using voltage sag as indication of bad battery

Hummina Shadeeba · Mar 17, 2019

can cells be relied on to perform consistently enough that a software program could pick up on an increased voltage drop under a given current draw from a complete battery and indicate a problem? It would first have to get to know how the battery behaved then if a cell or cells were damaged or disconnected or overdrawn the slightly increased voltage sag produced would be recognized by the program The goal being monitoring the whole pack accurately enough it could be done acumulatively as apposed to watching each cell with a bms

neptronix · Mar 18, 2019

Yes.. voltage sag will increase as IR goes up and capacity goes down. The biggest variable will be temperature, as temperature has a sizeable influence on IR. So the before and after tests need to be done at roughly the same temps.

Hummina Shadeeba · Mar 18, 2019

it seems a simple program (to someone who doesn't know how to program). If it were to work it would save space, time, money possibly replacing a bms if it were good enough, no? surprised no one has tried making it or probably someone has.

what happens with current and voltage in the WHOLE battery when a single cell goes bad? (the rest in the p group and the rest of the pack)

spinningmagnets · Mar 18, 2019

In the DIY home power-wall community, there is a growing use of thermal cameras to identify a hot cell, long before a fuse might blow from an internal short.

Of course the most affordable thermal cameras are not as precise, and have a coarse resolution, but...they appear to be "good enough".

billvon · Mar 18, 2019

Hummina Shadeeba said:
it seems a simple program (to someone who doesn't know how to program). If it were to work it would save space, time, money possibly replacing a bms if it were good enough, no?

No. The primary reason you need a BMS on a li-ion battery is to monitor for cell imbalance, not for remaining life.

what happens with current and voltage in the WHOLE battery when a single cell goes bad? (the rest in the p group and the rest of the pack)

Depends on how the cell goes bad. A common failure is a weak cell that sees extremes of discharge and charge starts weakening, making the ESR of the entire battery go up - and causes the battery to age even faster.

Hummina Shadeeba · Mar 18, 2019

billvon said:
Hummina Shadeeba said:

it seems a simple program (to someone who doesn't know how to program). If it were to work it would save space, time, money possibly replacing a bms if it were good enough, no?

Click to expand...

No. The primary reason you need a BMS on a li-ion battery is to monitor for cell imbalance, not for remaining life.

it wouldn't be monitoring for remaining life but more so integrating into the program the drop in capacity and subsequent voltage drop at full discharge.
if the cell is over discharged it will increase its internal resistance and will sag more and maybe a program could tell if individual cells within a p group had gone beyond that point. the bms likely has other safety features to stop extreme discharges like in a short and they would assure all P groups are fully charged but even doing that the bms is not monitoring individual cell voltages and if one were to die a not dramatic death that would be picked up on by the bms it woudnt be known about but with a program monitoring minute amounts of voltage sag maybe it could be spotted. maybe

liveforphysics · Mar 18, 2019

Every BMS I've worked with in the last 8 years tracks both the state of health in capacity as well as the state of health in impedance.

I've not been working with alibaba ebike BMS's though, which are lucky to just balance a pack without killing it themselves.

Hummina Shadeeba · Mar 18, 2019

the bms and even the vesc will typically shut down when the whole pack gets to a certain low voltage but it doesn't take into account the amp discharge that's being done to hit that low voltage cutoff so I could go up a mountain with a crappy battery and have huge sag and the vehicle will shut down, but if the bms or vesc incorporated the wattage output with the sag it could recognize how strong the battery really was and maybe to such an accurate degree it would recognize an abnormal change. you guys all get it. It would have its limitations in not really being a bms and many possible safety benefits there but it would take no space, have no dangerous wires (or possibly dangerous bms), surely be cheap, and give security in knowing that it would keep you from over-discharging ANY cell or cells in theory. It would be a crappy balance job keeping you just a step above the cell voltage/capacity cliff, or maybe you would potentially fall off and it would be that causing the a recognizable sag and a subsequent alarm. for me I like it as I don't use a bms and check cells periodically (or obsessively). Maybe the temp variability would be too much but otherwise it seems feasible, so far.

billvon · Mar 18, 2019

Hummina Shadeeba said:
it wouldn't be monitoring for remaining life but more so integrating into the program the drop in capacity and subsequent voltage drop at full discharge.

