Digital phase converters for running 3-phase motors

[Guineafowl]

TL;DR Summary

DPCs are sold primarily for running 3-phase motors on a 1-phase supply. How do they generate the output, and what does it look like?

To be clear, I’m not asking about variable frequency drives (VFDs, inverter drives). These rectify the 1-phase onto a DC bus, then use PWM to generate three phases 120deg apart. They can vary the output frequency.

DPCs, also (I think) called digital phase shifters, apparently feed the 1-phase straight through, and generate a third leg from a DC bus electronically. Their output frequency is fixed to whatever the input is.

I can’t see how they approximate real 3-phase - the mains input peaks would be 180deg apart, so interspersing an artificial leg would give a double waveform, with two +ve peaks then two -ve peaks, all 90deg apart. Perhaps this is ‘good enough’ for running the motor?

They seem to be a solid-state version of a rotary phase converter; thread here: https://www.physicsforums.com/threads/trying-to-understand-rotary-phase-converter-output.836577/

In short, what does the output waveform of these devices look like?

 

[.Scott]

I just did a quick Google image search on "220v single phase to 3 phase digital converter electrical schematic". Here is the URL one of the first results.

Based on this and some other search results, most of them (perhaps all) follow the same basic steps: Convert to DC then modulate each of the three phases digitally.

 

[Baluncore]

Convert to DC then modulate each of the three phases digitally.

That is the basis of the VFD solution.
Alternately, by using a phase shift capacitor to drive the third phase, the circulating field that sets the direction of the motor is decided. The motor is not receiving true 3PH and must be downgraded.

 

[Guineafowl]

I just did a quick Google image search on "220v single phase to 3 phase digital converter electrical schematic". Here is the URL one of the first results.
View attachment 364472
Based on this and some other search results, most of them (perhaps all) follow the same basic steps: Convert to DC then modulate each of the three phases digitally.

No, that’s a VFD.

 

[.Scott]

I have examine more schematics. Apparently, power conversion dealers feel comfortable calling anything short of coupled motors as "digital".

When I compare the diagram presented in the OP with numerous converter electrical schematics, it seems very likely that the "motor" in the OP's diagram presented is an "Idle Motor" and is part of the power conversion circuit. When I first looked at it, I interpreted it as an example load.

If so, then that powered oscillator is only providing enough off-cycle phase to keep the motor spinning in the right direction. The motor itself is providing power to the third phase most of the time.

 

[Guineafowl]

I doubt that your schematic would be called a "DPC" since there is nothing digital about it.

As far as I can find out, the ‘digital’ refers to the PWM switching of the third leg.

Any two legs of a 3-phase power source will appear to be 180 degrees apart.

Say we were to scope the star/wye secondary of a 3-phase transformer, common to star point, one probe on phase A, one on phase B, would the waveforms not be 120deg apart?

 

[.Scott]

Say we were to scope the star/wye secondary of a 3-phase transformer, common to star point, one probe on phase A, one on phase B, would the waveforms not be 120deg apart?

You would only see the 120 degree separation if you had some place to attach to ground. If you attach probe ground to one and the probe to the other, you just see a sine wave.

 

[Averagesupernova]

I can’t see how they approximate real 3-phase - the mains input peaks would be 180deg apart,

This is technically not correct in the context of what is happening. This same concern has been brought up previously with rotary phase convertors here on pf.
-
Read the following link. It should clear up your misconception.
-

J

Thread 'Trying to understand rotary phase converter output....'

Oct 8, 2015

I am looking to build my own rotary phase converter for utilizing 3ph devices however I am having a difficult time wrapping my head around the aspect of producing 120* out of 180* power. PracticalMachinist forums have a plethora of threads on this topic but after reading many of them I find that a third of the people are left clueless, another third make convincing arguments but lack any visual demonstration (waveforms) and the final third simply resolve that it just works and because it just works it must be producing clean, 120* separated, AC signals.

The windings are physically...

• JTraik
• Converter Output Phase Rotary
• Replies: 11
• Forum: Electrical Engineering

 

[.Scott]

Alternately, by using a phase shift capacitor to drive the third phase, the circulating field that sets the direction of the motor is decided. The motor is not receiving true 3PH and must be downgraded.

Even though the OP's diagram does not show it, I think that (your "alternately") is basically what the it is doing.
The OP's diagram does not show any direct connection between "L" and his "DC Powered Oscillator". But without such a connection, that "oscillator" would have of being in (or staying in) a useful phase.
I would guess that that "oscillator" is actually a power amplifier with its input coupled through capacitors to L and N.

...

