• Scam Alert. Members are reminded to NOT send money to buy anything. Don't buy things remote and have it shipped - go get it yourself, pay in person, and take your equipment with you. Scammers have burned people on this forum. Urgency, secrecy, excuses, selling for friend, newish members, FUD, are RED FLAGS. A video conference call is not adequate assurance. Face to face interactions are required. Please report suspicions to the forum admins. Stay Safe - anyone can get scammed.

VFD Output Filtering

Susquatch

Susquatch

Ultra Member
Administrator
Moderator
Premium Member
I was looking at my VFD Output Voltage Characteristics on an oscilloscope recently and noticed some very high harmonic spikes. The spikes are not surprising - in fact expected given the square wave characteristics of the VFD output voltage. But I didn't expect voltage level spikes over 2000 volts.

So I sent an Email to TECO - the manufacturer of my VFD to ask about it. Rather than reply to my email, they called me.

The fellow said such spikes were typical of short cable runs between the VFD and the motor and especially in higher quality motors such as inverter ready motors like I have because they don't dampen the spikes and instead are designed to tolerate them. I'm not really surprised by any of that either. It's really just the very high voltage levels that surprised me.

But what he said next totally caught me off gaurd. He recommended 3 phase OUTPUT filtering! At first, I was stunned. I have always assumed that the VFD measures current flow and back EMF from the motor and uses that information to set output parameters to optimize motor performance. He didn't disagree. However, he said that designing a filter with a cutoff frequency at least twice the carrier frequency would kill the spikes but still allow the current flow at the carrier frequency voltage to reach the analytical portion of the sensorless vector circuitry in the VFD.

So that's what I plan to do. I guess we will see how well that works.

I'm also curious to know if anyone else has had similar experiences or has tried output filtering. Please note that this is not input filtering on the single phase input circuit which is very common and in fact I am already doing that. This is output filtering on the 3 phase power from the VFD to the motor.
 
You probably want to look at 'line reactors'. As i understand it, they are a 1:1 transformer. Drive setup is done before installing the line reactor, to set motor parameters.

I was going to install one on my 10hp lathe due to some cogging issues i was having, but i was able to tune them out. I have a big input filter due to burning out light bulbs in my house whenever i'd use the lathe.
 
Darren said:
You probably want to look at 'line reactors'. As i understand it, they are a 1:1 transformer. Drive setup is done before installing the line reactor, to set motor parameters.

I was going to install one on my 10hp lathe due to some cogging issues i was having, but i was able to tune them out. I have a big input filter due to burning out light bulbs in my house whenever i'd use the lathe.
Click to expand...

Yes, I have a good input filter too. There is nothing on the input circuits at all.

That's a very good comment about letting the VFD set the motor parameters before installing anything into the output. I would not have thought to do that.

I will look into line reactors. I'm guessing they are basically a simple transformer that would no doubt kill the harmonic spikes. But I would be worried that they might also kill the carrier frequency. I suspect they are mostly meant to be used in standard voltage/frequency control mode or in sensor feedback mode rather than in sensorless vector mode. They might even be intended for use in a non-frequency speed control mode where the VFD is only used to create 3phase output but doesn't change the frequency to change speed. I'll let you know what I find.
 
Darren said:
Line reactors go before the vfd, LOAD reactors go between VFD and motor.
Click to expand...

OK, my read of that article goes like this.

Said load reactor is the same thing as the output filter I described earlier.

However, the article suggests that the need for output filtering doesn't arise until output wiring lengths exceed 100 ft for non-vfd ready motors and 300 ft for vfd ready motors.

The TECO guy specifically said that my spikes were the result of short wiring. Mine is about 6 ft. On the other hand, the article suggests that it is reflections on long wires that cause the spikes and harmonic noise.

I suppose it's possible that my VFD Ready motor doesn't filter spikes and just reflects them instead and it's also possible that I am just concerned about nothing.

Nothing in the article mentions the VFD control mode. I had thought perhaps it is an old publication, but although the article is not dated, it references other documents as recent as 2017. I do not know when sensor less vector control technology was developed. I'd guess at 2010 or so, but perhaps much earlier. Nonetheless, it strikes me as very odd that the article doesn't mention it or its implications.

It turns out that the filter that TECO recommended is also called a "Line/Load Reactor". Apparently the one device handles both input and output. But it is $250 without a cabinet. I don't think that's in the cards.

It doesn't appear to be a high enough low frequency cutoff to serve as an effective sensorless vector device either. I'd rather try and make one first. Who knows, that might become even more expensive.

Thanks for the info Darren. I'll keep this thread up to date with anything else I find or do.
 
If he's seeing 2000v spikes, the concern would probably be potential damage/breakdown of the motor insulation and possible damage to the drive.
 
David_R8 said:
Help educate us; what problem does the output filtering solve?
Click to expand...
The high voltage spikes but don't follow what the idiot cough42 says. "We're not going to wrap them around just run them through the cores".

Oh but do click on my links because I get paid when you do that. And I make these very long and boring because I get paid more if they are long.

Take a look at TECO's data sheet for their filters.

The reason the cores are as large as they are is because if the current that generates the magnetic field is too high then the core saturates and the inductance created by the winding drops to near zero and then the filtering effect drops to zero. So for example running the motor on the bench with no load is not an indication of the noise under full load. (Another reason why that video is silly)

I had to experiment with coil sizes back in the 90's when I designed this Sequential Fuel Injection and capacitive Ignition system for Honda 1.5L Engines modified with a belt drive to serve on Home Built EIA aircraft and Hovercraft.

