Sinusoidal filter on VFD does not cure all 4

[Marke]

Recently, I was asked to have a look at an installation with a VFD where there were major problems with equipment failure when the VFD was running.

The VFD is a well recognized brand on a 400V MEN system. The main switchboard is about 250M from the supply transformer which supplies this switchboard only.

The supply transformer is about 30KVA, delta / star 11KW / 400V neutral earthed.

The main load on the switchboard is a 15KW VFD with in built EMC filtering and DC bus chokes. Other load on the switchboard comprises local lighting and control equipment.

The output of the VFD is connected to a sinusoidal filter which in turn is connected to a submersible pump by a 300M length of neutral screened cable.

When the drive is operated, the carrier frequency is superimposed on the supply voltage at the input to the switchboard. The level of "noise" voltage on the input is in the order of 150 volts relative to neutral.
Most single phase electronic equipment connected phase - neutral has failed due to the high noise voltage.

If we disconnect the output from the VFD, there is no problem. If we connect the sinusoidal filter only, there is no problem.If we connect the motor via the screened cable with the screen disconnected at both ends (not legal!) the noise voltage on the input is reduced to about 10 - 20 volts superimposed on the phase - neutral voltage.

Part of the problem appears to the the sine wave filter which comprises a three phase reactor with three capacitors on the output connected between the three phases.

The output waveform phase to phase is a good sinewave, but the output voltage between an output and earth, is a reduced PWM waveform. This PWM waveform is applied to the cable capacitance core to the screen and results in a high earth current.

In addition to the sine wave filter curently installed, it is obvious that we need additional filtering to remove the PWM waveform to earth.
I am tempted to try additional capacitors from the outputs of the sine wave filter to the positive and negative rails on the DC bus to provide a return path for the HF signals.
Any comments?

Best regards,
Mark.

Mark Empson
L M Photonics Ltd

 

[Skogsgurra]

Tough case, Mark.

I know you have talked to the sine guys. They have done this for Grundfos. And you are right at the end of their power range.

One step further and they couldn't help.


Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[roydm]

Do you think it might be a ground loop problem with the screen grounded at both ends, Would it be possible to ground the pump with a ground external to the shield?
I don't think a line filter would help, I suspect the noise is getting to the supply through the grounding, not through the line.
Interesting problem
Roy

 

[itsmoked]

You should just use a water valve instead of a VFD..

Bwahahah..

This really sounds too bad for common VFD issues. It makes me think maybe there is a bad cable or perhaps the motor has an issue so one phase is closer to ground?

Got another motor laying around you could hook up instead?

Could the sine filter caps be interacting with the drive in a pathological manner.

I'd be pretty nervous about the cap addition you're proposing.

Have you tried a different modulation frequency? You may have a system that is resonate at just the operating point your system is set up at.



Keith Cress
kcress -

http://www.flaminsystems.com

 

[Marke]

Hi Keith

Yes, played with the carrier frequency and all the usual.

The total circuit is acting like a boost mode SMPS. The DC Bus voltage rises and at one stage, it took out MOVs on the DC Bus.
I have come across installations before where the total system behaves like a totally different circuit.

The sine wave filter works very well as long as you look between the phases.

Sine wave filters are promoted as the way to sort problems with long cables, reducing overshoot issues with motors and also eliminate the requirement for screened cables.
In this installation, the conducted emissions are real bad, so the sine wave filter does not sort those!!
Additionally, issues such as bearing currents are not corrected because the PWM to earth still exists.
A sine wave filter outputting a sinusoidal voltage between phases and between the phases and earth would address these problems and eliminate the need for a screened cable.

I believe that if I can provide additional filtering of the output relative to the DC bus, then the problems will go away, but at this stage I have not tried it.
Another suggestion is to fit an isolating transformer to the output of the VFD.

If there was significant load current at the switchboard, or the transformer was very close, this problem would probably not show.

Best regards,
Mark

Mark Empson
L M Photonics Ltd

 

[waross]

Tough case, Mark.

