Tweaking a China SMPS battery charger to manually tune voltage?

[viper]

So here is the deal, I have a brand new, free to me 36/48V ""smart"" battery charger....lmao.....  I say that because there is nothing smart about it.  It is actually a pretty darn well built circuit I think but it was a total miss for AGM batteries and at least operated in the 48V mode, I can confirm the voltage is too high to charge AGMs safely.  It was at about 59V and I need that at 57.6V, or a touch less. 

The current sense (shunt) in the device seems to reasonably accurate (tested), but I am not very sure how it set up the output stage, and if it is hard programmed, how I might lie to it.  I realize everyone will ask for the "schematic"....lol  If you know China, even if one existed, there are probably 15 variants by now, but you never get a schematic for this stuff.  They steal ours, but we don't get theirs!   

Anyway, the best I can do is pictures and start to reconstruct this dude.  If I can throttle the voltage, I have many more uses, including rolling it back to a safer float voltage so I can put it on ignore for days. 

Edit, grabbed a few pics of the one I started repairing.  I can do better but had to shrink to fit the 5meg limit here. 

I also have another one of these that had a direct short at the base of the front end filter caps.  Oddly it did not blow a fuse but toasted the board and caps a bit, as well as the small shunt, which I thought was odd.  Anyone, I replaced those components so far and she now powers up great, but will never commence actual charging....so I must still have a gremlin in that circuit from the short circuit affair.  That whole thing started from what I suspect was a burr on the bottom of a capacitor that was jammed right into the trace at assembly.  The issue was found and the circuit was first test fired with only one filter cap and it was stable and fine. 

[TheMG]

I see they've omitted all the EMC filtering components (common mode choke, X capacitors) top right corner.

More interference-spewing junk to pollute the airwaves.

Also, I can see traces running under an opto-isolator and a lack of any clearly defined spacing between primary and secondary side, so I'd say electrical safety isolation is suspect at best!

Not a nice design at all actually.

[viper]

LOL...   When I said "well built circuit", I just mean this one actually has straight components that don't appear soldered by a blind toddler. 

Obviously these don't even get flash tested because the capacitor issue would have blown up in their face, yet there was this nice "inspected by" sticker on it.  I think Wong Lee Chang was having bad day.. 

Any thoughts on the circuit mod?  I sort of expected a trimmer in there, but I guess those cost money and you would have to light the circuit and actually tune it...

[Kim Christensen]

Well, I would investigate / reverse engineer the circuitry on the LED side of the opto-isolator (low voltage side). That's where the current sense, voltage sense, and charging logic circuitry appears to be. Looks to be only x3 8pin SOIC chips there. Can you see the part numbers on them? Or is there more on the flip side?

[viper]

I hope these come through good enough.  If not, I will have to get to the microscope in the lab. 

[viper]

If those 3, 8pin chips were not clear enough, what I am making out are ST Electronics 2904 op amps?  I'm off searching for the MZ data and I think that is mfg data?


Oh, for the question on components on the back side.  There are none and it does not appear to be a triple layer board.   

[Kim Christensen]

Yea, they look like some Op-Amps. This partial schematic is what I think you have... Can't verify because I can't ring it out with a meter but this is what I can see so far. If my guess is right, you may be able to decrease the resistance of R64 to lower the output voltage. For a simple test, you could bridge another resistor across R64 to change the resistance by about 10% and then see if the voltage has changed. R64 is 68k so a 680K across it should be enough to see if I am correct but not enough to cause much trouble if I'm wrong.

[viper]

Wow, really appreciate that Kim!  I did not expect that, and I believe you are dead on.  The pin 3 network you presented seems to ring out correct.  The back side of R64 connects to the +out through the K1 main relay. 

What is odd to me though is testing R64 directly across it, it is reading about 7.8K but it is also behaving like a capacitor.  IE, my Fluke meters do not lock onto the value, the value climbs a bit like it's trying to charge. 

I only mention because the circuit we are looking at here is the one that had the capacitor short but has yet to actually charge for me.  When connected to a battery bank, it will recognize it is connected and display the correct current voltage.  At that time K1 should activate, but it never does.  No other errors are experienced. 

As I understand it, you want to reduce the R64 resistance, to raise the feedback voltage presented at pin 3? 

[Kim Christensen]

As I understand it, you want to reduce the R64 resistance, to raise the feedback voltage presented at pin 3?

Yes. That should make the LED in the opto turn on more, which should make the switcher circuit on the input/line-voltage cut back on it's output, thus lowering the voltage to the battery.

[viper]

Kim, I did some testing.  Installed a 470k resistor plus a 0-200k pot to get to about 670k, paralleled over R64.  I installed the pot so I could make some adjustment if things looked promising, but nothing changed.  I turned the pot down only slightly hoping to see any change at all.  I tested batteries of nominal 36v and 48v, both with the same results.  I was unwilling to turn the pot down any more until checking in for your thoughts. 

I did note the current presented on the charger is not right (no surprise), and was reading about .7A higher than actual.  I tested voltage with my good Fluke DMM that has good resolution but not even a twinkle of voltage change. 

The charger displayed voltage was reading pretty much correct, or within .05V of actual. 

Obviously if I turn the pot all the way down to 0, I would be at 470k but did not try it yet.  Seeing the actual resistance value is not possible due to the capacitor feedback but I did test my device before soldering in.  Right now I have two small wires attached to each side of R64 as I figured we might have to play with this. 

[Kim Christensen]

Kim, I did some testing.  Installed a 470k resistor plus a 0-200k pot to get to about 670k, paralleled over R64.  I installed the pot so I could make some adjustment if things looked promising, but nothing changed.  I turned the pot down only slightly hoping to see any change at all.  I tested batteries of nominal 36v and 48v, both with the same results.  I was unwilling to turn the pot down any more until checking in for your thoughts.

Hmmm... Because of the way D15 & D16 feed the LED of the opto, only one of the parameters (Current or Voltage) will dominate at any one time. I would check to see what's happening on pins 1 & 7 of U8: if the voltage on pin U8-7 is higher than U8-1, then the current limiting circuit is dominant and playing with the pot across R64 will do nothing. If so, you may have to wait a while until the batteries are 100% charged and gone into float mode.

But if it turns out that the voltage on pin U8-1 is higher than U8-7, then it would be interesting to see what the voltage at U8-1 does when you adjust the added pot.