Reverse P-MOS

[throbscottle]

I tried it with a beefy diode between the pass transistor and the feedback resistors, and it works OK. Haven't checked for noise and stability - too busy checking the two current limiters, and trying to chase down a strange fault (may be due to a failing o/a), but looking quiet and stable. Bit of voltage sag under load but that may be due to something else.
Here's the complete psu schematic - I've only built channel A for testing so far. Have tested the voltage tracking with a battery, need both channels built to test the current tracking.

 

[crutschow]

If you needed a lower voltage drop you could use an ideal diode such as the LTC4358.

 

[throbscottle]

Nice, but the operating voltage range is too restricted since it's a 0-38V supply (hence the CC source in my sim). I think I'll stick with what I've got for now. It looks as though the voltage drop might actually be a marginal advantage, not a disadvantage. Possibly a more linear version of a diode would help. I just want to get the thing finished and in a case now, then I can start developing the MKII version!

 

[dr pepper]

A said beefy diode across a pass transistor in reverse bias can be used to protect the pass tranny, however with a pre reg thats no a brilliant idea.
I have a prototyped lm723 linear psu which I was thinking of adding a switching pre reg to.

 

[throbscottle]

Yeah, the varying input voltage to the pass transistor has all sorts of implications that aren't obvious! I did have diodes like that, then took them out, then put them back but so they do the job in series with the other diodes. The principle is that the pre-reg is much more robust than my linear section, so it's ok...

 

[throbscottle]

Tried simulating the Raspberry Pi circuit in LTSpice, works really well (what else would you expect?) Resistance goes up when input goes down to about 1V or so, though. I might give it a try. Thanks, Bob

 

[KeepItSimpleStupid]

I'm using crowbar protection against reverse voltage applied to the input.

You can't do that. If it were a battery and the crowbar activates, don't you get short circuited battery?

 

[simonbramble]

For MKII, have a look at the LT4363. This limits the current in the forward direction, on startup only, so you limit the inrush current to the output caps. It also protects against under voltage, over voltage and reverse voltage. It also has a separate current limit once the circuit has started up. And you can simulate it all in LTspice which will save you hours of time

 

[throbscottle]

KISS - it would be in conjunction with a fuse in series with the output so it blows the fuse. Though now I think about it, - if a series diode is present it's only the feedback resistors which experience applied reverse voltage, and it doesn't look as if it would do any harm.
Simon - looks a very interesting device, definitely something to look at in the future!