Running older induction motors on VFDs

[Guineafowl]

First find the A_L value for the core.
Wind a test coil of say 10 or 20 turns, then measure the inductance.
A_L = uH / turns².
uH = A_L * turns².
While measuring A_L, use a test frequency closest to the intended operating frequency of the inductor.
Solve for the number of turns needed to get the required inductance.
Select a wire size with sufficient ampacity. I use 5A / mm².
Check that you can pass the required number of turns through the toroid.

Avoid thick plastic wire, it gets hot and melts. Use magnet wire. Find an old transformer and salvage what you need.

Experiment while measuring A_L. Place two cores flat against each other, then wind the test coil through both. That doubles the magnetic section available, so may reduce the chances of saturation. Let us know what A_L values you get.


If possible, switch the probes to x10, then neutralise them to show a square wave while the probe is connected to the probe adjust terminal on the front panel.

Thanks - I’ll get on with building the inductors when I get a chance.

I have adjusted the probes using the CAL terminal, and they were on x10 for that picture.

 

[Guineafowl]

I found some 1 mm$^2$ magnet wire. My LCR meter (actually an ESR meter with L as a side function) is quite limited, so I ended up with 5 turns around the Mn-Zn toroid giving 196 $\mu$H and so $A_L = 7.84$

28 turns was all I could fit, and certainly highlighted my carpal tunnel syndrome, but should give around 6 mH. I’ll report back after scoping.

 

[Baluncore]

28 turns was all I could fit, and certainly highlighted my carpal tunnel syndrome, ...

How to wind a coil faster, and more safely, with less pain. Get a nerve conduction study to confirm CT. If CT get the surgery. If RSI is not CT, notice if it is less during the day after NCS. If so, you can test it by holding an electric fence in that hand, a traditional farmer's remedy for RSI, treated every morning, before the milking. If that relieves the pain for a day, you have a chemical neurological problem, maybe a small part of long covid, or similar. Then, if you still need to wind coils, evaluate a low-dose SSRI, which has the same effect on neurotransmitters as does the electric fence, but is less shocking. Invest in your neurological and coil winding future by getting your B12 checked, as neuropathy is a longer term co-symptom. This is coil winding advice, if you need medical advice, consult a doctor (of medicine). That way you can wind more turns, through smaller cores, to get square-law higher inductance, all without the pain.

 

[Guineafowl]

Thanks! I did have CT surgery about ten years ago. This is more a side-issue of deep thenar pain, maybe tenosynovitis or basal joint arthirits. A career-ender (and guitar-playing ender) for me.
The electric fence trick is good neuroscience - gate control, endorphin release - and the basis of TENS. Incidentally, I was taught by a dairy farmer to grasp the electric fence line firmly. Much less painful than trying to hold it delicately, as you would instinctively.

Edit: that’s got me thinking. I have a Megger for IR testing, limited to 4 mA and offering 50-1000 V ripply dc. I wonder if this could be deployed as a home TENS unit? Probably not a good idea.

 

[Guineafowl]

Try as I might, I couldn’t get the scope to trigger properly. This is channels 1 and 2 on U and V terminals of the motor, scope set to CH1-CH2, 10x probe setting. The changing image was hard to get on camera:

Next the filter circuit was added:

However, the VFD threw an overcurrent error on start-up until I removed the capacitors, and replaced them with Vishay VY1, 2.2 nF, X1/Y1 rated ones. The VFD manual does mention that very long motor cables can have enough distributed capacitance to trip the protection, so maybe this is related.

Again, I had triggering problems with the subsequent scope shot, but it still looked ‘steppy’ like the first one, with or without the alternative Vishay capacitors. Might the chokes be saturating?

I think I need to sort the triggering before I can do any more meaningful experiments. Would a modern digital scope cope any better?

 

[Baluncore]

I think the two channel amplifiers are saturating before the subtraction is done.
You need more attenuation.
Do you have a neutral available on the 3PH output?

 

[Guineafowl]

No neutral. The VFD output is 240 V P-P. How come it displayed the real 240 V mains ok?

 

[Baluncore]

No neutral. The VFD output is 240 V P-P. How come it displayed the real 240 V mains ok?

240VRMS is 340 Vpk = 680 Vpp centred on zero.

