1
\$\begingroup\$

I am trying to isolate control signals that are used to remotely operate a variable frequency drive. One is a 0-5V analog voltage signal that give the VFD its set point. I understand that optocouplers can be used to transmit digital signals as well as current across the optocoupler which may in some applications be leveraged to isolate an analog voltage signal if you set up the circuit correctly to convert and pass current rather than voltage. I was hoping for a detailed explanation for someone who is a novice in circuitry that could explain how to set up this circuit ideally with non specialized optocouplers, or if specialized optocouplers are required, how to evaluate a specific optocoupler for its appropriateness. (In my case specifically, there are 2 types of optocouplers available at my uni so would like to know how to select one rather than a specific recommendation).

Also, I know there may be other, simpler ways to achieve my purpose of signal isolation for my VFD control, but I am interested regardless in the answer to this question using optocouplers

\$\endgroup\$
10
  • 4
    \$\begingroup\$ Isolating analog signals is a pain so know that there aren't really simple neat and tidy solutions. If you're looking for one of those you won't find one. Even pre-packaged analog isolators are kind of ungainly not to mention rare and expensive. There are linear optos and what sets them apart is they have two matched photodiodes so you can use the other for feedback to achieve linearization. I tend towards isolating the ADC and using digital isolation to transmit the data back but that isn't applicable in your case. Perhaps reconsider voltage for long-distance analog signals? Use current. \$\endgroup\$
    – DKNguyen
    Commented May 10, 2022 at 17:31
  • \$\begingroup\$ Thank for your comment- Like I said, I am interested in how this might be done using optocouplers independently of the problem I am trying to solve in my application, so I hope someone can post a good answer. That said I am curious if you can elaborate on your last point- that it would be simpler to just convert the signal to digital then isolate the digital signal via optocouplers, then I would convert back to analog? \$\endgroup\$
    – user74671
    Commented May 10, 2022 at 17:36
  • \$\begingroup\$ I didn't fully read your post properly. WHen I said easier to isolate the ADC that was just because when I normally run into the need to sample signals that need to be isolated for safety rather than long-distance comms. In your case you might be able to get away with a PWM DAC (just an RC filter+buffer) and transmit an PWM signal to it through digital optos. Must you transmit in analog? If you have to both transmit and receive in analog then you probably would just stay analog though you could double-convert if you only had digital optos. \$\endgroup\$
    – DKNguyen
    Commented May 10, 2022 at 17:39
  • \$\begingroup\$ Yes the VFD was bought off the shelf (its a WEG). The VFD can take a voltage or current signal though it must be analog for the speed control that I am controlling remotely. The other end that is sending the signal is just a NI DAQ that only is capable of sending voltage outputs. It can also transmit in digital but like I mentioned I think I'd need to convert to analog before passing the signal to the Vfd in that case \$\endgroup\$
    – user74671
    Commented May 10, 2022 at 17:47
  • \$\begingroup\$ You really don't want to be dealing with linear opto circuits as a newbie (I haven't really been able to understand them either). In your case, I recommend placing an RC low pass filter with a buffer and opto at the VFD and transmitting a PWM signal to it. The buffer could be an op-amp buffer but a comparator would work in this case too. You might want to have a tightly regulated voltage supply/reference to power all this as well since that will directly influence you analog voltage accuracy \$\endgroup\$
    – DKNguyen
    Commented May 10, 2022 at 17:48

2 Answers 2

2
\$\begingroup\$

Isolating analog signals is a pain so know that there aren't really simple neat and tidy solutions. If you're looking for one of those you won't find one. Even pre-packaged analog isolators are kind of ungainly not to mention rare and expensive. There are linear optos and what sets them apart is they have two matched photodiodes so you can use the other for feedback to achieve linearization.

You really don't want to be dealing with linear opto circuits as a newbie (I haven't really been able to understand them either). But you don't have much choice if you must transmit and receive in analog voltage.

If you need isolation for signal integrity, and not safety reasons, and need to both transmit and receive in analog but it does not need to be voltage, consider using current signals. You can transmit current-to-current or current-to-voltage by using a "current-to-voltage converter" at the VFD (which is just a resistor of the appropriate value). This won't provide isolation for safety though since there is no isolation.

If you can transmit in digital and receive in analog then I recommend placing a low-pass RC filter and buffer at the VFD to serve as a simple PWM DAC and transmitting a PWM signal to it through an opto. You probably also want a tightly regulated voltage supply or voltage reference powering this since that will directly influence the accuracy of the analog output.

enter image description here

Note that for maximum effect the power supplies also need to be isolated otherwise it just provides a sneak path for whatever it is you're trying to isolate to travel between systems. Therefore the leftmost GND in the above schematic is a separate ground from the others. There just weren't more ground symbols to choose from and Vopto and Vref should also be isolated.

Also note that this displayed configuration inverts the duty cycle. If you swap the the opto output with R4 it no longer inverts.

You will need to choose the RC time constant and PWM frequency based on what bandwidth you require in your analog signal.

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

Optoisolators do a great job of isolating kV, DC and AC. The Photo Diodes (PD) are very consistent and linear but LEDs and Phototransistors are not very linear and have a wide tolerance. The non-linearity can be corrected with error correction using negative feedback with dual detectors. There are many designs using this chip and some worse than others, but this is why they use only one LED with two PD's The LED will have an emission of X mW/mA and the the PD's have a reverse bias sensitivity of Y mA/mW with some current loss ratio loss k. So the transfer function is Av = kXY [mW/mA*mA/mW] where the units cancel out with current control. So the design consists of a voltage controlled current on the input with sensor feedback and the detector uses a pull-up resistor to convert current back to voltage and k includes the resistor ratios Rout/Rin and the voltage swing ratios Vo/Vin and the XY optical loss which is matched for both detectors.

Here is the App Note for the IL300.

To control any motor you would like to have current sensing and RPM sensing in a VFD. But if you know the load inertia range and motor torque curve and only wish to limit the surge current relative to rated current rather than 5 to 12 x rated current then you limit the rate of change of acceleration by regulating the motor current.

Otherwise you program a controlled start/stop ramp and measure the results then correct.

You must be aware of all the variables for stiction, inertia, temperature and high step torque effects to avoid problems. This ultimately determines how you want to control RPM and max current. You cannot control the motor's peak/no-load power ratio which can easily be 50:1 using voltage-frequency control unless you compromise the time to get to target speed with a slow ramp.

Start with all your expectations then decide what SNR and linearity you need. You should be able to achieve >40 dB SNR and < 1% linearity error with 0.5% resistors. But first you must decide on yours specs to choose the complexity of design to meet or exceed this. This includes gain error, offset error and linearity errors and I/O Range and signal BW limit.

Digi-key also has some simple examples for multiple vendors.

\$\endgroup\$
1
  • \$\begingroup\$ thanks Tony. This is not really addressing my question which is about isolating control signals that are being sent to the VFD. The VFD then controls the motor with its own control algorithm. As such I am only sending a analog signal to change the set point on the VFD and not worrying about the stiction, current sensing, etc, which is handled on the VFD itself and tuned well by internal parameters that I set. The first part answers the question but if you could elaborate more on its implementation that may be more helpful \$\endgroup\$
    – user74671
    Commented May 11, 2022 at 16:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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