Mini Rogowski Coil Current Probe

[Weston]

The mini Rogowski coil current probes that can fit between the leads of a TO-220 transistor that are made by CWT and also re-branded by Tek and Agilent always seemed so nifty. Unlike the DC current probes with a thin film hall effect sensor, there is no "special sauce" in these current probes that is hard to make. I got fed up with waiting for a low cost clone to appear so I made my own. This was also my first project trying out the low cost PCBA services.

The rogowski coil produces a voltage proportional to the dI/dt and requires an integrator, and the integrator low frequency cutoff plays a dominant role in determining system noise. This probe is optimized to be useful for typical SMPS frequencies while being able to resolve relatively low currents. A huge portion of the design effort went to the actual coil design: both figuring out a nice mechanical solution and the optimal wire, coil form, and heat shrink combination. If you want to fit between the leads of a TO-220 package there is a hard cutoff of < 1.7mm OD. I tried about 20 different heat shrink types, 4 types of wire, and 3 different coil forms before coming up with the ideal combination.






The full project details are on the project repo, but the specs are: 800Hz - 25MHz, 0.1V/A, 6mA RMS noise. https://github.com/westonb/rogowski-relief

The dv/dt rejection of the coil decent. If I tape the coil to a floating heatsink that slews ~20V/ns and 250V there is a ~ 1A blip on the output, and there is improvement if you try to keep the coil a bit shielded. The most recent work I have done on the project relates is on reducing the interference from wifi/LTE cellular that gets rectified by the inputs of the opamp.

Verifying the amplifier was my first time trying out noise simulation in LTSpice, its pretty cool to see that the LTSpice model matches pretty closely with my measurements.

The design uses a non-inverting integrator, which is advantageous because you can roll the cable capacitance into the integrating capacitance, and a servo high-pass architecture with a second op-amp.
The servo high-pass is advantageous because it rolls off the low frequency, reducing low frequency gain and nulling any DC offset.



At some point I had a vague goal of making these to sell, but all the regulatory requirements are a pain to deal with and the coil takes a lot of labor to assemble the coil. So the whole project is just open source on github instead.

[moffy]

Impressed with the project. I have also been playing around with a Rogowski current sensor, but using one of the coils from Ebay. Much lower frequency though from 20Hz to about 700kHz, the high frequency limit, I think, due to the sensor cable capacitance. Low frequency noise seems to be the limiting factor because of the high integrator gain needed at low frequencies.

[PartialDischarge]

Nice project. The OPA2822 is probably not the best choice since it has a low open loop gain and the low freq noise is not even quoted so probably it is very bad, as this is the are where the most gain, and the most output noise will come from.

All the relevant 3 opamps should be inside a shielding can. Also can't remember exactly why but the inverting configuration is preferred in rogowski integrators.

[Slh]

Very nice! Excellent write up as well.

Making the coils is a bit of a pain but also a bit relaxing (no good for profit obviously). After making a couple you soon realise why the commercial ones are so pricey. I like the o-ring. I've been wrapping my recent ones around a thick bit of copper for stiffness.

At some point I'm going try a PCB coil but I need a better integrator first... Yours looks good so looking forward to being inspired by it

It's cool that we've had three totally different rogowski coil projects in the last couple of months. 

[moffy]

Very nice! Excellent write up as well.

Making the coils is a bit of a pain but also a bit relaxing (no good for profit obviously). After making a couple you soon realise why the commercial ones are so pricey. I like the o-ring. I've been wrapping my recent ones around a thick bit of copper for stiffness.

At some point I'm going try a PCB coil but I need a better integrator first... Yours looks good so looking forward to being inspired by it

It's cool that we've had three totally different rogowski coil projects in the last couple of months. 

Yes that is quite the coincidence.

[PartialDischarge]

Also can't remember exactly why but the inverting configuration is preferred in rogowski integrators.

Now I remember why. The input low pass filters in your circuit have a very high cutoff frequency, which means that HF components will have a high peak-peak amplitude at the non-inverting opamp input (rog coil gives the derivative, easily +-10, +-15V). So your integrating opamp will easily saturate due to the limited rail voltages and won't be able to sense high current high frequency signals.

