Re: CT Current measurement

[gg339]

Hi all,

I'm looking for some advice in relation to a project I'm working on at the present. The idea is to use a 1:3200 CT to measure the energy in a 230V mains supply.

I have experimented with various ways of achieving this but thus far have not succeeded. I'm looking, where possible to measure RMS V and I as this will lead to correct power factor results, however would settle at this point for relatively close non-RMS measurements. The mains in my house is mostly resistive so the RMS aspect is not as critical.

My first iteration used a diode bridge feeding a 100R burden resistor, but this was inexact due to the forward voltage drop on the diodes.
Second attempt was w3ith an AD736, with the idea to convert the AC to a DC RMS and read same with a PIC adc - had one AD736 to hand but somehow managed to damage it, so awaiting another.
I'm wondering if anybody has had any success, or ideas for a schematic for the CT, AD736 and the PIC - I'm worried that some misconnection on my part was the cause of blowing my original 736.

I have thought of increasing the burden resistor to closer to the CT core saturation point in order to gain some more current out, and then revert to the diode bridge, feeding the burden resistor lastly - any thoughts on same?

At one point with the 736 displaying what seemed "nice" voltages, I connected same to the PIC, and since that point the 736 has gone south - I had forgotten the PIC ground on the circuit and can only suppose the ADC portion of the PIC did something to the 736.

Thanks in advance for your time.

Regards,
Ger

 

[Roff]

If you want to convert AC to DC, Google "precision full wave rectifier". You will find several circuits you can use.
I'm assuming you know that you can get extremely high voltages from your CT if the burden resistor is not connected.

 

[crutschow]

The 736 seems like a good circuit for your application. Certainly it could have blown because of a lack of a ground connection. A good ground connection is needed before you connect the 736 and PIC together. You might also use a 1kΩ resistor between the 736 output and the PIC input to limit the current in case you lose a ground.

And you need to be careful when measuring the mains voltage since it's not isolated and obviously lethal. Use at least 100kΩ, ½ watt resistors where you connect to the lines to minimize any fault currents.

Of course to accurately measure AC RMS power you need to simultaneously measure both current and voltage and multiply them together. Analog Devices makes a number of chips that do this https://www.analog.com/en/analog-to...gy-measurement/products/index.html?ref=ASC835 but that may be more elaborate than you require.

 

[gg339]

Thank you all for your comments. Yes I'm aware of the high voltages with leaving the CT without a burden resistor, and I'm careful to ensure one is connected to it at all times - once testing and design is complete, I'll return one into the body of the CT, and this will ensure that no high voltages can occur open circuit.

I have googled a number of precision full-wave rectifiers - however to be able to get a single device with little external components would be more practical for me as it removes the need in some cases to ensure matched resistors etc that are sometimes required in the case of a precision rectifier.

In relation to the proper method of connecting the CT - is it the case that I should connect one end of the CT to ground (assuming the burden resistor is present) and only connect the other pin to say the 736?

Thanks again

 

[crutschow]

In relation to the proper method of connecting the CT - is it the case that I should connect one end of the CT to ground (assuming the burden resistor is present) and only connect the other pin to say the 736?

One end of the CT should be connected to the same common (ground) as the 736 common. If it also connected to the circuit that measures the mains voltage, then the circuit common should be connected to the mains common. In other words, all commons need to be common (hence the name). If not, you may zap something.

 

[gg339]

Thanks again crustchow,

I take your information regarding the CT connection to common (ground) below.
Can I ask where this then leaves the ADC input - obviously the output of the 736 (assuming no opamp buffer etc) goes to the ADC - should it be then that the gnd connection to the PIC should also be connected to common - I don't want to blow another 736 - for their size they are quite expensive over here!

Regards,
GG339

 

[crutschow]

Can I ask where this then leaves the ADC input - obviously the output of the 736 (assuming no opamp buffer etc) goes to the ADC - should it be then that the gnd connection to the PIC should also be connected to common - I don't want to blow another 736 - for their size they are quite expensive over here!

As I stated in my first post, the commons(gnds) must all be tied together. All circuits (expect for those with specific isolation connections) require it. A signal always has to have a return path. That's normally achieved through the common ground. It's the first rule of circuits. An improper or noisy ground is the source of many circuit problems and poor performace.

 

[gg339]

Thanks Carl,

While waiting earlier tonight, and as I didn't have another AD736, I reverted to trying the circuit with a schotty-based bridge rectifier, before the burden resistor. This has turned out quite well, as I increased the burden resistance to compensate for the diode forward-voltage drop. Obviously this arrangement does not account for RMS values, but I could allow for same within the maths routines in the PIC - don't know how easy that will be.
In any case I'm now getting nice readings on the LCD, which seem to be in relative agreement with a clamp meter I also have for reference on the live feed. Only issue with the present arrangement, notwithstanding the issues at low values which the schottys may introduce and the RMS aspect, is the fluctuating values on the PIC - even though I have a decent smoothing cap in parallel with the PIC input.
I'm going to give the AD736 another shot once I get one, or even try again to build a precision op-amp rectifier, but at least some progress has been made.

Thanks again for your input, it is much appreciated.

Kind regards,
Ger

 

[gg339]

On another note I was rereading some of your post earlier, and indeed you are correct in stating that to fully measure AC RMS I will need to measure both voltage and current - ideally I would need to measure these many times per cycle and sum the result. However, for the moment I'm cheating, and broadening my error by hard-coding 230V RMS into the PIC - if circumstances permit i'd ideally like to make this project as accurate as possible.
As you also stated I could use a device such as the ADE7755 or equivalent which would offload all such processing from the PIC, and would be probably many times more accurate, however my PIC programming skills are not the best - I've only gotten my hands dirty with them in the last few weeks, so programming suitable interfacing between such a chip and the PIC might be more than its worth at the present. Maybe later I'll try something like that. Overall though I have to admit that the project has been interesting, and I'm learning a lot about electronics. My background is in experimental physics, and while I have some electronic knowledge it was always just from tinkering since I was younger with radios, computers, interfacing etc, so its nice to be able to somewhat design implement something that requires putting theory into action and learning from the DMM and scope what is going on.
Also the help from forums such as this is extremely valuable, and I thank you all for your time in responding to my post.

Regards again,
Ger

 

[MrAl]

Hello,

You may also want to check into "synchronous rectification".
These introduce little error into the measurement.

 

[Roff]

Why can't you do A/D conversion of the CT output, and do a peak detector in software?

 

[crutschow]

If you do an an oversample A/D conversion of the AC signal (at least 10 times the line frequency but the higher the better), you can perform an RMS calculation of the samples (say 1 cycle's worth). A running average of these calculations will give the true RMS of the waveform, even if it's significantly distorted.

What is the sample rate of the PIC A/D converter?

 

[eblc1388]

There is some useful information in this article, in particular the hardware front end that measures voltage and current of the load:

AVR465: Single-Phase Power/Energy Meter