Ideas about generating current for testing CT

[ali_asadzadeh]

Hi,
I want to test and bin CT's, the input range is 5mA to 10A, and the desired frequency is 40 ~ 100Hz, Basically would be used like this a single turn wire would pass trough the CT under test, and test currents would be generated to measure the Output of CT, so I need to apply an ARB sin wave, this part is easy I can use 16Bit DAC to generate a sin wave with the desired frequency and an amplitude of say 3.3V, Now I should convert this to a constant current, this simple circuit would came to my mind, But I think it would  add some DC content to the current, and the output would not be a pure sine wave, so What do you suggest that I would do?

[ali_asadzadeh]

I made progress in simulation
For the 5V I propose to use industrial 5V 40A switching power supply that would drive LED's (since they are dirt cheap), But if you suggest a way to lower the voltage, since it would be make dissipation a lot lower! But what's the cheapest way to do it?
This is my proposed circuit, How can I improve it? Is there a away to at least remove -12V supply? How can I increase the output to be able to handle 30A,


The DAC values are centred around mid supply (2.5V)
Out put wave forms,

[Terry Bites]

Basic opamp circuit theory shows you how create a set current in the primary of the CT.
The burden resistance is reflected into the primary, it will appear to the opamp as regular feedback resistor.
At 50Hz or 60Hz its all pretty straight fowards. THD good enough, DC accuracy woes vanish, happy days!
The "opamp" needs to be able to generate the current range you need and there are many current booster/ buffer schemes illustrated on the interweb.
Some complementary Darlingtons a couple of 4148s, resistors to bias and prevent meltdown might be enough.
You can easily use an audio power-amp as the current buffer.
If you look into its schematic you'll quickly see if it can be operated as the complete power opamp.
A class d might make a good buffer, but doesn't easily convert.

I'd be tearing down that subwoofer box mañana

[inse]

What parameters of the CT do you want to test, CTR, linearity, frequency response?
What equipment do you have at hand.
Multimeter, HD oscilloscope, highspeed DAQ, function generator?
What kind of CTs do you have- clip-on, cores to feed a wire through or solderable?
An other idea comes to my mind: use a suitable inductor together with a freewheeling diode and a transistor as switch to generate a triangle waveform by applying a pulsed voltage.
The pulse length determines the peak current and by both capturing the primary current and the secondary voltage you can test for linearity and CTR.
Just an idea…

[ali_asadzadeh]

What parameters of the CT do you want to test, CTR, linearity, frequency response?

All of them, I have a SDS2104X scope and a SDG2042X function gent too.

What kind of CTs do you have- clip-on, cores to feed a wire through or solderable?

My CT's are like ZMCT116A

I have modified the Circuit, so the current can be monitored and regulated, The updated circuit is attached, it's in Tina-TI software, The output current has some DC bias offset, how can I remove the DC offset?

The modified circuit for monitoring the output current


The output has some DC current offset to the output.

[Jeroen3]

Would an GW-Instek ASR-2050 arbitrary ac/dc generator fit in your lab? It can be very helpful in situations like this. It can do current limited AC (not clipped!), however you can also get a big few-ohm power resistor in series to help with that. Imposing DC is also not a problem.

It won't get you to 10A, but a few extra windings may do that for you, it goes up to 5A.

[petemate]

Why do you need both a positive and negative value for testing the current transformers? You can go from zero to Ipk, then flip the CT current path and do the same thing. That would be equivalent to a negative voltage through the original current path.

In any case, you can easily produce the desired voltage with a (powerful) audio amplifier. It can probably be purchased for much less than you'd spend in time and money trying to develop your own circuit.

Your circuit looks similar to an H-bridge, with one side tied to VCC/2, to give you that artificial ground level. But there are easier ways to do this. The obvious is a half-bridge with a dual supply. The second is a full-bridge(H-bridge) with a single supply(at half the voltage). By varying both sides, you can get the equivalent of a dual supply across your resistor/CT, because you basically flip the voltage. (this is what many class D amplifiers do)

One thing to note is that your transistors will become very hot when you are in their linear operating regions. With the opamp controlling the gate voltages, you'll basically burn the whole supply voltage at 10A in the FETs. For this reason, you can look into class D amplifiers, which turn the FETs on/off and then low-pass filter the resulting square wave to get a sinusoid.

The gist is that you should just get a powerful audio-amp and use that to test your CTs. Everything else is a waste of time and money, unless you are doing this as a learning experience. In that case you should build a half-bridge and flip the CTs. If you want an even harder learning experience, you should build a full-bridge class d amplifier.

[dietert1]

Why not use a second current transformer to generate the high current output? Then we have one driver transformer and one sensor transformer (the one to test). Both cores get linked by one common wire and that wire needs a shunt to measure and linearize drive current.
I think this way one can make the setup with opamps only. How many turns are on the sensor core? If it has enough turns to do the measurement with an opamp, you can reach the test current on the driver side also with an opamp.

Regards, Dieter

[ali_asadzadeh]

Thanks for the feedback, Jeroen3 I can not Use ASR-2050 arbitrary in my lab.

petemate thanks for the feedback, I can not use your Class D methods, because I want to generate precise current to calibrate the CT's under test, also I should be able to generate harmonics to my generated current, so I think this method That I'm going is better than your suggestion. But thanks for the tips.

[petemate]

petemate thanks for the feedback, I can not use your Class D methods, because I want to generate precise current to calibrate the CT's under test, also I should be able to generate harmonics to my generated current, so I think this method That I'm going is better than your suggestion. But thanks for the tips.

