2
\$\begingroup\$

I am working on a high-side current sensor, where the high-side is around 3kV. I have come to the conclusion, with some help from you kind folks, that it is best that I float the majority of my circuit up at that 3kV level. However, I need to get the signal out to ground level, so that it can be input to a microcontroller. I am having a hard time finding any isolation devices who isolate at such a high working voltage, rather than transient voltage. Does anyone have any suggestions of such a device? I guess I also need an isolated DC/DC converter, that will let me power the 3kV floated circuit from a power supply at the ground level...I also am struggling to find such a thing.

Any help or suggestions would be greatly appreciated.

Thank you so much.

\$\endgroup\$
3
\$\begingroup\$

Specialty optoisolators that can handle 50KV working voltage are available off-the-shelf, if you don't mind doing a little more work to properly isolate the I2C signal:
enter image description here
Yeah, that's an optoisolator. That's what you need for 50 KV isolation.

Slightly lower isolation rating parts (10 KV) are also available for considerably less money:
enter image description here

Note that these are all discrete optoisolators. If you can switch to a SPI ADC, it would make the isolator considerably easier to design, though you can isolate I2C with discrete isolators: enter image description here

Here is a whitepaper about I2C isolation topologies.

Depending on the I2C data-rates you want to achieve, it may be fairly involved.


Further follow-up:

Here is an excellent white-paper about safety concerns in power-supply designs.

\$\endgroup\$
2
\$\begingroup\$

Medical DC-DC converters, some, are rated at 4kV, but I suspect that's not good for continuous duty. Can you power it with batteries? For really high voltage isolation it's possible to use a photonic supply (laser diode and fiber), but they are not cheap, and the amount of power you can get is fairly limited.

For the signal, I suggest a fiber optic link, which is good for almost unlimited voltage (just keep the separation sufficient). You may be able to find a commercial I2C to fiber adapter.

enter image description here

\$\endgroup\$
  • \$\begingroup\$ Great, thank you. I found a few nice I2C isolation chips, but their working voltage is low. I think a fiber optic link is probably my best bet. Thanks again! \$\endgroup\$ – John Jan 24 '14 at 17:32
2
\$\begingroup\$

An alternative to isolating the actual I2C in some cases is to isolate a whole I2C controller at the HV side. This is ideal if the actual job looks like: I want to read/write come I2C devices which are at high voltage from my pc / ATE / microprocessor.

http://www.i2cchip.com/pdfs/I2C2PC_FibreOpticAdaptor.pdf http://www.i2cchip.com/pdfs/I2C2PC_Datasheet.pdf

We have made a fibre version for users needing extremely high, visibly sound isolation safety, and also for very low RF noise coupling test systems.

A standard galvanic isolated unit is off the shelf for lower voltages.

When you are using isolator chips (AD, Silabs etc) that highlight quite decent voltages at the top of the datasheet: - The continuous withstand voltage is much lower, and sometimes well hidden in the datasheet, or even completely unmentioned. i.e. a 1kV isolator is only really rated for 1kV transients.

  • If the highvoltage is something that flys up and down (ie you are isolating the swinging part of a convertor) the dV/dT currents through the few pF of C can be enormous, and completely destructive. This is where fibre really shines.
\$\endgroup\$
1
\$\begingroup\$

We have done a few projects where we needed high-voltage isolation. We have successfully used 1mm plastic optical fiber with transmitters and receivers from both HP and Industrial Fiber Optics.

This provides a robust and extremely reliable communication path in several fairly nasty Industrial environments.

The great advantage to this technique is that the communication link also provides the required galvanic isolation. The link is good for several hundred feet.

We were able to obtain the power required to operate the link from the devices sitting at that high voltage.

\$\endgroup\$
0
\$\begingroup\$

A very similar question came up last week but i can't find it. I suggested Hall-effect effect sensors for measuring the current. These don't have to make contact with the busbars and therefore are easily made to have large isolation voltages. One that I located had a rating of 5kV and sounds suitable.

OK, just found it here and the hall effect sensor was a CLSM-25M.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.