Organic Aluminum Polymer capacitors in series for higher voltage?

Err, I get a few like this:

Digikey have ones up to 15uF:

What is your space concern - height?, area?, both? There are ways to reduce both.

Dave.

In theory you could use zeners in parallel with those caps. As long as the zener voltage is lower then the Vmax of the cap, and the sum of the zener voltages higher then the total voltage.... Have not tried, maybe it would go poof.

I have done basically the same thing. I put resistors across the capacitors to hold the voltages in place with time. I used high enough voltage capacitors so that even with the most extreme value mismatch the voltage rating would not be exceeded. I also placed some

with ripple current at 100 kHz of about 3 amps RMS. The

I tested some 12-volt aluminum poly caps. Charging them at low current, through a resistor, leakage was very low until about 25 volts, when they instantly failed shorted. This is different from a regular aluminum or tantalum cap, where leakage increases with voltage so there is some free series-stack equalizing effect. So I'd think they need equalizing parts; Jan's zeners, maybe.

We stock this ceramic one...

PART: 1814 CAPACITOR CHIP 1206 10UF 25V

STOCK UM MIN ALLOC USED OUT BUY PRICE TOT VAL 996 EA 0 404 737 0 0 0.33 328.03

Mfr1 MURATA GRM31CR61E106KA12L Mfr2 DIGIKEY 490-3373-1-ND SAME AS M1 Mfr3 MOUSER 81-GRM31CR61E106KA12L SAME AS M1 Mfr4 Mfr5

which you could stack up in various permutations.

I really like the alum polys for switcher output filter caps, as long as you allow for their very low esr re: loop stability.

John

Have you looked at film capacitors? e.g.

The 68 uF, 75 volt 6.2 amp 1.7 m ohm unit runs about $21 (recently raised from $15) at Digikey, quantity 1.

This part is 31 x 17 x 43 mm and will not fit. The 33 uF version is 31 x 13 x 22 and is just barely too large to fit. The PCB is 0.95" wide x 2.5" long and sits in the middle of a 25 mm ID tube, so the 12.5 mm round can just barely fits. A rectangular footprint is more difficult.

Thanks,

Paul

I am using a PIC16HV616 or PIC16F684 for the PWM, and the output is actually current regulated at about 800 mA to a string of 7 or 13 high power LEDs. Using the A/D to sample the output current every 1 mSec or so, I am relying on the output capacitor to slow the output response enough to avoid a large overshoot, although this would usually not be a problem unless the input voltage rose suddenly, which is unlikely with an SLA battery. Ripple at 100 kHz on the LEDs is not a factor. My new design also senses the MOSFET source current, using a comparator which can react much more quickly to such conditions, and I can probably get by with a much smaller capacitor. I was able to run the prototype (with an older board) with only 2 uF output capacitance, but that was on a current limited supply. I am afraid that with an SLA battery it could ramp up to destructive current levels before the PIC can sense it and throttle back, at least with the old design that did not have a drain current sense.

I am already using a 1.5 uF 100 volt MLCC in SMT1210 that is only $0.53 ea/100, and there is also a 1.0 uF 100 volt in SMT1206 for $0.31 ea/100. So I could lay out maybe 10 of these in parallel on the bottom of the board, and that works out to only about $3 for 10 uF or $5 for 15 uF, which is probably enough. I will have to consider the extra cost of multiple SMT parts on the bottom of the board, which are not easily done by pick and place, and will add extra labor cost if hand soldered. But it won't cost anything to add the pads as an option. Maybe I could make a small separate board for the capacitors and wire it to the main board as a module.

Using the capacitor you suggest works out to about $5 for 10 uF at 100V, and would require 16 pieces as well as resistors or zeners for series connection.

Thanks for checking this out. Adding zeners or resistors will be a labor-intensive operation if the lower voltage series capacitors are used.

Paul

Oh, on your ceramic cap, check the actual capacitance *at 100 volts*. Some of these hi-K dielectrics can lose half their C or more as voltage goes up.

John

I'm sure I have seen 10uF 100V X5R ceramic multi layer SM caps for about 10$, so 5 of them in parallel should work and be small enough. Not a thousand dollars either.