Sizing a choke (or. . . ?)

Hi Folks

I'm working on a project that involves controlling about 42A worth of LEDs at 12V via PWM. The LEDs are split into three banks, switched 120° apart at 3906Hz. I'm currently testing with an SMPS rated at 48A. The problem I'm having is that even with only one bank of LEDs (14A) on at ~50%, my power supply is tripping out, I presume due to the inrush current when the output FETs turn on. The FETs are driven fairly hard with the goal of reducing switching losses, and should have switching times of <60ns. I suppose I need to do one or more of the following: 1)Add a choke to the input and/or outputs to limit inrush currents, 2)Add more bulk capacitance to the board (currently ~900μ), 3)switch the FETs more slowly, 4) suck it up and buy a bigger power supply.

For 3), I can remove the FET drivers from the board and drive the FETs directly from the 5V logic level PWM outputs, however that also means a lower gate voltage and hence higher Rds(on). So I'd rather avoid that.

I suspect 1) is my best bet, but I'm not sure how to go about selecting a part. With only one bank on, the switching frequency is 3906Hz, and load fluctuates between 0-16A. With all three banks at 50%, it's 11718Hz and 16-32A. But I suspect I should be looking at the rise time, effectively a much higher frequency. I'm seeing what looks like 1.2V/25MHz ringing, however I fear there's more going on than my 100MHz scope can show me.

The only resources I've found in this area so far are for audio amps or VFDs, and looking at some of the formulas found in the latter (for example), based on a frequency of 12KHz, I'm looking for a ~33nH inductor, is that in the right ballpark?

I'm a little out of my depth here, so any advice would be appreciated.

 

I don't see why it would be inrush current since, with a resistive type LED load, the inrush shouldn't be any higher than the steady-state current.

You could try an inductor in series. It's value is not critical for your application.

Post a schematic of the complete circuit.

 

I'm not sure what else it would be, but I'm open to suggestions. Perhaps rather than inrush, it's simply the rapid, drastic change in load. I haven't yet seen the power supply trip when all three banks are running at 50% (load fluctuating from 14-28A at 11.7KHz), only with one bank running.

The schematic is not terribly complicated, but I can post one when I get home tonight if you really want. It's 48 N channel FETs arranged in 8 groups of six. Each group has two FETs on each of the three phased banks, an independent connection back to the PSU, two 47uF capacitors, and three dual FET drivers. The FET drivers each have their own decoupling caps and are arranged so that each of the two inputs comes from a different phase. In the current test setup, each FET is switching 1A of LEDs. In the final arrangement, any given FET will carry between .25A and 2.2A.

Come to think of it, I didn't account for the FET drive current in sizing the power supply, thinking that it would be so brief as to be negligible when averaged over time. Perhaps that was a mistake, but I don't think it accounts for the behavior I'm seeing. I can split the FET drive onto a separate power supply just to see what happens tonight.

 

Most inexpensive SMPS have really crappy response to load transients. I'm not sure there is a good fix for your problem except a better supply. The usual response to a change in current demand is for the voltage to fold-over (drop) until the control loop can adjust the switching waveforms to deliver more current. Do you think that supply output current is available immediately in any quantity up the supply's maximum? I don't think it is unless the designer intended for that to be the case and took steps to make it so.

 

Output schematic is attached--it's representative of the 8 identical output groups. Not much to look at. The +12V FET drive supply and V+A and V+B currently all come from the same external SMPS. I have a new, beefier PSU on the way, should be here Friday.

 

If it is the slow response of the SMPS causing the problem then try increasing the power supply output filter capacitance. Try around 10,000μF.