We're working on a design for a 28VDC system. The main supply power input is 28VDC, and there is some holdup bulk capacitance on the frontend. I was surprised to see the aluminum electrolytic caps were rated to only 35V.

I suggested these be increased to 63V rating, and received the following comment from another engineer:
"Wrt the capacitor working voltage it is a quirk of this capacitor type to gradually attain a voltage “memory” equal to the actual applied voltage, so the recommendation for Aluminium Electrolytic is to go as close as you can to the working voltage."

I am not sure what they are referring to. Maybe they are talking about recovery voltage (polarization effects) on a disconnected cap? Does anyone have any insight here? I am not sure if this is a teachable moment for me, or of the other engineer is off base.

I'm concerned about transients that would exceed the 35V rating, as well as any excess leakage due to operation too close to the rated voltage.

 

35V seems way too close for comfort, 50V would be my choice. My guestimate is the system voltage is about +/-15% tolerance, so there normally can be up to 32.2V.

 

Depends how stable and accurate the incoming 28V supply is. 35V working will be OK if its a good quality supply (say from a regulated power supply). If it's from a vehicle alternator for example, that may be noisy or have voltage spikes, 50V or even 63V might be better.

 

I suggested these be increased to 63V rating, and received the following comment from another engineer:

"Wrt the capacitor working voltage it is a quirk of this capacitor type to gradually attain a voltage “memory” equal to the actual applied voltage, so the recommendation for Aluminium Electrolytic is to go as close as you can to the working voltage."

In over 40years in the "will do circuits for food" racket I've NEVER heard anyone express such a belief; If I had, the fellow saying it would get a severe dressing-down. He's just plain wrong.

IMO, 35 WVDC is cutting it awfully close; 50V or 63V would be more appropriate. So your suggestion is correct and wise.

 

Okay, thanks everyone for your confidence-building comments

 

I am not sure if this is a teachable moment for me, or of the other engineer is off base.

I suggest you discuss with the other engineer before posting questions on the internet. I suspect it’s a teachable moment for both of you.

 

We're working on a design for a 28VDC system. The main supply power input is 28VDC, and there is some holdup bulk capacitance on the frontend. I was surprised to see the aluminum electrolytic caps were rated to only 35V.

I suggested these be increased to 63V rating, and received the following comment from another engineer:
"Wrt the capacitor working voltage it is a quirk of this capacitor type to gradually attain a voltage “memory” equal to the actual applied voltage, so the recommendation for Aluminium Electrolytic is to go as close as you can to the working voltage."

I am not sure what they are referring to. Maybe they are talking about recovery voltage (polarization effects) on a disconnected cap? Does anyone have any insight here? I am not sure if this is a teachable moment for me, or of the other engineer is off base.

I'm concerned about transients that would exceed the 35V rating, as well as any excess leakage due to operation too close to the rated voltage.

Your engineer is at least partly correct. Aluminum electrolytic caps do "de-form" slowly over time to whatever voltage they are experiencing. However, the voltage rating is still relevant as there is no guarantee that a cap will reliably re-form to a voltage greater than its rating. Since the 28V may reasonably be expected to sometimes run greater than 28VDC, a rating somewhat greater is reasonable. There is no reason to go higher than 50V (assuming the only choices are 35V, 50V,...). In general such caps do give longer life if operated well below their rated voltage.

Alum electrolytic caps will de-form in storage as well (the operating voltage then being 0V) but will usually reform reliably once voltage is reapplied. During the reforming period, leakage current will likely be much higher than after forming has completed. If a cap has seriously de-formed, then it is often advisable to limit the re-forming current via a resistor, but the need for the resistor depends on how the cap is used in the circuit.

 

We're working on a design for a 28VDC system. The main supply power input is 28VDC, and there is some holdup bulk capacitance on the frontend. I was surprised to see the aluminum electrolytic caps were rated to only 35V.

If the power source is well regulated and the caps are operating within their temperature range, 35V rated caps would seem sufficient.

This is from a Cornell Dublier application guide: