2019-08-01
2019-08-04
Finally success. FINALLY. I finally have inrush current management under control. In case a refresher is needed, here are a few posts discussing the problem.
https://www.simkus.com/2017-10-01/
https://www.simkus.com/2018-07-21/
https://www.simkus.com/2018-08-24/
To recap:
Attempt #1 – relay. When the relay would turn on, the power supply would dip causing havoc.
Attempt #2 – PWM drive of the low side. Electrical current is like love – it finds a way. By disconnecting the low side (negative rail), the positive side found all sorts of fun ways to find it’s way back to the low side.
Attempt #3 – PWM drive of the high side. Absolute inductive spike hell.
My fix to #1 was to use PMW (Pulse-width Modulation) in order to slowly raise the average voltage that the load (a capacitive load that are the boards) sees. Well reality got in the way. No mater how short I made the period I could not completely get rid of the inductive spikes.
Fun side story.
I thought I had the problem solved. As I was screwing around with the PWM drive, I said lets see what happens if I just turn on the output FET. Just boom – on. Like a relay. To my dismay it … just worked. The FET turned on. The boards turned on. Nothing rebooted. No smoke. All this time trying to get PWM working properly has been wasted.
I then hooked up everything to a proper industrial power supply – a 5 amp MEAN WELL – and typed the ‘go’ command. The fuses on both boards blew. Fuck. My lab power supply was treating me with kid gloves and protecting me from my self. The real power supply is much stiffer than my lab power supply and didn’t put up with my foolishness. Back to the drawing board.
As I was analyzing my failures I had a world-changing idea – what if I increased rise time of the output FET gate? That would increase the time that the FET is in the ohmic region thus reducing the inrush current and so forth. Sure the FET takes a beating, but that’s why I have a nice beefy FET on there. As a mater of habit I put resistors on the pull up/pull down pins of a FET drive, but until just now I didn’t really get WHY I put them there. Sure I knew it was to control the rate of gate voltage rise/fall, but WHY. Well, I just figured out why.
I stuck a 10uF cap between the gate and the 100 ohm gate drive resistor. Turned on the FET and blew two fuses again.
Having exhausted all ideas I read the manual for the LTC7004. And what do you know. They have a reference design for switching large capacitive loads. It addresses the exact problem that I am having. The only thing I left out was a resistor between the capacitor and the gate which cause high frequency oscillation just as the manual states. I put a resistor in there and boom – everything worked exactly as I need it to.
Don’t know how I missed it before. Well I mean I know how – I was blindly pursuing the PWM drive idea.
So here we are. The last major hurdle has been cleared. Time to send out the final revision of the board.