240 volts RMS, makes 415 VRMS per line pair, or phase winding, with peak phase voltages over 580 volts, about the neutral or zero volts, giving 1160 volts peak-peak vertical.

The oscilloscope has 5 volts per cm, x10 probe, and 8 cm, giving a 400 volt vertical range. That cannot display 1160 volts PP.

You will need to make a compensated attenuator adapter to go between the 3PH supply and the oscilloscope.

 

[Guineafowl]

But the VFD isn’t a step-up type - the output is three-wire, effectively delta, 3ph at 240 Vrms phase and line voltage. It should, as I understand it, take the form of 340 Vpk (680 Vpk-pk) pulses of varying width. Could it be that the very thing we’re trying to get rid of, voltage overshoot and ringing, is preventing the scope from triggering? This article shows some scope shots that suggest it might be:

https://www.power-electronics.co.nz...anaging-dvdt-in-ac-variable-frequency-drives/

 

[Averagesupernova]

It's pretty clear to me that you don't have the input attenuation set up correctly on the scope. As far as triggering goes, if you're worried about correct triggering set the scope to trigger off of the power line and set the VFD to output the same frequency as whatever the power line frequency is.
-
You also say "No neutral". The output of the VFD must somehow be referenced to neutral/ground. If it's floating your scope may have a hard time.

 

[Guineafowl]

Thanks, the line trigger trick worked better, but the trace is either incomplete or slowly redrawn, presumably due to variations in line and VFD output frequency. Not camera-friendly.
This is a typical VFD topology. In my case, the input is single phase, and the motor in delta. It is mains-referenced, but there is no output neutral.

I’ve just measured the output with a Fluke true-rms meter with low-pass filtering for VFD measurements:
A-B, A-C, B-C: 230 Vac
Any phase to earth: Interesting. It appears to oscillate between 260 Vac and 30 Vac over about 4 seconds.

 

[Averagesupernova]

Any phase to earth: Interesting. It appears to oscillate between 260 Vac and 30 Vac over about 4 seconds.

That's because it's not referenced to earth. Can't expect the scope to give you anything readable.

 

[Baluncore]

I would expect the 3PH lines generated by the VFD should each look like a 240 Vrms relative to the neutral input or ground.
Each line will vary between +340 V and -340 V.
You cannot view that with the oscilloscope using only the x10 probe switch.

5 volts per division, x10, over 8 divisions = 400 V total, yet you need twice that to show the whole ±340 V wave on the screen. Without more attenuation you will not see a whole cycle of 240 Vrms, let alone the difference between two lines.

In your post #35 you show stepped sine waves, but they are being clipped by the input amplifiers. The subtraction is clamping the voltages at the x-axis.

In dual trace mode the single beam CRO either alternates sweeps, or switches rapidly between the traces. You may need to take control of that dual trace mode.

 

[Averagesupernova]

I would expect the 3PH lines generated by the VFD should each look like a 240 Vrms relative to the neutral input or ground.

In the USA 240 volts 3 phase would generally be delta with 240 between the corners of the delta. Typically one winding center tapped for ground/neutral but not always. Wye would be 120 volts hot to neutral.
-
The fact the he gets readings between the outputs of the VFD and ground that are not stable implies the output is floating. Wonder if that is configurable.

 

[Baluncore]

In the USA 240 volts 3 phase would generally be delta with 240 between the corners of the delta.

I assumed the UK with 240 Vrms single phase. 415V 3PH.

 

[Averagesupernova]

I assumed the UK with 240 Vrms single phase. 415V 3PH.

Actually I'm not sure where the op is. So my comment was informative but not necessarily helpful.

 

[Baluncore]

Actually I'm not sure where the op is.

Cambridge (UK).

 

[Guineafowl]

That's because it's not referenced to earth. Can't expect the scope to give you anything readable.

It was the multimeter measuring. The supply neutral is referenced to earth at the pole transformer, probably around 100m of cable length away, overall:

You cannot view that with the oscilloscope using only the x10 probe switch.

Looking again at the scope shot in #29, I was able to move the trace down and see the peak of that waveform, which would be +340V. This should be the same for the PWM waves, not counting voltage overshoot. It’s probably that causing the trouble.

