This Analog Devices technical article describes a motor controller that uses their digital isolators to provide an isolation barrier between the 3 phase bridge/DC bus and the microcontroller. It is mostly about how digital isolators are better than optocouplers, but they do mention this:

The motor power electronics are floating at high voltage potentials while the ADSP-CM40x processor is referenced to earth ground, thus the need for isolation.

I have designed similar circuits, but I always tie the negative DC bus voltage to the same ground as the microcontroller. In this way I can measure the DC bus directly, and my pwm outputs directly drive the FET drivers.

Edit: In other words, I have earth referenced the micro and the DC bus minus. Now the motor works from earth gnd to whatever DC+ is.

Can anyone discuss the advantages of providing this isolation? I realize you would protect your microcontroller if you had some failures in your power stage, but you'd be replacing so many components at that point it isn't likely to matter. Someone could probably also make an argument for not putting all the motor noise into your microcontroller ground, I have seen this problem trying to communicate over usb while driving a motor.

What else you got?

  • \$\begingroup\$ The isolation is for safety reasons. You can see their controller has an LCD and a bunch of USB ports that present a shock hazard to the user if the 300V motor bus is not galvanically isolated. Whether you need this or not depends on the application architecture. \$\endgroup\$ – Jon Jun 3 '15 at 15:57
  • \$\begingroup\$ You can't reference a signal to ground (which is what a micro will do) without severe damage/explosions if the incoming signal is sitting at hundreds of volts in comparison with ground - even though for the floating circuitry the signal may only be a few volts with reference to the floating 'ground' reference. This is the case where there may be digital circuitry sitting at both potentials. If it's just a power driver stage, then isolation (with optocouplers for FET driving etc) is just for safety \$\endgroup\$ – KyranF Jun 3 '15 at 15:57
  • \$\begingroup\$ you can if you have a resistor divider...Possibly, I misunderstood your comment. The isolation would make it difficult to measure the dc bus. With grounds tied together it is easier. \$\endgroup\$ – Matt Jun 3 '15 at 15:58
  • \$\begingroup\$ @MattAnderson I think the grounds are still tied together, but interfacing a 500V signal with a 5V signal needs isolation. Often it's done by photocouplers/optoisolators. But that's only if there is any data to transmit, if it's a one-way drive signal it can be stepped up and isolated using various means (like transformer, or optoisolated relay switches etc) \$\endgroup\$ – KyranF Jun 3 '15 at 17:03

In a modern AC motor drive such as depicted in that document, the input power supply passes through a full-wave bridge rectifier and is used to charge a high voltage DC capacitor bank from which the drive electronics then synthesize variable frequency and voltage 3-phase waveforms. Higher power industrial versions may rectify a 3-phase input, but the idea is the same.

Without an isolation transformer, the lower side of that capacitor bank will not be at ground potential - rather, the capacitor voltage will normally be approximately symmetric with respect to ground. Needless to say this requires care to work on!

The size, weight, loss, and cost of an isolation transformer increases with power requirement - but the cost of a mechanism for isolating the control circuitry is basically fixed.

  • \$\begingroup\$ Ok, so this is simplified in my case where the DC bus is supplied as a separate power supply or battery bank. This is what I meant by earth referencing the micro and the dc bus. Does the same need for isolation even exist then? \$\endgroup\$ – Matt Jun 3 '15 at 17:12

Usually the microcontroller need to communicate with the outside world (beyond the VFD output stage), so you need RS-485, USB, analog input 0-10V, LED or LCD display, membrane keyboard and other things connected to the microcontroller. It's also useful for EMC reasons (susceptibility in particular) to have the micro at ground potential (at least for AC). Most of these things are easier of the micro is ground referenced and not connected directly to the mains. Ethernet comms are transformer isolated, but they are not provably safe to use for mains isolation.

If the HMI of your VFD consists only of a knob labeled 0 to 100% RPM then you might well decide it's worth saving a few dollars on the BOM and live without the isolation (but use a pot that is rated for mains voltage isolation!).

If you're making more of a high-end product, then isolation would be called for.

  • \$\begingroup\$ Are you saying the worry is that during failure the high voltage will be present on the USB/RS-485 data lines. In other words some failure that places any/all of the microcontroller pins at high voltage levels. \$\endgroup\$ – Matt Jun 3 '15 at 16:33
  • \$\begingroup\$ You have to assume that a direct connection to the mains can be at mains voltage even if it's to the neutral. So, yes, that means that the RS-485/USB that should be at GND +/- a few volts at most (less for USB) could be at full mains voltage, which would blow the cr*p out of whatever it was connected to as well as being hazardous to people. I've seen every single chip in a computer destroyed by such an occurrence. \$\endgroup\$ – Spehro Pefhany Jun 3 '15 at 17:00
  • \$\begingroup\$ I am probably confusing the issue because I am considering the DC bus voltage as isolated from the mains. So as if the high voltage DC supply was the input and whether we should isolate from that. Although, if I understand you correctly the same logic applies. \$\endgroup\$ – Matt Jun 3 '15 at 17:07

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.