Capacitive Backup for Power Supplies

Dave,

If you increase the voltage at which you store the charge Q, the capacitance reduces as Q=CV^2. Is your power supply a typical SMPS which rectifies the mains? If so, use a large capacitor after the rectifier, rather than trying to support the load directly.

The only issue might be the low voltage ratings on supercapacitors, but fortunately, you can get thousands of farads:


So you'll need a boost converter to get it back to 24V. The good news is that your downstream devices won't see any droop.

TTFN

Thanks for the responses. We have a size limitation that will preclude using a lot of 60mm dia caps. The size limit is still being determined.

We also do not plan on using any boost at this time on the 24V, but rather just diode switchic to bring the cap on-line. We can allow some droop, maybe 20%. This will raise the capacitance needed since we only draw the first 20%.

Another possibility is to use the cap charged to 48VDC and down convert to 24, but now the array of caps is twice as large.

Size seems to be the limiting factor here. Alternative is a lead acid sealed battery with its associated headaches.

More thoughts welcome!

Dave at USI

The supercaps in the link above are only rated for 2.5V, hence the comment about boosting. So if you don't want to boost, you're stuck with conventional caps, which are large.

Then again, given a requirement for sustaining only 2.5 seconds, you could use NiMH rather than lead-acid, since you're only looking for about 750J of energy, although, pulling 50A from a 6V NiMH might be a wee bit challenging, but not impossible, and you only have to do it for 2.5 seconds, so maybe not it's that bad at all. Since the overall energy requirements are rather minimal, you could easily get by with a camcorder battery, although a laptop battery isn't much bigger and it puts out 16V, instead of 6V. You may need to do a ESR comparison to determine which one is better overall.

TTFN

Here is what we need to supply:

24 VDC 14A continuous 25A Peak 350 W

12VDC 6.5A continuous 70W

5VDC 5.0A continuous 25W

If the 12 and 5 are derived from DC_DC modules off the 24V, then there is the module efficiency to consider so the value to be supported 2.5 sec will be roughly 500 Watts
with some current peaks due to e-m actuators.

We are looking at stacking super caps in series with a resistor tree to split the voltage. There are some app notes on this subject I need to read.

Thanks again!

Dave at USI

IRstuff: I now see what you meant about boosting from the low cap Voltage. That would be a cost/complexity trade off but an option that may solve the size issue.

Will look at some discharge curves and try to firm up ideas by next week. Enjoy the weekend!

Dave at USI