VFD trips GFI

[InverterDuty]

I am trying to run a 1hp, 3p motor with a vfd. I need the motor to run on a gfi. I have tried several different brand of motors as well as drives. Every configuration has caused the gfi to trip. With some testing we determined the trip was caused by current leakage from the motor. I contacted marathon motor and they had a solution. They have made a special motor that has one extra lead that you connect the the negative side of the DC brake output. Since the DC output is isolated, the result was almost a complete elimination of leakage current. Since it is a special motor, lead times can be long and it is only offered in a few frame sizes. My question is... What attributes of a motor contribute to motor leakage and has anyone found an off the shelf motor that will work in conjunction with a vfd on a gfci circuit? The odd thing is a competitor of mine makes a machine that uses a standard motor with a vfd that does work on a gfi. I tried there control circuit on my motor with no luck, so I know it is a motor issue. The problem is that the motor they use is made in china and does not have certification on its own to be imported into the us. In addition the equipment I would like to use USA sourced parts as much as possible. Are there any filters or devices that can help with the leakage current of a motor.

Thank you in advance.

 

[mikekilroy]

interesting post....

did you ask marathon EXACTLY what that 4rth wire ties to in the motor? If not, then do so? All they can say is no....

I will suggest that they supply a wye configured winding and the 4rth wire is the Xo or neutral center point..... tying that back to one of the DC bus lines probably is the key to prevent leakage back to ground instead.....

that said, YOU can make your own Xo to do the same thing I would guess and then still be able to use a std motor..... here is my idea you could try out if you want....

get a small 3ph xfmr with a 230v wye winding on it (460v if u r on 460 vfd) and tie it across the 3 ph vfd output. Tie the Xo back to the dc bus (minus sounds 'cleaner' that plus so go there altho it does not matter). This will reference all 3 phases to a low Z path back to dc bus so leakage may not go back to ground instead.....

Idea 2: if you are into experimenting, might work just as well to simply tie your dc bus minus side to ground directly and see if that does it? If you try this, stick it thru a small fuse rated say 1/10th the vfd current output for safety. We used to tie the dc bus minus to ground 100% of the time in the 1980's with the thought it reduced switching noise. In reality it did nothing to help but also hurt nothing - other than providing a path to ground to blow input diodes if one of the input lines or output lines of the vfd got tied to ground by mistake. So if your vfd is supplied by an old fashion corner grounded delta, you cannot do this or you have a direct short.

These experiments are just to give you some more ideas and help understand why your competitor can do it and you cannot.

 

[InverterDuty]

Mikekilroy,

I have attached documentation from marathon explaining what they have done. Thank your for the suggestions, I will play with them a bit.

 

[jraef]

What do you mean by "GFCI" by the way? Are you referring to trying to use a Class A GFCI for personnel protection on a 3 phase motor circuit? Why? I seriously doubt you will EVER get it to work, a Class A GFCI has to trip on no more than 6ma of current, you common mode noise from a VFD driven motor is likely going to be more than that. You may need to rethink this entire concept.

If instead you are looking for equipment ground fault protection, I think most good quality VFDs come with that built-in now. What VFD are you using? If your's doesn't have that, maybe it's time to consider not feeding off the bottom...

"Dear future generations: Please accept our apologies. We were rolling drunk on petroleum."
— Kilgore Trout (via Kurt Vonnegut)
For the best use of Eng-Tips, please click here -> faq731-376

 

[mikekilroy]

They have made a special motor that has one extra lead that you connect the the negative side of the DC brake output. Since the DC output is isolated, the result was almost a complete elimination of leakage current.

not what their doc says! I assumed your neg side of dc brake was typo and you meant dc BUS but now not sure..... their doc says they simply put in some metal sheets in the windings like capacitors to drain off some of the dv/dt spikes to GROUND not dc bus....

 

[itsmoked]

Besides the point jraef makes this is what I'd do.

Your motor is leaking AC to its case from the waveforms sent to the motor. With a classic hookup you have the motor grounded locally by having it touch machinery or a ground wire from somewhere else. When running there is current lost to this ground that causes a GFI trip. Instead of any 'ol ground you should run a single ground from the motor back to the case of the VFD. The only ground available to the motor should be the VFD case. Now the high frequency current will close its circuit with the VFD instead of thru quasi-unrelated wiring. You still ground the motor you just do it from a specific location - the VFD.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[rhatcher]

Marathon's paper describes the source of the problem as being high frequency leakage current. Their solution is pretty ingenious but there is an alternate approach that may not require the use of a 'special' motor.

