Multimeter accuracy needed for charger/battery measurements

[Chriull]

Unfortionately I stated in a couple of posts that any cheap multimeter is fine to check charger/battery voltages.

I just discovered that some of the cheap offers have +/-1.5% accuracy! Thats +/- 1.26V at 84V! Such equipment is not usable for calibration/measurements!

But there are many cheap +/-0.5% available - this are still +/-0.4V. Enough for a rough check and almost about enough for calibration?! One should be consious about this inaccuracy and pay attention, so the chargers voltage cannot be to high.

Really recommendable multimeters (+/-0.1% ~~ 0.1V at 84V) start to get expensive...

... But i did just a quick google search for some models - maybe there are great price/accuracy/quality deals available?

[Seba]

8 hours ago, Chriull said:

But there are many cheap +/-0.5% available - this are still +/-0.4V. Enough for a rough check and almost about enough for calibration?! One should be consious about this inaccuracy and pay attention, so the chargers voltage cannot be to high.

It's slightly more complicated. First, it's important to determine the measurement range and measurement resolution. Why? Because multimeter can have accuracy of 0,1% but measurement range of 0 to 400 V with 0,1 V resolution (four digits). What does it mean? It's accuracy will be one digit. What does it mean for cell measurement? It will give us 100 mV of measurement resolution and accuracy. This equals to... 2,3 % !

Let's take a Brymen BM231 multimeter as an example. Lower shelf multimeter of generally proven brand - price tag of about 70 EUR. It's specified to have an accuracy of 0,6 % + 3 digits. Display is defined as 3 and 5/6, what effectively means that voltage of 4,2 V in ideal conditions will be displayed as 4,200 V.  Taking all this into account we can assume that when measuring cell voltage of 4,2 V it may display values from 4,167 V to 4,228 V. Not that bad  Trust me, BMS circuitry doesn't get that accuracy of measurements ;-)

[Chriull]

Getting the real accuracy out of specifications is almost an own "sience".

11 hours ago, Seba said:

Let's take a Brymen BM231 multimeter as an example. Lower shelf multimeter of generally proven brand - price tag of about 70 EUR. It's specified to have an accuracy of 0,6 % + 3 digits. Display is defined as 3 and 5/6, what effectively means that voltage of 4,2 V in ideal conditions will be displayed as 4,200 V.  Taking all this into account we can assume that when measuring cell voltage of 4,2 V it may display values from 4,167 V to 4,228 V. Not that bad 

That sounds very good!

But from the specs i found at a link with a datasheet for BM231,233,235 specifying dc voltage range 6.000 even with +/-.3% +2d leading even to +/- (0.0126+0.002)=+/-0.0146V!

For 84V at the 600.0V range with +/-(0.3%+2d) would be +/-(0.252+0.2)=+/-0.452V

Seems to be as good as it can get in this price range?

Just not with the Surpeer av4?  currently not available but specced at incredible 200mV/2V/20V/200V ±(0.05%+5d) and ?4 1/2? digits for ~30-40US$. Input protection circuits shall be absolutely not state of the art but "usable good accuracy". But hard to tell if one got a good or bad one without an reliable multimeter to compare...

Here the 084.0V at the 200V setting would make +/-(0.042+.5)V=+/-0.542V Or do the 4 1/2 digits mean that at the 200.0V range 84V are displayed as 84.00V and the accuracy is +/-(0.042+0.05)=+/-0.092V?

.... The +/- digit are a crude way to "hide" inaccuracy in leaflets... And "hit" especially hard if one needs to measure this 84V...

11 hours ago, Seba said:

cell voltage of 4,2 V it may display values from 4,167 V to 4,228 V. Not that bad  Trust me, BMS circuitry doesn't get that accuracy of measurements ;-)

From one curcuit i saw once of "common chinese" BMS they use comperators with fixed build in reference voltages for something like 4.27V as upper voltage threshold and they have +/-50mV specified. So, yes, they have almost twice the inaccuracy.

Afair there was also comment that many comperators/voltage references like this used have "horrible" temperature drift...

 

Edited May 14, 2019 by Chriull

[Seba]

1 hour ago, Chriull said:

For 84V at the 600.0V range with +/-(0.3%+2d) would be +/-(0.252+0.2)=+/-0.452V

Seems to be as good as it can get in this price range?

