Capacitor Specs

[xtraelectronics]

When replacing electrolytic caps, as far as voltage and capacity are concerned, it is permissible to "overate" the originals, to an extent. Ex. 680uF -ok to sub in 1000uf. 16V rating, ok to sub in 25V. (Keeping space requirements in mind) Sometimes the manufacturer will run a 15V rail and use 16V caps. THIS IS NOT ACCEPTABLE as the part in question WILL FAIL in no time. When I was taking electronics training, the instuctor (and text books) stated the rule of thumb for capacitor voltage ratings should be 50% higher than the expected circuit voltage. I have alaways followed these guidlines, and have had no susequent failures. Just my 2 cents.

Dan

[PlainBill]

Re: Capacitor Specs

With all due respect to your instructor (and the author of the textbook), that is a gross oversimplification. A great deal depends on the circuit design. The advice ignores the fact that many circuits are designed with a specific capacitance in mind. Look at the design discussions for inverters; many of them choose components to form a tuned circuit.

The 50% higher rule is also suspect. Certainly I would agree to allowing a generous margin on a capacitor filtering the rectified AC input, but 50% seems rather extreme. Using a 16 volt cap on a 15 volt rail is going too far in the opposite direction.

PlainBill

[xtraelectronics]

Re: Capacitor Specs

Yes you are absolutly right about specific capacitance values for tuned circuits. That was my oversimplification. I should have specified this for filter circuits. I.E. low voltage DC as used in the final filter for ripple. My bad.

A 50% margin may seem excessive, but some of the stuff today runs right on the edge of its maximum ratings. Do we blame shoddy parts, or poor design practices for all the failures? It's enough to drive ya to drink

[PlainBill]

Re: Capacitor Specs

Yes you are absolutly right about specific capacitance values for tuned circuits. That was my oversimplification. I should have specified this for filter circuits. I.E. low voltage DC as used in the final filter for ripple. My bad.

A 50% margin may seem excessive, but some of the stuff today runs right on the edge of its maximum ratings. Do we blame shoddy parts, or poor design practices for all the failures? It's enough to drive ya to drink

A LOT of it is shoddy (or marginal) design. It's not uncommon for a designer to overlook the capacitive load of a FET gate. Or to allow for variations in components.

About 20 years ago I was working for an electronics control manufacturer. Someone cooked up a series of modules for signal inputs and outputs. The guy was good at digital logic, but the power module he designed was crap! It was notorious for flaming out when power was applied (no thermistor to limit surge), when powered up with no load, or when powered up with maximum load. They took it back to the lab and reworked a few of them until they were reliable. More than half of the components had been changed before they got it working.

The actual data collection part of the design featured an on-board switching power supply (very unusual for the company - most of the power supplies were analog monsters). One large order presented dozens of failures that we gradually worked out, but a persistent problem was that the switching power supply would shut down, then restart for no obvious reason. Finally some genius noticed that:
A. There was no problem with the systems after the doors of the cabinets were closed,
B. A number of cards would shut down at the same time in an open cabinet, and
C. This would occur only when they were talking to the guys in the control room - using walkie-talkies.

We wound up adding a few 100 pf caps to every power supply and that problem went away. It's not the problems you anticipate, it's the ones you don't expect that will drive you nuts.

PlainBill

[Wizard]

Re: Capacitor Specs

Apparently that guy didn't have a AoE book. This book covers so much in detail about design issues and what to NOT do.

Cheers, Wizard

[PlainBill]

Re: Capacitor Specs

Apparently that guy didn't have a AoE book. This book covers so much in detail about design issues and what to NOT do.

Cheers, Wizard

I think the phrase that applies here is "You can lead an EE to information, but you can't make him think."

Back in the 80's I'd have recent EE graduates looking at circuit boards (in particular logic boards) and wondering what all those little .1uF caps were spread across the board. "Why not just use one or two 10 uF caps?"

PlainBill