Ekco AD76. Same capacitor keeps failing.

[Filament]

I think I have been convinced that a motor run capacitor is the way to go, being the simplest and cheapest option. The problem is which one. RS are offering a possible one under part number: 911-9403, rated at 7.5uf at 450v AC. Although slightly below value, it is suitable dimensionally. I think it will squeeze in the box together with the other two with a bit of dexterity. I would be very grateful for any comments on this choice.
On a philosophical note, isn't it strange that a capacitor made 85 years ago, set in a wax block together with two others, wrapped in cardboard, was up to the job whereas modern “equivalents” are not?

[Sideband]

Ahh but technology has changed! It's only idiots like us that keep these old sets going and need low value high voltage capacitors rated for suitable ripple at 50 or 100Hz. Modern capacitors are normally used in modern equipment and in the case of switch mode power supplies will be running at much higher frequencies so ripple factor is different (where needed) typically 100khz.

Having said that, so far I've not had a problem when replacing any smoothing caps.

[Herald1360]

RS 882-9357 8.2uF 450V metallised polypropylene would likely also cope and might be a better shape/size fit being rectangular.

[beery]

Hi Fillament,
I've never had a problem with ripple current causing failure of electrolytics, though it could be possible, I've never known it in 30 years of collecting radios and TVs.

I've used motor run capacitors in pre-war television sets with a high HT voltage and they work very reliably there.

Aside from double checking the polarity of the caps, I wonder if there is something else going on.
I noticed that you have changed all the valves, do you still have the original rectifier (V5, Brimar ID5)? If so, I suggest swapping it back. I have had problems with NOS valves occasionally and in this case, you might have one that has been tested for emission, but not for heater-cathode insulation. Being an AC/DC set, even a slight breakdown of the heater-cathode insulation of the rectifier will cause AC to be presented at the cathode and this will destroy the electrolytic smoothing caps.
I once had a PZ30 go heater-cathode short on an Invicta T102 television, where it is used as a voltage doubler. The result was that one of the electrolytics quickly ejected its entire contents onto the bench!

Cheers
Andy

[PJL]

It is difficult to get high ripple current electrolytic capacitors in small value high voltage ratings. Like others I tend to replace them with the cheapest film capacitors I can find and have had no troubles with the red polyester types from ebay.

[murphyv310]

Hi.
I've never ever had ripple current issues with modern caps in vintage sets and have been repairing sets like this for over 50 years..
I've also done a couple of AD76's and never had a cap failure. I think you need to look elsewhere for the issue.
Is the rectifier going short circuit and feeding the cap with AC? Is there another component fitted in the wrong place that could be feeding AC to the cap. Have you looked at the waveform on the cap that is failing? Is it getting hot quickly? Don't rebuild into the can initially to do some checks.
It's possible the 1D5 has an intermittent heater cathode short that could give the issue you have.

[Station X]

I've always doubled the HT current and taken that as the ripple current. For most domestic sets suitable capacitors are easy to find, it's more difficult for comms receivers with many more valves.

I've restored a Bush AD65, restuffing the capacitor block and have had no problems so far.

On the other hand simulations of the AD76 have been done and suggest a capacitor with a high ripple current rating is needed.

[murphyv310]

The HT on the AD76 is 195v incidentally.

[murphyv310]

Hi.
I've done a simulation online and TBH the results don't really add up. I'm currently looking at an AC DC Philips and if I were to believe the simulator the reservoir cap would be a great heater right now and it's stone cold. The ripple is about 5v not the nearly 100v the simulator says. So either I'm doing something wrong or the results are just incorrect.
I'm with Andy Beer here and suggest it's not anything to do with the cap. Look at the rectifier, check the ripple on the reservoir with the scope and keep a close look on the waveform if it's low to see if it suddenly rises. I'd though change the 1D5 though.

[ms660]

You could work it out from Schades curves with an assumption or two.

Lawrence.

[bionicmerlin]

I have a AD 76 . About 5 years ago I changed the same caps ,I have had no problems. If I remember they were just bog standard replacment ones I changed them for Andy

[Radio Tech]

Could I suggest getting all of your capacitors from Jellyfish Audio, I have used this company for years and never had any problem with ripple or other issues relating to HT voltage or currents

Ken

[murphyv310]

Hi.
Personally I still doubt it's anything to do with ripple rating. I also in the main buy caps from a reputable source in Canada and most likely the same company as the OP. In fact I've had two recent failures with RS yellow polypropylene axials, the electrolytic caps from Canada are first rate quality items and in 15 years using the same company I've had not one single failure.
I have no interest in the company and all I'm doing is to say that I'm satisfied with their components.

