MB Air 2017 can no longer start from battery

[Minipudding]

My 13" 2017 Air suddenly died today from power loss while running on battery with ~80% charge, and now it will not power up at all from battery. If I remove the battery it powers up fine from the power adapter. If I measure the battery's voltage when disconnected (unloaded) I get nominal 7.8 volts. The battery is a third-party battery with about 300 charge cycles on it, giving me some 6-7 hours of use, and up until now it has never showed any issues.

I've tried resetting the SMC with the battery connected, though it changes nothing. I don't know if the battery spontaneously died or the battery charge/regulator circuit on the mainboard died. Either seems likely at this point.

Anyone with a similar experience?

 

[real_mon2]

Check the voltage to ground of each pin on the battery connector. Interested to know the values of the smbus SDA and SCL lines which are used by the logic board to extract details of the logic board. Each of these pins should be ~3v as they are open drain pins with local pull up resistors. Post your measurements.

 

[A Hobo]

The battery is a third-party battery with about 300 charge cycles on it, giving me some 6-7 hours of use, and up until now it has never showed any issues.

It is now..
not shaming you but these batteries are usually junk fabricated in a recycling centre in china. I'm surprised it gave you 300 cycles as it is, it probably already had 900 cycles wiped on the chip before resale.

 

[real_mon2]

As per suggestion above, do test with a known good battery - Amazon is great for this. In general, 3rd party batteries are often rated to operate for ~500 charge cycles where as the original (Apple) is ~1,000 charge cycles. Not carved in stone, but the original will last longer and of course will be more expensive. Otherwise, you will need a multimeter to diagnose other logic board faults which may be due to the SMBUS pin communication.

 

[Minipudding]

Check the voltage to ground of each pin on the battery connector. Interested to know the values of the smbus SDA and SCL lines which are used by the logic board to extract details of the logic board. Each of these pins should be ~3v as they are open drain pins with local pull up resistors. Post your measurements.

7.8V on the three voltage pins. SCL, SDA and SYSDETECT to ground measure 0V (on disconnected battery)... The battery's microcontroller appears to have died, or possibly the grounded resistor on the SYSDETECT line has fried and the circuit is open.

 

[Minipudding]

It is now..
not shaming you but these batteries are usually junk fabricated in a recycling centre in china. I'm surprised it gave you 300 cycles as it is, it probably already had 900 cycles wiped on the chip before resale.

It certainly did not have 900 cycles on it before shipping. If that were the case it would not have given me 8-9 hours of usage during the first 9 months. It's obviously not comparable to Apple's original batteries, but that was never expected from a Chinese $30 battery. The cells themselves on this battery are still fine, but the microcontroller on it seems to have died.

 

[real_mon2]

The SMBUS lines have a local pull-up resistor on each leg to park the lines to ~3 volts each. Now measure the battery connector on the logic board while it is powered. It is ok to connect the battery since that is where the fault occurs.

Do you have ~3 volts on the SCL / SDA lines? If not, then the local ESD transient diode is likely damaged but can confirm after your next reply.

 

[Minipudding]

The SMBUS lines have a local pull-up resistor on each leg to park the lines to ~3 volts each. Now measure the battery connector on the logic board while it is powered. It is ok to connect the battery since that is where the fault occurs.

Do you have ~3 volts on the SCL / SDA lines? If not, then the local ESD transient diode is likely damaged but can confirm after your next reply.

Pardon the delay. 3.4V on SCL, 0.03V on SDA. Definitely looks like that one diode is out.

 

[real_mon2]

Yes indeed. Often it is the dual ESD tvs diode that is present to shunt the ESD transients. It is sacrificial and if blown, it dies but has protected the transient from going further onto your logic board. I am not at my desk but will review once at the office. Do you have schematics? From memory, look for D6950. This part if my memory is correct is the part that should be removed off the board. You will need tweezers and suggest flux to remove with a hot air tool. Will post back after confirming against the schematics and boardview files. PS: took us 3 days to locate this fault on a similar unit on our first case. We documented the full process on badcaps forum.

 

[Minipudding]

Yes indeed. Often it is the dual ESD tvs diode that is present to shunt the ESD transients. It is sacrificial and if blown, it dies but has protected the transient from going further onto your logic board. I am not at my desk but will review once at the office. Do you have schematics? From memory, look for D6950. This part if my memory is correct is the part that should be removed off the board. You will need tweezers and suggest flux to remove with a hot air tool. Will post back after confirming against the schematics and boardview files. PS: took us 3 days to locate this fault on a similar unit on our first case. We documented the full process on badcaps forum.

