2
\$\begingroup\$

I need to isolate an input from my processor. What is a reasonable isolation voltage to look for optocoupler/digital isolator? I've pretty much decided to go with an RF isolator at this point because they use much less power and are easier to implement that traditional optocouplers. Also - from what I've read, I need a DC/DC converter with isolation to make sure there is an isolated power/ground for the isolated side of the isolator. Does this sound right?

\$\endgroup\$
  • \$\begingroup\$ It does sound right. However, why do you want to have galvanic isolation in the first place? There is no universally reasonable voltage rating for galvanic isolation. The rating is determined by the application. Either some standard dictates the isolation rating (e.g. 4 kV rms for patient isolation in medical devices). Or, the high voltage threat dictates the isolation rating. \$\endgroup\$ – Nick Alexeev Nov 26 '13 at 7:02
  • \$\begingroup\$ Basically there are two cases I'm worried about - 1) user putting in wrong voltage by accident (my thought is this is the main concern) and 2) power spikes caused by lighting hitting connected equipment. \$\endgroup\$ – Pugz Nov 26 '13 at 7:12
  • \$\begingroup\$ I know most human ESD models are 10kV to 15kV for ESD resistance testing - I'm not really worried about this. I just want isolation from noise and over-voltage, possibly surges on external equipment. \$\endgroup\$ – Pugz Nov 26 '13 at 7:14
  • \$\begingroup\$ As Nick says - in some cases a standard sets the required minimum. Abset that, 1000V is "low" but will work in practice in most cases. 2000V+ is usually adequate. Note that in extreme cases it is not JUST isolation per se that counts. eg coupled dV/dT can cause triggering of devices if rise times are fast enough. This is not usually an issue. \$\endgroup\$ – Russell McMahon Nov 26 '13 at 7:19
  • \$\begingroup\$ On the same note - for isolating the power supply with a DC/DC converter - is it sufficient for the converter to match or exceed the isolator value or is there some math to do? \$\endgroup\$ – Pugz Nov 26 '13 at 7:25
2
\$\begingroup\$

It does sound right.

There is no universally reasonable voltage rating for galvanic isolation. The rating is determined by the application. Either some standard dictates the isolation rating (e.g. 4 kV rms for patient isolation in medical devices). Or, the high voltage threat dictates the isolation rating.

Threat 1. User applying wrong voltage. Take the highest wrong voltage that the user may possibly apply (e.g. 380 V rms) and add 1 kV rms.

Threat 2. Lightning strike. The trick is to use gas discharge tubes (GDT), which would shunt the lightning strike into the earth ground. GDT take care of the bulk of the energy, of the lightning strike. Add some TVS Zeners in parallel with the GDT.

\$\endgroup\$
  • \$\begingroup\$ So basically if it seems like it'd be uncommon for a lightning strike to affect this device then the lightning solutions comes at unjustifiable expense. \$\endgroup\$ – Pugz Nov 26 '13 at 7:34
  • \$\begingroup\$ @Pugz It's your decision to make. Hazard-vulnerability-impact assessment is yours to make too. You could check what the UL standards say about lightning protection for your type of device. \$\endgroup\$ – Nick Alexeev Nov 26 '13 at 7:40
3
\$\begingroup\$

If you want isolation from the mains you'll need 1000 to 2000 V isolation voltage. Most optocouplers will provide this. Just keep an eye on the pinning: a standard DIL package with standard pads won't give you the required 6mm clearance. You'll find packages with the rows of pins further apart:

enter image description here

\$\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.