Detecting bad connections

[upnorthandpersonal]

I was looking for a way to detect bad cell terminal connections in multiple parallel battery packs. I found out that this is actually very easy to spot if you graph the cell delta on the packs. I introduced a bad connection in one of my packs and let it run for a day. It went through charge and discharge, and seemed fine. No hot spots or anything. The arrow in the following screenshot show when the bad connection was fixed. Both screenshots are of the same data, just zoomed in.






Having bad connections in multiple parallel packs also show up with this kind of 'oscillating' behavior. I'm just posting this here in case it's useful for someone later on, or if you want to add this kind of check to your monitoring system.

 

[efficientPV]

The unseen world, there is always telltale behavior. I have an inverter which only runs a refrigerator. The startup current is fairly high and consistent. Whenever the fridge starts it stores five data points in the first second of battery slump. It has caught things a couple times. My first introduction to analysis was in the late 70's with a Japanese paper which could detect a signal 85dB below the average noise.

 

[upnorthandpersonal]

The unseen world, there is always telltale behavior.

Tell me about it. I build radiation detectors (as in alpha, beta, gamma and neutron) with one of my companies.

 

[740GLE]

What’d you do to correct the connection? Just tighten down a little bit? Doesn’t look like it was disassembled cleaned then reassembled.

How often do you retorque?

 

[svetz]

So what's the recommended technique to find a bad connection? Jumper over a connection and see if the oscillations disappear? Or is your BMS reporting per cell data?

 

[upnorthandpersonal]

What’d you do to correct the connection? Just tighten down a little bit? Doesn’t look like it was disassembled cleaned then reassembled.

How often do you retorque?

I introduced a bad connection by slightly unscrewing the nut on one of the terminals and later fixed it bu re-torquing.

Since my batteries are stationary, I never really check the connections - I never had issues.

 

[upnorthandpersonal]

So what's the recommended technique to find a bad connection? Jumper over a connection and see if the oscillations disappear? Or is your BMS reporting per cell data?

Data per cell is reported, and I calculate the cell delta from that. I introduced a bad connection on purpose to test by slightly unscrewing the nut. It's not something that is apparent from just a glance at cell voltages, or just even in regular use, but once you graph the cell delta over time it shows up.

Edit: to add, the 'bad' connection in this case was not the kind of thing you see as when people add a washer in between the terminal and ring terminal. It doesn't heat up, or show on thermal camera, or even shows up just in use. That was the idea of the experiment: can these small deficiencies that might turn into bigger problems later on (especially in mobile applications where you have vibrations) be detected early somehow. The answer appears to be yes: by looking at cell delta over time. Also note that this appears only to be present when there are multiple parallel battery packs - it's some kind of interaction between them, yet, the oscillations only show up in the pack that has the problem.

More research is needed. I'll likely publish these results in a paper since I have not seen this effect described anywhere. I'll run more tests first and gather more data...

 

[Skypower]

Tell me about it. I build radiation detectors (as in alpha, beta, gamma and neutron) with one of my companies.

So you’re a zoomie hunter?

 

[Skypower]

Data per cell is reported, and I calculate the cell delta from that. I introduced a bad connection on purpose to test by slightly unscrewing the nut. It's not something that is apparent from just a glance at cell voltages, or just even in regular use, but once you graph the cell delta over time it shows up.

Edit: to add, the 'bad' connection in this case was not the kind of thing you see as when people add a washer in between the terminal and ring terminal. It doesn't heat up, or show on thermal camera, or even shows up just in use. That was the idea of the experiment: can these small deficiencies that might turn into bigger problems later on (especially in mobile applications where you have vibrations) be detected early somehow. The answer appears to be yes: by looking at cell delta over time. Also note that this appears only to be present when there are multiple parallel battery packs - it's some kind of interaction between them, yet, the oscillations only show up in the pack that has the problem.

More research is needed. I'll likely publish these results in a paper since I have not seen this effect described anywhere. I'll run more tests first and gather more data...

Perhaps the number one reason people start having cell imbalance is a bad(or very subtle) connection. So small they can’t find it or won’t believe it’s of consequence.

 

[upnorthandpersonal]

umber one reason people start having cell imbalance is a bad(or very subtle) connection. So small they can’t find it or won’t believe it’s of consequence.

That is my thinking as well in many cases.