Need help with Low voltage protection circuit

[flinx]

Re: Need help with Low voltage protection circuit

You should put down Thermal Vias for the MOSFET.
Is the copper plane as heatsink for the MOSFET large enough for it to handle 20A or are you going to use heatsink to keep MOSFET junction at safe temperature?

Also consider VIA Stitching for high current.
https://www.microtype.io/high-power-...n-tips-tricks/

While i said it could handle 20 amps I would very likely not be running this much past 12.

[eccerr0r]

Re: Need help with Low voltage protection circuit

Warning: I did a quick reextract of the new PCB and noticed you didn't remove the diode, and in fact put it in the worst possible position that I feared. Please REMOVE the diode completely, or if you must use one, use a schottky diode which is still very bad. If you really want the diode there, you _must_ use either the LM6142 or use the three BJT version, else you risk running into weird leaking behavior.

Also, you will likely need a heatsink even at 12A, though this is at the borderline. If you have enough Cu foil and the aforementioned vias specifically for thermal conduction, you might be able to safely get away with no heatsink. At 20A the heatsink will be required.

[flinx]

Re: Need help with Low voltage protection circuit

Warning: I did a quick reextract of the new PCB and noticed you didn't remove the diode, and in fact put it in the worst possible position that I feared. Please REMOVE the diode completely, or if you must use one, use a schottky diode which is still very bad. If you really want the diode there, you _must_ use either the LM6142 or use the three BJT version, else you risk running into weird leaking behavior.

ok, it was only there because it was on the original board no other reason.

now removed

[flinx]

Re: Need help with Low voltage protection circuit

before any one says it, I changed the component labels to make them less confusing.

[eccerr0r]

Re: Need help with Low voltage protection circuit

and I take back what I said about the assumptions. As I didn't know the characteristics of a IRF5305 I was assuming it was similar to the 75N06 which I am somewhat familiar with that is also quite high in current capability. The 75N06 will be marginal on the board with no heatsink at 12A.

The IRF5305 will roast at 12A without a heatsink. Your current maximum is more like 6A.

(Note that the 75N06 is n-channel which is not appropriate for this application, however, same principles can be applied.)

[flinx]

Re: Need help with Low voltage protection circuit

oh fiddlesticks this is what happens when I don't play with discrete P-channel devices and assume it works like PNP. You're right, there's a problem there, really need to have simulated it and got my current directions wrong. I'm still not going to simulate since there's no warranty expressed or implied :-P

Now I was hesitant on doing the lower left no-BJT version in the first place, though it may well just work. Do not use a "logic level" MOSFET here, IRF9540 and your IRF5305 should work fine. If you you have to use a logic level or the no-BJT version doesn't work, try the upper left version with three BJTs.

I must thank you profusely for taking the time to redesign this.

[flinx]

Re: Need help with Low voltage protection circuit

and I take back what I said about the assumptions. As I didn't know the characteristics of a IRF5305 I was assuming it was similar to the 75N06 which I am somewhat familiar with that is also quite high in current capability. The 75N06 will be marginal on the board with no heatsink at 12A.

The IRF5305 will roast at 12A without a heatsink. Your current maximum is more like 6A.

(Note that the 75N06 is n-channel which is not appropriate for this application, however, same principles can be applied.)

I figured out how to add thermal vias, and can add them in
square 6, 8, and 10mm sizes.

How much would I need?

[eccerr0r]

Re: Need help with Low voltage protection circuit

Once again, the question is - to accomplish what?

You cannot put enough of them to satisfy the fact copper clad and vias simply do not conduct and radiate heat as well as a bona fide anodized aluminum heatsink...

[flinx]

Re: Need help with Low voltage protection circuit

just for S&G's I decided I would make an SMD version.

After accounting for thermal vias and the ability to add a heat sink the resulting circuit board is no smaller than the first one I made.

[eccerr0r]

Re: Need help with Low voltage protection circuit

Looks like a lot of whitespace...