The leakage current that is tripping the GFCI is not at the 'power frequency' of the VFD output (0-60hz) since the low frequency impedance of standard winding insulation is more than adequate to prevent 5ma of leakage current at normal power frequencies. The leakage current is at the high frequency generated by the switching frequency of the VFD. For high frequencies, the relative impedance (capacitance) of the winding insulation is very low, thus allowing high frequency leakage current to flow.

My suggestion is to set the VFD for the lowest possible switching frequency. The relative impedance of the winding insulation will increase proportional to the decrease in switching frequency. As an example, reducing the switching frequency from 10khz to 2khz will increase the relative impedance of the winding insulation by 5 times. This will reduce the amplitude of the high frequency leakage current by the same proportion. This may be the way that your competitor can operate a similar design without tripping the GFCI.

MikeKilroy's and itsmoked's responses also bring up a good point. Many VFDs have two grounding points, an 'equipment safety ground' for the VFD and a 'motor ground' for the motor frame. (The terminology may vary according to different maufacturers.) The 'equipment safety ground' is attached to the incoming power system ground (or to a building girder or other external ground) and the 'motor ground' should be connected to the motor frame. The 'motor ground' provides a low impedance return path to the VFD for high frequency leakage currents that prevents them from flowing in the power system ground. You should examine your VFD instructions and ensure that the motor is properly grounded according to the instructions. If there are two separate ground points on the VFD, make sure that they are connected according to the instructions and do not externally connect, or jumper, them together.

In any case, I am confident that the reason that your competitor's system does not have this problem is not because they are using a Chinese motor. This is not an offense to the Chinese. I am simply suggesting that with the exception of special features like the one used by Marathon, all motors will be subject to the same high frequency leakage currents regardless of where they are manufactured.

 

[mikekilroy]

rhatcher's idea is well worth trying. I believe the actual capacitive coupling tho is more due to the 'frequency' of the on/off transition of the igbts rather than the actual pwm freq. changing pwm freq wont change this on/off time so it is my swag that changing pwm freq will not help - but do try!

it brings up point tho that an output reactor in series with the std motor WILL slow the on/off times and thus also possibly reduce the capacitive coupling. prob worth buying a $ 100 inductor and trying it too.....

 

[LionelHutz]

I see 1 anwer with a bad error and another incomplete answer.

Do not ground either bus terminal in the drive as suggested. That will cause a line to ground short on a grounded power system and it will cause you to alternately ground each phase 60 times a second on a floating power system. Definately use a fuse when experimenting like this or you'll have a blown-up drive. Overall, it's a very bad idea.

The suggestion to take the ground of the motor back to the VFD probably won't help unless the GFCI detection is being done on the input side of the VFD. The VFD would also need some type of connection to feed the motor ground currents from the VFD chassis directly back into the DC bus of the drive. Most likely, this connection would be capacitors from ground to the positive and negative of the DC bus. Overall, I would not expect to find these capacitors on a cheap small hp VFD. However, it's possible your competitor has a drive with these or has added them himself. He may also be using something like a Schaffner FN5030 combined with one of their sinewave filters. This couples the common mode high frequency switching energy directly back to the DC bus, which should eliminate it from travelling into the cable and motor and then coupling through the cable and motor capacitance into ground.

More details on the GFCI you are using would be helpful. I don't typically see 6mA GFCI equipment in the 3-phase world.

I guess as another question - Is your competitor using a NFO Drives Sinus model?

 

[InverterDuty]

Rhatcher,
I have tried turning the carrier frequency down with no luck, but it did reduce the leakage current.

Mikekilroy,
I spoke with an engineer at marathon, he told me to connect it with the negative side of the DC brake output and that's the only thing that will work

Itsmoked,
I have tried every possible ground configuration, It definitely made changes in current leakage, but never enough to make it run on a gfi.

Jraef,
These machines run on a class A 110V GFI the inverter takes care of the protection on the output side. Yes it does work with the special motor or this Chinese motor I have.

LionellHutz,
I am using every GFI available to my knowledge, they are all the 2006 5ma trip standard. Single phase 110V input.

I cant find the article I was reading but it talked about the way the rotors are constructed and how it effects the current leakage. The depth and configuration of the outer bars to be specific.

 

[InverterDuty]

Here is the article. I found it.

 

[LionelHutz]

Take the Chinese motor to a motor shop and dissect it.

Did you read what I posted about grounding? Hooking the motor ground to the VFD is useless unless the VFD has some way to couple the current in the ground back to the bus.

