Reducing the LiIon charge termination voltage

[Stereodude]

I have a MightyHat R3 and it charges a Li-Ion cell to just short of 4.3V.  I measured the resting voltage of the 18650 I connected to it a few days ago at 4.28V.  The battery was just under 3.8V when I connected it.  Charging to this level is a bad practice for several reasons.  1) Most Li-Ion / LiPo cells should only be charged to 4.2V.  They're not supposed to breakdown and have catastrophic issues below 4.3V, but that's the maximum voltage, not the working limit.  2) Cell life will be degraded by leaving the voltage of the cell so high.

The practical limit should be 4.2V and frankly 4.15V would be better considering cell life.  I measured the voltage at 4.2V when the charging light went out, but didn't realize it would continue to trickle charge the cell even further rather than stopping.  I will not be using the charging circuit feature any further until I dig into the datasheet and determine how to modify the charging circuit for better behavior.

[Felix]

It's a MCP73831 4.2V charging voltage controller so it should not go above 4.2.
In my own tests, after the charge-LED goes off, the vbat is slightly less than 4.2 (usually 4.15 or so). I use this charger on all my LiPo powered boards.
So this is unusual. I can take a look at it if you want me to, I can replace the MCP with a new one if I find it's defective in some way. If there's no other hardware mistake or fault then swapping the charger chip would be the fastest route to see if there's a difference.
Or you can let me know what you find further and we can go from there.

[Stereodude]

It's a MCP73831 4.2V charging voltage controller so it should not go above 4.2.
In my own tests, after the charge-LED goes off, the vbat is slightly less than 4.2 (usually 4.15 or so). I use this charger on all my LiPo powered boards.
So this is unusual. I can take a look at it if you want me to, I can replace the MCP with a new one if I find it's defective in some way. If there's no other hardware mistake or fault then swapping the charger chip would be the fastest route to see if there's a difference.
Or you can let me know what you find further and we can go from there.

What is the open circuit voltage on the charger if you don't have a battery connected?  The red charge LED flickers on mine with no battery connected and I measure 4.29V.

Edit: I see from the datasheet for the MCP73831 that there is no way to trim the charge voltage.  I will check it again today with another DMM or two.

Edit2: I checked the open circuit voltage from the charging circuit at the leads on the board to the connector with 3 multimeters.  I got readings between 4.285V and 4.295V.

[Felix]

With USB connected:

- With no battery connected, it is 4.3 as you say
- With battery connected, it reads the battery voltage, and slowly charges up to < 4.2

I just put 2 types of batteries (a samsung smartphone battery and a lipo chinese generic battery) to charge, will monitor them for this and come back with results once charged LED turns off.

I also checked all my orders this past year and in all cases it was one of these 2 chargers, both "4.2V" rated:
https://www.mouser.com/ProductDetail/579-MCP73831T-2ACIOT
https://www.digikey.com/product-detail/en/microchip-technology/MCP73831T-2ACI-OT/MCP73831T-2ACI-OTCT-ND/1979802

[Stereodude]

With USB connected:

- With no battery connected, it is 4.3 as you say
- With battery connected, it reads the battery voltage, and slowly charges up to < 4.2

I just put 2 types of batteries (a samsung smartphone battery and a lipo chinese generic battery) to charge, will monitor them for this and come back with results once charged LED turns off.

I'm really interested in what the battery voltage is in the hours / days after the charge light turns off.  After the LED turned off my 18650 cell was 4.202V.  About 36 hours later it was 4.28V.

[Stereodude]

I decided to retest.  I connected up the same Panasonic/Sanyo 18650 cell, now at ~4.15V after discharging it some with a flashlight.  The red charging LED immediately turned off.  However, I noticed the voltage at the battery connector on the MightyHat is 4.178V and the voltage at the battery itself is 4.158V.  I realize it's only a difference of .02V, but that would strongly suggest that current is still flowing into the battery charging it.  I will check the voltage at both points again today after work.

Edit:  After a few minutes the red charge LED turned back on.

[Felix]

I think it may depend on the battery too and how it behaves (quality?).
I have not seen evidence that the charger is charging to a higher voltage than designed.
It is perfectly normal that the voltage is >4.2 during charging, that means charger is in constant current stage and current flows INTO the battery.

[TomWS]

Given that the voltage drifts slightly above 4.2V without the battery installed, it may be that the protection circuit on the battery is simply 'opening' up to prevent overcharge of the battery, allowing the voltage at the battery connector to rise above spec.  This could happen if the battery is fully charged.

If you have a LiPo battery where you can get to the actual battery terminals, this would be one way to check to see what is actually happening in this case. 