That may be one of your more notable understatements Gunnar.
This is the kind of question that has me asking myself, "I wonder what Mark will have to say?"

I have some questions. The answer to the questions may lead to a solution, or may prove that I should have kept quiet!

Is the screen used to ground the motor? Is it possible to use a conductor to ground the motor and ground the screen only at the supply end?

The PWM waveform is, of course, both current and voltage. Is it possible that the screen is passing enough PWM current to drive up the PWM voltage at the DC bus?
The problem as I see it is PWM voltages reflected back onto the supply side.

I have an idea that I am struggling to express.
It goes something like this:
PWM voltages are transmitted on the motor conductors. (Fact of life.)
These voltages are capacitively coupled to ground through the screen. The size and length of this installation is causing this coupling and the resulting currents to be more than normal.
The resulting PWM currents are causing higher than normal PWM currents to flow in the supply side.
The higher than normal supply side PWM currents are causing higher than normal PWM voltages on the supply side.

Now, the cause may be in the output but the symptom is in the input.
To treat the cause;
Can you consider more reactance instead of more capacitance in the filter (or both). The idea is to filter the PWM current. I guess that I am suggesting that the filter may be undersized for this challenging application.

Capacitors to ground on the filter may make the issue worse by causing more current to ground.

To treat the symptoms;
Have you considered capacitors to ground on the supply side to provide a low impedance path for the reflected PWM voltages?

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[Marke]

Hi Bill

This is an interesting problem and a bit of an eye opener in that the sine wave filter is promoted as the cure all by many suppliers where in fact it only achieves part of what it is expected to do.

When the sine wave filter is not installed, the three phases operate independently in terms of the pwm output and have bot a differential component and a common mode component to the PWM as seen phase to earth. Once the filter is installed, the PWM is now all common mode only so there is no cancellation in the cable.

The effect of the sine wave filter is to provide a common mode PWM signal relative to earth so it is now a "single phase" waveform and it is capacitively coupled into the earth and that is of course tied to the neutral.

Disconnecting the screen at both ends significantly reduces the problem.
Disconnecting the screen at the drive end reduces the problem by about 50% +/-.
The screen is the earth conductor for the motor so we can not disconnect it.
Running another earth wire is not really a practical solution, it is all buried, has been concreted over in places, goes under a stream etc.

I connected a 230V 100W lamp in series with the connection to the screen and it glowed quite bright so there is some energy there!!

There is a connection from the screen directly back to the drive earth and there are capacitors on the input as well, but these are small.
The connection of large capacitors from the input to ground will possibly introduce other issues as these will cause 50Hz currents to flow in the neutral circuit so are providing a capacitive KVAR affecting the power factor, but of greater concern is the ability of these capacitors to resonate with the supply reactance. I suspect that there could be some very high ringing voltages developed in response to supply transients. I have experienced this before.

The cause of the problem is complex but stems from the fact that there is a common mode PWM voltage at the output of the sine wave filter.
My suggestion is that if we can eliminate that, we will eliminate the problem.
One solution would be to add additional reactors after the sinewave filter and follow these by capacitors back to the DC bus of the VFD thereby closing the loop.
I am sure that this will work, but am also curious to know if we can close the loop safely without the additional reactors.

Best regards,
Mark.

D

Mark Empson
L M Photonics Ltd

 

[itsmoked]

How about a serious input EMC filter as apposed to the marginal internal types?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Marke]

Hi Keith

The internal EMC filter is OK. The unit is compliant with our local CTick requirements.
I think that the only EMC filter that will work is a switch (Isolator).

Unfortunately, the installation is 4 - 5 hours away from here so it is difficult to pop out to site and make a few more measurements, and I did not have a spectrum analyser with me when I was on site, but I expect that the sinusoidal filter will have taken the edge of the PWM waveform, reducing the rate of rise of voltage. This will reduce the effectiveness of the internal return path between the DC buss and the earth.
This is usually a capacitor from one side of the DC bus to earth and is adequate for the high frequencies usually causing problems.