IGN200-Inside.jpg


In this case there are two toroids, stacked with, IIRC, 88 turns of wire to handle at least 10A at 8V without saturating with enough inductance so when the current stopped the capacitor would charge up to almost 300V. I went through multiple toroid combinations until I found something that was not too heavy but also met the specifications.

If you find an old transformer with a current rating of your motor and remove the windings you can create something like the ones you see in the above tagged PDF.


My two cents...
 
Yes @Darren . That plus I just don't like the idea of spikes that high. I was shocked to see that.

TECO didn't seem overly concerned or surprised about it though.
 
jcdammeyer said:
The high voltage spikes but don't follow what the idiot cough42 says. "We're not going to wrap them around just run them through the cores".

Oh but do click on my links because I get paid when you do that. And I make these very long and boring because I get paid more if they are long.

Take a look at TECO's data sheet for their filters.

The reason the cores are as large as they are is because if the current that generates the magnetic field is too high then the core saturates and the inductance created by the winding drops to near zero and then the filtering effect drops to zero. So for example running the motor on the bench with no load is not an indication of the noise under full load. (Another reason why that video is silly)

I had to experiment with coil sizes back in the 90's when I designed this Sequential Fuel Injection and capacitive Ignition system for Honda 1.5L Engines modified with a belt drive to serve on Home Built EIA aircraft and Hovercraft.

View attachment 27057

In this case there are two toroids, stacked with, IIRC, 88 turns of wire to handle at least 10A at 8V without saturating with enough inductance so when the current stopped the capacitor would charge up to almost 300V. I went through multiple toroid combinations until I found something that was not too heavy but also met the specifications.

If you find an old transformer with a current rating of your motor and remove the windings you can create something like the ones you see in the above tagged PDF.


My two cents...
Click to expand...

Might only be 2 cents from your perspective but totally spot on.

I had thought about using three transformers directly, but it requires testing like you describe to determine their real world performance for the range of loads my VFD and motor might see. I think it's prolly better to do some analysis and calculate the required parameters and then just buy that. It's either that or spend the money on the OEM filters.

I guess my hesitance is driven by the fact that there has to be thousands of similar systems to mine out there and they probably have the same problem but nobody even knows it because they don't have an oscilloscope.

Perhaps as @David_R8 might be implying, I'm just being a worry wart for nothing.
 
Last edited:
Oh and if you are going to roll your own note the comment that a 5% filter is reduced to 3% if the current is 60% of the rated current.
Here's a schematic of the basic filter.
emitech.co.in

EMI / RFI / EMC Filter manufacturer

EMI / RFI / EMC filter manufacturer from India, with over 30 years of expertise in EMI Filter and EMC consultancy.
emitech.co.in
 
As I said it's been a few decades since I did the ignition. I've forgotten how to do all the calculations but I'm sure the information is out there. I'd tear down a surplus transformer like one out of a microwave. If it is the classic E Type transformer core
Then just wind say 10 turns on each vertical leg like that RL50003 photo in the Teco data sheet. I'm not sure you need the capacitors but I'd connect the frame to earth.
 
@Susquatch Here's an example of a simple circuit to measure inductor current and coil saturation. You have a scope. If you don't have any sort of pulse generator here's a perfect little project for an Arduino to tickle the transistor to turn on the current.

electronics.stackexchange.com

How to measure Inductor maximum current?

I have an inductor and seller claims that inductor supports a maximum of 5A current and he asked me to test it myself. Can someone help how to test that how much this inductor can support/ draw max...
electronics.stackexchange.com electronics.stackexchange.com
 
jcdammeyer said:
Oh and if you are going to roll your own note the comment that a 5% filter is reduced to 3% if the current is 60% of the rated current.
Here's a schematic of the basic filter.
emitech.co.in

EMI / RFI / EMC Filter manufacturer

EMI / RFI / EMC filter manufacturer from India, with over 30 years of expertise in EMI Filter and EMC consultancy.
emitech.co.in
Click to expand...

I guess my only problem with the filter in your suggestion is the low frequency cut-off of 60 Hz. I would think it needs to be more like 16KHz to allow the carrier frequency to get through even if it is a bit rounded. Perhaps 60KHz would be better to maintain more of the square wave structure. I didn't really look at how fast those high voltage spikes were but they were definitely well above the carrier frequency as they were just spikes sitting on the edges of the carrier square wave - perhaps something like 100KHz or even faster with a tiny bit of ringing.
 
With a bit of experimenting with a transformer wound the way I suggested you can determine e the inductance znd therefore the reactance at 16 kHz.
 
Some quick math. Assuming you want a bit of room above yhe 16khz frequency I used this web page for a pi filter

Simple LC Low Pass Filter Circuit Design & Calculations » Electronics Notes

Design considerations, circuit and formulas for a constant-k 3 pole LC low pass filter for RF applications.
www.electronics-notes.com www.electronics-notes.com

So freq is 18khz and I guessed at an inductor with value 0.000050 Henrys or 50uH.
That works out to impedance of 2.83 ohms.
Using that value then gives me 0.00000625 Farads or 6.25uF.
A 6.3uF 900V cap runs about $5.86 at Digikey. C4AQOBU4630M11J

Ferrite cores aren't that expensive but single coils won't work as well as all three on the same core. Hence winding on a transformer core.

I think that's shy you want to make one like a common mode choke but for thee phase.

Been a long time since I did anything with this sort of stuff.
 
jcdammeyer said:
But single coils won't work as well as all three on the same core. Hence winding on a transformer core.
Click to expand...

You lost me on that one. Why do you say that?

Do the phases feed on each other in some way that's not obvious to me?
 
You must log in or register to reply here.
Back
Top