I'd be surprised if these battery managers overcharge.  The manufacturer (Microchip) is a good one and the Vreg spec is really tight (4.232 volts max).

Tom

[Stereodude]

I think it may depend on the battery too and how it behaves (quality?).

I'm using a very high quality 18650 cell.  It's a Sanyo/Panasonic NCR18650GA 3500mAh cell.

It is perfectly normal that the voltage is >4.2 during charging, that means charger is in constant current stage and current flows INTO the battery.

No it's not.  A proper Li-Ion charger should switch from CC to CV before the voltage exceeds 4.2V.

If you have a LiPo battery where you can get to the actual battery terminals, this would be one way to check to see what is actually happening in this case.

I'm charging a bare unprotected 18650 cell in a 18650 battery holder.  There is no protection circuit.  I have access to the actual battery terminals.  The voltage at the battery's terminals is over 4.2V.

I'd be surprised if these battery managers overcharge.  The manufacturer (Microchip) is a good one and the Vreg spec is really tight (4.232 volts max).

So would I, but the measurements seem to speak for themselves.

[Felix]

I'm using a very high quality 18650 cell.  It's a Sanyo/Panasonic NCR18650GA 3500mAh cell.
So would I, but the measurements seem to speak for themselves.

I dont know what else to do on my end, everything looks good. I had several lipos charge since yesterday. They settle around 4.15 after led shuts off, USB plugged or not.

If you want me to investigate more, you can send me the board you have, along with the LiPos you're testing. I will try my batteries with the board, then try your lipos with that board and other boards with the same charger.

Otherwise, you may want to bring this up with Microchip.

[Stereodude]

I dont know what else to do on my end, everything looks good. I had several lipos charge since yesterday. They settle around 4.15 after led shuts off, USB plugged or not.

If you want me to investigate more, you can send me the board you have, along with the LiPos you're testing. I will try my
batteries with the board, then try your lipos with that board and other boards with the same charger.

Well, I'm repeating the test today.  I will let you know the outcome.  Depending on the results we can look at next steps.

Note, I am not using a Lithium Polymer battery.  I am using a cylindrical Lithium-Ion cell.  However, they charge the same way and are interchangeable.

Edit: The battery was at 4.281V 12 hours after completing.  I'm in the process of running it back down to a lower voltage.  I guess I'm interested in getting you the board and the battery to test since I won't use this as is.

[Felix]

Check out this interesting video of some cells with/without protection circuitry.
Although the datasheet specifies a typical 4.2V regulated voltage for the MCP7383X-2 with a MAX of 4.232V , maybe the protection circuitry (or lack of it) has to do with this.

[TomWS]

Check out this interesting video of some cells with/without protection circuitry.

I'm not sure what this video adds to the discussion.   I think it's worthwhile checking the charger circuit with a discrete protection circuit connected to a bare battery so that voltage at the battery terminals and charger terminals as well as current flow can be simultaneously monitored.

Tom

[Felix]

I'm not sure what this video adds to the discussion.

My thought was maybe the MCP73831 assumes there is a battery protection circuit that drops some voltage. (since I cant find anything in the DS). But that seems to contradict the regulated voltage spec. Otherwise I have no clue how to explain the 4.3v vs 4.2 claimed.

[Stereodude]

My thought was maybe the MCP73831 assumes there is a battery protection circuit that drops some voltage. (since I cant find anything in the DS). But that seems to contradict the regulated voltage spec. Otherwise I have no clue how to explain the 4.3v vs 4.2 claimed.

That doesn't really make sense to try to account for that because the voltage drop would be variable.  The less current you're pushing or pulling into the battery, the less voltage drop it would have.  By the time you're in the CV stage and at 10% or less of the charging current about to switch off the charger, the voltage drop would be quite small.

Do you want me to send you a PM to get your address or how do you want to handle my sending the mightyhat + battery to you.

[Felix]

Before we do a return, do you happen to have other LiPos to test this with?
(Just for comparison sake)

[Stereodude]

Before we do a return, do you happen to have other LiPos to test this with?
(Just for comparison sake)

I have a number of cylindrical Li-Ion cells of different sizes from several makers.  I'll have to see if I have any bare single cell LiPo batteries.  I have a number of 3S and 4S packs, but that doesn't help me here.

[Felix]

I meant lipos with protection circuitry, the ones commonly found online.
Even batteries from cell/smart phones would work (not fakes like in the video I linked) and you'd know they have protection.
Let me know what you want to do.

[Stereodude]

I meant lipos with protection circuitry, the ones commonly found online.
Even batteries from cell/smart phones would work (not fakes like in the video I linked) and you'd know they have protection.
Let me know what you want to do.

