I never did this before and had always the feeling that the Victron shunts were spot on regarding measurements, but today I’ve conducted several measurements on the Smartshunt 500A and it did not perform as expected. At very low currents (1A) the Smartshunt was 5% off (1.0001A; Shunt shows 0.95A). At 5A and 10A it was more or less 2% off (5.002A, shunt shows 4.89A; 10.002A, shunt shows 9.80A). So, is the resistance of the shunt off or is the calculation of the smartshunt wrong? I measured with different currents the voltage across the shunt and with all three measurements the conclusion was that the resistance was a neat 0.107mOhm. That’s within it’s specs. So, why is the shunt regarding it’s measurements so off? Has someone measured on the shunts too and what were your ideas regarding the accuracy? Specs say it should be acting within 0.4%.
And when we measuring, we are measuring. Sometimes that’s fun, isn’t it? So, also measured the own consumption of the shunt. Should be less than 1mA, according to the datasheet. Measurement showed more than 3mA. How come?
(measurements were done with professional equipment, Fluke’s and DMM’s with 55000 counts)
Insights? Is this unique and a bad version of the Smartshunt?
Let me try to explain it to you a little bit more. This topic is about misreading of the Smartshunt at different currents. In the openingspost I tell what readings I did. It was not about not measuring a current on low level. So topic has nothing to do with the treshold. The treshold is a parameter when the shunt should start reading. With a setting of ‘0’ the treshold is ‘off’ and everything what is measured is also being displayed and calculated. The inaccuracy of the smartshunt at e.g. 10A has nothing to do with this.
I have just measured voltage loss over the Victron Smart shunt (1000A) with 0.02% accuracy…
Over brass Shunt blocks 2.17 mv at 78 A charge current.
But over the cable connectors/ lugs of the shunt I measure 4.78 mV… substantial difference, because the Voltage loss (resistance) in the connection.
How did you measure the current? It’s difficult because of the resistance in the connection.
If the shunt itself has a resistance of 0.05 m Ohm but over connection 0.07 mOhm the current measure is waaaay off… depending where you measure the voltage.
Is the connection between the shunt pcb and the shunt good? No corrosion etc?
So the clamp meter or coil over the body of the shunt at 90°?
The shunt is basically a resistor, (the resistor value is set in the misc setting) the pcb measure volt drop over it under load.
Other currents in the system if being measured over cable have a different method of measurement so can be different. Plus the move through lugs and cable which have added resistance and that would affect readings.
It’s a Lab test setup guys. The shunt is brand new. I just was curious if the shunt would meet it’s specs. Apparently not. I might be the first one who’s done this checks, but I’m probably not. That’s the reason for checking here. My guess it’s not a failure of this specific one, but the series has a problem with accuracy. But that is just a guess without measuring more devices.
(for the technical people here, measurements has been done in several ways to confirm the results. First of all a known source of current. You don’t know the exact resistance of the shunt. It should be 0.1 mOhm, but if something has gone wrong with the trimming of the shunt you have your problem there. That can be checked with the known current and a very precise DMM. I used a Fluke and bench DMM with an awesome resolution and an accuracy of up to 0.025% DC voltage measurements. With this setup the exact resistance of the shunt can be checked. Current measurements can be done with this setup too with a 10A max. That’s where the figures came from. Both the Fluke and bench DMM agreed on the current, the shunt did not. I’d trust on the Fluke and high precision DMM. So expect the shunt to be wrong and are very interested whether more Smartshunts don’t meet there specs.)
Does the fact that you are trying to measure the accuracy of a 500A device over the range of 1 to 10A ( 0.2% to 2.0%) of the equipment range have any bearing on this. I am not an electronic specialist, I am a Chemical Engineer and in my experience of measurement on petrochemical plants is that you can not expect the same accuracy at the very low values compared to what the meter is ranged for. I do not know if this applies in electronic measurement or not. It is a question / comment seeking an answer rather than an answer itself.
Will be keen to see if it repeatable with higher currents.
Since a clamp reading is based on magnetic field ensuring 90° as angle and orientation matter and is important and making sure nearby cables are not affecting the readings either.