 

[jraef]

So now more of the story comes out; you have a 110VAC single phase input. That means the VFD has what is called a "Voltage Doubler" ahead of the diode bridge, which is a circuit that rectifies the AC to DC, then uses caps to boost the voltage to 325VDC to feed the DC bus of the VFD so that it can put out 230VAC to the motor. Some of these voltage doubler circuits can add a particular form of nasty to the input power, they are notoriously cheaply made. That right there may explain why one works and another does not. The main reason may be that your mfr uses a half-wave voltage doubler, the other mfr uses a full-wave version. Lots more noise on a half-wave, but they are cheaper to make and smaller. People use the half-wave version when they just slap it onto an existing VFD design because they will have the caps in the VFD to filter out the extra ripple anyway. The noise is an unwelcome but largely unnoticed side effect, however in your case it might be proving problematic.

"Dear future generations: Please accept our apologies. We were rolling drunk on petroleum."
— Kilgore Trout (via Kurt Vonnegut)
For the best use of Eng-Tips, please click here -> faq731-376

 

[mikekilroy]

How about try a 230v input vfd? if that solves your issue as it may, then you could consider using the lower cost 230v model with an input 110->230v isolation transformer for maybe the same overall cost? the xfmr isolation probably would totally solve the issue by itself too so you could use any motor....

Grounding one side of the DC bus will certainly not short anything out if it were 230v 3 phase wye or ungrounded delta input; being 110v input is a different story - yes, one side of the ac input is ground so you cannot ground one side of the dc bus on that system.

I spoke with an engineer at marathon, he told me to connect it with the negative side of the DC brake output and that's the only thing that will work
So do you have a dc brake output? what is that? never heard of such a thing, tell us more or call that engineer back and get the confused wording corrected?

 

[LionelHutz]

You could put a transformer between the GFCI and the VFD but nothing after the tranformer would be GFCI protected.

It would be OK to put a capacitor from the ground to one or both of the DC bus terminals. Make sure to pick suitably rated capacitors. I would expect you'd want ones that are 10x or 20x the winding the ground capacitance of the motor.


Mike - Did I post that it would cause a short on an ungrounded system??? Still, you do realize that every time the negative diode on a phase conducts the conducting phase will be grounded? What do you think will happen when there is another ground fault in the power system with the negative DC bus of the VFD grounded?

 

[rhatcher]

It may be that the 'DC Brake' InverterDuty refers to is the 'dynamic brake' (resistor)that is normally connected to the DC bus and used to prevent DC bus overvoltage during braking.

 

[LionelHutz]

It could be. Typically, the braking resistor connections have the positive connected to the bus and the negative connects to a transistor, which is then connected to the bus. So, it would be best going to the positive DC braking resistor terminal if trying to feed that coupling energy back into the DC bus inside the drive.

 

[mikekilroy]

I assumed the 'brake' might mean this but did not want to confuse the issue but instead said just try thru a protective fuse to dc bus MINUS side. I think I also said plus side like this is same exact difference as tying to minus side of bus.... there are details to be considered but for the all important testing ideas, a fuse covers protecting tying either plus or minus bus to motor ground wire like this. details like whether an input is already grounded - like it is if using 120v wall outlet power (so even THIS test cannot be done to dc plus) rather than 230v house power or 3ph why or non-corner grounded delta, the additional potential issue of getting a ground fault in the system, etc. It also may be the OP saying it must work on GFCI means what he said: he is feeding it from GFCI circuit so it must work this way, not that it must have GFCI MOTOR protection itself. And any decent vfd will have GFCI equiv built in anyway - called ground fault protection shut down, so this additional GFCI input is really not even required in my opinion - stick a 120->240v 1 phase isolation transformer in and save money with a lower cost 230v input vfd. Clarification on the actual reason for GFCI would help.

 

[InverterDuty]

I work for an oem manufacturer. We make hundreds of units per year. These machines go in schools and damp environments. The customers need the machines to run on 110v. Yes I am referring to dynamic braking resistor outputs. My company buys hundreds of vfds per year so I get them cheap. I have tried the following brands: KB,Hitachi,Teco,Allen Bradley, Emerson,Danfoss,Fuji and you know what... The best ones we have ever used are the Teco Units, over 2500 in the field with very little problems compared to 1000 KB units in the field with about a 10% failure rate. With my experience 220V drives are no cheaper than 110 Volt drives. I am paying under 100 for each 110V inverter, so I cant complain. Since this is an OEM production, I cannot do the easy fixes, like an isolation transformer. It just adds too much cost and I have Chinese competition to deal with as well.

 

[itsmoked]

Very interesting.

So where are the tripping GFIs? Are they random ones on the customer premises or one inside your product?

Keith Cress
kcress -

http://www.flaminsystems.com