Karcher window vac WV2 not charging

[throbscottle]

This is more for the benefit of others who may have similar problems, since I've "fixed" mine (read - modified slightly so it works) even though I still don't understand what's actually wrong.
I have a Karcher WV2 window vac. It came to me without it's original charger. Got a USB substitute cable as the chargers are pricey! Means it charges at 5v instead of 6v. It hadn't been used for a long time and would not charge it's battery. Light flashes for a bit, no current is drawn.
Replaced the battery and it charged once, refused to do so since.
Otherwise works correctly, turns on, motor will run, turns off, light flashes when battery starts to go.
Re-worked the charger board. Tried another battery. Tried the original battery.
Drawn a schematic of the charger board so I can understand what's supposed to happen. Not sure if it's 100% accurate. My initial sketch has a lot of crossings out on it!
Anyway, the first of the P-MOS transistors is supposed to get turned on by a signal to the transistor connected to it's gate, which is itself turned on by pin7 the ATTiny24A. That signal is missing. If I bypass the ATTiny with a resistor directly to the charger + input, it will charge, and once it's started charging it seems to latch and continue to do so without the resistor.
The 2nd P-MOS turns it off when the battery voltage gets too high. I think the FET is a J301 (marked !301)
When the charger is first switched on, pin7 goes high very briefly. So I think the ATTiny is sensing something it doesn't like, but I can't figure out what.
Q6 and Q7 form an interlock so the ATTiny can keep itself powered on when the motor's running, since the switch is momentary.
Then there's that weird arrangement with the TL431 type reference, set to 2.5v. All it seems to be doing is regulating the supply to a PD that has the low end tied to the battery side of a low-side current sense resistor. Is it just acting as a level shifter for the current sense voltage? Or something else?
At one point during my investigation it started working. Put it back together again - stopped working and refused to do so again.
The LED is supposed to turn green when charging is complete but it no longer does so.
So my "fix", based on the experiment turning on Q6 with an extra resistor, was to connect a bootstrap capacitor (100n) to it's base, directly to the charger + input. So far it's working, even after I put it back together. That's my entire weekend gone, better darned well keep working now...

 

[Nigel Goodwin]

Two things, as it hadn't been used for a long time the batteries may be too flat, and so the charging circuit refuses to charge them as they are ruined. Secondly though, applying only 5V when it needs a 6V power supply means it may not work at all due to the voltage been too low.

You shouldn't go modifying Li-Ion charging circuits, as it's an extremely critical procedure, and could cause the battery to set fire.

 

[Diver300]

I guess that one of Q2 and Q3 should be opposite ways round, one to stop the battery powering the charge connector and one to control the rate of charge.

The TL431 is simply providing a regulated voltage to the ATtiny microcontroller.

R15, R16, R17 and R18 provide an offset on the voltage that is generated across R1, the shunt resistor. That offset allows the microcontroller to measure positive and negative currents.

 

[Nigel Goodwin]

I guess that one of Q2 and Q3 should be opposite ways round, one to stop the battery powering the charge connector and one to control the rate of charge.

I would suggest they are the right way round, Q3 controls the charge, and Q2 is a hardware over voltage shut-off, to turn the charge current off if the battery is over charged. IC1 is presumably a special IC for detecting the over voltage.

The TL431 is simply providing a regulated voltage to the ATtiny microcontroller.

R15, R16, R17 and R18 provide an offset on the voltage that is generated across R1, the shunt resistor. That offset allows the microcontroller to measure positive and negative currents.

The 0.01 ohm resistor seems remarkably low, bearing in mind it's feeding directly in the processor - however, in differential mode you can add a x20 amplifier to the input, presumably that is used?.