It's good result. For example LG MJ1 cell charging voltage is specified as 4,2 +/- 50 mV. For 20S string it's 84 V +/- 1 V. So full-charge voltage difference as in datasheet is over twice of multimeter overall accuracy. Over 100 % of safety margin. Of course you can get much better multimeter for 150 EUR - Brymen BM857s has 0,03% + 2 digits of accuracy at 5V measurement range. It's less than 1,5 mV !

1 hour ago, Chriull said:

The +/- digit are a crude way to "hide" inaccuracy in leaflets... And "hit" especially hard if one needs to measure this 84V...

It's normal. This is because analog-to-digital converters ane not ideally linear and have non-zero quantization error. So they have offset error specified in bits LSB.

1 hour ago, Chriull said:

From one curcuit i saw once of "common chinese" BMS they use comperators with fixed build in reference voltages for something like 4.27V as upper voltage threshold and they have +/-50mV specified. So, yes, they have almost twice the inaccuracy.

Afair there was also comment that many comperators/voltage references like this used have "horrible" temperature drift...

Yes, sometimes temperature drift may be well over declared reference range  And we cannot forget about voltage divider that is built with two resistors. They also have limited precision and have temperature drift...

[Chriull]

27 minutes ago, Seba said:

It's good result. For example LG MJ1 cell charging voltage is specified as 4,2 +/- 50 mV. For 20S string it's 84 V +/- 1 V. So full-charge voltage difference as in datasheet is over twice of multimeter overall accuracy. Over 100 % of safety margin. Of course you can get much better multimeter for 150 EUR - Brymen BM857s has 0,03% + 2 digits of accuracy at 5V measurement range. It's less than 1,5 mV !

You are right! If i want for example adjust the charger, so that the batteries are at maximum 84V i'd have to adjust it so that (84-0,452)V=83,548V are measured at the batteries. In reality the voltage could then be between 84V and  (83,548-0,452)V=83,096V leading to 4,155V for one cell in average up to 4,2V.

Or if one wants to stay within the 4.2V+/-50mV spec (85V max) one could adjust it to 85V-0.452V and has as lower limit of (85-0,452*2)V making 4.2V per cell for using the max capacity.

That sounds very acceptable!

 

[esaj]

On 5/14/2019 at 10:26 AM, Chriull said:

Getting the real accuracy out of specifications is almost an own "sience".

That sounds very good!

But from the specs i found at a link with a datasheet for BM231,233,235 specifying dc voltage range 6.000 even with +/-.3% +2d leading even to +/- (0.0126+0.002)=+/-0.0146V!

For 84V at the 600.0V range with +/-(0.3%+2d) would be +/-(0.252+0.2)=+/-0.452V

Seems to be as good as it can get in this price range?

Just not with the Surpeer av4?  currently not available but specced at incredible 200mV/2V/20V/200V ±(0.05%+5d) and ?4 1/2? digits for ~30-40US$. Input protection circuits shall be absolutely not state of the art but "usable good accuracy". But hard to tell if one got a good or bad one without an reliable multimeter to compare...

I was looking at Brymen BM859s  https://brymen.eu/shop/bm859s/   a couple of years back, seemed pretty good for its price, but then found a second-hand HP34401A (6½ digits bench meter) for 240€ with shipping. I then faced the same problem, how to tell how good it actually is (ie. how much it has drifted from last calibration, which was several years ago)...

What I ended up doing was to buy precision resistors & voltage references, make boards for them with measurement points and power the references from a battery (to prevent any "noise" from a linear or, even worse, SMPS power supply) and see if the meter reading was within the tolerance of the part. I got down to 0.05% in voltages and 0.01% in resistors with no problem (using good probes, they're just as important as the meter itself, for example:  https://www.tme.eu/de/details/pp-bm10a/messleitungen-komplettsatze/brymen/pp-bm-10a/#  , I got horrible jumpings in the readings with cheapo Chinese probes ), beyond that, no idea. Good enough for me 

You can get 0.01% 5PPM/C (SMD) resistors in singles for around 1€ per piece for "common" values (I simply got 10, 100, 1k, 10k, 100k, 1M, 10M), don't remember what the 0.05% voltage references were, a couple of € maybe. After that, it starts to get expensive, 0.005% resistors are something like 20€ per piece. 

Edited May 19, 2019 by esaj

[Flying W]

I've been adjusting my ewheels charger lately but was worried about the meter accuracy. 

My charger was reading 83.9, battery voltage at the connector was reading 83.2. I turned it up to 84.4 and now battery reads 83.7. 