An EMC filter on the input only becomes effective when there is an alternative low impedance path back to the DC bus. Increasing the HF impedance in series with the input causes more noise to take the alternative path.

Unfortunately, due to the distance, and the amount of work required to reinstall the VFD and filter, I need to provide the cure on next site visit rather than making more measurements. I may be able to set up some simulation in the lab, but I a sure that someone has struck this before and cured the problem!!

Best regards,
Mark.


Mark Empson
L M Photonics Ltd

 

[Skogsgurra]

As I said, Mark. There are inverters that are intrinsically void of any noise and do not have a filter in the classical sense. They have been used in wells with long cables where no other inverter could do the job. I hesitate to put the brand here, but if you google "sinus frequency inverter" you will get hits that lead you to a solution.

I am absolutely sure that you have been in touch with the company. At least, that is what they told me a couple of years ago.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Marke]

Hi Gunnar

Yes, That is a good solution, but the customer has already purchased and so we are stuck with another unit made in a Neighboring country.

That solution would be a more cost effective solution than the existing with the filters, but I was brought in by the equipment supplier looking for help, so I best find a solution using their gear!!

What are you doing reading these posts at this hour of the morning?? Didn't Karin tell you it is sleeping time??

Best regards,
Mark.

Mark Empson
L M Photonics Ltd

 

[itsmoked]

Yes Skogs but changing VFDs always means "someone screwed up" and no one will go with that solution.

It's too bad this VFD wasn't placed at the well head. Sounds like that might have shaved off 100m.

Thanks for the explanations Mark.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Skogsgurra]

Well - you know. The P word and such things.

BTW, Karin had a fall last night. Went right through the coffe-table. Not hurt at all. But table was less lucky.. :-(

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[waross]

Thank you for your kind response, Mark.
Not a suggestion as much as a question.
Would an isolation transformer provide extra filtering as well as isolating the currents returning on the screen from the VFD? Cost of purchase and installation?

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[Skogsgurra]

BAW, there is a very simple solution. If you remove the two first and the two last letters in the brand you have - then you will get the brand you should have!

Not sure if that works in practice, though..

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Marke]

Hi Gunnar
I hope that you told Karin not to be so clumsy in future!! Good to hear that she is OK!

Yes your formula is correct. I am tempted to get one over and loan it to the end user to get them going as at present they can not run things. May prove a point as well.

Best regards,
Mark.


Mark Empson
L M Photonics Ltd

 

[Marke]

HiBill

Your suggestion is to silly. One of the proposals that I put very early on, was to connect a delta primary, star secondary 1:1 transformer to the output of the drive. The flux in the iron will be almost sinusoidal (same as a motor) and so the voltage in the secondary will be a sinewave and it should be relatively easy to filter any additional PWM leak through.

Best regards,
Mark.


Mark Empson
L M Photonics Ltd

 

[Skogsgurra]

I would not say she was clumsy. She fell off a ladder when cleaning a filter in a heating device. She is more of a daredevil, I think.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[ozmosis]

Hi Mark
Happy New Year!
Not sure if this drive is one of ours but a potential solution is a combination of LC filter and common mode output filter as developed by Schaffner. This filter connects also to the DC link of the VFD and has improved performance in the past. I haven't used it in this hemisphere but did in Europe to pretty good effect.
Be aware of the losses in the complete system though when adding additional components on the output.
Regards
Patrick

http://www.schaffner.com/components/en/_pdf/Datasheet FN5030 e 83.pdf

 

[Marke]

Hi Patrick

Happy new year to you also.
This is probably the sort of solution that I was looking for and I think the connection to the DC bus is very much the answer.
I expect that they are coupling the output of the sinusoidal section to the DC bus via capacitors, possibly with additional series L in each phase.

I will put this filter forward as a possible solution.

Have a good year, must catch up sometime.
Best regards,
Mark

Mark Empson
L M Photonics Ltd