I think i have killed maybe 3 or 4 smps using them as battery charging sources. But, i feel nearly useless in diagnosing my own failures because i buy the absolute cheapest things i can get my hands on, use them in ways they aren't intended to be used, and don't understand enough about how they work to diagnose the failure modes. I'm just screwing around! I have bought at this point probably nearly a dozen of the ~300-600w form factor 'meanwell knockoffs' in the ~$12-40 range. Of the ones that have failed, the actual most common failure is the NTC on the input exploding (like, case splitting, white powder everywhere) but in those cases i have mostly just bypassed it unless it failed immediately in the exchange window. The 2nd most common failure is blown fets in the back of the case (i have looked at smps block diagrams but still feel unsure how to correctly ID these) in this area:
And i think only once have i blown fets in the front/side area:
One time i smoked a large resistor, which i thought was odd. So i can't really identify any pattern to the failures other than that i
assume most of them are related to a lack of current limiting since the only thing that is going to physically 'ventilate' a sealed piece like a fet or resistor is the local expansion of material, caused by heating, caused by overcurrent or lack of sufficient cooling for an otherwise acceptable amount of current. That's how my brain works it out, anyway.
So this is me admitting that im a bull in a china shop over here with this stuff. I don't feel bad about it, that's why i buy the cheapest things i can, because this is 'learning while playing', not seriously studying a topic with intent to do anything important for anyone else, for money, etc.
As far as the PWM sccs, i have broken a few of the extremely cheap ones, again always blown fets. I repaired a couple by harvesting matching fets from others and fully 'populating' all the empty fet spots in the boards to turn them into the higher rated versions of the design (if that is the only difference?), and that worked, although it is a silly thing to do.
At this point the only PWM scc i am using with SMPS as input is actually a fairly expensive ($70) and bulky piece that is rated for 80a and actually has a selectable current limit. I started with a 20a model of it because i noticed the selectable current limit feature and wanted to test the theory that just limiting a SMPS below its rated current would make it live 'forever' while pwm'ing into a battery, and that seemed to basically work out to be true with the caveat that forever is i've only been doing this stuff for like 7 months. So i bought the 80a model, and currently have 4x30a psu's paralleled into its PV ports, charging ~460ah of lifepo4 for the past month or so (it has existed in smaller form with less PSU hooked to it for the past 6 months or so). It is a big dumb setup and i am not calling it a good setup or even claiming that im not slowly murdering the lifepo4. But at that point my components are not failing 'quickly', so i feel like current limiting the PSUs makes them not die quickly while being PWMd. On the other hand, I also have PSUs that are hooked straight to battery and NOT pwm'd and they seem to live just fine putting out their max current rating continuously. So i think the PWM'ing is having some effect on the control scheme of PSUs and making them briefly overcurrent themselves during pulses which gradually physically damages the fets.. Weak theory!
As far as whether it hurts the lifepo4, i think the main thing is that when the lifepo4 is near full and cannot accept as much current, it also cannot clamp down voltage and can be pushed up to damaging voltages, 4.0v/cell etc. So if you have your PSUs set to something which equates to >4.0v/cell then yes, unless you are using some sort of shunt-based scheme to actually interrupt the psu>battery connection then when it gets close to full the voltage of the 'pulses' will rise to cell-damaging levels. However, if your PSUs are set to an acceptable voltage, then the only question that remains which i could answer for my dang self if i scoped it, is does the PWM result in the PSU voltage spiking to cell-damaging levels? If not i think my current setup can work long term. If so, then the followup question would be: would hooking the PSU output to my lead bank through several diodes in series, cause any brief voltage spikes to be 'clamped' and diverted harmlessly into the lead? If the answer is yes, great. If the answer is no, then i would have to do a generally smart thing and use a shunt to implement charge termination based on tail current, at which point the PWM SCC serves no function other than current limiting the PSUs which they may not even need if they weren't being PWM'd!
LOL at everything i do, what a mess.
This is what an auto mechanic learning about electronics without actually
studying electronics, looks like.