You should use a cheap 1-turn pot for the hysteresis control, the resistor in series already compensates for most of the invalid region...

[budm]

Re: Need help with Low voltage protection circuit

No need to have B- and OUT - so far apart, move them to be closer as possible so less resistance and so you can have room for more copper plane for more heatsinking area since you are not using through hole MOSFET which you can use bolted down heatsink.
BTW, you should also add bypass cap.

[eccerr0r]

Re: Need help with Low voltage protection circuit

did someone have too many lm393's?

Redesigned from scratch but once again NOT simulated/tested and YMMV if this actually works. This should work down to a bit above 5 volts limited by MOSFET selection and then the reference. Power draw is about a quarter of the the previous design, though I think I'm really pushing it without detailed calculations, and may be running up the limits of the amplifier.

IMHO LM358 is better suited for the task, but if all you have is a LM393, all you have is a LM393.

[flinx]

Re: Need help with Low voltage protection circuit

I got my 36v power supply yesterday, got it all hooked up to the control head and was able to make some more detail measurements.

It takes a bit of fiddling to dial in but I have it currently set to turn off at 18.02 volts and back on again at 20.68v. A spread of 2.66v.

The hysteresis does have an effect but it will take a bit more fiddling to figure out exactly how. When it is adjusted the cut off has to be re-adjusted. this is not a big deal and after I can test it under load I can decide what is the best voltage spread.

I may just determine that voltage and hard set it on the board with some resistors instead of a pot.

I will have to try the 393 circuit when I get my bench cleaned up.

[flinx]

Re: Need help with Low voltage protection circuit

ok full travel the hysteresis pot changes the voltage .45v. nice.

[eccerr0r]

Re: Need help with Low voltage protection circuit

btw the 393 version I think I chose R2 poorly. It needs to be a lot lower, else the gain will be really bad. Probably 20K is good for it too.

[flinx]

Re: Need help with Low voltage protection circuit

Just got my boards back from JLCPCB
IT works perfectly. though see if you can spot the error (on my part).




Yes I forgot to join a copper area to a network.

[eccerr0r]

Re: Need help with Low voltage protection circuit

dammit hope I don't get spoiled by kicad having built in schematic-layout compare...hopefully reduces the chances of a mismatch when the pcb comes back.

and I really should get my grid tie inverter project finished, been lazy because of frustrations of not getting it fully single sided so I can etch my own board...nevermind needing to write firmware and ordering mosfets for it, as well as add *sigh* jumpers. Other issue being the board is huge, like 6"x9", really need to put it on a diet.

even made my own homemade ferric chloride with old nails...

[budm]

Re: Need help with Low voltage protection circuit

Just got my boards back from JLCPCB
IT works perfectly. though see if you can spot the error (on my part).




Yes I forgot to join a copper area to a network.

So it works pretty well and no oscillation due to no filter cap for IC VCC and Ground?
Good Idea that you made separate board for MOSFET so you can scale up or scale down as you like.

[eccerr0r]

Re: Need help with Low voltage protection circuit

Honestly though they beat us into our heads not to omit bypass caps but the PSRR of a lot of well designed analog circuits it's not critical to include them,..

[flinx]

Re: Need help with Low voltage protection circuit

I setup a test where I powered a purely resistive load of 4ohms which pulled about 7.5amps.

Using my thermal imaging camera I measured the temp of the mosfet at about 120°f 49°c. for about 5 minutes. Well withing the 175°c 347°f operating temp. I have the capability to test it to 10.5a resistive maybe 21a for a few minutes.

the whole thing was in free air with fans blowing about.

I did run in to an issue and it may require some tweaking of the hysteresis settings and resistors. If you turn off power to the circuit and the voltage that is powering it is below the upper level cutoff voltage the device won't power the mosfet.

My test setup is a bit janky and I need to devise something better that won't follow over, burn down, and sink in to the swamp.