"multisection" capacitors help

[g1]

[ATTACH=CONFIG]55215[/ATTACH]

It's a bit hard to tell, but it looks like that (-) lug arced over to the chassis. I assume it's one of the caps that needs it's shell insulated from ground.

[Helmholtz]

My cap meter reads leaky caps as if they have higher than expected values, say 40 to 200% over.

Agree. Markedly increased capacitance of E-caps is most probably a measurement artefact. Simple meters just interpret the total magnitude of impedance, while better LCR meters are able to identify resistive and reactive components.

Typically E-caps lose capacitance over service time because they dry out. Less electrolyte volume means less active capacitor area means less capacitance.

[Enzo]

Dumb little cap meters like the ones in a DMM send a small charging current out to the DUT, and reads the rate the voltage rises. It then translates that into capacitance. Well, if your cap is leaky or worn, it charges at a slower rate. That fools the meter into thinking the cap is larger than it should be.

[vintagekiki]

When checking old and new capacitors, an ESR meter may be the right solution, as it indicates, among other things, the state of dryness.

Personally, before any capacitor testing, I use an old good analog ohm meter at range x1000, on base of the instrument pointer in permeable and non-permeable direction with 90% accuracy can be judged capacitors condition.

The real solution is the RLC bridge, but this measurement is thing of the past, the digital C meter has suppressed it because more that sufficiently shows the value of capacitor capacity.

For demanding testing H.V. capacitor (the first C after the rectifier) I measure the leakage current through the capacitor by connecting the capacitor with series resistor 100k / 2W direct to + HV.
If, over a period of time, the current starts to rise or is unstable, this indicates a loss factor. I allow 0.5 - 1 uA ^Correction) and then is "peace in house"

EDIT 190915
^Correction)
instead: I allow 0.5 - 1 uA ... ...
should: I allow 10 – 50 uA ... ...

[soundmasterg]

Maximum filter for a GZ34/5AR5 is 60uF, and 47uF is less than 20% over 40uF, so that should be fine.

Depending on the vintage of B15N, the first cap after the rectifier is a 30uF. Then there is a 1k 10W, and then the 40uF sections of the can cap. I have a 1967 B15N on my bench right now. Someone had replaced the can with a CE several years ago, but didn't replace that 30uF cap, so when it went bad it took the power transformer with it and the transformer leaked a bunch of black tar like goo all over the inside of the amp. I had to take everything out of the amp carefully and clean it all up, then replace the power transformer and that 30uF cap. I say carefully because that black goo stuck to everything and it was hard to get the PCB out without breaking it. Anyway, the can cap is bad also. The 1k resistor will start smoking even on the light bulb current limiter with only the rectifier tube in the amp. These amps don't have much space inside so this is one that benefits from having the can instead if discrete caps. In my case the customer is going to pay for the can, so I'll just replace it instead of using discrete caps and trying to find space for them inside. Even if I went with a dual 50uF can there isn't much room inside, and the customer would likely be a stickler for originality, knowing him.

Greg

[boroman]

Thank you all, I really appreciate that. I want to keep the solution as simple as possible and do not mess with other parts of the amp. Bias cap have already been changed. The susupicious resistors and little caps, too, along with 20+20 cap that I replaced by 40uf. OK, so is this "SIMPLE" solution will do the job?

- REPLACING 1ST CAP: oriignal was 70+40+40 but third part was not used, so I could take 70uf and 40uf caps so everytjhing can be connected the same way - just all the grounds would have one point (joined grounds of 40 and 70 cap) right?
- REPLACING 2ND CAP: original was 40+20+20+20 and JJ is maiking identical multisection so I can throw it just in place and connect.

Am I thinking right and simple? Especially asking for the ground (-). As you see on the picture I posted in the last post, each old Mallory cap has 4 grounds and lot of things connected there. But those grounds are internally connected to each other (checked), so this was just the one common ground that just has the 4 terminal for better soldering/hooking up things, right?

[Mick Bailey]

Just to be sure on terminology, do not confuse ground with negative when looking at caps. The 40+20+20 negative terminal is not a ground - it connects to the +ve terminal of C20 70uf section and the junction of R56/R57.

The cans have one negative connected to 4 terminals. The mounting is the twist-lock type and from a manufacturing perspective you have 4 tabs connected to negative, so why not punch them all and make them all solder terminals for convenience, rather than just have one negative and 3 redundant tabs. I can't recall ever seeing a multi-section cap in a guitar amp that had separate negative terminals for the different sections.

With some multisection caps, the negative terminal connects to the can and/or mounting tab, with others, the negative is isolated so that when mounted it's not connected to the chassis ground.

[boroman]

Just to be sure on terminology, do not confuse ground with negative when looking at caps. The 40+20+20 negative terminal is not a ground - it connects to the +ve terminal of C20 70uf section and the junction of R56/R57.

The cans have one negative connected to 4 terminals. The mounting is the twist-lock type and from a manufacturing perspective you have 4 tabs connected to negative, so why not punch them all and make them all solder terminals for convenience, rather than just have one negative and 3 redundant tabs. I can't recall ever seeing a multi-section cap in a guitar amp that had separate negative terminals for the different sections.

With some multisection caps, the negative terminal connects to the can and/or mounting tab, with others, the negative is isolated so that when mounted it's not connected to the chassis ground.

Thanks for the explanation! My bad, I was talking about negative when I wrote ground... of course.

[Mick Bailey]

My cap meter reads leaky caps as if they have higher than expected values, say 40 to 200% over. If I'm reading an old cap say 40+ y/o I interpret "bonus" capacitance as leakage, not a magic advantage by far. And those caps are destined for replacement anyway, so if I read their value, it's mostly for amusement. Occasionally I find one that reads a small fraction of its rated value, of course that means it's headed for the rubbish bin, for sure!

I have a software-based tester that does a slightly better job of measuring cap parameters, better than my Fluke tester and dedicated LCR tester under most conditions. The problem with most high-voltage electrolytics is the ESR can increase along with leakage, so leakage increases the value and ESR reduces it, resulting in a totally useless cap reading the correct value. I find the only good way to measure a cap with a low-voltage tester is to firstly eliminate leakage as a factor. I have a HT supply that I use in series with a DMM to measure leakage current either at the operating voltage or the cap's rated voltage. With older caps you can see the cap 'forming' as the leakage drops, then eventually stabilizes.

I find that there are very few instances other than personal interest to check the value of an electrolytic capacitor in repair work, though I had a piece of test equipment for repair where all the caps had house part numbers and no values, so that was probably the last time I needed to do this. Other than that it's mainly SMD caps or older low-value caps with unclear markings where actually reading the value is necessary.

[Helmholtz]

The problem with most high-voltage electrolytics is the ESR can increase along with leakage, so leakage increases the value and ESR reduces it, resulting in a totally useless cap reading

The µC based LCR meters I used gave correct C readings independent of leakage and ESR.

[The Dude]

It's a bit hard to tell, but it looks like that (-) lug arced over to the chassis. I assume it's one of the caps that needs it's shell insulated from ground.

^^^^That. If you zoom the pic, it looks like arcing to the chassis. I'm not sure the cap is even bad or in need of replacement. It looks like connections are just too close (maybe even touching) to the chassis.

Edit: Just looked back and saw the photos of the caps leaking, so obviously they need replacing.