I have cylindrical Li-Ion cells with integrated protection circuits also.  However, I won't be able to measure the true battery voltage as I don't have access to the cell side.

What exactly do you want me to test and measure?

[Felix]

The VBAT of the MCP73831, during and after it's charged, of a LiPo that has its own protection circuitry.
You can bet the internal voltage is lower if it has such a circuit.

[Stereodude]

The VBAT of the MCP73831, during and after it's charged, of a LiPo that has its own protection circuitry.
You can bet the internal voltage is lower if it has such a circuit.

So a protected battery was 4.205V at the terminals after charging.  As soon as you flip the switch the voltage drops to 4.19V.  The protection circuit is stopping the battery from being overcharged.  The charger circuit is still trying to charge the battery.

Also, the red LED never came on while charging the protected cell and the LCD got all wonky and scrambled.

[Uncle Buzz]

Since the diode of the P mosfet doesn't block current from VUSB_OR_VBAT to VBAT if it's higher than VBAT + the mosfet's Diode forward voltage , (which should not be the case with common values), maybe some current are flowing to the battery through the mosfet without be controlled by the charger MCP73831 ?

[Felix]

Uncle Buzz,
You mean PMOS1, when USB is active and PMOS1 supposed to be off?
The PMOS1 body diode voltage is -0.8 with a max of -1.2. Add to that the schottky and we're above 5v already, so I don't think there is a reasonable case that 5VUSB is charging an unprotected battery unintentionally.
The other thread about inverted PMOS1 suggests this could cause a bit of extra voltage in the battery if its unprotected.
But I dont see that with any unprotected batteries, which is only what I have used.

[Felix]

Stereodude,
Given the comments by UncleBuzz, if VUSB was much higher than 5V maybe that can explain the overcharging. Can you check with the 4.3V on the Lipo, what the VUSB is?
In any case with a protected LiPo this should never happen. Plus the protection circuit offers other protections not just overcharging, which is why I would recommend anyone to use a protected cell, more is better when it comes to protection circuits and Lithium cells.

As a "fix" for this edge case (unprotected cell, VUSB>>5V causing VUSB drainage into VBAT):
Inverting the PMOS1 should avoid this problem, or maybe using a schottky instead of the PMOS1 source->drain.
I can do either for you, if you really want to use the unprotected cell, and are willing to do the return and wait a few days.

[Uncle Buzz]

The PMOS1 body diode voltage is -0.8 with a max of -1.2. Add to that the schottky and we're above 5v already

It's what I meant when I said it wouldn't be the case with common values, but is a lower body diode voltage possible on this board in particular ? higher VUSB ? It's just a possible scenario you didn't check even if it's unlikely.

0.8V (minimum body diode voltage) + 0.3V (forward shotky diode voltage, don't know the real value of yours) + 4.2V = 5.3V, so any VUSB > 5.3V will flow current into a full charged battery through PMOS1 (with those values)

But it could be less because of the forward shotky diode voltage which is not a constant value but depends of the current flowing through it, and with some mA or even µA, this value could be lower (but needs lot of time to overcharge the battery)

EDIT : for the low current through the shotky, it's possible only if you can shutdown or sleep anything powered by the mightyHat, but maybe on the same way, the body diode voltage of the PMOS could be lower with very low current ?

[TomWS]

Any leakage through PMOS1 will charge a battery in CV mode and, since PMOS1 body diode is forward biased, there will be some leakage, no matter what the current.  In this case, the MCP73831 is completely off and is not responsible for the increased charge.

A protected battery will prevent overcharge, but you'll still get a higher voltage at VBat (during Charging).  I would not use an unprotected battery with this circuit.

Tom

[Stereodude]

Plus the protection circuit offers other protections not just overcharging, which is why I would recommend anyone to use a protected cell, more is better when it comes to protection circuits and Lithium cells.

Yes, over-discharging and overcurrent protection.  Given that you've designed the circuitry and the base sketch I would hope that you've already taken those into account.  Neither will offer thermal protection on the battery.

Also, I didn't see any mention anywhere on the MightyHat pages that a protected LiPo/Li-Ion battery was required.  Perhaps you should add that.

I can check what VUSB is on my board later when I get home, but your board has the same 4.3V open circuit voltage on the battery terminals too.  So it's not just something specific to my setup.

[Felix]

Also, I didn't see any mention anywhere on the MightyHat pages that a protected LiPo/Li-Ion battery was required.  Perhaps you should add that.

Good point, I should add something. It's not really required, except until the next revision when this is fixed in the circuit. I would say it's recommended. And yes the MCP73831 is just a nice basic charger, does not include thermal protection.
5.2V+ on USB is not normal either

[TomWS]

5.2V+ on USB is not normal either

But within USB standard spec...

