Using bench PSU instead of a battery -- getting low battery indicator

[morfi717]

Hello folks.

So I've been trying to debug a faulty circuit of a bicycle flashlight. It was powered by a double li-on cells in parallel of 3.6V without any smart battery circuit -- just plain positive/negative terminals.

Now for the purpose of debugging, I soldered out the batteries and connected a bench PSU. I've measured the voltage across the batteries terminals and applied the same voltage on the PSU.
However, the tested circuit was flashing one of its LEDs indicated that the battery is low. I tried to slightly increase the voltage on the PSU up to 3.85V but no budge. Once I connected the batteries back again, there was no low battery indicator anymore. I've probed the voltage on the pcb and there wasn't any voltage drop across the wire (that is, the psu <--> pcb wire).

Unfortunately I had to give the flashlight back to the owner, but it keeps bugging me what property does a battery cell have that the bench PSU doesn't, which could cause the battery health circuit to refuse to work with the PSU.
I know that without a schematics it's a guessing game but perhaps someone had experienced that himself in the past.
Even wild guesses would be appreciated as I might encounter the same issue in the future and a heads-up ideas would help a lot :-)

[IanB]

A normal lithium ion cell would be close to 4.2 V when fully charged, and around 3.6 V when discharged. So if you set the bench power supply to 3.6 V you are essentially simulating a discharged cell.

A second thing a battery monitoring circuit may do is look at the internal resistance, or the voltage drop under load. Many bench supplies will not be able to match the observed behavior of a lithium ion cell when a load is applied, and will tend to go into current limiting mode, making the circuit think there is a faulty cell.

In future if you repeat the experiment, look at the manufacturer code on the cells and check the data sheet, then try using the full charge voltage on the power supply. On the other hand, you said the circuit was faulty. If the circuit is faulty, it may not respond correctly.

[BlackICE]

The typical 18650 cell is around 4.2 or 4.3 volts fully charged and fully discharged maybe 3.3 volts depending how far down you want to go. So it could be that the flashlight is a little conservative on the discharge voltage and prefers to see something above the 3.85 volt you provided

[radiolistener]

A normal lithium ion cell would be close to 4.2 V when fully charged, and around 3.6 V when discharged. So if you set the bench power supply to 3.6 V you are essentially simulating a discharged cell.

4.2 ±0.05 V is the max allowed voltage during charge.
2.0 V is the min allowed voltage during discharge, but it's not recommended to discharge it to 2.0 V, it's better to stop at about 3-3.7 V.

100% charged li-ion cell has about 4.15 V.
90% charged li-ion cell has about 4.06 V.
80% charged li-ion cell has about 3.98 V.

20% charged li-ion cell has about  3.74 V.
10% charged li-ion cell has about  3.68 V.
5% charged li-ion cell has about 3.45 V.
0% charged li-ion cell has about 3.0 V.

As you can see, below 10% charge voltage drops down very quickly.
So it's better to stop charge at about 3 V to prevent more fast cell degradation.

[wasedadoc]

So I've been trying to debug a faulty circuit of a bicycle flashlight. It was powered by a double li-on cells in series of 3.6V without any smart battery circuit -- just plain positive/negative terminals.

If really two cells in series you would need to set the bench supply to around 7.2 Volts.

[radiolistener]

If really two cells in series you would need to set the bench supply to around 7.2 Volts.

7.2 Volts is close to a full discharge - something about 5-7% charge.

For 2 cell li-ion 100% = 8.4 Volts.

[wasedadoc]

If really two cells in series you would need to set the bench supply to around 7.2 Volts.

7.2 Volts is close to a full discharge - something about 5-7% charge.

For 2 cell li-ion 100% = 8.4 Volts.

If the "low batt" indication was not happening when using cells which each measured 3.6 Volts then it should not happen when the PSU is set to 7.2. That is a separate issue from emulating two fully charged cells.

[morfi717]

Sorry I meant in parallel not series:-)

so 3.85 shouldnt show a "dead" battery (not fully charged either).

I couldnt see any load on the PSU during the battery level check though perhaps the current draw was too low for my low end psu to show that on the display.

The flashlight doesnt seem to be conservative as the cells were ~3.64V at the time of experiment and the low bat was not flashing when cell was plugged in.

Dont know about internal resistance of the battery, I thought voltage check is more common/standard? Does it have anything to do with psu having different negative reference (ie ground) compared to what battery cell negative terminal is like?

[radiolistener]

so 3.85 shouldnt show a "dead" battery (not fully charged either).

For a signle cell li-ion battery 3.85 V is about 55% of charge, it depends on the device what it should show at that voltage.

[SmallCog]

Possibly a current issue? Bench supply may be sagging as the LED turns on?

[tunk]

You can find many discharge curves here (by eevblog member HKJ): https://lygte-info.dk/
Do you know how much current it draws?

[morfi717]

The bench supply was drawing little to no current and had set a limit of 3A. If it was about too much current being drawn by the flashlight, I would see on the display that the voltage has been dropped.

The low battery indicator is being shown even before the actual (ie. lamp) LEDs are turned on.