Most BMSes with a "remaining capacity" display do a form of coulomb counting to gauge remaining capacity; it's generally easier and faster than a formulaic voltage-based approach (and to get a good measure of ESR you have to measure current anyway, which is the big "cost" with a coulomb counter.)

Very cheap batteries sometimes have a four or five LED bargraph based only on resting voltage. Cheaper.

That being said, many systems do both. My bike uses a Lunacycle battery with a built in BMS that does only one thing - shuts down when any cell goes out of range (or there is an overcurrent.) There's also a cycle analyst that does a gas gauging algorithm that looks at 1) a coulomb counter, 2) resting pack voltage and 3) setup info on # of cells. This means that there has to be two shunts (=more cost, more voltage drop, not elegant) but that's pretty minor.

the bms likely has other safety features to stop extreme discharges like in a short and they would assure all P groups are fully charged but even doing that the bms is not monitoring individual cell voltages and if one were to die a not dramatic death that would be picked up on by the bms it woudnt be known about but with a program monitoring minute amounts of voltage sag maybe it could be spotted. maybe

That would be nice. Could it tell between the gradual change in ESR that all cells see and a more dramatic change of a single cell within a good pack? It would be tough, since those "signatures" are very similar. But maybe.

Here's a thought. In a typical 14s pack, any one cell group failing will change the midpoint voltage, since either the 'top' or 'bottom' group of cells will contain the bad group - and the midpoint voltage will no longer be exactly half. You might be able to use that as an easier option than monitoring all 14.

markz · Mar 18, 2019

Wouldnt one just get like an Adruino, and a ton of balance wires, along with some precisely placed temprature probes. Then have a neat little LCD screen. You can either eyeball the sag, or be even more fancy about things and have a program to monitor it then give you a visual indication with a l.e.d. if things are going smoothly.

A more fancy version of the Hobby King 8S meter with lcd display.

Hummina Shadeeba · Mar 18, 2019

Mark I guess with that method you would be comparing the voltage by eyesight of the different p groups. that would work and be pretty simple but I figured a computer that knew what sag to expect and incorporated the temperature effects and the cycles of the battery would be better than me looking at a screen to spot a discrepancy between p groups and then wouldn't need to wire up each p group to a voltage monitor.
...which bring up how a pack will behave when a cell goes bad or is over-discharged? lets say the battery is 10s 4p and one cell in group 10 becomes over-discharged or completely fails yet the complete battery is putting out 20 amps. What will happen to the bum cell and what will happen to the other 3 in parallel and what would happen to the other 9s?

dogman dan · Mar 19, 2019

Being an electronic dumbass, I have found it fairly simple to watch the thing charge to judge which P group is weak. It charges fastest, being the low capacity one. Once you see which cell or cell group fills first consistently, you can watch just that group for both sag and resting voltage with a simple voltmeter, and know all you need to know.

This does require a plug you can put a cellog 8 on, or similar device though. You don't get that in a commercial e bike pack though. I used to tell Jason we needed this, but behind a sticker that voids the warranty if you remove it. The idea was that a bike shop tech could then just plug a device into a 14 pin plug, and then watch each groups voltage as a charge happened. The results would then be used to justify warranty replacement of the pack, or not. This diagnostic port would let even a fairly inexperienced bike shop tech tell if a battery was bad, or just needs a balance.

I put some 8s balance plugs on a ping pack once, just so I could see if it was going bad or not, at 2 years old. You could see the bms working, or not, at top of charge, and track if one cell was consistently the first to fill.

Part of why I decided to stick with lipo after my garage fire is that it comes with the balance plugs. The other was low cost, for the c rate. After that fire, I definitely lost all trust in cheap 18650's, and cheap bms's keeping them from flaming.

using voltage sag as indication of bad battery

Hummina Shadeeba

1 MW

neptronix

Administrator

Hummina Shadeeba

1 MW

spinningmagnets

Moderator

billvon

1 MW

Hummina Shadeeba

1 MW

liveforphysics

100 TW

Hummina Shadeeba

1 MW

billvon

1 MW

markz

100 TW

Hummina Shadeeba

1 MW

dogman dan

1 PW

Similar threads