Power Supply Design help

Mar 22, 2006 7 Replies

Hello,



Im an entry level electrical engineer. I have been designated the task to design a power supply that meets MIL standard 1399. Basically, I need to develop a switching power supply with the following specs:



INPUT



60/400Hz single phase 120V AC

OUTPUT



+12/-12/+5/-5 200W

Pretty simple up untill the difficult constraints



Max THD(Total Harmonic Distortion) is 5% PF .9 leading and .85 lagging.



Is it possible to make such a perfect power supply with such a low THD? If it is possible, is it too difficult for someone with limited knowledge of power supplies?



Thank you for your help, Matt



Yes, you just need high enough switching freqency and big enough capacitors.

It's just a very technical term to filter out newbies.

I have built high power factor supplies with a two layer approach. The first layer regulates the current drawn from the line, to keep it proportional to the voltage, with the proportionality factor determined by what it takes to keep an energy storage capacitor charged roughly to a constant voltage. I say, "roughly", because this regulation has to be sloppy enough to not try to change the proportionality constant within a line cycle at the lowest line frequency. This slow response is also what provides the soft start ramp.

The output layer draws power, as needed from the energy storage capacitor, and must tolerate its necessary ripple voltage (that results from it being charged with a current that is proportional to the absolute line voltage).

In effect, there are two regulators in series. One regulates the energy storage level, and one regulates the outputs.

Max THD(Total Harmonic Distortion) is 5% PF .9 leading and .85 lagging.

Is it possible to make such a perfect power supply with such a low THD?

================== Search on Power Factor Correction PFC..

Mark

Yes.

It is not difficult if you're willing to learn, but when you come out the other end you will no longer have such a limited knowledge of power supplies! Think of it as just another senior thesis, only you'll be graded every time one goes through manufacturing, or a service tech installs it -- and while you'll never hear about your 'A' grades, your 'C's, 'D's and 'F's will come to you quite loudly.

Tim Wescott Wescott Design Services http://www.wescottdesign.com Posting from Google? See http://cfaj.freeshell.org/google/

yep, and there are plenty of ways to do it.

do you have size and weight constraints? if not, the simplest approach would be a whacking great input filter, followed by a switcher.

if you are very smart, learn quickly, and have access to someone who knows what they are doing, then you'll be OK, but expect it to take a

*LONG* time.

otherwise, yes. This is a typical management trick - give an extremely complex smps design to someone who's never designed one before (power supplies are easy, right?). Betcha theres a tight time frame too.

If I were you, I'd start with a pair of vicor bricks - a PFC and a dc-dc. once you've got that working perfectly, then design replacements for each part.

Cheers Terry

The degree of difficulty is increased because of the 60/400Hz . Vicor PFC bricks will not operate at 400Hz input power. Look at the PICO bricks, they are the only ones that are rated full power at 400Hz. The OP did not mentioned EMI but he probably has to pass MS-461 and that is a task in itself. Half way thru the design the input requirement will be changed to: VF. (VF means Variable Frequency.. 50Hz to 820Hz). If it was me, starting out like the newbie, go for a full discrete design with all the bells a whistles, no bricks. Work, study, design as hard as possible, you will learn much in a short period of time. Keep your boss informed of all problems so you sleep well at night. If he fires an enthusiastic worker then he is a jerk. Worst case you get fired but get another job designing power electronics, you now know more than 95% of the engineers in this field, get fired again and you will be as smart as Terry G, who I would guesstimate he is well worth $135/hr in the US. But JT take the cake at $210/hr, so big whoop! Cheers, Harry

I mostly meant for him to get some experience playing with pfc and dc-dc before trying to build one. it wouldnt surprise me if the vicor stuff wont do the trick. I reverse-engineered one once. ratshit topology/control, BUT awesome thermal management (beware the heatsink requirements though) and fabulous emi control - the faraday-shielded transformer core was a work of art. cute tricks like high-side FETs in a push-pull so the drains dont wobble up and down etc.

The OP did not

thats normally the case :)

Hi Harry,

I managed to do that without being fired, although I did quit one job after 2 months - I sat on the floor for a week, which was bad enough (FFS how hard is it to organise a chair?), but the clincher was the guy who asked me one friday for the XYZ. Que? sez I. "We had a meeting 3 weeks ago, and you were assigned this task" - first I heard of it :) So I went and designed LED video screens instead.

I got 2 new power electronics jobs this week - a 200kVA active harmonic compensator, and a 200kW motor controller. Yay, back to things that explode catastrophically :)

Cheers Terry

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required