BMV's showing double midpoint percentage since last firmware update

Good morning,

Since last update for the BMV702 to FW3.11 (2021), I have noticed that during charging, BMV’s are showing double percentage of the real midpoint deviation.
As mentioned in the manual;
SS_-_Picture34.JPG

SS_-_Picture35.JPG
Where:
d is the deviation in %.
Vt is the top string voltage.
Vb is the bottom string voltage.
V is the voltage of the battery (V = Vt + Vb).

During charging;
The BMV shows top 13.2V and bottom 13.1V, shows deviation 0.06V and shows 0.5%.
So, 0.06/26.3x100=0.23%, e.g. 0.2% as it rounds off hundreds.
Or, (13.2-13.1)/26.3x100=0.38% e,g 0.4% as it rounds off hundreds.
We measured at various charge states and BMV’s and we come to the conclusion that it calculates double.
Since a long time we never watched this as it is always very low on the installations we do and maintain, but as some installations are now getting older we suddenly noticed this.
Our installations are still running under 0.07V (on the BMV’s) after e.g. 15+ years of abuse, but when we see this, it is of course looking like a failing cell, which is absolutely not at 0.3% really.
As we went to these sites to measure for ourselves to measure to confirm.
What to do?

Regards, Jeroen.

Maybe V is the nominal voltage of the used batteries?

V = ( vt + vb ) / 2

Good afternoon,

@Ludo, it is not, read the BMV manuals (HTML and PDF), either one 702/712/smartshunt has the same formula as described above (actually copied from the manuals).
But it is doing as you describe, indeed.
And, when checking other brands BMV’s (example Alpha SD08, Eaton SC200, NASA Lithium, etc), they use the same formula as the Victron ones do, so the firmware is not calculating correctly as we discovered.

Regards, Jeroen.

1 Like

Too lazy to check what makes more sense…

I only need the deviation of one battery in relation to the true middle, best in mV, not in %.

Good afternoon,

I do not believe that NASA copies others…
The given percentage is over the total battery voltage, not over half of it.
So, that formula in the books is correct (already for more than half a century).

Many brands set their alarm level on 0.5V standard for 24V or 48V alike.
That’s about 1.7% for a 24V system and about 0.9% for 48V systems, see instruction manuals.
Well, one cell will mishap first and that 0.5V is really the maximum a cell can be doing less than all the others during absorption/equalizing, 24V or 48V systems alike.

Still the BMV’s percentages are showing twice to high!

Regards, Jeroen.

I know they do, my father and my brother in law both worked at JPL in Pasadena

Good night,

@guystewart,

Please check the above issue.

Regards, Jeroen.

Hi @Jeroen2

I just checked the calculation in the BMV and it is calculated as given in the manual. The confusion comes from the reported deviation in Volt. What is reported there is the deviation of the bottom voltage compared to the center voltage of the total battery bank (bottom - ((top + bottom) / 2)). So the voltage difference between top and bottom is actually 2 * 0.06V = 0.12V. Then calculating the percentage we get 100 * 0.12 / 26.3 = 0,45627376425855513307984790874525 which rounded to a single digit becomes 0.5 %.

With kind regards,
Thiemo van Engelen

1 Like

Good afternoon,

@thiemovanengelen
We measured 3 BMV’s, and the respected differences of the batteries themselves during charging.
One measuring example with a Fluke multimeter;
Top 13.21V, bottom 13.14V, difference is 0.07V.

The bmv showing, top13.2V, bottom 13.1V, difference 0.07V, difference 0.5%.
Meaning the bmv is calculating the 0.07V perfectly but calculates the % double?

0.07/26.35x100=0.266% as per the standard formula.

We use also the Alpha SD08 (and others alike), this one is doing exactly as we measure with the Fluke.
The dial alarm can be set maximum to 1.6V.
The equal alarm level of 0.5V in a bmv I have to set now double to 3.4%, that cannot be?

What to do?

Regards, Jeroen.

Good afternoon,

@thiemovanengelen

We just had a meeting with some collegues and we all come to the same conclusion.

