Overwhelmed with trying to select replacement capacitors for old test gear

[kelchm]

In the past few months I've amassed a small collection of old test gear including the following: 4x Fluke 8050A, 2x HP 3478A, 2x Sorensen XT15-4. While diagnosing a misbehaving XT15-4 I came to realize just how badly capacitors degrade over time. I've of course heard of the 'bad caps' plague, but I honestly didn't realize that otherwise healthy looking capacitors could degrade and cease to function correctly.

I'm now on a mission to replace all the electrolytic capacitors in this gear, but the more reading I do the more overwhelmed I am with selecting replacement capacitors. I've collected all of my notes on the basic specs such as general dimensions, capacitance and voltage ratings. Where I start to go off the rails is how to take a 20 or 30 year old capacitor and figure out what a modern equivalent would be. These devices certainly weren't designed with a modern low ESR capacitor in mind (ex: the well-regarded Panasonic FM series), and I don't want to introduce issues rather than fix them.

Are there any tricks I can use to track down specifications for the old capacitors that were used in these devices so I can more easily find a modern equivalent? Am I wasting my time? Should I just pick a run-of-the-mill series like the Nichicon UPW and hope for the best?

[tautech]

Simply, break it down into the application types.

Rectification: linear or SMPS
Bulk capacitance or remote (to the PSU) bulk capacitance
Coupling

That helps heaps with replacement selection but you need take into account frequency of operation in some applications. Keep in the back of your mind the BEST cap was not selected for the OEM. Cost vs acceptable MTBF guides manufacturers on product selection, historically and now.

A little circuit knowledge and schematics will help make good choices.

[Vic20]

If most caps on the circuit, with same value and voltage rating, are the same kind, I expect that run-of-the mill will do if you use same voltage rating capacitors. I will suspect of a capacitor that is much different of the others.
In most designs, having a better ESR than specified does not harm, but it will not always need to be true.
If the equipement is old, current generic use capacitors are probably better than the ones you have.
If you have a switch mode power supply I will take more care and select caps for this application.

[David Hess]

For power supply bulk input and output capacitors, it will be difficult to go wrong.  I usually use long life 105C replacements and if necessary, a higher voltage rating to get the right diameter or pin spacing for mounting.

[GerryBags]

I've got a Marconi R C Oscillator from the sixties that some numpty fired up to "see if it works". It must have been quite exciting when it revealed the fact that it didn't. There is flash over all over the area where the 6080 completely popped the bottom of it's envelope. There was also the charred stump of what turned out to be a thyristor once I tracked down the service manual...... and that is my point (yes, I have one!) The old test gear laid out every schematic, component list and cal procedure right there in the documentation that shipped with the product. The cap list (as most of the 60 odd are getting replaced) lists the ID, the composition, the capacity, the voltage and an upper and lower capacity tolerance for it's use in the device. All I had to do was make a list of replacements as I found them, I have an order cued up at Farnell that I'm adding to as I work through.

If only the more modern gear came with the same level of trust. I understand why they can't in today's litigation culture, but it's still a damned shame.

Interestingly, the documentation also warns about the leaky output electrolytics they had to use. They advise unplugging the DUT everytime any of the range or output switches are to be used because of the switching transients. There is also a type written note varnished to the front above the output selector, just in case you forgot! 

[daedalux]

I've not seen much the electrolitics being used when the actual precise capacitance value is important, if it's so and it is you better replace it with poliester or polipropilene even if they are big.
If you are trying to choose capacitors for replacement in a device like a linear supply almost anything legit with the same voltage rating and capacity will do. If you buy off the shelf regular stuff you'll have less ESR than 20 years ago which will usually be good. Lots of times more capacitance will be even better. In SMPS the problem may be life expectance for the cap, but it depends more of actual quality that is not easily measured in a new cap, and that's why people trust brands in electrolitics

Once the voltage rating is enough you may think of increasing it, but a electrolitic with a much higher voltage rating, for example 200V in place of 35V, will have more ESR than the small rated one. So you just may think of increasing the rating to the inmediately available above. It will be bigger (or not if you are replacing very old ones) and will self heat less due to its higher size.

On the other hand i've tested lots of pulled out electrolitic caps 30 years old good in capacity and ESR. I would not rate their life expectance in age but in hours in use. They are much more prone to be bad in something on 24/7 the last 10 years even if industrial quality that in something consumer grade mostly on storage 30 years.

[TheUnnamedNewbie]

More capacitance will not necessarily be better. It can lead to stability problems, will slow down the loop response of the system meaning it might not be able to react to changing load situations as well - this applies especially to linear regulators.

Lower ESR can also give stability issues with regulators, though I don't think that is common on modern regulator ICs. But some rely on the internal ESR to push away the pole to get the loop stable.

When replacing regular electrolytic capacitors in devices, I try to just get the cheapest capacitor with similar ratings and lead spacing, from a reputable company (Nichicon, Vishay, Panasonic, Kemet, etc). I don' t try to get too fancy capacitors unless it is clear that the ones I'm replacing are fancy ones. Keep in mind that if the capacitors in there are 20 years old, they were good enough back then - and since the technology was less advanced back them modern low-end capacitors will surely be good enough as well. There is no point overdoing it, and might inadvertently do more harm than good.

[David Hess]

Once the voltage rating is enough you may think of increasing it, but a electrolitic with a much higher voltage rating, for example 200V in place of 35V, will have more ESR than the small rated one. So you just may think of increasing the rating to the inmediately available above. It will be bigger (or not if you are replacing very old ones) and will self heat less due to its higher size.

ESR generally decreases up to 100 volts and then has a step increase at 160 volts and increases after that.  I think this is because of a change in the electrolyte to support higher voltages but maybe it is due to construction.

More capacitance will not necessarily be better. It can lead to stability problems, will slow down the loop response of the system meaning it might not be able to react to changing load situations as well - this applies especially to linear regulators.

Lower ESR can also give stability issues with regulators, though I don't think that is common on modern regulator ICs. But some rely on the internal ESR to push away the pole to get the loop stable.

Low ESR leading to instability is a real problem but only when changing capacitor types.  Standard aluminum electrolytic capacitors all have about equally poor ESR which is an advantage for regulator stability.  Solid tantalum capacitors are even better in this regard because of better consistency.  Replacing a standard aluminum electrolytic with a polymer aluminum electrolytic is an example where they could be a major problem and if I did this, I would also do transient response testing of the regulator to confirm stability.

I have never seen increasing the aluminum electrolytic output capacitance cause a stability problem.  This lowers the ESR but it also lowers the cutoff frequency.  The lower ESR limits ripple and the higher capacitance gives the regulator more time to recover.  Increasing the output capacitance quickly reaches diminishing returns so there is little reason to do it.

If the regulator is designed for minimum output capacitance, then at low capacitance, control of the ESR becomes critical as part of the frequency compensation but you will not find aluminum electrolytic capacitors used in these applications.