Induction Motor Switch-off Transients

[rowan.bradley]

I have been experiencing some problems with a power supply unit failing due to transformer faults. These 230V AC input 9V DC output PSUs are installed close to an induction motor which can be from 1/2HP to 10HP depending on the installation. It's been suggested that possibly the transformers are being damaged by transients on the mains supply caused when the induction motor switches off. There is no surge suppressor or filter in the PSU.

Can anyone comment on this?

What do you see on the mains when an induction motor is switched off?

Should the motor itself be fitted with some sort of surge absorption device?

Should I add a surge arrestor to the PSU? If so, what sort of surge does it have to deal with (how long, how big?)?

Thanks - Rowan

 

[Resqueline]

Are the PSU's of the old heavy transformer type or are they the modern lightweight switchmode type?
Have you verified that the mains RMS voltage is within normal boundaries, and that the transformers have not failed due to overloading on their output?

Spikes will appear over the motor windings as it's turned off. These will also appear over the switch and may (will) arc over to the mains.
The spikes are very short but may very well reach kV values if the mains wiring has a high impedance at high frequencies (no capacitors).
Varistors (transient absorbers) can be fitted both across the motor, the switch, the motor mains, and of course also at the PSU mains.
They need not be very big (20mm). The stored energy is not large and is probably dissipated within microseconds.

 

[rowan.bradley]

Thanks for your helpful reply.

The PSU is linear, i.e. it uses a toroidal transformer, rectifier and linear regulator. It's not particularly big or heavy (or old ) since it's only 9V at 3A.

I haven't checked the mains voltage, but it's just plugged into a 13A socket, so I can't imagine the voltage is greater than 264V other than in short transients.

I think it's very unlikely that the transformers have failed due to overloading. They are rated at 4.17A secondary current, and there is a 3A fuse in the DC output. I can't think how I would overload this if I wanted to.

I will think about adding a varistor. How does using a varistor compare with the use of gas arrester diodes such as the EPCOS? We use those (albeit at a lower voltage) elsewhere in the system, mostly to protect against lightning strikes and other nasties on the data transmission lines.

Thanks - Rowan

 

[Resqueline]

Gas arresters are not diodes. They have a lower impedance than varistors but when they ignite they practically short out the rest of the mains half-period (not good).
They behave almost like thyristors; when triggered they remain higly conductive (only a few volts drop) until the current drops blow a certain treshold. (Bad across mains.)

Then there's transient absorbing semiconductor diodes, some unipolar and some bipolar. The bipolar ones can be used instead of varistors, and also across mains.

 

[rowan.bradley]

Thanks for your reply. As you say, the gas arrester does not sound appropriate across the mains then! I think I'll try the varistor as you first suggested, and see if that does the job. The EPCOS B72220P3271K101 (Farnell part number 1781938) seems about right. Presumably, since we know that if these transients are coming from the induction motor they are differential mode, we only need a varistor between live and neutral (not between live and ground or neutral and ground)?

It's a bit of a shot in the dark, because we've only had a couple of these failures in over 100 systems over 2 or 3 years. I've basically just got to add the varistor and see if, after a few more years, any more similar failures occur.

Thanks again for your help - Rowan

 

[Resqueline]

Yes, I figure those should do the trick (if the problem stems from spikes), wired across live-neutral. May I ask what kind of fault was ailing the transformers?

 

[rowan.bradley]

The one that's just failed has, according to the company that supplied it, been overheating "for some time". They blamed a shorted turn, but then shot themselves in the foot by saying that even one shorted turn in a toroidal transformer (which this is) would cause it to draw enough current to blow the input fuse. I don't know whether this is true - do you?

I have therefore concluded that they don't know exactly what the fault was or why it failed. I think it very unikely that the fault was due to an external overload, since the transformer secondary is rated at 4.17A and the DC output fuse is 3A (and in most installations the actual current will be substantially less than 3A). I suppose there could actually be some other fault inside the PSU, in which case hopefully we will find it when they have replaced the transformer (which is happening now). I haven't physically seen this unit so I can't give any more info, I'm afraid.

Thanks for your help - Rowan

 

[Resqueline]

That's a footshot allright. A shorted turn would (if not blow the fuse immediately) burn the insulation immediately around it to crisp within a minute, and then blow the fuse.
A general overheating "for some time" can only come from a serious long-term mains overvoltage, or an overloaded secondary (which you say isn't possible).
It's not always easy to distinguish exactly what chain of events lead to it's demise after the initial fault has evolved into a catastrophic failure.
Did it blow the primary fuse eventually? How big is that fuse? Oh, and 3A DC is actually "the same as" 4.24A AC btw..