I'd like to try for more but this thread confirms my suspicion that I shouldn't count on the bms shutting down the charging. 

When my batter at the connector shows 83.7 wheelog shows 82.6v. Oddly my original board would show 83.4 when full on wheelog. Changes boards at 340 miles and wheelog has showed 82 and change ever since (for the last 2300 miles) so I'm guessing the boards report with differing amounts error to wheelog. 

This is on an mcm5 800wh. Wheel has 2640 miles on it. Never stored fully charged and durring commuting charged to 85% with a 100% charge once a week.  I'm not noticing capacity loss but there must be some with that many miles. 

I'd like to see the battery read 84.2 at the connector but with the margin of error (my meter is rated at +- .5%) it may be foolish to chase a few tenths.

[RockyTop]

31 minutes ago, Flying W said:

I've been adjusting my ewheels charger lately but was worried about the meter accuracy. 

My charger was reading 83.9, battery voltage at the connector was reading 83.2. I turned it up to 84.4 and now battery reads 83.7. 

DC voltage sags given distance and wire size. I am used to seeing a loss of 3 volts at 200-300 feet. It takes 26v to get 23.5v just 150 feet away. Also if the wire is undersized and loaded down during changing you are likely to see a drop at the other end of the wire. 

Also, the battery gives a real DC signal while the charger sends an artificially generated DC signal. Some meters will measure the average a little different. Kinda like measuring the ocean shoreline. The more point you use the longer it gets. 

Edited November 3, 2019 by RockyTop

[Chriull]

1 hour ago, Flying W said:

I'd like to see the battery read 84.2 at the connector but with the margin of error (my meter is rated at +- .5%) 

0,5% is +/-0.4V. In the right tempetature range, humidity and timespan since last "calibration". And normaly dmm have some digits error on top.

1 hour ago, Flying W said:

it may be foolish to chase a few tenths.

If it's within a volt it's about ok.

1 hour ago, Flying W said:

I'd like to try for more but this thread confirms my suspicion that I shouldn't count on the bms shutting down the charging. 

What one should "reach" with the charger voltage is to be as minimal as possible just above the balancing voltage threshold and stay below the cell overvoltage shut off voltage. That should be somewhere about between 4.2Vx20=84.2 and 4.28Vx20=85.6V.

These voltage threshold (example from other BMS thresholds) are given there +/-5mV. How much one want to trust chinese ali... Spec sheets...

...and there is some voltage drop from the BMS input protection circuitry.

[alcatraz]

The more you use a voltage meter the more you'll get a feel for it's deviation.

I've measured several chargers, some high precision and the end voltages are verified by the factory at +/-10mV. They measure 4.201V in their pictures.

With my voltage meter I then measure 4.21-4.22V.

Then after all 20 cells individually were charged by the high precision charger the pack voltage is exactly 84.0V so the wide range seems to be working ok, but low range is overreporting.

[Aneta]

7 hours ago, Chriull said:

0,5% is +/-0.4V. In the right tempetature range, humidity and timespan since last "calibration". And normaly dmm have some digits error on top.

If it's within a volt it's about ok.

What one should "reach" with the charger voltage is to be as minimal as possible just above the balancing voltage threshold and stay below the cell overvoltage shut off voltage. That should be somewhere about between 4.2Vx20=84.2 and 4.28Vx20=85.6V.

These voltage threshold (example from other BMS thresholds) are given there +/-5mV. How much one want to trust chinese ali... Spec sheets...

...and there is some voltage drop from the BMS input protection circuitry.

4.28V? Where does this come from? I've seen 4.18V as a threshold for BMS to start balancing, but not 4.28. Isn't anything above 4.20V per cell "black death" territory for Li-ion cells?

[Seba]

3 hours ago, Aneta said:

4.28V? Where does this come from? I've seen 4.18V as a threshold for BMS to start balancing, but not 4.28. Isn't anything above 4.20V per cell "black death" territory for Li-ion cells?

No, 4.2 V is standard charging voltage. It's safe and used as a reference voltage for 100 % state of charge.

[Chriull]

5 hours ago, Aneta said:

4.28V? Where does this come from?

From  aliexpress BMS specifications as cell overvoltage threshold where the BMS cuts the charger off. 

Quote

've seen 4.18V as a threshold for BMS to start balancing,

Most i've seen have 4.2V for start balancing. Could eadily be that there are some with lower thresholds available, too?

I've no idea how the thresholds are for the different BMS used in our wheels.