Actually, there is a way that the OP's circuit could work without a direct connect to L. The line from the oscillator to the motor could provide the oscillator with a reference to the phase on "L". It would be a tricky analogue circuit, but I think it could be made to work.
But, I don't think that's it. Such a circuit would become very confused by the wrong kind of load.

 

[Baluncore]

A 3PH supply is just three single phase lines relative to neutral, the phase differences are 120°. The phase voltage, is the voltage between any pair of those lines. Each phase voltage, measured between any two lines, is the sum of two phasors, and that is a sinewave.

When a single phase A and N is connected to two lines of a 3PH motor, the active phasor rotates about the line that is connected to neutral. The third line needs to be driven so that the two remaining phases are at +120° and -120° relative to the A and N lines. Once the motor is turning, the coupling between the 3PH windings may provide that phase requirement.

 

[Guineafowl]

Thanks, the phasor explanation makes sense.

The terminology is confusing, but here is an example device:

https://www.ebay.co.uk/itm/283812096505

It looks like the driven motor is used somewhat as an idler, as suggested above, given the requirement to match the rating so closely, talk of swapping phases around till it works, and balanced currents/voltages only above 50% load. Also, the third phase being only present with a motor connected.

More confusing is the description, “digital type phase converter” alongside “doesn’t make any high frequency”. If the third leg is PWM, you’d expect some, as you get from VFDs, which require shielded cable on the output.

I’m fairly sure these things are distinct from:
- Static phase converters, purely capacitor-based.
- Rotary “” , with their own idler motor.
- VFDs

 

[DaveE]

More confusing is the description, “digital type phase converter” alongside “doesn’t make any high frequency”. If the third leg is PWM, you’d expect some, as you get from VFDs, which require shielded cable on the output.

The HF switching noise can be easily filtered, if they want to. In many motor drives users don't care enough to pay for that.

 

[Baluncore]

This is one way to convert a single phase, active with neutral, into three-phase lines with 120° separation.

The neutral becomes the Yellow line, Y, while active, (source V1), becomes the Red line, R. The DPC then only needs to generate (source V2), the fake Blue line, B, such that it has a 120° phase separation from both R and Y.

Since the R and B lines are both referenced to Y, they are being measured in different directions around the phasor diagram, so one is reversed in time, or inverted = 180°. For that reason, we need a phase shift of 180° – 120° = 60°, not the obvious 120° shift when faking the B line.

The R, B and Y lines are referenced to the common at the centre of the motor phasor diagram, not to the neutral input. The motor is represented by the three resistors in star connection. The actual phase voltages across the motor, in delta connection, will be R-B, B-Y, and Y-R.

For this simulation, the line frequency has been adjusted so 1ms = 10°. Phase angle can therefore be read from the time axis. The line voltage input is assumed to be 230 Vrms, so the peak amplitude relative to neutral is almost 400 V. Spice assumes peak voltage is specified, hence the root two in the amplitude of voltage sources V1 and V2.

 

[Guineafowl]

This is one way to convert a single phase, active with neutral, into three-phase lines with 120° separation.
View attachment 364520
The neutral becomes the Yellow line, Y, while active, (source V1), becomes the Red line, R. The DPC then only needs to generate (source V2), the fake Blue line, B, such that it has a 120° phase separation from both R and Y.

Since the R and B lines are both referenced to Y, they are being measured in different directions around the phasor diagram, so one is reversed in time, or inverted = 180°. For that reason, we need a phase shift of 180° – 120° = 60°, not the obvious 120° shift when faking the B line.

The R, B and Y lines are referenced to the common at the centre of the motor phasor diagram, not to the neutral input. The motor is represented by the three resistors in star connection. The actual phase voltages across the motor, in delta connection, will be R-B, B-Y, and Y-R.

View attachment 364521

For this simulation, the line frequency has been adjusted so 1ms = 10°. Phase angle can therefore be read from the time axis. The line voltage input is assumed to be 230 Vrms, so the peak amplitude relative to neutral is almost 400 V. Spice assumes peak voltage is specified, hence the root two in the amplitude of voltage sources V1 and V2.

That’s very clever. I’m trying to understand the circuit in terms of two sinusoidal voltage sources, V_RY and V_BY. Both are of magnitude 340V, the mains peak in a 240V country:

$V_{RY} = 340V <0°$
$V_{BY} = 340V <-120°$ ie, reflected from 60°

Should the final result should look something like this, with the resultant voltage $V_{RB} = 340V <120°$, and the common/star point ‘com’ in the centre of a delta?:

I’m not entirely convinced I have the $V_R, V_Y, V_B$ right.

