Archived Forum

How expensive is your commercial meter, and is it a true RMS meter?  What sort of equipment is drawing the 1.8A when the kettle is off?

I've got a fairly popular $100 clamp meter that doesn't do at all well with signals that aren't nice sine waves.  The attached trace is the current going into my on-line UPS, which has a whole bunch of electronic devices with nasty SMPSs on the other side of it.   You can see it's a pretty ugly signal.  My cheap scope, with even cheaper (but calibrated) current probe, thinks the current is 1.808A RMS.  

My precision front-end with 3 calibrated CTs wrapped around the feed all think the current is extremely close to 1.801A RMS

My clamp meter flips between 1.2A and 1.3A RMS.

For kicks I just fed a 1.8A RMS pure sine wave through my clamp meter and it read 1.6 - 1.7A.  At an 11A pure sine wave it read 10.8 - 10.9A.   So mine seems to naturally under-read a bit, but particularly so if you don't feed it a sine wave.

Few meters will ever be correct on every wave shape that you can think of. Most budget and mid-range meters read "rectified average scaled rms" - which means they respond to the average value of the waveform after rectification, but the scale is then correct only for a pure sine wave. The form factor (ratio of rms to average) of the wave shape is what matters. Even true rms meters will only be accurate over a limited range of form factor, since they run out of range if the waveform becomes too 'peaky'.

Thanks for your replies.

The bought clamp meter is a mastech ms2101. So not expensive compared to what's out there, but expensive compared to this home built one (maybe not). Not sure if it's true rms.

the 1.8A is just what the house is using normally (so fridge, computers, things on standby, maybe some lights).

so what's each of your advice to calibrate my arduino? Buy a power supply that can accurately output a set current, and work off that ? Or just do some maths to calculate my calibration constant in the code...?

my arduino project is logging current values over time, so I was hoping it be accurate! For eg, I want to see what my 5kw solar system actually produces at its peak production in different seasons.

But if it's off by even a few amps, I'll get skewed results.

and is it weird that the meter reads 1.8A & 11A (using my initial example) and my arduino reads 1.8A and 9A on the same currents? Like it's not a linear error?

so my results will only be wrong either at high current OR at low current.

or maybe the results are correct, and it's my electrician's meter that's wrong! Arrrgh. Any advice appreciated

You could rule out my theory (that it may be the non-linear standby loads confusing your clamp meter) by replacing those loads with a known load.  You could hack up an extension lead so that you can clamp it, and then re-do your tests with the kettle  Vs some other nice simple resistive load at about 2A, and again compare each with your clamp meter.  

So you need a ~500W load that can plug into the end of your extension lead, and is nothing more than a heating element (or incandescent light bulbs).

As far as your commercially produced meter is concerned, you should have details of the accuracy they claim for it. Most probably, it met that specification when it left the factory, whether it still does is open to question.  There's a Building Blocks article about the accuracy of meters, that also tells you how to interpret their spec "±(2% + 5)". As the spec doesn't claim "True rms", I think you can take it that it's rectified average, scaled rms.

For your Uno, we know a little more. Experience of problems other users have reported here suggests that interference from the digital circuits can get into the analogue front end and give you a reading of current when there is none. That is easy to determine - put the CT well away from any cable or appliance that generates a magnetic field and see what you get. The other error is one of scale factor, and that is covered in a Building Blocks article. If you know the tolerance of each of the components that you used, you can calculate the limits for "your" meter. Bear in mind that while most errors will be linear, your CT will not be and the curve "bends", so it will tend to read high at medium current and low at high current. You can read more in the report on the YHDC CT.

If the two ranges overlap (at whatever current you choose) then both are "right" to within your ability to measure. If not, then you don't know how far wrong either is, and you need a third reference. If you are that concerned, have you considered borrowing or hiring a good quality calibrated meter to calibrate against?