You can easily generate "precise current" to an arbitrary degree using a few different methods of current sensing. The same goes with harmonics. You just create the signal, including harmonics, that you want to send through the CT and pass it through the amplifier. Its a question of placing your switching frequency sufficiently high above the largest component in your signal. Audio switch-mode devices typically work at around 400kHz. They are meant to reproduce up to approx 20kHz. If you want to go to 100Hz, plus harmonics, 400kHz should be more than enough.

Again, easiest solution is a audio amplifier. That is basically what you designed.

[dietert1]

The correct and most efficient method is using one current transformer to generate the test current for the other one (DUT).

Regards, Dieter

[jbb]

To expand Dieter’s good suggestion:
- 10A is quite a bit of current out of an amplifier
- you probably need much less than 1V
- why not generate something more like 1A 10V with the electronics and put that through a transformer inside the tester?

Also, IF the CT has a big enough hole in the middle to pass a multi-way connector, putting a 2-turn (or more) test winding in would reduce the required current to 5A, which is a bit easier.

I note that if you want to do some tests @ 10A and other tests @ low current,  it’s probably good to build a current source with a range switching relay.

Now might also be a good time to check up on:
- What sort of test precision do you need? There are trade-offs and hitting 5% uncertainty is a lot easier than 0.5%
- Are you expecting to test a lot of CTs? If so, the mechanical design of the test setup could use some attention, especially if you have access to a mechanical engineer

[ali_asadzadeh]

Guys thanks for the feedback, The CT's are inside Protection relays that I have designed for the smart grid, and each Relay has at least 4 of them, So they are a lot of CT's for calibration and testing, also Note that I need reasonably low currents like 10mA to at least 10A for calibration, as the normal conditions would suggest up to 30A primary current too, which for this phase and making things easier I omit for specs for now. so the CT's has a non linear curve from say 10mA to 10A, and I need to calibrate the amplitude and phase, so I need precision enough in my design.

Here is an updated circuit, I managed to remove DC bias, But if I increase the output current, the circuit will some how oscillates and generate harmonics, But in lower range currents, everything seems perfect!

New Circuit


Up to 1A it seems it's doing Ok


Higher currents like 5A makes the circuit unstable!

[TopQuark]

https://www.eevblog.com/forum/projects/gt1mhz-bandwidth-electronic-load/

Hate to be that guy that constantly direct people to their own project, but I think my circuit can do everything you want and more. 

The circuit shows a working 0-50A DC load with >1MHz control bandwidth. Pair this with a good voltage source and some calibration work, it should suit your needs.

[CosteC]

My CT's are like ZMCT116A

I do not see stated stability of test current, so maybe you can use transformer say 5-6 V one and load it properly to get desired current?
With transformer designed for this purpose I got 400 A, which I measure with reference CT and compare it to tested CT. You can do it at your lower current level.

Another way to get 5-10 A relatively easily at 40-100 Hz is audio amplifier loaded by resistor, driven by signal generator.

[ali_asadzadeh]

https://www.eevblog.com/forum/projects/gt1mhz-bandwidth-electronic-load/

Hate to be that guy that constantly direct people to their own project, but I think my circuit can do everything you want and more. 

The circuit shows a working 0-50A DC load with >1MHz control bandwidth. Pair this with a good voltage source and some calibration work, it should suit your needs.

Thanks for sharing TopQuark, which version of your design do you suggest? Also I have these questions

1-your output is ground referenced, it means I have DC current in my AC currents, how can I make it mid supply referenced like my proposed circuit to remove the DC component.
2- your output is a burst with uS ON times, But my design should be alive and active cooled by a fan, like 1 minute or longer for testing various scenarios, because After calibrating my CT’s I want to use this circuit to test the functionality of the Protection Relays too.
3-Have you made a simulation file, if the answer is yes, can you share it here?

[TopQuark]

https://www.eevblog.com/forum/projects/gt1mhz-bandwidth-electronic-load/

Hate to be that guy that constantly direct people to their own project, but I think my circuit can do everything you want and more. 

The circuit shows a working 0-50A DC load with >1MHz control bandwidth. Pair this with a good voltage source and some calibration work, it should suit your needs.

Thanks for sharing TopQuark, which version of your design do you suggest? Also I have these questions

1-your output is ground referenced, it means I have DC current in my AC currents, how can I make it mid supply referenced like my proposed circuit to remove the DC component.
2- your output is a burst with uS ON times, But my design should be alive and active cooled by a fan, like 1 minute or longer for testing various scenarios, because After calibrating my CT’s I want to use this circuit to test the functionality of the Protection Relays too.
3-Have you made a simulation file, if the answer is yes, can you share it here?

1- Have the DC load drive a transformer, and hook your CT to the secondary coil?

2- Well obviously drive the load with as low a DC voltage applied to the load terminals as you can, without creating distortions in the current waveform. You can then calculate the power dissipation for the shunt and MOSFET, and size them and heat sink it as appropriate. If by your calculations you can't continuously dissipate the power without something blowing up, you can just build multiple copies of the circuit and run it in parallel to share the heat. Do not attempt load sharing by hooking multiple MOSFETs to a single control circuit, the heat will not be evenly shared and will pop one by one.

3- No, I just kinda winged it. I even had to reverse engineer what I built to come up with the paper schematics you saw. Use the schematics shown in Reply 8 for the advertised 50A 1MHz load.

[ali_asadzadeh]

Have the DC load drive a transformer, and hook your CT to the secondary coil?

By adding a CT, I could make the situation worse, Because I want to have precise currents, so I can calibrate for amplitude and phase, we should find a better way to remove the DC offset.