 

[Baluncore]

I’m not entirely convinced I have the right.

Your diagram is correct, but the voltages are confusing because you do not indicate if a voltage is Vrms or Vpk, so it is hard to reconcile with the plots of voltage.

When I started to model the converter, I decided the line voltage was 230Vrms, so I used that as the 1PH input, that was correct.

However, the single phase line voltage is used immediately to make the 3PH phase voltage, R-Y, which must therefore be 230Vrms.

The effective 3PH line voltages are therefore 230 / √3 = 132.8Vrms, which will then be 132.8 * √2 = 187.8Vpk, and that is what the model shows.

That explains why the 3PH line voltages do NOT peak at 230 * √2 = 325Vpk.

 

[Guineafowl]

Your diagram is correct, but the voltages are confusing because you do not indicate if a voltage is Vrms or Vpk, so it is hard to reconcile with the plots of voltage.

When I started to model the converter, I decided the line voltage was 230Vrms, so I used that as the 1PH input, that was correct.

However, the single phase line voltage is used immediately to make the 3PH phase voltage, R-Y, which must therefore be 230Vrms.

The effective 3PH line voltages are therefore 230 / √3 = 132.8Vrms, which will then be 132.8 * √2 = 187.8Vpk, and that is what the model shows.

That explains why the 3PH line voltages do NOT peak at 230 * √2 = 325Vpk.

They’re all peak voltages, which seems to work:

Using geometry, the distance D of each vertex of the delta to its centre, ie $V_R, V_Y, V_B$ is calculated from the side length A, 340V: $D=\frac {\sqrt3 A}{3}=196.3V$ as shown.

Also, 340V peak across two motor phases gives a peak phase voltage $V_{pk, phase}=\frac {340}{\sqrt3}=196.3V$

Now, the more common usage of one of these DPCs would be with a 415V star/240V delta motor configured in delta. Would that affect the reflection of the phasor $V_{BY}$? What would the simulation and phasor diagram look like?

 

[Baluncore]

Would that affect the reflection of the phasor ? What would the simulation and phasor diagram look like?

The phasor diagram does not change with voltage, the voltages are simply scaled. An autotransformer could be used to boost the input voltage.

To minimise starting current, a motor would be started star-connected, then once it was spinning, it would be switched to run in delta-connection. The supply would not be changed, only the internal motor connections would change from star to delta.

My simulation showed the motor as star connected resistors, because I needed the midpoint voltage to plot the 3PH line voltages. I could have added a delta of resistors to the simulation, but it would have been cluttered. Resistors connected between low impedance voltage sources do not change those voltages.

240V 1PH and 415V 3PH are now, in theory, replaced by the international standard of 230V 1PH and 400V 3PH. That is why I use those numbers. The actual voltage being delivered seems to remain the same, it is all within the wide permitted tolerance.

 

[Guineafowl]

The phasor diagram does not change with voltage, the voltages are simply scaled. An autotransformer could be used to boost the input voltage.

To minimise starting current, a motor would be started star-connected, then once it was spinning, it would be switched to run in delta-connection. The supply would not be changed, only the internal motor connections would change from star to delta.

My simulation showed the motor as star connected resistors, because I needed the midpoint voltage to plot the 3PH line voltages. I could have added a delta of resistors to the simulation, but it would have been cluttered. Resistors connected between low impedance voltage sources do not change those voltages.

240V 1PH and 415V 3PH are now, in theory, replaced by the international standard of 230V 1PH and 400V 3PH. That is why I use those numbers. The actual voltage being delivered seems to remain the same, it is all within the wide permitted tolerance.

Yes, same here in the UK - real 415V/240V is deemed within tolerance of the nominal 400V/230V.

Star-delta starting, over here, is usually for motors above around 5kW. What we typically deal with in the home workshop are motors designed to take 3ph 400V in star, which can be reconfigured in delta to take 3ph 230V. This is provided via 1ph 230V by ordinary AT1 VFDs and these new DPCs. In some older motors, the star point is not available at the connection box, but must be dug out of the windings and brought out.

So, the typical application for a DPC would be a motor configured in delta. Would it look something like this?:

 

[Baluncore]

So, the typical application for a DPC would be a motor configured in delta. Would it look something like this?:

Yes.

 

[Averagesupernova]

The bottom line is that if you have anything besides single phase you can make any number of phases you like through vector addition. If you take single phase and synthesize another signal that is anything but the same vector as your standard single phase (or the 180° opposite) you have the ingredients to do it.
-
Now this is more easily said than done. Technically I can synthesize a signal that is one degree different than the single phase source and that is good enough. But if you start doing the math it will soon be realized that there is a much more practical choice concerning the phase difference between the single phase source and the synthesized one.