[Herald1360]

Originally Posted by murphyv310

Hi.
I've done a simulation online and TBH the results don't really add up. I'm currently looking at an AC DC Philips and if I were to believe the simulator the reservoir cap would be a great heater right now and it's stone cold. The ripple is about 5v not the nearly 100v the simulator says. So either I'm doing something wrong or the results are just incorrect.
I'm with Andy Beer here and suggest it's not anything to do with the cap. Look at the rectifier, check the ripple on the reservoir with the scope and keep a close look on the waveform if it's low to see if it suddenly rises. I'd though change the 1D5 though.

Did you measure the ripple voltage at the rectifier cathode/8uF point or on the HT rail downstream of the smoothing choke? Total HT draw is about 90mA, over half being the parallel fed field coil draw.

I had no info for the effective series resistance of the rectifier, which would damp down ripple current somewhat it is significant compared with the 50R surge limiter.
Just had a thought.... I was seeing about 235Vdc at the rectifier cathode in my sim but the datasheet suggests 190V actual- I'll try upping the surge limit R to drop cathode volts to 190V and see what that does for ripple current.
TBH, genuine ripple current alone does seem unlikely to cause such rapid failure of a modern cap- their ripple rating is at max operating temperature of 85 or 105C and what they can take at lower temperatures is considerably higher. Even a life reduction to 10% of expected would still suggest several hundred hours operation not a few minutes.
It would be informative to temporarily sub a 1N4007 + 100R for the rectifier- lift the wire off the cathode pin of the rectifier and hook the diode + R between rectifier anode pin and the lifted wire, leave the rectifier in situ to complete the heater chain.

The h-k leak suggestion does sound promising.....

[crackle]

I was been getting rather worried whilst progressing through this thread. I have restored a few AC/DC radios and if the smoothing cap was not reforming I simply replaced with out much heed to "ripple".
Luckily KB did not produce so many AC/DC sets but there were examples for many of the models, and I have done quite a few DAC90A's and such.
I feel a little more reassured now that others have said they too have not had any problems with using modern electrolytic caps in AC/DC sets.

I think I would be suspecting the rectifier by now.

Mike

[ms660]

Originally Posted by ms660

You could work it out from Schades curves with an assumption or two.

Ripple results as per Schades curves if I've read the log graphs right....

Resistance in series with the anode of the rectifier = 50 ohms
Reservoir capacitance = 8uF
Supply Frequency = 50Hz
HT at reservoir = 190 volts
An assumed mains voltage of 230 volts AC
Calculated load resistance = 2,175 ohms approx.

Ripple voltage = 47.5 volts

Ripple current = 120mA

That's disregarding the resistance of the mains filter inductors (the resistance of those isn't given in the manual)

Lawrence.

[Herald1360]

Hmm, that was interesting.....


Adding an extra 220R of rectifier series resistance in the sim brought the HT at the rectifier cathode down to the 190V suggested by the datasheet and reduced the cap ripple from 208mA rms to 155mA rms.


Also found some interesting life data for a particular Vishay cap with a quoted life at 85C of 5,000 hours at nominal ripple current suggesting a life for the same cap at 40C and 1.4*nominal ripple of 120,000 hours. Not necessarily directly applicable here, but it does rather add strength to the argument that excessive ripple current at temperatures way below max rated will not kill a capacitor in minutes!

[ms660]

Ref. my calcs, the 190 volts HT I used is what was given in the service manual.

Diode resistance at peak came to 224 ohms.

The effective RMS resistance of the diode came to 241 ohms.

Lawrence.

[murphyv310]

Thanks Lawrence and Chris.
In all honesty this fault can really only be a rectifier issue or perhaps a very leaky valve base.

[ms660]

Just to note, the method I used can be seen in RDH4 starting at page 1170:

http://www.tubebooks.org/books/rdh4.pdf

The curves/graphs I used can be seen in this publication starting at page 186 of Schades lecture:

http://www.nj7p.org/Manuals/PDFs/Books/RCA_1940_VTD.pdf

I used those because for me they were a bit better resolution wise than the ones in RDH4.

Lawrence.