D-Class for testing CTs?

[steviec7]

Hello Everyone!!

I wanted to try and build a test source for checking Current Transformer accuracy (CTs).

I therefore need to be able to provide a variable AC current 0-5Amps were the load would just be wire passed through the CT so ~0.01ohm. The voltage would be vey low as the resistance is so low. The easiest way to achieve this is to use a variac on the primary side of AC supply but I want to do it another way.

If I used PWM & MOSFET Bridge supplied from a DC source would this be able to give me a good current resolution/accuracy. I've seen the Class D amplifiers, with a 50Hz signal produce X watts through say 4 or 8 ohm but can these systems handle such a small resistance load as 0.01ohm and still provide an accurate variable puresinewave current waveform? :-?

Many thanks in advance for any helpful information

 

[svhb]

Is this for testing in a production environment?

A class D amplifier can be seen as a switching regulator, I don't know if you really want a switching regulator for a test system. You can have problems with noise generated by the test system itself, and this is for sure not a good thing.

Also, for a test system, I don't know how important the efficiency of the system is. How big is the current you need to generate? Maybe you can use an off the shelf programmable power supply, or a simple linear current driver (opamp, power transistors and BIG heatsink), and a function generator.

Stefaan

 

[steviec7]

Testing would not be in a production environment. Its for an example kit to show how ac current is measured using Current Transformers. I need to make a Primary supply for it.

The efficiency of the system is not important. The reason I looked at Class D was the quality of the AC current wave form which needs to be as close to 50hz current sine wave as possible, even if current quite high.

Will a simple opamp -> Mofet with negative feedback and sine 50hz at positive input provide good control with out sacrificing the AC current output waveform? Wont the feedback Voltage to opamp be too small if the load is only 0.01ish ohms?!

The original current level I wanted was upto 5A but lets say 10A to give it more range.

Thanks again for your advice!

 

[svhb]

If you start with a state of the art class AB amplifier, you can change the feedback to be on a small resistor in series with the primary of your sensor. This way, you don't depend on the resistance of your primary winding.
For example a 0.1 Ohm resistor will generate 1 Volt with 10Amps. This should be more than enough for a feedback signal. (it will however dissipate 10Watt).

Somewhere I have a schematic of something like this. Give me some time to search it.

Stefaan

 

[dick_freebird]

A long time ago I was faced with the problem of making an
accurate pulsed current source. With the secondary problem
of measuring it accurately. CTs of the day were about 5%
datasheet, no guarantee,; part had to be sub-%.

What I came up with was a variation on the transmission
line pulser like people use for ESD testing. The charged
line puts out a nice square pulse (if you tweak the
termination right). I used a mercury relay for the switch.
A current transformer placed in the termination resistor
path sees the line current. The resistor value and the
pulse voltage can both be measured accurately (DMM,
'scope; 'scope can be calibrated back to a DC source).
So from these you can know the true pulse current and
derive the correction for your CT-measured pulse.

Easy. Only took me weeks of late nights to build the
thing (bench automated test for 5 channels times 4
current sources apiece, laser trimmed). Then the
fixture ate itself in a few months from the 28V supply
and icing at low temp, electrolytically chowing all the
pins in the $10K Kelvin pogo contactor. Big fun.

Some CTs have a lower frequency limit and this scheme
would not apply there.

However if you filter your Class D amp adequately and
make it look like a sine wave voltage, and can get
confidence in the measurement of said voltage and a
load resistor, the same chain of calibration idea ought
to work. The main thing is to have the source clean
to better than a small fraction of your error budget.

To which point, you are better off not betting on a
very low resistance, a 0.01 ohm sense resistor will
likely be dwarfed, or compromised at least, by PCB and
wiring resistances. More voltage and higher resistor is
going to be more reliable I would bet, for the same
current. However if you are able to get Kelvin sensing
on a low value resistor, it could work (though your
fidelity at lower currents will suffer from voltmeter
resolution I expect - what, 200mV scale, 1mV accurate,
0.01 ohms is 100mA?).