 

[Averagesupernova]

Yes, same here in the UK - real 415V/240V is deemed within tolerance of the nominal 400V/230V.

Star-delta starting, over here, is usually for motors above around 5kW. What we typically deal with in the home workshop are motors designed to take 3ph 400V in star, which can be reconfigured in delta to take 3ph 230V. This is provided via 1ph 230V by ordinary AT1 VFDs and these new DPCs. In some older motors, the star point is not available at the connection box, but must be dug out of the windings and brought out.

So, the typical application for a DPC would be a motor configured in delta. Would it look something like this?:
View attachment 364532

This is what is known as open delta. Look around in small industrial neighborhoods in North America. You will see a large transformer with a center tapped connection which is neutral. Then you will see a smaller transformer with one lead that is connected to one of the leads from the larger transformer. The remaining lead on the small transformer is what is known as the wild leg. V1 and V2 in the above drawing are the transformers. The triangle of resistors are the three phase loads. Typical connection where the majority of loads are single phase and there are a few smaller 3 phase loads.

 

[Guineafowl]

This is what is known as open delta. Look around in small industrial neighborhoods in North America. You will see a large transformer with a center tapped connection which is neutral. Then you will see a smaller transformer with one lead that is connected to one of the leads from the larger transformer. The remaining lead on the small transformer is what is known as the wild leg. V1 and V2 in the above drawing are the transformers. The triangle of resistors are the three phase loads. Typical connection where the majority of loads are single phase and there are a few smaller 3 phase loads.

A temptation to discuss the seemingly complicated US electrical distribution system… But no, I won’t ;)

I assume there would be no 60° phase shift between the two transformers as with our DPC. Or perhaps ‘RY’ and ‘BY’ run in quadrature? I could see that working geometrically.

 

[Averagesupernova]

A temptation to discuss the seemingly complicated US electrical distribution system… But no, I won’t ;)

I assume there would be no 60° phase shift between the two transformers as with our DPC. Or perhaps ‘RY’ and ‘BY’ run in quadrature? I could see that working geometrically.

Feel free to discuss the North American system. May want to start a new thread.
-
The phase shift with open delta is no different than a closed delta. It's just a matter of leaving one transformer out. It is an example of one method to obtain 3 phases by using only two.

 

[Baluncore]

https://en.wikipedia.org/wiki/Scott-T_transformer

 

[sophiecentaur]

https://en.wikipedia.org/wiki/Scott-T_transformer

I was on an engineering induction course at the BBC Engineering Training Dept in 1967. I remember (fuzzily) being shown a demo of a Scott(?) transformer for producing three phase from single phase. Transformers are hard things but could the T arrangement in the wiki article be worked the other way round?

 

[Averagesupernova]

I was on an engineering induction course at the BBC Engineering Training Dept in 1967. I remember (fuzzily) being shown a demo of a Scott(?) transformer for producing three phase from single phase. Transformers are hard things but could the T arrangement in the wiki article be worked the other way round?

You cannot get 3 phase from single with just transformers. Scott Ts were common to go from two-phase 90° to the more commonly known 3 phase 120°.

 

[DaveE]

I was on an engineering induction course at the BBC Engineering Training Dept in 1967. I remember (fuzzily) being shown a demo of a Scott(?) transformer for producing three phase from single phase. Transformers are hard things but could the T arrangement in the wiki article be worked the other way round?

Since these transformers are linear passive networks, yes, they will work forwards and backwards. 3 phase 120 to 2 phase 90^o would work too. Although it wouldn't be common. In rural (low power) regions they will sometimes distribute 2 phase 90^o to save on copper & pole costs. Then they use these to recreate "good" 3 phase power if customers need it.

I had an old text book that had a bunch of these configurations for this and making 6 phase, etc. Basically it's all just vector addition in a 2D plane. If you have a source with at least two independent phasor vectors, you can scale and add them with transformer windings however you want.

 

[sophiecentaur]

3 phase 120 to 2 phase 90o would work too

I wonder just what was in that box, then. We were all grads on the course but with a wide range of knowledge. The training establishment provided a lot of (young) technician level intake and their courses were very 'basic'. So we may well have been a bit 'mislead' about what we were shown. Three Phase was something my Dad was involved with but I only came across it much later so I couldn't have spotted any 'fudge'. AFAIR, we saw three waves on a scope screen.

It amused me that an AI search suggested that Scott may have been the name of the tutor!