Quote,
What is reported there is the deviation of the bottom voltage compared to the center voltage of the total battery bank (bottom - ((top + bottom) / 2)).
Unquote.

It is not, it indicates 0.07V, not 0.035V.

Bottom,
(13.14-((13.21+13.14)/2)= -0.035V, bottom to centre, correct?
Now top,
(13.21-((13.21+13.14)/2)= +0.035V, top to centre, correct?
So, top + bottom = still 0.07V difference between top and bottom as measured with a Fluke and showing on the bmv.

Then, the formula in all the instruction manuals from Victron…(and other brands).

16703955f8d3f1

(13.21-13.14)/26.35x100=0.2656%.
But it shows 0.5%.

We are quite shure something is multiplying the % by 2.

Regards, Jeroen.

Hi @Jeroen2

I just did a measurement here where a fluke multimeter shows a total voltage of 13.51V and the mid voltage (measured at the shunt) of 8.83V. I know these are weird numbers but it is an artificial test setup with a big mid point difference to make the calculations very clear. The BMV is reporting a total voltage of 13.50, a bottom of 8.92V and a top of 4.58. It reports a deviation voltage of 2.17V, which is 8.92 - (13.50 / 2). So this matches what I mentioned in my previous answer. The deviation percentage is 32.1%, which matches 100 * (4.58 - 8.83) / 13.51. So this also matched the manual.

Can you add images of the display of the BMV of the main voltage, the top voltage, bottom voltage, deviation percentage and deviation voltage just we are sure that we are talking about the same numbers from the same moment?
And is the BMV also connected to a GX system (cerbo GX, ekrano GX, etc)? If so, can you give the VRM id and enable remote support so I can take a look?

Kind regards,
Thiemo van Engelen

Good morning Thiemo,

@thiemovanengelen

We will try today to make some photos.
We will be doing that test you did with a new bmv, we will do it with a “broken” 24V battery, just missing 1 cell in the bottom, that should give enough “error”.
Not with resistors over the shunt B1 and B2.

Regards, Jeroen.

Good morning Thiemo,

@thiemovanengelen

We went to my site quickly as nobody is on board with a Victron installation.
We used a “worn/dead” battery, lead acid, 24V, no charging and hooked it up to our 2 BMV’s, photos from 1 bmv.





We also checked cable resistance with a 12 Ohm resistor on the +B2, no difference in readings, so, eliminating cable resistances on the operating battery while charging.

On very low voltage differences of the top and bottom till about 0.10V, the % is still double.
We think that in the formula these smaller values in hundreds are rounded off?
So, at bigger values that shows less effect and then it becomes closer to the real (correct) value, can that be maybe?
Or we suspect the +B2 capacitor on the shunt PCB which is connected to the load shunt side instead the battery shunt side, can that be?

Above 0.10V difference the bmv starts to indicate correctly as can be seen above photos, on the “old” battery.
There must be a resistance creating that reading, therefore the 12 Ohm test, no issue there.
The ripple over the battery terminals from the site is 0.004Vac, so also no issue there.
We checked the small PCB (new) on the shunt, both the +B1 and +B2 have a 0.1microF capacitor (difficult to read the text) to the load side zero, these might cause this, specially the +B2 one…
In the supply line of the +B1 is a 3.5 Ohm resistor (difficult to read the text), no issue there.
We used a new original cable between shunt and bmv, measured the pins of the cable at the bmv side, voltages are correct, no issue there, of course no mA currents flowing over +B1/2, and so no voltage drop .
We have 2 BMV’s with midpoint there, both do the same, exactly the same below 0.10V difference, indicating double %.

If you want us to do more tests today, let us know, we will be standby here till about 1600.

Regards, Jeroen.

Hi @Jeroen2

So the calculations do indeed seem correct, at least for “larger” values.

When you are talking about problems below 0.1V difference, is this then the difference reported by the BMV itself or are these multimeter measurements? It could be that the calibration of the BMV is not that accurate anymore and thus that the values it measures (and reports) are different from the multimeter measurement. And if the top to bottom difference is so small, a small difference between the BMV measurement and multimeter measurement can have quite the impact on what is reported. So it would be good to know the exact reported battery, top, bottom and deviation voltages and deviation percentage when the midpoint difference is small.

