise the caps were polarised! The original caps had a groove running around the body of the cap at one end, and so did the new ones, so I installed them the same way around for what it was worth, but I assumed it was cosmetic, like orienting resistors all the same way. How can I test for polarity? These are the bias
caps I installed: https://www.watfordvalves.com/product_detail.asp?id=450

[greengriff]

I opened up the amp and checked. The bias caps are positive to ground, so it's not that. Thank you though. Any other suggestions?

[greengriff]

OK so I bodge wired in one of the old Daly can caps in as well in parallel and the problem went away. Although I could leave it hidden in the chassis I'd rather not. Does the fact that this cured the point to one of the ARS being faulty, or just that they naturally have a lower level of filtering ability and needed 'topping up' with an extra cap?

[Helmholtz]

Your result shows that the replacement cap either has too low capacitance or a bad connection.

An ecap measuring high with a DMM doesn't necessarily mean it's bad. Old ecaps had large capacitance tolerances and often were on the high side. Capacitance never increases with use. Higher than specified capacitance (say +50%) doesn't cause problems and actually improves filtering.

The problem with DMMs is that they often can't distinguish between capacitance and leakage. So a cap testing high might be fine or have high leakage (bad).
Such meters only show if a cap is bad (measuring way too low) but not if it's good.

[greengriff]

Your result shows that the replacement cap either has too low capacitance or a bad connection.

An ecap measuring high with a DMM doesn't necessarily mean it's bad. Old ecaps had large capacitance tolerances and often were on the high side. Capacitance never increases with use. Higher than specified capacitance (say +50%) doesn't cause problems and actually improves filtering.

The problem with DMMs is that they often can't distinguish between capacitance and leakage. So a cap testing high might be fine or have high leakage (bad).
Such meters only show if a cap is bad (measuring way too low) but not if it's good.

Oddly that contradicts the instructions that came with the cap meter which state that if a cap reads way too high it's knackered.

[Helmholtz]

Oddly that contradicts the instructions that came with the cap meter which state that if a cap reads way too high it's knackered.

Exactly that statement shows that it is no reliable C-meter. A dedicated LCR meter (that's what I use) would not show increased capacitance when a cap is leaky.
A cap has three essential quality parameters: capacitance, leakage resistance and ESR. Each of them impacts performance. A good C-meter should be able to measure each of them separately.

What were your actual readings?

[greengriff]

Exactly that statement shows that it is no reliable C-meter. A dedicated LCR meter (that's what I use) would not show increased capacitance when a cap is leaky.
A cap has three essential quality parameters: capacitance, leakage resistance and ESR. Each of them impacts performance. A good C-meter should be able to measure each of them separately.

What were your actual readings?

I can't recall precisely but the bias caps were around 14/15uf and the bad side of one can cap read 60 something uf.

[Helmholtz]

I can't recall precisely but the bias caps were around 14/15uf and the bad side of one can cap read 60 something uf.

Assuming these C-values were real, they would be absolutely no problem for the amp. As said even 50% (sometimes up to 80%) above nominal is fine. (Too low C is far more likely to cause problems than too high.)
Nevertheless there could be excessive leakage, which would be a problem. Leakage tends to increase C readings on DMMs.

[greengriff]

Assuming these C-values were real, they would be absolutely no problem for the amp. As said even 50% (sometimes up to 80%) above nominal is fine. (Too low C is far more likely to cause problems than too high.)
Nevertheless there could be excessive leakage, which would be a problem. Leakage tends to increase C readings on DMMs.

However the amp had less prominent ghost notes with these old (potentially faulty) caps in. All the new caps tested fine on my meter (which as you pointed out earlier, might actually be meaningless) but now I have ghost notes. If there was any other change in sound after the recap it's that notes on the low strings became slightly more 'piano like' with better separation and a more immediate attack, but it was only a small change.

So to eliminate the ghost notes am I stuck with changing out the new caps for the old ones one at a time to find the culprit? Or is there a better way?

[Helmholtz]

So to eliminate the ghost notes am I stuck with changing out the new caps for the old ones one at a time to find the culprit? Or is there a better way?

All I can add is this:

- New ecaps are not necessarily good caps. Especially when shelf life is exceeded.
- Even 50 years old ecaps sometimes are surprizingly good, but only a precise C-meter which also reads leakage resistance and ESR could verify.
- Ghost notes often are the consequence of insufficient filter capacitance or high ESR(or bad electrical contacts) and your test with paralleling another cap seems to verify this.

Did you check the connections of the replaced caps?

[greengriff]

All I can add is this:

- New ecaps are not necessarily good caps. Especially when shelf life is exceeded.
- Even 50 years old ecaps sometimes are surprizingly good, but only a precise C-meter which also reads leakage resistance and ESR could verify.
- Ghost notes often are the consequence of insufficient filter capacitance (or bad electrical contacts) and your test with paralleling another cap seems to verify this.

Did you check the connections of the replaced caps?

Yes, all connections are solid. Looks like my only choice is to desolder and return the new caps. :-(

[greengriff]

Forgive my bad manners - I meant to say thank you for your help with this! I'm very grateful.

[g1]

Do you have any test clip leads? If so, you could remove one of the old caps, and try all the new ones just by clipping them into that position. It would save you a lot of work and may find the bad one.