How to check if capacitor is bad in bank of many / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

How to check if capacitor is bad in bank of many

Move Thread

LAN_403

HV Enthusiast

Thu Mar 21 2013, 02:22PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

Does anyone know of a method to determine if a single decoupling capacitor is bad on a digital board if there are say 50-100 other decoupling capacitors on the same circuit?

Not sure if this can be done, but perhaps there is a method.

When i say decoupling capacitors, i'm talking about 0.1uF SMT capacitors.
Thanks!

Steve Conner

Thu Mar 21 2013, 04:05PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

To find a shorted capacitor, hook the board up to a high current power supply and follow the smoke!

(or use a thermal imaging camera if budget allows)

Finding an open is trickier. You might like to try something like forcing high frequency current into the power rails, and using a near field probe to look at the current in each capacitor.

HV Enthusiast

Fri Mar 22 2013, 02:42PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

Thanks. I do have access to IR camera so that will probably work if i don't see any smoke.

Sulaiman

Fri Mar 22 2013, 06:29PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

The above method does work
BUT
sometimes a s/c capacitor can carry more current than pcb tracks
If the pcb has both + and - planes then the above technique is ok,
if you don't find the fault at 1A or 2A don't continue as a blown inner-layer trace is impossible to find without diagrams or a working comparison unit.

You can find ANY short-circuit on a pcb with this method;
put about 1A dc through the power lines
with a meter on mVdc you can find the trace with the largest mV per cm
(normally from the + input as - is a ground plane.)

testtest

Sat Mar 23 2013, 12:46AM

Registered Member #3271 Joined: Mon Oct 04 2010, 02:29AM
Location: Canada
Posts: 159

You are talking of the multitude of decoupling caps found on the power rail at each (or nearly each) digital chip on large pc boards I assume.

If so:
have a look to see if the are tantalum drop type. These are notorious for twikle or going dead short. When in the twinkle mode they are intermittent low resistance and cause noise on the power rail. An IR camera is a good idea. I have found several over the years using my hand or nose to find the warm ones. Once shorted they will emit smoke signals but often they will take down the power supply. The short is so good that you might even burn a trace before the culprit heats up. In that case I used the divide by two method, cutting the pc board trace in several segments to isolate a few candidates and then physically removing them for testing. One time on a large board TDR gave me a "far" indication that was valuable. You need a comparison board for this however to interpret what you see. AND in that case it was NOT a capacitor but the actual chip that was vcc to gnd shorted!!! How do you know it has to be a capacitor (or all the chips were in sockets and have been removed?)

If they are electrolytic look for leakage or a bulged can first. Apply voltage and the culprit will become warm.

In both cases be careful to not exceed the chip voltage and have the right polarity as all the rest of the circuit is still there..... should the short go away you do not want your voltage to shoot up and ruin something else.

If this is an open cap the circuit will/should still function sort of OK, or might be erratic or noisy. Individual caps can be tested with an esr meter but that will not help if there are several on the buss at the same tine. In one board I isolated a single open failure by using a high speed differential probe with the storage scope to isolate the few four or five chip area where the supply rail was obviously noisy while exercising the DUT and finding a 7474 not triggering in time with the master clock (logic analyser).

Second possibility: if you are talking about coupling caps fanning out a signal to multiple parts of the circuit a short will upset that stage mostly/only. An open can be spotted with a scope measurement for a low throughput and confirmed by a large esr on that particular cap.

It would help if you could give more details on the circuit topology and symptoms you are seeing.

Cheers.

HV Enthusiast

Sat Mar 23 2013, 04:41PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

Thanks for the info.

Basically these are decoupling caps on a large digital board which has multiple FPGAs. There are probably about 100-200 decoupling capacitors in all fanned out across the board for all types of various digital ICs on a digital bus (planes for voltage) The capacitors are ceramic chip.

testtest

Sat Mar 23 2013, 04:48PM

Registered Member #3271 Joined: Mon Oct 04 2010, 02:29AM
Location: Canada
Posts: 159

In that case you might find this link about ceramic chip cap failures interesting:

Much like the tantalum case in some ways. Look at fig 8 for "twinkle" before total failure and fig 1 and 2 on the sudden shorting out behaviour.

I would think the deep IR camera would be the first try here if you have one. Still could be a chip or a solder bridge also?

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob