Chargery BMS Cell Readings Inaccuracy. Should I be concerned?

[Tony]

Hi,

I noticed my BMS16T cell measurements are not accurate as some of the cells report higher or lower voltage than my DMM. I have a 2.5V reference standard and have adjusted the DMM numbers to correct for its 0.00271V low readings:

Cell #	BMS Reading	DMM Reading	BMS vs DMM
3	3.325	3.325	0.000
4	3.329	3.329	0.000
9	3.300	3.322	-0.022
10	3.359	3.327	0.032
16	3.360	3.324	0.036

I included cells 3 and 4 comparisons to show that the BMS and DMM do measure the same voltage in some cases. Comparing cell 9 to cell 16, there is a difference of 58 mV. That means when I try to charge the bank to 58.4V, cell 16 hits HVD while cell 9 is still around 3.592V. I guess that's not so much of a problem given cell 9 is still above 99% capacity. However, I plan to charge the bank to 56V (3.5V/cell) normally, so cell 9 will only reach 3.442V. Since this is in the flat part of the curve, the stored energy in that cell could be much lower than cell 16. So, this is a problem, right?

Just to anticipate the question, I thought maybe the connector terminations were defective so I replaced some of the ring terminals with the highest discrepancies with no change in the results.

 

[BiduleOhm]

Let me guess, 1 to 8 are correct but 9 to 16 aren't? Or it's just 9, 10 and 16?

 

[Tony]

Let me guess, 1 to 8 are correct but 9 to 16 aren't? Or it's just 9, 10 and 16?

Here's the full list of readings:

Cell #	BMS Reading	DMM Reading Calibrated	BMS vs DMM Calibrated
1	3.315	3.325	-0.010
2	3.328	3.328	0.000
3	3.325	3.325	0.000
4	3.329	3.329	0.000
5	3.331	3.328	0.003
6	3.340	3.325	0.015
7	3.318	3.327	-0.009
8	3.326	3.328	-0.002
9	3.300	3.322	-0.022
10	3.359	3.327	0.032
11	3.326	3.328	-0.002
12	3.346	3.327	0.019
13	3.316	3.330	-0.014
14	3.329	3.326	0.003
15	3.329	3.330	-0.001
16	3.360	3.324	0.036

After thinking about my concern, I realized I was not calculating the impact correctly. For cell 9, the BMS is reporting lower (3.300V) than the DMM (3.322V.) The DMM readings are all within a couple mV of the average of 3.3265V. So, unless my "new" logic is wrong, the issues would be that on HVD the BMS cell 16 would report being at 3.65V when actually at 3.624V and on the LVD the BMS cell 9 would report 2.5V when actually at 2.522V.

 

[Craig]

If you look at 7,8,9 their difference almost adds up to #10's overage. I would say there is a measurement problem and it compounds as it moves along. I don't know a solution other than checking connections. I also do not know what the acceptable tolerances are. But at least if it gives bad readings it is in the safer direction

 

[Tony]

If you look at 7,8,9 their difference almost adds up to #10's overage. I would say there is a measurement problem and it compounds as it moves along. I don't know a solution other than checking connections. I also do not know what the acceptable tolerances are. But at least if it gives bad readings it is in the safer direction

Craig, thanks for the feedback. I checked all of the connections when I installed the ring terminals. As a first step, I also replaced 9-and 9+ ring terminals. and then 16 ring terminal. No change observed. So, you don't usually experience misreadings on the cell voltages in the Chargery BMS? I was going to ask Jason, but he seems to be quite busy right now.

 

[Craig]

I do see some variances. But I'm not sure what is expected and the tolerances are supposed to be.

 

[BiduleOhm]

The errors definitely are bigger and more frequent in the top half which suggest the main cause is due to common mode voltage (its hard to measure 3 V with a few mV accuracy when it sits on top of 50 V...) and also some of the bigger errors are on cell 1 and 16 which suggest there's some error due to being near the power rails.

Can you calibrate the voltages of each cell on the chargery? because they are mainly offset and gain errors so if you can calibrate them you should have single digit mV accuracy.

 

[Tony]

The errors definitely are bigger and more frequent in the top half which suggest the main cause is due to common mode voltage (its hard to measure 3 V with a few mV accuracy when it sits on top of 50 V...) and also some of the bigger errors are on cell 1 and 16 which suggest there's some error due to being near the power rails.

Can you calibrate the voltages of each cell on the chargery? because they are mainly offset and gain errors so if you can calibrate them you should have single digit mV accuracy.

BiduleOhm, Yeah, generally it does go up in the second half, but the BMS cell reporting for 12 and 13 have larger variances than cell 1. There's a ground on cell 9-. I'm wondering what the BMS does internally with that additional - reference point. I am not aware that there is a way to calibrate the readings. Seems like they could do that in software. They could have a section to store the mV adjustments and then software to add/subtract those adjustments before storing. Will that happen, I don't think so.