I don't have the schematics, would appreciate them or just a picture of a logic board with the zeners marked.

 

[real_mon2]

I don't have the schematics, would appreciate them or just a picture of a logic board with the zeners marked.

Will do. In the meantime, can you post the 820-**** markings of your logic board? If not known, post a pic of the logic board to help us identify which schematic to study. This markings is etched in copper on the board itself.

Update:

Post the A*** marking from the bottom cover of the laptop. Is your model A1466 ?

Does your logic board shape look like the attached?

If yes, then the TVS ESD diode is NOT stuffed on this model so the fault is elsewhere. Disconnect the battery and confirm the voltage to ground again at the battery connector on the logic board. Do you see the SDA & SCL pins being pulled up to ~3v42. This should be the voltage reading unless the SDA rail is defective or not being pulled up. The pull-up resistors for the SDA pin is @ R5380.

See attached for the schematic & board views.

Let us review the DIODE mode readings of the same SDA / SCL pins. Remove all power. No battery.

Meter in DIODE mode.

Red meter probe, yes red to ground. Can be a metal shield.
Black meter probe to the point to test.

Check the diode mode reading of the SCL pin. We know this pin is OK so it should be ~0.5 or perhaps higher. If the diode mode reading is too low then there is a short on that line.

Repeat for the SDA pin which is really of interest since it is parked at 0v.

Respectively, wondering if the SDA pin tanks to 0v (stuck) once the battery is mated. In this case, the battery pack is clearly defective (BMS board is damaged and is pulling down this line).

Without the battery pack, both SDA & SCL must return to ~0v5 or higher to denote there is no short condition. When power adapter is connected (without the battery), these lines should be ~3v42 each.

The same once the battery is connected.

 

[Minipudding]

Post the A*** marking from the bottom cover of the laptop. Is your model A1466 ?

Does your logic board shape look like the attached?

Board revision isn't visible at all on the exposed side of my board. Maybe on the flip side? I cannot disassemble the laptop at this point as I'm actively using it, but the model is A1466 (EMC 3178). Same contours as in your picture.

If yes, then the TVS ESD diode is NOT stuffed on this model so the fault is elsewhere. Disconnect the battery and confirm the voltage to ground again at the battery connector on the logic board. Do you see the SDA & SCL pins being pulled up to ~3v42. This should be the voltage reading unless the SDA rail is defective or not being pulled up. The pull-up resistors for the SDA pin is @ R5380.

With disconnected battery and power adapter plugged-in both SDA and SCL measure 3.42V exactly. With battery connected only SCL measures 3.42V while SDA drops to 0.0-something volts.

Check the diode mode reading of the SCL pin. We know this pin is OK so it should be ~0.5 or perhaps higher. If the diode mode reading is too low then there is a short on that line.

Repeat for the SDA pin which is really of interest since it is parked at 0v.

Yes, both zeners show a drop of 0.5V in diode mode.

Respectively, wondering if the SDA pin tanks to 0v (stuck) once the battery is mated. In this case, the battery pack is clearly defective (BMS board is damaged and is pulling down this line).

Without the battery pack, both SDA & SCL must return to ~0v5 or higher to denote there is no short condition. When power adapter is connected (without the battery), these lines should be ~3v42 each.

The same once the battery is connected.

All of this checks out. Measurements without any power, and with power adapter only, match. As soon as the battery comes in SDA goes to 0 volts.

 

[real_mon2]

Aha moment. The battery is the fault but ironically not for the reason we suspected. It may very well be that the battery chemistry is fine but the BMS (controller board) is making it look bad as a whole due to the short on the SDA line. This breaks the communication between the main logic board and the battery pack. If the battery BMS board is visible on the battery pack, you can study it to locate the TVS diode which if removed, should fix the battery pack. Simple solution is to just replace the battery pack.

 

[Minipudding]

I've ordered a replacement battery, and will have a look inside the dodgy one just out of curiosity to see if i can revive it as a spare. Huge thanks for all the valuable information and help with troubleshooting, @real_mon2.

 

[rampancy]

So my main "mission critical" machine (2017 13" MacBook Air, A1466) has been experiencing the same behaviour, and it's puzzled me quite a bit: My battery unexpectedly dies at ~80%, even though the system thinks the battery is fine. It used to die at 50-60%. It's interesting to see this deep dive into the nuts and bolts of the battery itself – I'm thankful it's most likely an issue with the battery controller, as opposed to my motherboard.

I've ordered a replacement from iFixit, which hopefully gives me a better guarantee of a higher quality battery vs. an Amazon replacement.