# Why did C2 fail?

I'm trying to run a Raspberry Pi as a remote controller and timer for a diesel water pump but it keeps rebooting when the engine cranks. I checked and the supply doesn't drop below 10.5V during cranking so it must be dirty power.

I'm trimming the supply to 5V using one of these switched mode power modules.

SMPs are sometimes sensitive to supply spikes, so I thought I'd filter the power with a textbook Pi filter.

I built this with C1 and C2 both 4700μF and L1 being 470μH and as soon as power was supplied, C2 boiled, yet C1 didn't even get warm.

From the speed with which failure occurred I imagine there was some kind of resonance between L1 and C2. Before I trawl through old textbooks relearning network analysis I'm hoping the voice of collective experience can tell me whether there ought to be some resistance in the C2 leg, or even whether there are standard values for a Pi filter when you want to arrest spikes from an alternator. VIN is rectified but dirty 12V.

• I would guess that $C_2$ is a polarized electrolytic or tantalum capacitor and you got the polarity backwards. They fail pretty spectacularly when that happens. Jan 28 '19 at 23:58
• How did you check that you did not drop below 10.5V? You need to use an oscilloscope. The RPi will notice milliseconds/microseconds of no power. A voltmeter is not fast enough. Jan 29 '19 at 0:14
• Try a series diode .Then your cap wont fail. Jan 29 '19 at 1:16
• @spacecampreject because there's a huge capacitor across power and ground. Millisecond drops will be supplied by the cap. Would you take your oscilloscope out on a pontoon in a river with reckless people in powerboats whooshing past? Jan 29 '19 at 1:19
• @Autistic - a diode on the C2 leg? Jan 29 '19 at 1:21

Here with large caps and possible low ESR and low DCR of the pump the motor DC current could be all supply by C2 with an RMS ripple current that exceeds it rating. Worse yet the impedance at resonance near the pump frequency can amplify the reactive current increasing cap conduction losses as fast as a light bulb.

This is due to X(f)/(DCR(ESR) amplification at resonance.

A proper design needs the Motor specs for DCR, RPM, Amps, and regulator specs.

The back EMF of the pump now becomes another power source and resonance avoidance is possible by design.

• By motor specs what do you need to know? From an electrical perspective there's a 12V 17AH lead acid wet cell battery, a 12V alternator, a regulator/rectifier in a galvanised heatsink package of the kind often found on small motors for outdoor use, and a 12V starter motor. I will add this to the question info plus how-it-works for my entire system. Jan 29 '19 at 22:26
• Motor DCR RPM and load rating Jan 30 '19 at 2:36
• It's a Chinese diesel. I don't have that kind of info unless you can tell me how to measure it myself. Jan 30 '19 at 3:30
• Ohm meter for DCR and is it gear reduced ? Jan 30 '19 at 5:37
• Oh you mean the R component of the motor coil RLC. Sorry, this is the first serious work I've done in 30 years. I'll measure it. As for whether it's gear reduced - only to the extent that the starter motor cog drives a much bigger ring gear on the flywheel. This is a gorgeously well made diesel that's easy to pull start. With the pressure relief valve open there is quite a light load on the starter. Jan 30 '19 at 6:00