And what is the current running through the shunt while doing these measurements? This is also important when it is a really large current as then the shunt will also have a drop of upto 50mV and the BMV measures the voltage between the load side of the shunt and the B1 connection.

Regarding rounding:
All displayed values are rounded to the shown decimals. So when the BMV reports 0.10V deviation, this might be 0.095V .. 0.105V. So this would not explain the double percentage, but can explain a bit of the difference.
I am not sure what the measurement accuracy is of the BMV and how thus how large the rounding error is in the measurement.

Kind regards,
Thiemo van Engelen

Good morning Thiemo,

Thank you very much for the understanding, but in the meantime we talk about 5 tested bmv’s.
3 in the field my colleagues discovered and 2 at my site.
Next week we will try to use a 24V to 24V isolator with a 24V to 24V power supply (both from Phoenix contact, the ones we use on board a lot) and one adjustable absolute linear resistor (certified) to see per 0.01V difference what is going on exactly.

Below 0.10V; we measure 0.01-0.02V real difference (top/bottom) with the flukes we use, but the difference is monitored as 0.03-0.04V to midpoint (x2=0.06-0.08V top/bottom) on the bmv’s, actually more than double sometimes when very low differences.

Regards, Jeroen.

Good afternoon Thiemo,

@thiemovanengelen

We have been busy as promised.
We now used a BMV702 which was never measuring the midpoint before.
And a Phoenix power supply exactly set to 24VDC, no isolators.
It is at 0.20V between top and bottom starting to indicate as should be, below it is about half the real value now?
No current going through the shunt.
So with no current it shows half till 0.20V (top to bottom) and with current it shows double till 0.10V (top to bottom).
Below a curve to make it more visible.

And the raw data;

BMV BMV BMV BMV BMV Multimeter
V total V top V bottom dV d% dtop-bottom real d%
24,00 12,00 12,00 0,00 0,0 0,00 0,00
24,00 12,00 12,00 0,02 0,1 0,05 0,21
24,00 12,00 12,00 0,03 0,2 0,10 0,42
24,00 12,00 12,00 0,03 0,3 0,12 0,50
24,00 12,10 11,90 0,08 0,5 0,20 0,83
24,00 12,10 11,90 0,10 0,8 0,20 0,83
24,00 12,10 11,90 0,13 1,1 0,30 1,25
24,00 12,20 11,80 0,18 1,5 0,40 1,60

Regards, Jeroen.

Hi @Jeroen2

This is a nice overview. I think you are running into the measuring accuracy of the BMV here. The calculations of the BMV seem correct given the data reported by the BMV (except for the 0.08 dV, 0.5% measurement, but perhaps the midvoltage changed between switching the display items).

The funny thing is that the reported percentage is lower in this data than the real d%, which contradicts what you initially reported, so there probably is also some other effect that causes the differences.

With kind regards,
Thiemo van Engelen

Good morning Thiemo,

Indeed, with no battery and no load connected it flips opposite (bench test), we were also surprised.
We discovered this during charging initially and were focussed on that.
When turning that potmeter you see that the Fluke registers it but barily the BMV.

What we see is correct on the installations, with zero load to 50-60% load and no charging the midpoints are very low (too low) as we checked yesterday some quickly.
So, we think that it is correlated with the direction of the current through the shunt and or the temperature of that shunt.

What we can do is a test with a 2000AH 12V battery and a seperate charger connected.
Then we can do the charging data, in rest data and with a load data.
This we put again in a curve, that should give then all the data.

Regards, Jeroen.

Hi @Jeroen2

Out of curiosity, what was the impedance of the potmeter connected as midpoint voltage? If this is not very low, the it might also be of influence on the actual measurement of the BMV, as its filtering circuitry might be influenced by the output impedance of the potmeter in combination with the input impendance of the BMV.

But to be frank: because the calculations are correct, we do not see a need to take any action regarding this midpoint voltage deviation at the moment. So if you want to do more measurements then that is just for you or for the shared knowledge.

Kind regards,
Thiemo van Engelen

1 Like