[Felix]

But within USB standard spec...

Quite barely. It's actually:
Limits: 5.00V+/-0.25. But this is upstream from a device. Cables drop some voltage (ball park it 50mV). So you end up with ~5.2 (or less with china cables?). Measure 5.2+ at the device and I call it out of spec.

[Stereodude]

Vusb is 5.193V with the Rpi turned off and just the MightyHat on.  The voltage seems to rise slightly with it turned on to 5.205V.

Also, the LCD seems permanently garbled now. 

Edit:  The LCD went back to normal after cycling through the backlight brightness settings.  It previously stayed garbled through power cycles, resets, etc. 

[Felix]

Vusb is 5.193V with the Rpi turned off and just the MightyHat on.  The voltage seems to rise slightly with it turned on to 4.205V.

Also, the LCD seems permanently garbled now. 

Edit:  The LCD went back to normal after cycling through the backlight brightness settings.  It previously stayed garbled through power cycles, resets, etc. 

You might be right on the edge. If the schottky drops 0.2V and the PMOs1 another 0.8V through the body diode, then we're around 4.2 which can explain the slightly higher 4.205 in your battery.
The solutions were already mentioned, but for your trouble, I can offer to fix it for you if you want to return it.

[Stereodude]

You might be right on the edge. If the schottky drops 0.2V and the PMOs1 another 0.8V through the body diode, then we're around 4.2 which can explain the slightly higher 4.205 in your battery.
The solutions were already mentioned, but for your trouble, I can offer to fix it for you if you want to return it.

I had a typo there that I've fixed.  However, I will see how the MightyHat behaves with a 5V different power supply (closer to 5V) before I do anything else.

[Felix]

Ok no problem. Let me know..

[Stereodude]

Ok no problem. Let me know..

After connecting a more tightly regulated 5V supply connected to the barrel jack of the MightHat the behavior of the board changes noticeably.  The wall wart I'm using measures 5.06V open circuit / 5.03V when connected to the MightyHat with the RPi3 turned off / 4.89V with the RPi3 on.  The open circuit voltage of the charging circuitry with the RPi3 off is only 4.13V.  It drops even lower and bounces around with the RPi3 turned on.  It settles around 4.03-4.04V.

What is the Red Charging LED actually supposed to do?  With no battery connected it's now off (with the RPi3 off) instead of flickering like it did before.  However, once I turn the RPi3 on, the charge LED turns back on (with no battery connected).

I will charge a Li-Ion cell and report the results back later.

[Uncle Buzz]

What is the Red Charging LED actually supposed to do?  With no battery connected it's now off (with the RPi3 off) instead of flickering like it did before.  However, once I turn the RPi3 on, the charge LED turns back on (with no battery connected).

While your RPI is off, the MCP73831 "see" a "battery" voltage through the body diode of PMOS1 at about 4.13V which is almost the value of a full charged battery and there is no current flowing into the battery, since there is no battery and PMOS1 is off. So the charging led is off, but when you powered your RPI, your USB voltage drop and the voltage seen by the MCP73831 follow, so it could try to charge this lower battery voltage and  lights the charging led

[Stereodude]

I will charge a Li-Ion cell and report the results back later.

So, with the RPi3 off the battery charges to about 4.13V.  With the RPi3 on the battery charges to 4.205V (the protection cell is not limiting the voltage as this battery will charge to 4.25V before the protection circuit opens).

Edit:  Removed incorrect conclusion

[Stereodude]

It's what I meant when I said it wouldn't be the case with common values, but is a lower body diode voltage possible on this board in particular ? higher VUSB ? It's just a possible scenario you didn't check even if it's unlikely.

0.8V (minimum body diode voltage) + 0.3V (forward shotky diode voltage, don't know the real value of yours) + 4.2V = 5.3V, so any VUSB > 5.3V will flow current into a full charged battery through PMOS1 (with those values)

But it could be less because of the forward shotky diode voltage which is not a constant value but depends of the current flowing through it, and with some mA or even µA, this value could be lower (but needs lot of time to overcharge the battery)

EDIT : for the low current through the shotky, it's possible only if you can shutdown or sleep anything powered by the mightyHat, but maybe on the same way, the body diode voltage of the PMOS could be lower with very low current ?

So after thinking about this more...  Is the already charged battery the worse case scenario with this circuit design?

What happens after a power failure when the battery is depleted (from powering the Rpi), the Atmel chip has turned off the RPi (due to battery depletion), and then power is restored?  You'd now have a battery at something like 3.0V and 5V on the input.  Since there's a voltage differential dependent current path from the input to the battery around the MCP73831 couldn't there be quite the current surge into the discharged battery?