In dual trace mode the single beam CRO either alternates sweeps, or switches rapidly between the traces. You may need to take control of that dual trace mode.

I did experiment a bit - the traces shown (#29 and #35) are in single-trace mode, CH1-CH2, with CH2 turned off (you can see it’s in lowercase). Adding it in or out didn’t seem to make much difference, nor changing add/chop with it turned on.

 

[Averagesupernova]

Any phase to earth: Interesting. It appears to oscillate between 260 Vac and 30 Vac over about 4 seconds.

This needs to be addressed before we go any further. Those measurements indicate the outputs are not referenced to earth. You say the power supplying the VFD is referenced to earth which is likely true. Until we know how the internal power supply of the VFD is designed we are somewhat in the dark.
-
It would be interesting to take three resistors wired in a wye confirmation connected to the VFD output. Measure from the node of the wye to earth with your voltmeter. Do not connect the scope ground to this. Measure with the voltmeter both AC and DC. Of course you will have to make appropriate selections of resistors and wattages. You could use incandescent lights bulbs for this with matched wattages.

 

[Guineafowl]

Ok. I have to be careful about the selection of the resistors - running a VFD with the output open circuit can ruin it. Using high-value resistors may have the same effect.

The motor windings (in delta) have a dc resistance of 6$\Omega$. This is one winding in parallel with the other two in series.

I was rather hoping someone would insist I get a digital scope, then I’d have an excuse. With one of those, I could single-shot trigger until I hit upon a good waveform, or just run a recording and scroll and zoom to get the picture, no?

 

[Averagesupernova]

I was rather hoping someone would insist I get a digital scope, then I’d have an excuse. With one of those, I could single-shot trigger until I hit upon a good waveform, or just run a recording and scroll and zoom to get the picture, no?

Well I'm certainly not one to talk someone out of acquiring more tools. :) But your voltmeter readings don't make sense. What have you been powering to obtain the scope readings you have posted? No reason you can't have the light bulb trio wired in at the same time.

 

[Guineafowl]

What have you been powering to obtain the scope readings you have posted?

This, a 1961 Beaver VBRP:

 

[Averagesupernova]

This, a 1961 Beaver VBRP:
View attachment 373198

Ok. So you are assuming the voltage spikes won't damage the motor during testing I assume.

 

[Baluncore]

The VBRP was the much improved version.
VB = Victor Balding, the founder and owner of Balding Engineering.
RP = Richard Prime, the machine tool designer, who redesigned the machine.

PAL was a smaller mill, named after Victor Balding's golden retriever.

 

[Guineafowl]

Ok. So you are assuming the voltage spikes won't damage the motor during testing I assume.

Yes, like most small-time machine users, I’ve set up the VFD and hoped for the best. This one’s been in use for a year; I have other old machines on VFDs for years. The risk of insulation failure is apparently a real one, so it’d be nice to make a simple, cheap filter for peace of mind.

The VBRP was the much improved version.
VB = Victor Balding, the founder and owner of Balding Engineering.
RP = Richard Prime, the machine tool designer, who redesigned the machine.

PAL was a smaller mill, named after Victor Balding's golden retriever.

Good knowledge! It’s usually described as a Bridgeport copy, although slightly heavier at 1.5 tonnes. Note very long bar in the picture, for shunting it into position.

 

[Guineafowl]

Well I'm certainly not one to talk someone out of acquiring more tools. :)

If you insist, then. I’ve ordered the Rigol DHO804, which has some good reviews. Having four channels on the Fluke has been helpful, eg when scoping a switch-mode power supply - I can watch the feedback, oscillator, chip power supply and output voltage all at the same time.

Also, two cheapish 100x P4100 probes, which might come in handy for other things.

The wye incandescents gave a phase voltage of around 130 Vac, with the same sort of oscillation to earth. It might be something to do with the VFD’s EMI filters.

 

[Averagesupernova]

The wye incandescents gave a phase voltage of around 130 Vac, with the same sort of oscillation to earth. It might be something to do with the VFD’s EMI filters.

I assume the 130 volts was measured from the wye node to each VFD output? Sounds about right. You will probably not be able to makes sense on any scope with the voltage to earth moving around like that. I wonder what would happen if this wye node were tied to neutral through another incandescent or maybe a pair in series in case the voltage exceeds the bulbs rating.