Hi there,
I have made a PCB Rogowski coil, which is 180 turns with an active integrator using THS4631, but there is high noise when sensing at 200 Hz to 1000Hz (A photo is attached). Any
suggestions for me?
Yellow - PCB Coil & integrator output
Red - FFT of ---> PCB Coil & integrator output
Pink - PEM Rogowski output
The open loop gain of that Opamp is very low for a Rogowski integrator, and with a coil of only 180 turns your are going to need it.
Anyway without the circuit I don’t think much help can be given. Have you at least simulated your circuit?
The photo is of TI reference circuit (PDF attached to this MSG
)
Well, as you can see the OPA2188 has 70dB more of open loop gain than your op-amp, and there is a reason for that...
So, Do you mean the (70dB) open-loop gain is not enough for the integrator circuit?
So, Do you mean the (70dB) open-loop gain is not enough for the integrator circuit?
The OPA2188 has 136db and yours has 70dB less. The gain of a rogowki coils is very low at low frequencies, so you need a higher gain opamp to compensate.
Now the noise seems coming from your power supply or maybe the coil itself since it is probably a good EMI antenna.
Actually, You are right about the power supply, (Previously I have a direct supply with through-hole PCB and had a lot of noise after then I added LDO and de-cap near opamp).
------------------------------------------------------------------------------------------------------------------------------------------------------------------------
About the Open-loop gain
1. The PEM guys are using The same Opamp IC - THS4631 but have much superior results than mine.
2. Even the OPA2188 at my operating freq. (300Hz to 1KHz) has an open-loop gain of around 80dB (Photo attached).
So, I do not think this is a major problem
------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Yes, maybe the coil itself since it is probably a good EMI antenna.
We need proper shielding for the PCB coil -- Any idea about it?
1. The PEM guys are using The same Opamp IC - THS4631 but have much superior results than mine.
I'm interested in how you know what Opamp the PEM guys use - do you have a unit to open up for reference? (If so, posting pictures of the inside would be most appreciated!)
This thread over on HVF may be of interest, the OP makes another DIY rogowski (not a PCB one though) towards the end, though I'm not sure what parts he used for opamps:
https://highvoltageforum.net/index.php?topic=1228.0;allBTW I think the integrator opamp from the PEM unit in that thread (basically unreadable in the pics) is an OPA627, which has an open loop gain of >100dB
The scope trace looks like the amplifier is oscillating. At least there is a lot of high frequency crap that should not be there.
For the low frequency signal the fast OP is a poor choice. There one should more look at something like OPA141.
1. The PEM guys are using The same Opamp IC - THS4631 but have much superior results than mine.
I'm interested in how you know what Opamp the PEM guys use - do you have a unit to open up for reference? (If so, posting pictures of the inside would be most appreciated!)
I really doubt they are using that op-amp, because of the gain, and the high-bandwidth. HF integration is not done actively but by passive components next to the coil, so usually op-amps with less than 20MHz are enough.
Would you mind posting a printable image for making a DIY rogawski coil on a PCB (I can do vias)?
Could you also post your circuit with the resistor- and capcitor values? Do you already have a PCB or is it still on a breadboard? Photo?
About the Open-loop gain
1. The PEM guys are using The same Opamp IC - THS4631 but have much superior results than mine.
2. Even the OPA2188 at my operating freq. (300Hz to 1KHz) has an open-loop gain of around 80dB (Photo attached).
So, I do not think this is a major problem
It may be a problem or not, but that is not how you get the gain of an integrator. You need to trace an straight line starting from the LF point of the rogowski integrator, not a curved one. So again with a straight line the THS4641 has a much lower gain that the OPA2188.
Ya, I do have a photo of the PEM current sensor (photo attached).
OPA627 - 100dB about at high frequency, THS4631 has a better open-loop gain than OPA627 (Photo attached)
1)The THS4631 is used as a output driver NOT as a integrator.
2)The open loop may be higher in the THS4631 vs the OPA627, BUT the integrating gain is higher in the OPA627, because remember you have to trace a straight line to determine this one, and the flat gain of the THS4631 from DC to 10kHz makes it very low to use as an integrator
Is the THS4631 doing any integration, or is it just acting as a line driver? I see what looks to be a THS4031 in there too (plus an OPA130, but that'll be there to eliminate any DC from the integrator).
If you're able to get some higher resolution photos of the lower half of the circuit (and of the bottom of the PCB too, if possible) it would help with understanding what PEM are doing exactly, and what the different opamps are there for. I assume this is the ultra mini (based on CWTUM written in the solder mask)?
Note that the OPA627 was not from an ultra-mini, so having a different circuit isn't a surprise.
RE your noise issues on the DIY unit, as a test have you tried running from batteries to eliminate any possible PSU noise? And yeah checking for oscillation is also worthwhile.
Edit: also worth noting that not all the PEM CWT ultra-mini models are optimised for LF performance (the high sensitivity ones have -3dB points in the 10Hz-100Hz region), so the choices for this may not represent the best options if you want to measure 50Hz for example. (I didn't see a marking to show which model yours is)
I meant like one that I can print out on a transparency to do a double sided PCB etch (a scaled up image file will be enough), I am sure I am not the only one that will build one if its easy.
I think you should have the filter network on the PCB btw, but I don't care (I can dead bug the parts with epoxy), and also use a twisted pair there instead of twin lead I think.
One more doubt related to PEM current sensor and related reference paper, Could anyone explain about the T-shaped Integrator circuit? As I tried simulating it in LT-Spice, but could not see any difference.
That limits the LF gain of the integrator, so that noise does not get too high, dc offset does not saturate the amplifier and also in the area where the opamp runs out of gain. PeM does not use that method
Summarizing, you should, 1) put a limit to the low and high frequency integration band. 2) shield the integrator, 3) filter the output with a low pass filter, 4) get a low noise power supply
That limits the LF gain of the integrator, so that noise does not get too high and also in the area where the opamp runs out of gain. PeM does not use that method
Hmmm.. But PEM guys published this paper wf-ray from (uni - Nottingham)
http://www.pemuk.com/how-it-works.aspx
I kind of wonder, what if you just drill it out on a piece of copperless FR4 and thread wire through the holes, to avoid 5000 vias...? Will it be that much different?
Well sure, but you could also at that point just wrap a piece of stripped-back coax or something. The point of the PCB is that 'P' there.
Tim