There would also need to be temperature compensation for the calculation throught the shunt. So there is a possibility that the shunt could be correct in resistance but the calculations are not.
A further idea would be to alter the shunt value in the settings, since that is possible to do and see if you can marry the figures. And repeat that through several amperage test ranges. Since measuring the resistance on the shunt is not really possible with most meters since they can’t resolve it properly.
That might be very well the case indeed. With the specs of Digital Multimeters every range has it’s own accuracy and it differs in every range. However, Victron just gives an accuracy of 0.4% and doesn’t mention anything regarding the range. I’ve conducted more measurements today, will follow up on the results.
I have conducted the test in the range 0-10A just because of the reason I can measure with high accuracy in that range. So, did not use a clamp for current readings because they have less accuracy. Because of the very low currents there was no thermal heating in the shunt, also checked with a thermal camera.
Regarding altering other values in the shunt value; I was not yet aware of that setting. Where can you change the shunt value? And then I think you can edit the internal resistance?
We think it has to do with temperature differences only.
Henge, in the winter (20-25 degrees C in the tech room) showing quite good but in the summer (30-40 degrees in the tech room) quite wrong.
This due that temperature of the shunt is not compensated.
Normally all electronics should read correct between 20-25 degrees C (as we set the parameters correct then), above and below it has some error.
Then, the position how it is mounted, e.g. on the wall or on the floor.
On the wall is better as it “Cools" itself better, on the floor the top plate is getting hottest as the other ones below heat it!
If this problem persists, place a “mini” ventilator which blows between the plates, issue solved and temperature almost constant with the tech room temperature.
At 500A and 50mV, it creates only 25W heat loss, so you do not need a huge fan.
Hello @mystudio,
considering that the measurement accuracy is specified without a measurement range, the accuracy of ±0.4% should refer to the full scale, in your case to 500A. This results in ± 0.4% * 500A = ±2A. This means that your shunt far exceeds the specifications. This is logical, as the 2A applies to the entire temperature range of -40°C to +50°C and -500A to +500A, and you are probably measuring under ideal conditions.
I’m wondering why I’m writing this, because anyone who has access to such high-quality measuring instruments should be aware of these things.
I have to admit that the high resolution of 0.01A suggests for sure high accuracy, but this is in fact not the case.
Thx Ulfilas. You might be right. That can be the case. I’m not a know it all, but I like to be a learn it all. So, I learned something today. Normally, a percentage regarding the sensitivity/accuracy is based on the measured value. You say it’s based on the full 500A. That would mean it can be around 2A off at very low scales, which makes no sense I think. But I’m not sure. Still also looking for the explanation for the fact that the power consumption of the Smartshunt is almost 4 times the specs. Instead of 1 mA, it’s almost 4….even without bluetooth active communication.
This is what I’ve found so far. I’ve measured the resistance of the shunt in several times to get a more precise result and it looks like it it just the resistance which is off. It’s not 0.1 mOhm, but less. That gives missreadings. If you look at the electronics of the shunt to calculate the current, that that follows very well the voltage over the shunt. So the electronics seem to be very precise, it is just the shunt which is not trimmed exactly. If there WOULD be an option in the shunts software to trim/adjust the value they are working with (the 0.1 mOhm), then you could calibrate the shunt very precise.
Like said, this is a shunt in a test setup. I have always had the feeling that the shunt in my production setup was spot on. Now I have to do the same tests with that one I think to see if my feeling was right.
Always good to see someone check stuff, instead of just trusting it.
However I do not think it’s fair to state “it does not perform as expected” or “bad version of the Smartshunt” . At these low levels (1 mA instead of 3 mA) it is very difficult to do accurate measurements. I would genuinely love to see how you would set up a measurement that precise. Especially for the resistance measurement.
Also, your SmartShunt can measure -500 to + 500A and costs €90. Most Fluke DMMs can only measure 0-10A at a much higher cost. Not really fair to compare them. The SmartShunt should be within specifications, and it is at the higher power levels that it is designed for. At 10A you are at 2% of the measuring range. (or is it 1%? -500A to +500A is a 1000A range)
It would interesting to know if energy in is measured exactly the same as energy out. So if you reverse the current, is the measuremen exactly the same?
