bq24620 /PG and STATUS indications problem

Hi Daniel,

Would you mind sharing a schematic with me to look over? If you don't feel comfortable  sharing it here, you can send it to d-wiest@ti.com and I'll give it a look.

If you can also get a scope  shot showing  battery voltage, input  voltage, /pg,and stat that  may give  us  some insight.

What is the charge voltage you are using?

hi David!

sure, no problem - in fact schema and also board layout are posted on a previous post here:

e2e.ti.com/.../556261

I have a feeling the STATUS output is simply blown - maybe I managed to short the pin to Vcc or something while I was testing the prototype. But the /PG output being unstable comes and goes - I can't figure what makes it start and stop. Often changing the battery load (e.g. from my capacitor bank "test" load to a set of real batteries) will make it go - but then a while later it will start again (usually after connecting/disconnecting power).

Power supply has either been a linear bench PSU set to 25V, with current limit at 2A, or a plugin type supply (made by XP) which gives just over 24V at just over 2A).

I'll take some scope pics a bit later and post them.

thanks!

I should point out that the /PG LED is now pulled up to +24V via 2k2, not to 3V3 as per the schema. 

Of course I can't get it to happen now - when I do I will take some scope pics.

thanks!

Daniel,

Thanks for re-sharing the schematic! I think looking for the /pg issue is the first step here - if the charger does not recognize a good power supply, issues will pop up.

For /pg to be off, there are only a few conditions that can lead to this:
1) UVLO (doubtful)
2) ACOV - it may be that some noise spike is periodically putting the charger into ACOV. This would cause the STAT LED to be off
3) Sleep - it could be that noise (VCC or SRN) is causing the sleep comparator to trip. a scope shot would prove/disprove 2 and 3
4) The /PG and STAT FETs are shot. This is entirely possible. You can try changing out a fresh IC, and I would increase the 2k2 resistor, maybe to a 10k. That should better protect the FETs from large transients when turning on a 24V psu.

Daniel,

Good deal! I agree- finding those current limitations would be helpful for the STAT and /PG Fets. I will work on finding those, but I did a rough calculation and figured there would be ~10mA of sinking current. This should be handled pretty easily by the FETs, so this makes me think that it is a transient from turning on or swapping power/battery that is creating an overvoltage condition for the FET.

Hi again David,

Well, I swapped out the chip today and the battery absent indicator worked straight away, also went into charge mode with no problems when I connected the battery.

I haven't seen any sign again of the odd instability problem that I encountered, so for now I won't worry about it. It could perhaps be a side effect of having sot the chip? I don't know. Or maybe there was/is a bad solder joint someplace.

Anyway, I have to spin the board again and in the new rev I will make some slight improvements to the layout (get the MOSFETs closer to the chip which I didn't pay a lot of attention to on this version). The circuit seems to working pretty well, which is what I wanted to prove for this revision.

best regards and again thanks for the help

Daniel

Hold it! Here we go again.

here's what happened.

I had the board with new chip, it topped up the charge on a 6 pack of LifePo4's just fine.

When the charge light went off, I removed the DC PSU (an XP power 24V power pack) - and the light flashing started.

Sorry for the bad pic - was holding the scope probe while using my cam phone to snap the trace. That waveform is 10ms/div, 5V/div. The lower line gets close to 0V (about 0.6V or so).

After a minute or two, this stopped, for no apparent reason. Started again on another hotplug.

No pics, but i can tell you that neither the incoming DC, the 3V3 ref, or the REGn pin (which is at 6V) have any sign of this low frequency stuff. All are absolutely clean when I look with scope.

it does indeed seem to be a hot pluggng issue - but what? the input filter I used is straight from the datasheet.

Unfortunately I don't have a nice storage scope to catch input transients ... let me think on that ...

here is something : it only does it if I hotplug with the (fully charged) battery connected. If I remove the battery - never does it. tried this four or five times.

I have a theory.

The aiming potential for the charger on this prototype is 23.7V (680k/56k). Incoming DC is at about 24.7V.

The problem only seems to occur with battery fully charged.

Could it be that the aiming potential is right on the edge of what is possible for the incoming DC? So the chip cannot quite decide whether DC is sufficient or not?

I will try dropping the upper resistor to 630k or 560k and see if that stops the problem.

Hey Daniel,

Thanks for all of the follow up information!

So if my understanding is correct - it is STAT that flashes only when a fully charged battery is inserted? In other scenarios, it works correctly?

Do you know if your battery pack has any internal protection? If it does, there may be that the gauge in the pack is turning off the protection FETs when the battery is full, effectively removing the battery from the charger. This would cause STAT to flash. In these scenarios, the protection in the battery pack will allow discharge through a FET body diode, but remove it from a charging circuit.

no, STAT is behaving itself fine now with the new chip. It is /PG that is doing this flashing on/off business. That's why I am wondering if the aiming potential set by the potential is borderline.

The issue seems to be that I was trying to set too high a charge voltage. Dropping the target voltage to about 22V with a 24.5V input seems to have resolved the problem. Thanks for the help!

Daniel,

I'm glad you got some resolution, and  that your theory was correct! 

It is interesting that lowering the battery voltage seems to have resolved the issue... before, it seemed as if there was enough headroom at VCC for this to be okay. With the 30ms time (this is the deglitch time listed in the datasheet) you indicated earlier and what we know now, i think we can chalk this one up to the sleep comparator. Now, the question is why? The earlier combination *should* have  enough headroom with a fully charged battery (Max value for sleep comparator is VCC>SRN by 150mV). 

Can I ask you to go back to the original setting, and measure  (or a scope shot), VCC, and SRN? That might give us some closure regarding what the root cause was.

Hi David,

The proto board is a bit messed up now for me to change any more components around unless I really have to. But I can tell you that incoming volts (Vcc) was around the 24.5 to 24.7V mark. SRP/SRN were at the battery voltage, which, if I had just completed a full charge with the old resistor values, was 23.7V, as expected from the resistor values of 68k/5k6.

Note that I seemed to need to fully charge the cells and immediately hot plug the external DC to reproduce this. Once /PG started cycling on and off, it did so constantly though, even when the cell voltage dropped. But if I waited until the cell voltage dropped by a few volts, the cycling happened rarely or not at all.

thanks!

Daniel

Daniel,


Hm, that is interesting. You are measuring VCC at the pin, and not the DC supply, correct?

I am wondering if the diode at VCC is causing the sleep comparator to trip (the voltages would be right at the borderline for this to happen), and then it is somehow latching at that state.

Maybe I'll play with our EVM when I get a chance, and try to replicate it.

ah, that was the DC supply coming in. Indeed, it's probably slightly lower, and there will be a small amount of ripple because of the LED current.

I linked out the diode between the sensing resistor minus pin and the battery, so SRN is connected directly to the battery (via the fuse of course).

I also have the EVM - i will try on that when I get some spare time.