Detecting current in an aluminum fusible link on a 18650 battery

[Mr Style]

Hello,
I am attempting to repair the high voltage battery in a 2012 Tesla Model S.
The battery pack consists of 16 modules.
Each module consists of 6 sections called bricks.
Each brick consists of qty 72 lithium batteries of the 18650 type.
All 72 batteries are housed in between 2 plates that act as bus bars.
Each individual battery cell has a short piece of aluminum wire connecting the anode to one bus plate and the cathode to another bus plate. This wire is ultrasonically welded.

One of the battery cells is going resistive and draining the other 71.
Since they are all connected in parallel, they all have the same voltage reading.
Cutting the wire would allow me to test each cell individually, but then I would need to replace 72 wires. I don’t have this wire or the ultrasonic welder.

I’ve attempted to charge the battery and look for a heat difference between the batteries with a FLIR one camera.  No difference.

The cells currently are at around 3.6965 vdc and drop about 10-20mV per day.
The other bricks in the pack do not change voltage, so they are fine.

The goal is to locate the bad cell and isolate it by cutting the fuses on each end.  The brick will have a slight less capability but only like 1.5%.


I am looking to design something that can do the following:
Detect DC current flowing into or out of a cell through this fuse wire.
The wire is short and close to the cell. I can clip onto it or press next to it, but there is not enough space to use a current clamp like on a CT or other toroidal current meter around it.
The only device I’ve found that can read current as a probe is the iprober 520, but it’s $800 new.
I was wondering if a circuit could be designed using a Hall effect sensor that could detect the current if the face of the IC Chip was pressed against the wire.
I’m not looking for a reading or accuracy. I’m looking for comparing against the other cells current.
I think this cell would be taking in current and the others are giving it. So the current direction would be different.

Let me know what you think.
A photo of the cells and the fusible links is attached. For those not familiar with the 18650 battery just think of it like an AA battery cell.

[J-R]

What's the best DMM you have?

I made some assumptions and came up with a range of 0.05A to 0.3A of drain via the one bad cell.  So maybe you could detect that current flow via the voltage drop in the wire.

The good cells each contribute a tiny amount that isn't very detectable, but the bad one(s) should be relatively easy to spot.

Making more assumptions about the resistance of the wire, I think you will want a DMM than can measure down to at least 0.01mV.  0.001mV would be better.

[J-R]

Update: I ran some extreme real world tests to check my assumptions.  I pushed 10mA through a piece of wire that looks about the same size as what is in the photo (24AWG?).  Using my BM869s in 500,000 count mode, I measured 0.004mV when the test probes were about 1/4 inch apart.  I could swap the leads and/or the current flow direction and see this change.  A 6.5 digit DMM with a 100mV range will make this a trivial operation, IMHO.

[Mr Style]

Thanks. I will give that a try. I appreciate you going the extra step of the real world testing too.

A search on the Tesla Motors club says it’s .28mm aluminum wire.
https://teslamotorsclub.com/tmc/threads/tesla-cell-level-fuse-bondwire-material.45892/

[J-R]

OK, 0.28mm is about 29AWG, so that is even better for this since the voltage drop will be higher.

Using an online calculator, 1/2" of aluminum wire at 0.28mm diameter has a resistance of about 5.5m Ohms.

So with my previous guestimate of 50mA to 300mA, you could see a voltage drop across that wire of 0.275mV to 1.65mV.  This is looking really good even with a basic DMM...

[Mr Style]

Thank you . This worked rather well. 
I ended up pulling a load of about 150 amps at 3.6VDC from the brick of 74 lithium ion 18650 cells that are connected in parallel.  So about 2 amps per fusible link. I used car battery jumper cables with a heat sink for the dump load.
I ended up with an average reading of 6-10mV per cell when reading across each fusible link.
Using ohms law that meant each fusible link has a resistance of 3 to 5 milliohms. (V/I=R so 6-10mV/2 amps=.003 to .005) so your right there with your resistance estimate!
The higher the voltage reading, the more current coming from a cell.
I did have one cell with a negative reading (drawing current in instead of giving current out). So I think I was able to find the bad cell.  There were several others that also had a lower voltage reading of 2 or 3mV so they are kind of “semi retired” cells coasting along but at least not working against the rest of them.

I appreciate your insight. 

I have purchased a used Agilent 6 1/2 digit DMM from eBay that will let me check cells without having to draw as large of a load.  Still testing at high current does show the “worst” performance of the cells so it’s a good way to test.

[J-R]

Oh, actually I was proposing testing while the pack was idle.  The bad cell is still drawing current from the good cells.

[PlainName]

The bad cell is still drawing current from the good cells.

How about a (magnetic) compass? If it's sensitive enough it will either show which wire is deflecting it or, if it's too sensitive, one wire will deflect it in one direction and all the other is the opposite direction.

[microbug]

A current sense amplifier like the INA181 might be a good way to do this with your existing multimeter.