What's new
What's new

Mitsubishi Freqrol FR-SF on Single Phase?


Feb 19, 2019
Has anyone overridden the input phase detection on a Mitsubishi Freqrol FR-SF drive to run it from single phase input rather than 3 phase?

I have connected 2 of the small wires (red and blue) that go to the control side of the drive to the same phase on the input connector, and the other (white) to the other phase (looking at the schematic in the service manual to see which of the 2 it made sense to connect together)

With the main breaker off (that feeds the rectifiers / dc bus), if I power up the control side, I get only alarm 24 sometimes (main breaker trip - it's intentionally open), and sometimes alarm 24 and 26 (26 is input phase loss detected). If the main breaker is on, I get alarm 26 every time no matter what. I don't understand why sometimes with the main breaker off, alarm 26 goes away.

Is there a trick to disabling the phase detection? Can I just use an RC delay circuit to simulate 3 phase (with 2 stages of 120 degree phase delay) to the phase detection pins on the control board? I know the 3 phase for detection goes to CON 102 pins 7 8 and 9 on the SF-CA board, and it appears that it goes through a resistive divider before it goes to the ADC, but I haven't fully traced out the circuit yet. Clearly it seems to be doing something more intelligent than just checking if there's voltage between R-S S-T and R-T. I tried re-ordering the phase detection pins to the control board in every combination that made sense, and I get the phase error no matter what.

I see there's a jumper on the board to disable the main breaker open alarm to test the inverter/converter waveforms, but no jumper to disable phase detection, and no parameter (unless it's one of the parameters listed in the manual as "do not use - set to 0")

I already know about "it's not a good idea", and "derate the vfd vs the motor" etc etc (it's already a 22KW drive running an 11KW motor)
So I heard on another forum, "AFAIK you cannot run that version/model on 1ph you will get alarm 26 if any phase is missing." Looking at that person's post history, they seem to know what they are talking about, but "cannot" translates to "challenge accepted" to me!

I reverse engineered the phase detection circuit, simulated it in SPICE to see what outputs it would provide if 3 phase input was available, removed some parts to isolate the phase detector from the actual input power, and generated the correct outputs using a microcontroller!

phase detector.jpg

SPICE simulation:
The top 2 traces are the LEDs in the optocouplers from the phase detector circuit. Essentially it's just a positive and negative peak detector on each phase. There are 6 optocouplers total, one positive, one negative in each phase. Pretty trivial to replicate using a microcontroller. The middle trace is the analog signal that goes to the ADC for input voltage monitoring. At first I thought the 0.2V pseudo-sinusoidal signal would be important, but I thought I'd try it just with a constant 1.5V DC signal to the ADC before I bothered implementing an arbitrary waveform generator to generate that crazy waveform in the microcontroller.

Logic analyzer trace of pulse generated by microcontroller:
The pulses going to the negative peak detectors are active low, and the pulses going to the positive peak detectors are active high, simply because on the board, the positive peak optocouplers are wired common cathode, and the negative peak detectors are wired common anode, so I took advantage of that.

Pictures of the "phase fooler" connected:

Youtube video of it powering up with no errors or warnings:

I have to finish reconnecting some other wiring before I can power up the axis servo drive, and the NC controller, and I don't have the connector I need to provide the signals to run the spindle independently, and it's missing the "machine ready" signal, but a lot of progress has been made in 3 evenings and one weekend day (this is a hobby project after all, it only gets time outside of work)!
Don't seem to be getting any feedback here, but for completeness or anyone interested in the future, it actually works*!


*Ok, there is a caveat - the regenerative breaking trips a current limit if I stop the spindle rapidly above 1200RPM. I know this is because my fake phase detection pulses aren't actually synchronized to the line, and the drive connects the DC bus to the line to dissipate extra energy on motor deceleration. When the pulses aren't synched to the line, it's not hard to see why it would trip a current limit, and shut itself down!

I see a couple of solutions:
1) Align my fake phase pulse generation I'm feeding into the drive with the actual phase of the input AC. Experiment with unequal phase pulses (if the drive will allow that) to better align the fake "3 phase" 120 degree separated pulses with the real 180 phase difference between my actual split phase 240VAC. At first I didn't get why they had 6 pulse channels in the phase detection, but now I see it's so they know when to switch on which IGBTs in the "regenerative" side of the drive to dump power from the DC bus back to the line when the motor is breaking. Makes total sense!


2) Replace "the line" on the regenerative converter side of my drive with a breaking resistor (aka, stove element), like a Haas spindle controller (and probably lots of others) use. Then the timing of the phase detection pulses and the line voltage doesn't matter. Basically just end up with 6 IGBTs dumping the excess power to the resistor (heat), rather than trying to be fancy and shove it back to the line.

And the saga continues...
Impressive, but have you tried just ganging two of the three inputs to one line? Will it error if it sees one input is out of phase?

Also, be aware the induction motors require excitation to generate, and just shunting between the motor leads and a resistor with an IGBT will not do much.