 

[Cal]

If Chargery is using a differential adc then there’s no common mode voltage error. The upper cell voltage difference from dvm measurement varies from + to -. Would that be the case if there’s a common mode error?

 

[BiduleOhm]

There's a ground on cell 9-. I'm wondering what the BMS does internally with that additional - reference point.

That really hints toward a similar type of topology I adopted (divided the 16 cells in 2 halves as it's too expensive or even impossible to find an op-amp with more than 40 V of Vcm).

I am not aware that there is a way to calibrate the readings. Seems like they could do that in software. They could have a section to store the mV adjustments and then software to add/subtract those adjustments before storing. Will that happen, I don't think so.

That's a bummer as it's easy to do and should solve most of the errors.

If Chargery is using a differential adc then there’s no common mode voltage error. The upper cell voltage difference from dvm measurement varies from + to -. Would that be the case if there’s a common mode error?

There is as it's precisely a common mode error on a diff input. It comes from the difference in resistors values and from the internal circuitry of the ADC/op-amp. You need both a high common mode rejection IC and well matched resistors to avoid common mode errors.

The error can be in either direction as it's the differences in values of components due to manufacturing tolerances who causes it (i.e. a positive common mode voltage doesn't always causes positive error, it's random and so should be roughly 50/50 between positive and negative errors).

If you know what you're doing and there's resistors on each channel you can compensate most of the errors by adding very high value resistors in the right places on the concerned channels. If it's an all in one chip software is the only place you can do something.

Errors due to being near the rails (only for cells 1 and/or 16) need software compensation thought (and depending on many factors it may not be compensable) but you may not have that kind of error, we can't tell that from just voltage measurements.

 

[Tony]

That really hints toward a similar type of topology I adopted (divided the 16 cells in 2 halves as it's too expensive or even impossible to find an op-amp with more than 40 V of Vcm).



That's a bummer as it's easy to do and should solve most of the errors.



There is as it's precisely a common mode error on a diff input. It comes from the difference in resistors values and from the internal circuitry of the ADC/op-amp. You need both a high common mode rejection IC and well matched resistors to avoid common mode errors.

The error can be in either direction as it's the differences in values of components due to manufacturing tolerances who causes it (i.e. a positive common mode voltage doesn't always causes positive error, it's random and so should be roughly 50/50 between positive and negative errors).

If you know what you're doing and there's resistors on each channel you can compensate most of the errors by adding very high value resistors in the right places on the concerned channels. If it's an all in one chip software is the only place you can do something.

Errors due to being near the rails (only for cells 1 and/or 16) need software compensation thought (and depending on many factors it may not be compensable) but you may not have that kind of error, we can't tell that from just voltage measurements.

Of course, I don't know what I'm doing so that's not an option for me.

 

[BiduleOhm]

Yea I wouldn't recommend that, especially as it's usually very small components, mistakes are easy to make. But it's more likely it's all integrated into a chip anyway so you can't do anything on the hardware level.

 

[Craig]

I have talked to Jason about calibrating the Chargery he claims that it can only be done by them and you need a high precision DVM one of these days I will try and reverse engineer the Firmware to see if things can be adjusted.

 

[BiduleOhm]

Well, @Tony has a pretty high precision DVM and even a standard to calibrate it so he knows what he's doing.

So if we decipher a bit, that means there's a way to do it but only people working at Chargery know how to do it. So hopefully it's something simple a bit of RE can reveal

 

[Tony]

Well, @Tony has a pretty high precision DVM and even a standard to calibrate it so he knows what he's doing.

So if we decipher a bit, that means there's a way to do it but only people working at Chargery know how to do it. So hopefully it's something simple a bit of RE can reveal

Thanks guys. I sent Jason a message about the issue and a link to this topic. He's busy I'm sure, but hopefully he can share his thoughts.

 

[Steve_S]

@Tony, out of curiosity.
When you installed the BMS Wire Harness for the cells, where did you start with Cell-1 & the Negative ?
When you installed the Ring Terminals on the BMS Harness, did you crimp, or crimp & solder and did you check continuity and if there is a resistance increase ? They obviously all connect but a weak connection or one with too much solder can potentially cause misreading at the mv level.

 

[Tony]

The black lead on the 1-8 wire harness is on the negative post of the cell 1 along with the negative cable going to the shunt. In trying to address the problem, I have replaced 9-, 9+ and 16+ wire harness connectors and crimped them (no solder) with no change in the results. When I originally installed the ring terminals, I did a continuity test and all were good. I have more ring terminals on order and will replace the rest on Monday.

 

[Ampster]

One of my BMSs went screwy when I forgot to unplug the connector from the BMS before removing some of the individual cell connections. I apparently damaged some of the electronics.

 

[Tony]

One of my BMSs went screwy when I forgot to unplug the connector from the BMS before removing some of the individual cell connections. I apparently damaged some of the electronics.

Ampster, yeah I have been worried about that, but I have been careful to disconnect from the BMS before doing any work.

 

[Ampster]

Are the new wires the same size and length?