0
\$\begingroup\$

I'm an electronic tech so I've got some background but not so much as an EE.

I work at a CM diagnosing boards to the component level where possible.

I have a bunch of boards that are malfunctioning. One of the possible root cause is power starvation. The final product has PWM LEDs so the surge current is likely several amperes. I don't know for sure how much given I lack the equipment to check. It's also powered by two lipo batteries in series, so overcurrent drives them into protection mode.

We use two 100 uF capacitors on the board to act as a reserve. They are 1210 X7R 16 VDC. I suspect they aren't performing as we would expect. Nominal maximum voltage across them is 8.4 volts but we usually get 8.2 volts in practice.

I was thinking about tacking on an aluminum capacitor in parallel.

What raised my BS alarm was the volume delta. The SMT capacitor is around 1/20 the volume of the nichicon capacitor.

Is it legit the SMT capacitor can be THAT much more efficient? I don't have the part number at the moment, but I do have it on my work computer.

\$\endgroup\$
12
  • \$\begingroup\$ What kind of malfunction? \$\endgroup\$
    – DKNguyen
    Jun 19, 2022 at 20:23
  • 1
    \$\begingroup\$ It's not about efficiency. It's about frequencies. Ceramics = low capacitance but large bandwidth. Electrolytics= high capacitance but low bandwidth. Fortunately, the higher energy the surge the lower the frequency it is likely to be. If you need high to support frequency currents, having all the charge storage in the world won't matter if you can't access that charge fast enough. \$\endgroup\$
    – DKNguyen
    Jun 19, 2022 at 20:25
  • \$\begingroup\$ A surge of several amps is usually too high energy for ceramics and too low frequency to need ceramics. But logic running with much faster transitions will malfunction if you do not support their small, but high frequency currents. The LED surge current is not the only current. The PWM logic is probably much faster. So-called "decoupling caps". \$\endgroup\$
    – DKNguyen
    Jun 19, 2022 at 20:27
  • 3
    \$\begingroup\$ Dig up the datasheet for your specific ceramic caps and find a graph or formula which tells you the capacitance derating under DC bias. It's possible you'll be surprised at the result. \$\endgroup\$
    – brhans
    Jun 19, 2022 at 20:28
  • 1
    \$\begingroup\$ "100 uF capacitor...They are 1210 X7R 16 VDC." I don't think this is physically possible. Not even with a dieelectric that trades stability for volumetric density and X7R is not such a dieelectric. Are you sure you did not misread the capacitance? \$\endgroup\$
    – DKNguyen
    Jun 19, 2022 at 20:36

1 Answer 1

1
\$\begingroup\$

We use two 100 uF capacitors on the board to act as a reserve. They are 1210 X7R 16 VDC.

There are no 100uF X7R 16VDC capacitors in 1210 smt footprint. Period.

I suspect they aren't performing as we would expect.

Correct, because they are not 100uF capacitors.

The smallest ones I could find were TDK C7563X7S1C107M280LE in 3025 footprint, $5.30 each qty 100.

What raised my BS alarm was the volume delta. The SMT capacitor is around 1/20 the volume of the nichicon capacitor.

That's a good intuition!

I don't know for sure how much given I lack the equipment to check.

I would argue that you do have all the equipment to check :)

You got a battery powered circuit, i.e. it's floating. Get a 0.25 Ohm resistor in series with the load, and connect an oscilloscope probe across it (ground one end, tip to another end). The resistor can be multiple larger value resistors paralleled. If you don't have any low-value resistors, you could use a piece of thin wire. 1 foot of copper AWG 40 wire is about 1 Ohm at room temp, so 3 inches of it would do the job.

With a 0.25 Ohm shunt resistance, the oscilloscope will measure 4A / volt across the resistor.

3 inches of such wire would have about 0.1uH inductance. You can get rid of most of it by folding the wire in half, so that the current flowing in one direction flows right next to a current flowing in the opposite direction. This will make the inductance negligible if the wires are touching each other (via an insulator of course - that would be enamel, since you'd be using magnet wire).

The folded wire should be twisted so that it stays in a low-inductance spatial configuration :)

You'll essentially be making a gimmick capacitor from magnet wire, and not cutting the loop open:

A gimmick capacitor prior to opening the loop

Image source: GIMMICK CAPACITORS by Harry Lythall - SM0VPO.

\$\endgroup\$
2
  • \$\begingroup\$ Holy shnikes, that's cool! Can you solve procrastination too LOL? \$\endgroup\$
    – Katherine
    Jun 19, 2022 at 22:03
  • \$\begingroup\$ GMC32X5R107M16NT is the mpn so my memory was correct. I thought it was a BOM error but it really is 100 uF. I didn't believe it was remotely possible. \$\endgroup\$
    – Katherine
    Jun 20, 2022 at 13:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.