So in practice 2 things are done: Input passive filter filter (LR or RC) with a low cutoff frequency (maybe 1-2MHz) that does the 20dB integration for the high frequencies and "trims" by integration these large pk-pk signals. And a inverting integrator that does the low frequency integration (where gain is actually needed). The inverting integrator capacitor will pass-thru the high frequency already integrated signal from the input. So in practice the integrator opamp is a low GBW, high open loop gain, low noise opamp. It does not make sense to actively integrate in a range where you need -40..-80dB attenuation and moreover passive components can easily handle 10's or even 100s of volts coming from the rog coil for integration.

[Weston]

Now I remember why. The input low pass filters in your circuit have a very high cutoff frequency, which means that HF components will have a high peak-peak amplitude at the non-inverting opamp input (rog coil gives the derivative, easily +-10, +-15V). So your integrating opamp will easily saturate due to the limited rail voltages and won't be able to sense high current high frequency signals.

So in practice 2 things are done: Input passive filter filter (LR or RC) with a low cutoff frequency (maybe 1-2MHz) that does the 20dB integration for the high frequencies and "trims" by integration these large pk-pk signals. And a inverting integrator that does the low frequency integration (where gain is actually needed). The inverting integrator capacitor will pass-thru the high frequency already integrated signal from the input. So in practice the integrator opamp is a low GBW, high open loop gain, low noise opamp. It does not make sense to actively integrate in a range where you need -40..-80dB attenuation and moreover passive components can easily handle 10's or even 100s of volts coming from the rog coil for integration.

I have not seen what you are describing for a inverting integrator, it seems like the pass-through would be dependent on the output impedance of the op-amp. However, if done right you could get away with an op-amp slower than the signal bandwidth?

The non-integrator I am using is pretty similar to what you describe. The opamp in the non-inverting operation has the gain asymptote to unity as frequency increases so an input RC filter is needed to continue to roll off gain. That input RC filter has all the advantages of what you described, and it can also provide the correct terminating impedance to the rogowski coil. This paper does a better job of describing it: https://gmw.com/wp-content/uploads/2019/01/ias_2000_pem.pdf

For this design I gave up any hope of having a decent low frequency cutoff, the small diameter coil does not provide a very large signal. A LF -3dB frequency cutoff of 800Hz is enough to see most SMPS waveforms with minimal distortion and keeps noise low.

For a 800Hz to 25MHz range the OPA2822 was one of the best opamps I could find. With the inverting integrator the op-amp is at unity gain above the integrator crossover frequency, so to avoid excessive peaking I would need an opamp with a GBW of 40MHz+ or so. There are a few lower noise opamps with a GBW that high, but they all have significantly lower slew rate. This is somewhat exacerbated by the fact that I am using the second opamp in the package as the output buffer. 

The noise for the OPA2822 can be extracted from the plot in the datasheet. The 1/F voltage noise corner is ~1KHz, which is a bit high, but not much of an issue because of my high cutoff frequency. With the high cutoff frequency the integrator has a >20dB gain margin from the minimum OPA2822 open loop gain.

The most recent version of the probe I have on the bench uses some conductive shielding paint inside the enclosure to reduce RF interference, which is pretty effective. I don't know if I am going to get around to building it, I am trying to wrap up the project and get around to building projects instead of tools, but the PCB files on github have the integrator in a metal shield can.

There is an endless amount of cool stuff to try in optimizing these Rogowski coils. If I were to spend more time on the project it would be cool to work on a differential input to reduce E-field pickup, like moffy is working on, or to modify the integrator to use a low noise and low speed opamp for low frequency and some alternative signal path for high frequency.

[mercurial]

Hi, yes this seems to be ages. But just needed to know, how many turns did you wind for the rogowski coil. I didn't find any information here or the github page where you have linked your project to

https://github.com/westonb/rogowski-relief

[dpenev]

Hello

About six months back, I stumbled upon this rogowski-relief project, and it really caught my attention.
Thank you very much to the original author of making this project open!
I thought, wouldn't it be awesome if this thing came in a more compact, product-like form.
Plus, it seemed like a great opportunity to brush up on my analog skills, which have been gathering some rust lately.
So, I figured, why not dive into the joy of creating a product?

I had the idea to extend the frequency band down to 50 Hz, but after playing around with a few zero-drifted Opamps, I have found out that the original author did a really great job selecting the components.
In the end, I decided to stick to the original schematic design, with just a few tweaks in the power section to fit everything snugly into a small extruded aluminum enclosure.

I consider this as initial version.
Any suggestions on how to make this design even better are welcome!

The initial batch is ready! Please see some pictures bellow:









If anyone's interested, please check out www.rogprobe.com for more details.
Worldwide shipping available via DHL.

Best Regards
Dimitar