Mains current monitor to ADC

[Mosaic]

Hi:
I need to design a 0 - 10A, 230V 50Hz mains monitor using this sensor:
https://www.newark.com/triad-magnetics/cst-1015/current-transformer/dp/46M5100.
4 identical channels.

I only have a 3.3VDC supply to use.

My approach is to use a quad opamp like this one:
https://www.farnell.com/datasheets/1669460.pdf

In the form of a voltage follower to output a max. 0-3V signal representative of 0-10A. A bit of voltage offset at the low end is ok, perhaps 0.2V max.

Design:
So the mains current sensor outputs into a schottky rectifier and then into a 5K ohm burden potentiometer (adj. to output 3V at 10A RMS) then to the voltage follower via a 10Kohm resistor, a 3.3V zener clamp, and a 22uf capacitor , low pass filter. Basically a DC signal will be buffered and output.

The weakness I see here is perhaps 0.2V of the signal is lost to the rectifier. This equates to a 200mA offset with the 5K burden.

Can anyone suggest a better approach to reduce that offset? Would a MOSFET active rectifier be the solution, burden current load is around 2mA.


thx!

 

[moty22]

Look at this thread; https://www.electro-tech-online.com...sformers-for-input-to-microcontroller.143734/
posts #7 and #15

 

[Mosaic]

Thanks for that moty22,
Now I need a non linear transformer model for LTspice to do a sim!

Or shall I just alter the ideal transformer turns ratio to match the V/A rating across the burden resistor?

 

[ronsimpson]

First you should use the 100 ohms as the part was built for. Don't use the 5k.

Here is simple hardware but your software will be hard. The CT is connected to 1/2 supply and swings above and below. The rectification and averaging needs to be done in software. There will be no diode drop because it is in software. You will need to sample 600 times/second or more.

15A will give you about 1.5V.

Post #7 shows that the diode drop is removed if you know where to put the diodes.

In spice you need a transformer with a K=1 or 0.98. The inductance is 500nH:500mH. That 1,000,000:1 inductance is for a 1,000:1 turns ratio.

 

[Mosaic]

Well, my spec calls for a 0-3V DC signal to be ADC sampled. Plus a low pass to knock out transients....
So here's what I have so far: The blue line represents the desired 10A sampling = 3V. The top line shows the voltage clamping and the lowest line shows the 0V rail offset.



The LT1677 (rail to rail) is a bit costly, I am considering this:
https://www.farnell.com/datasheets/1669460.pdf
As a viable substitute.
What do u guys think?

I am assuming that the specs of the CT are in RMS?
**broken link removed**

 

[crutschow]

The burden resistor should be no more than 100 ohms according to the spec sheet. Why are you showing 5k ohms in your simulation?
To get a higher voltage you need to amplify the signal, not use a higher value burden resistor.

 

[moty22]

Thanks for that moty22,
Now I need a non linear transformer model for LTspice to do a sim!

Or shall I just alter the ideal transformer turns ratio to match the V/A rating across the burden resistor?

I'm afraid I can't make useful contribution to this discussion. I like electronics nearly as much as I like sex and this is why I don't use simulators in these two cases.

The LT1677 doesn't do anything, you'll be better off without it.

 

[ronsimpson]

1) Please send your .ASC file. You might have to rename it .TXT to post it. I can't see the details. I don't think you are getting good results because of some errors.
2) Two of us don't like using 5k burden resistor. It is true that using 1k should give you 10x more gain. It will also cause core saturation sooner. There will be more VT on the core. (with 100 ohms you get 0.1V/amp so 15A = 1.5V) So if you do down this road change to 200 or 240 ohms. Your spice will not show core saturation!
3)You are measuring (more or less) peak current. Do you want RMS or average? Power line current is not sign wave. Many loads are like computers where the current is not sign wave.
4) Don't add a POT that reduces the voltage.
5) The amplifier has no voltage gain so why?

 

[Mosaic]

The spec sheet shows V/A specs for burdens of 100, 200, 2K and 5K. I am using the CST-1015 (15A rated) for a 10A RMS AC, thus saturation isn't likely to happen.
The opamp is a voltage follower buffer. It has a purpose of converting the higher impedance signal to a low impedance signal.

The pot allows me to calibrate the output to deliver 3V signal at a 10A on the primary. Linearity is not the issue.

 

[crutschow]

The spec sheet shows V/A specs for burdens of 100, 200, 2K and 5K. ..................

Could you post a link to that data sheet?
All the data I found on it refers to only a 100 ohm burden resistor.

 

[Mosaic]

Bottom of this pdf...although I may be reading it wrong.
**broken link removed**

Do u see the V/A for various loads mentioned in the middle?

 

[Mosaic]

Ok, I adjusted the burden to 100 ohms as the better approach and reorganized the OPA to provide the required gain.
Here it is:


Spice Asc file zipped.

 

[ronsimpson]

Looks good. D5 could be a diode to 3.3V.

 

[crutschow]

Bottom of this pdf...although I may be reading it wrong.
**broken link removed**

Do u see the V/A for various loads mentioned in the middle?

Yes, I do but I'm confused as to what the load resistor is as compared to the burden resistor.

 

[ronsimpson]

I think the load and burden resistor is the same thing.
You can can change the resistor value. As long as you know what will happen when you do.
1) If you want more voltage across the transformer you will have more (voltage time). Core saturation happens when there is too much VT on the core. Time has to do with 50hz or 60hz. Voltage is simple and related to current and resistor. I believe the core starts to saturate some where beyond 50hz and 15A and 1.5V assuming 100 ohms. I think you will be at the same point with 50hz, 7.5A, 1.5V and 200 ohms. Spice will not show this!
2) The frequency responses will change with the resistor. (your 1:1000 transformer will not show frequency responses because you did not include inductance) With my transformer, 100 ohms shows a -3db point at 20hz and -1db at 45hz. With a 200 ohm resistor the two points move up by 2:1. Using 1k moves the frequency responses to a point where 60hz is hard to see and is not accurate.