As a rider to dBC's answer, you can multiply the current by passing several turns through your CT (and your clamp ammeter), so a 1 A load can give you 1, 2, 3, 4, 5, 6 A etc so you have many scale points at which to compare the meters.

And by writing your own sketch to rectify and calculate the average (rather than rms as emonLib does) and then multiply by 1.11 - the form factor of a sine wave, you could have 'your' meter reading the same quantity as the commercial one.

your CT will not be and the curve "bends"

Interesting.  I knew we all had to deal with phase shift varying with current, but you guys have the added burden of amplitude errors as well?  Is that down to the relatively high voltages you try to produce across your burden resistor?

In my application I use the 333mV style CTs, and they give excellent amplitude response across a wide current range.  You inspired me to re-test that just now.  I tested a 20A (Ch6) and a 50A (Ch8) CT at: 20mA, 50mA, 2A, 10A, 20A, 35A and 50A.  

You can see in the attached readouts, the only disappointment was the 50A CT trying to read 20mA, but at that level it's only outputting 133uV.  Both CTs performed well all the way down to 1/1000th of their rated max current.  Of course, having 24-bit A/Ds helps measure that finely.

Take a 100 A CT up to 250 A and the curve is very nearly flat - see the report on the YHDC one! That was measured into a multimeter requiring a maximum of 200 mV to drive it - it's the magnetic material (ferrite in the case of the YHDC) saturating.

Thanks for your replies.

The only thing I could think of that I have around me that could help me is my solar system...my inverter reads out the wattage it's outputting. So I clamped my arduino clamp around the AC output, changed the code to display wattage, then compared the two at different output levels as the sun changed. i figured this would be a good test, as the sine wave wouldn't change with the changing current.

I set the calibration constant to a level that read about correct at 1800 watts. Then at about 1000 watts, the arduino was over reading by only about 50 watts. At 200 watts, the arudino was over reading by 150 watts.

These reading were quite acceptable by my standards! I was quite happy.

Interestingly, the commercial clamp meter was under reading the output by about 150 watts when it was outputting about 1250 watts. I should do some more tests to see it's accuracy at different currents...

Now that I think of it, I'm not sure if the inverter is displaying the AC wattage, or the DC input wattage.... Surely it must be the AC output...

I know mine displays AC wattage, and like you, I would guess that's the norm.

I assume you just used a nominal voltage to turn your CT reading into a wattage?  That will probably work fine for the purpose of this linearity experiment, since your inverter is probably only outputting real power, and no matter what voltage you chose, you would have calibrated away any error when you matched them at 1800W.

One big assumption you've made though is that the grid voltage is constant.  I don't know what it's like over there, but in these parts it varies by a good 10V, often quite abruptly, so if yours is similar for example, that would introduce an error of about 4% there alone.  At least some of that variation is down to the massive PV penetration here where you can end up with big residential areas producing massive amounts of power, and nobody home to use it.  It eventually finds it way to shops, factories and office blocks that want it, but the resistance along the way tends to push the voltage up at the residential end.  So it's conceivable that your grid voltage when you measured the 1800 watts was higher than when you measured the 200 watts, which would upset your linearity.  Does your inverter display the AC voltage as well?

There are some interesting graphs in this article:  http://reneweconomy.com.au/2013/confessions-of-a-network-operator-solar-changes-the-game-35541 .   We now have some major 11kV feeders that run backwards during the middle of the day.

the commercial clamp meter was under reading the output by about 150 watts

I've seen claims in these forums by reliable contributors that the current signal coming out of their inverters is not a sine wave, but rather sags near the peaks.  If yours is doing that, then again it could be confusing the clamp meter's RMS calculation.  I've never seen that with my inverter.  Whenever I've clamped it, it's always produced a very clean sine wave, perfectly aligned with the voltage.

Good, I hope mine is AC wattage too.

Yes the inverter outputs the AC voltage. It was pretty stable staying within 247-249v. So I was confident with my multiplication to wattage.