This site is now in read-only archive mode. Please move all discussion, and create a new account at the new Victron Community site.
I am looking to use the bmv712 and the supplied 500 amp shunt with the ve.direct output and reading the output with an arduino. My question is the bmv712 specs indicate amps accuracy of .01 Dc amps. I am trying measure very small currents say .25 amps up to a maximum 100amps, will the bvm712 measure dc currents that small and output them in the ve.direct port ? The dealer I contacted said it was only good for + or - 1 amp.
thanks
Tip: Remove 4 of the 5 resistance plates in the BMV shunt. This increases the sensitivity/accuracy by 400% ... which is particularly advantageous when measuring current. You can enter the changed sensitivity into the BMV712 via Bluetooth.
From what I understand you can simply replace the naked shunt part with a different one of your choice. I.e. remove the electronics from the shunt and fit it onto e.g. a 100 A shunt. No need for messing up the original hardware.
Furthermore the accuracy of your measurements depends on the precision by which you know the resistance of your shunt and that is much reduced if you simply cut away 4 of its plates. Thus you need to recalibrate it using very high precision tools and in the general case you are better off completely replacing the shunt with a lower rated one which is calibrated already.
I seem to remember that the resolution over VE.Direct is 1mA, so no need to start sawing shunts to pieces.
Maybe the BMV shows it, but the accuracy is far beyond this specification.
I removed 3 plates for myself. I am reluctant to measure 10A with a sensor suitable for 500A. But you're right that you have to think about the calibration. I do software-based 3-point calibration for every sensor I use anyway. This results from the fact that I don't use any standard components (BMS, etc...) but record EVERYTHING and Linux. This gives me much better options. It is very disappointing that Victron has not published any information about the VE.Direkt HEX protocol that "MPPT RS 450/100" has published so far. This means the device is unusable for me. Incidentally, I do not believe that a 500A shunt resolves 1mA (I read somewhere in the forum). There is no sensor with an accuracy of 0.0002%. A good shunt has an accuracy of 1% ie. from 5A. Then you get a maximum of 0.1% through calibration on the software or hardware side.
1mA over a 500A shunt is a resolution of 500000 (500000 1mA steps between 0A and 500A). The accuracy is something else derived from the specifications of the shunt and the ADC, certainly not 0.00002%. It's important not to confuse accuracy with resolution.
With a 24bit ADC (like the one in a BMV) you can theoretically resolve <30µA with a 500A full scale (500 / 2^24), but due to a multitude of ADC errors it would be very inaccurate. A 1mA resolution is easily achievable and with minimum of 5 LSB representing each 1mA step, ADC accuracy could now be pretty good.
The tolerance of this type of shunt resistor is typically around 0.25%. You can calibrate for this inaccuracy afterwards, like you say.
Of course, a higher resistance shunt and a lower FSD would offer higher resolution to suit your requirements, but the typical accuracy of the shunt and ADC would not change. However, with a lower FSD you'd end up with more LSB representing a certain resolution, so the ADC accuracy would be improved in that respect.
I understand your derivation .. but .. see it differently. I use a good 0.5% class shunt with 50A maximum current (48V batteries). I record the shunt voltage using a very good isolation amplifier and a 24-bit delta sigma ADC. All recorded voltage values are converted into a shunt current using a three-point calibration. Why this effort now? In order to determine the SOC as precisely as possible, I need a very precise integration over the current time series. In order to capture current ramps and peaks well, I use a sampling rate of 1 kHz. This gives me a satisfactory approximation of the SOC value over time and the need for calibration the batteries can be reduced. For the precise determination of the SOC, the current must be measured with the greatest possible absolute accuracy, especially since small currents are usually integrated. Of course, the absolute accuracy increases with the sensitivity of the sensor (shunt) and a (software) calibration. Incidentally, I also record the individual battery voltages (Lifepo4) with my own multi-channel isolated DAQ. This gives me much better opportunities to identify problems with individual cells, for example, or the criterion for switching from battery to mains, etc. etc.. The background is that I have been working on the www.dlr.de wind tunnels for decades and I was responsible for IT and measurement technology. I've been retired for a few years and have been running 3 PV systems for many years, one of them for self-supply.
Summary: A higher resolution doesn't solve my problem. Especially with small currents (98% of the time I have to measure currents of approx. 5A) I need a high absolute accuracy.