For lower balancing thresholds as 4.2V one imo should also use lower voltage chargers (like 4.18V x cell-count in your example) - my hypothesis is that the balancing circuitry bleeds the "overcharged" cells down to that threshold even after charging.

Quote

but not 4.28. Isn't anything above 4.20V per cell "black death" territory for Li-ion cells?

There is imo no hard threshold. "Just" everything above makes it worse, everything below a bit better.

Edited November 4, 2019 by Chriull

[esaj]

On 11/3/2019 at 7:36 PM, Flying W said:

When my batter at the connector shows 83.7 wheelog shows 82.6v. Oddly my original board would show 83.4 when full on wheelog. Changes boards at 340 miles and wheelog has showed 82 and change ever since (for the last 2300 miles) so I'm guessing the boards report with differing amounts error to wheelog. 

I think it was @DaveThomasPilot who pointed out a (common?) design mistake in the reverse engineered Firewheel mainboard in how the wheels measure the battery voltage (probably common in many wheels): the usual set up is using very high resistances (hundreds of kilo-ohms) to form a voltage divider (you can't exatly put up to 84V direcly to an analog converter-pin in a 3.3V microcontroller without magic smoke coming out of everywhere ), and then a small-valued multi-layer ceramic capacitor (MLCC) paralleled with the low-side resistor to filter the fluctuations:

The issue is that the MLCC will have a (very small, like micro-amperes, µA) current leak, but with large resistances this current leak becomes more significant vs. how much current the large-valued high-side resistor "allows" to flow, skewing the measurement lower. Furthermore, the resistors are just "plain" 1% tolerance, with 84V +-1% is already +-0.84V (not to mention if they're at the "far ends" of their tolerance error, ie. one is +1% while the other is -1%, although this is very, very unlikely, and probably the +- percentage error of the resistors themselves is smaller than 1%). But then again, that's not the only error source in the measurement, so the total error can be higher.

TLDR; the measurements done by the wheel itself using the divider in the mainboard may have 1% (or even more) of error in them, usually towards lower than the real voltage. Shouldn't be really much of an issue outside the owner being confused why the voltages seem slightly out of whack; If you measure your charger putting out the "correct" voltage with a fairly accurate multimeter and there are no issues with charging or riding, everything's highly likely just fine.

Edited May 7, 2020 by esaj

[Flying W]

@esaj, so in layman's terms it's a non issue . I like learning more about the electrical side of things though so thank you. 

On the charger side, to measure 83.7v output I need to turn the volt trim pot up to read 84.6v on the LCD screen (ewheels 5amp charger). This is on a middle of the road Klein meter (40usd from home depot)

[Chriull]

2 hours ago, Flying W said:

@esaj, so in layman's terms it's a non issue . I like learning more about the electrical side of things though so thank you. 

More or less. If one knows this to interpret the values right.

Bad is high deviation if one has many downhill rides - overvoltage alarm, tiltback and shutdown can come way too soon!

Quote

On the charger side, to measure 83.7v output I need to turn the volt trim pot up to read 84.6v on the LCD screen (ewheels 5amp charger).

You know the accuracy of the ewheels charger?

If it's in some 0.5% range (which would be quite good for such a product) 83.7V to 84.6V can be the same value and justify _absolutely_ no adjustment of the charger! It's some 50:50 chance to worsen things or improve them.

For serious calibration one needs measurement devices at least a magnitude more accurate!

Quote

This is on a middle of the road Klein meter (40usd from home depot)

Above are the calculation on what to ecpect from the specifications...

Edited May 8, 2020 by Chriull

[.....]

Anyone interested, I have TWO of the exact same wheels. I set my charger at 84.1v with my fluke meter. I charged them both to maximum and let them keep charging for about 2 hours after green light.The apps i checked with (DB, EUCS, KS) showed one wheel to be 84.5V  and another was 84.3V.  This just goes to show that you CANNOT rely on wheel reported voltage for exact figures. Unless the wheels are magically generating their own voltage AND at a different amount than the other, they are both wrong by .2v and .3v. I have 3 chargers and NONE of the 3 matched exactly. One charger was 83.9v another was 84v and the one I set beforehand  84.1V.  I trust my meter more than the wheel reportings. ALL my chargers are now set at 84.1v, so no matter if I use 2 at a time or grab any of the 3, my charging habit will remain. I just have to ignore that BS 84.5V and not be alarmed by it.