 

[daddles]

One thing I learned from a few decades in industrial R&D is "never trust the customer's statements", at least when it comes to failures. Get the unit that failed and do the failure analysis yourself. It keeps you from draining the wrong swamp. Even then, you only find out what failed -- not necessarily what caused it.

 

[duke37]

Toroidal transformers have one thin lamination with aligned crystal structure. This makes them more efficient but means they run into saturation more abruptly. Is it possible that the saturation voltage can be obtained? Can the transient votages be measured when the motor is switched?

 

[rowan.bradley]

Is it possible that the saturation voltage can be obtained?

Do you mean that the core could be saturated by the transient? I've no idea - what would be the effect if this were to happen?

Can the transient votages be measured when the motor is switched?

Well, it's not so easy to do this on the site where the failure occurred since it's 400 miles away. And doing it somewhere else is not so meaningful since each of these installations is different, with different motors, different arrangement of mains wiring etc.

You are of course right that I will have to try to do this sort of testing and analysis if these problems continue...

Thanks - Rowan

 

[rowan.bradley]

"never trust the customer's statements"

I agree - although in this case it's more the supplier that I don't trust. I know the only real solution is to get hold of the failed units myself, and I did try in this case, but unsuccessfully. Next time...

Rowan

 

[duke37]

When the core goes into saturation, the input current is limited only by the primary resistance which should be low for an efficient transformer. The high current then heats the wire and perhaps spoils the insulation between turns. Transformers can go into saturation during switch-on and I have fitted soft start to some of my home projects to alleviate this problem.

If the PSU has a lot of surplus capability, you could try a crude bodge by putting a resistor in series with the primary so limiting the current on saturation.

It should be fairly easy to estimate the saturation voltage with a Variac, high voltage transformer and a series bulb. Transients are not easy to measure since they may not be there when you look. There they was - gone!

 

[rowan.bradley]

I don't think core saturation can be causing the problem I'm observing. There wouldn't be enough energy in the motor switch-off transient to heat up the whole transformer, or its primary winding. If it is these transients that are causing the failures (not proven), I think they must be doing it through over-voltage, not over-heating.

Rowan

 

[rowan.bradley]

Did it blow the primary fuse eventually? How big is that fuse?

I'm told that it did not blow the mains fuse. The fuse is 3.15A anti-surge. This is obviously far bigger than it needs to be since we are only talking about a 27W PSU.

Oh, and 3A DC is actually "the same as" 4.24A AC btw.

I don't understand this. Surely a fuse will respond to the RMS current, not the peak, since it responds to a heating effect? I'm assuming its time constant is long enough to smooth out the 100Hz cycles of the heating effect of the mains. So a 3A fuse will surely take 3A DC, and 3A (RMS) AC?

This is actually a 6v-0-6v transformer with a full wave rectifier feeding into a reservoir capacitor, so the current will be quite peaky, but must average 3A. Since only half of the secondary is conducting at any one time, and they conduct alternately, the average current for each half of the secondary must be 1.5A, so (since the transformer is rated at 4.17A) it is only operating at a fraction of its capacity.

Rowan

 

[duke37]

Heathkit used to supply transformers which were a little light on the iron. On switch-on there was a thump on the iron cabinet due to the escaping magnetic field when the transformer saturated. On the principle that action and reaction are equal and opposite, there is quite a force on the wire which will vibrate. Eventually a shorted turn could develop.
This is not to say that saturation occurs in your case or that it is the cause of the fault - just a possibility.

 

[Resqueline]

Toroid transformers can have a very high current surge at turn-on due to saturation, thus the seemingly excessively rated mains fuse.
But if it didn't blow the mains fuse then what happened to it? A winding break, or an internal temperature fuse opening perhaps?
Transformers by themselves usually doesn't give up without a fight..

Transformer power: 12V * 4.17A = 50W
ACV to DCV: 6V AC * 1.414 = 8.5V DC
Max DC current: 50W / 8.5V = 5.9A DC
This is a somewhat simplified calculation but shows that the transformer can supply almost double the current of the fuse.
Notice the square-root-of-two relationship here; it ups the voltage but downs the current, and so the AC & DC power remains the same.