Electrician Talk banner
1 - 20 of 22 Posts

·
Registered
Joined
·
368 Posts
Discussion Starter · #1 ·
I am working on a project to power an environmental chamber that requires 208. We only have 240 single phase. The chamber is wired delta with fan motor on one phase, a chiller compressor on the second phase, and a resistive heater on the third phase. The loads are not balanced and the chamber specifies a maximum of 20A load on each phase. It is expected that there will be cases during operation where there will be no load current on multiple phases. We bought a 7.5 kW LSIS VFD , as well as some inductors and capacitors to use to filter out the VFD noise.

To check out the VFD and filter set up, we just connected the filter to the VFD with no load. Just after start up, the VFD trips at 1 Hz and 150A with either 120V or 240V input. We think the filter inductors are saturating and the transient current through the 2.2 uF capacitors is then getting very high.

Our inductors are toroid ferrite cores with 3 windings each 470 uH 25A rated. Each weighs about 0.4#. We put 6 ( about 2.4# total) in series between the CFD and the capacitors. The capacitors were wired from each phase to a common node and then across each phase with no difference in VFD tripping.

Most filter inductors you buy for this purpose are steel laminated e-cores with 3 windings. A 2400 uH 25A rated one of these bad boys weighs about 28#.


Could our ferrite inductors be saturating right away?

Ferrite saturates much easier than laminated steel, but we wanted to save weight and volume. We will connect the capacitors to the VFD output without the inductor string to see if the VFD still trips out. We also will measure the capacitor current to see if anything interesting is going on.

With no load, will we have to worry about very high voltages on that might cause damage?

The LC filter resonates at about 2 kHz and the VFD chops at around 4 kHz.

Will the VFD PWM frequency be constant so we don't have to worry about it running at the filter resonant frequency?
 

·
Chief Flunky
Field Service Engineer
Joined
·
3,305 Posts
I am working on a project to power an environmental chamber that requires 208. We only have 240 single phase. The chamber is wired delta with fan motor on one phase, a chiller compressor on the second phase, and a resistive heater on the third phase. The loads are not balanced and the chamber specifies a maximum of 20A load on each phase. It is expected that there will be cases during operation where there will be no load current on multiple phases. We bought a 7.5 kW LSIS VFD , as well as some inductors and capacitors to use to filter out the VFD noise.

To check out the VFD and filter set up, we just connected the filter to the VFD with no load. Just after start up, the VFD trips at 1 Hz and 150A with either 120V or 240V input. We think the filter inductors are saturating and the transient current through the 2.2 uF capacitors is then getting very high.

Our inductors are toroid ferrite cores with 3 windings each 470 uH 25A rated. Each weighs about 0.4#. We put 6 ( about 2.4# total) in series between the CFD and the capacitors. The capacitors were wired from each phase to a common node and then across each phase with no difference in VFD tripping.

Most filter inductors you buy for this purpose are steel laminated e-cores with 3 windings. A 2400 uH 25A rated one of these bad boys weighs about 28#.


Could our ferrite inductors be saturating right away?

Ferrite saturates much easier than laminated steel, but we wanted to save weight and volume. We will connect the capacitors to the VFD output without the inductor string to see if the VFD still trips out. We also will measure the capacitor current to see if anything interesting is going on.

With no load, will we have to worry about very high voltages on that might cause damage?

The LC filter resonates at about 2 kHz and the VFD chops at around 4 kHz.

Will the VFD PWM frequency be constant so we don't have to worry about it running at the filter resonant frequency?
Depends on the VFD as far as carrier frequency. Not sure which one you bought. Read the manual.

Ferrites have a band pass effect and effectively are limited to several tens of KHz to Megahertz ranges. Good for EMI but terrible for power filters. Yes you have higher mu but it’s very frequency dependent. Siliconized steel doesn’t have that problem. The thinner it gets the smaller the domains which helps reduce saturation but at higher expense and mu rarely gets over 100. If you truly want excessively high and high bandwidth then the best way is using nano crystalline cores which can get up to 30,000-80,000 with reasonably high bandwidth but it has serious saturation limits considering the extremely high mu. I’ve had to use huge stacks of the stuff in filters handling tens of Amps. So may want to do this in stages. Ferrites OK for say 50 KHz or higher. Iron cores good for low frequency but it’s hard.

AND do not for a minute discount the REAL inductor model. You can’t avoid the resistance. In conjunction with capacitors you can easily end up with effectively a dead short. Running a frequency sweep and measuring filter performance can fool you because magnetics are pretty nonlinear.
 

·
Registered
Joined
·
14,573 Posts
This sounds like a flaming trash receptacle in the making, actually sounds like the device was a dumpster fire from the start lol.

Was lets rig this onto a VFD the first stop on the idea train or was there a plan A? Are all the parts just line-line 208 single phase?
 

·
Registered
industrial E,I&C
Joined
·
4,230 Posts
Some of the powerflex drives can handle single phase output as they are used to control resistant heaters. This makes sense as you can use the vfd to control the heat.

I would rewire a machine rather than take this approach but its interesting so carry on and keep us updated as we may learn something.
 

·
Chief Flunky
Field Service Engineer
Joined
·
3,305 Posts
The loads are hot to hot; delta. Why are load side caps on the VFD a bad idea if they are after the series inductors?.

Aren't most post VFD filters series L shunt C?
It’s a general rule. The problem is that at some point eventually the VFD will excite the resonant frequencies of any LC circuit. True LC circuits that are used are typically RLC circuits that include a resistor to prevent resonance. That’s why you never see a pure “naked” LC filter used. Pure LC filters are OK for radio circuits and other things where ferrites belong, not power distribution.

A real (non-ideal) inductor has a significant series resistance that you cannot ignore. It also has significant parallel capacitance and a somewhat insignificant parallel resistance. This is in their ideal operating range without getting into significant frequency rolloff or core saturation. When working with inductors you need to be in communication with manufacturer or distributor technical support and there is always experimentation and tweaking involved I know the ferrite application guy for MH&W in this area quite well. He would have told you up front not to do what you did. You are way outside their design envelope taking what amounts to a very nonlinear RF device and using it at low frequencies. It is great for Megahertz or even tens if Kilohertz where the “big iron” cores will not work. That is why practically every computer power cord has a bug ferrite core molded into the cable.

This is a classic case of don’t reinvent the wheel. I have no problems with DIY filters and you can often build them better or cheaper than the commercial ones, or at least with far less advertising and pure snake oil. All of the commercially available filters are public information and well documented in the research literature. You will NOT get truth out of the commercial manufacturers though because they protect their profit margins.

Look at a “sinus wave filter”. They have a 3 dB corner around 180 Hz and they do what you want. They are very physically large and heavy because of the inductors that are most of the filter. The filter is about the same physical size as the VFD itself and eats about 5% of the output power so cooling is a consideration. It is probably cheaper to just buy one than build one. The exact design data is freely available in IEEE papers on IEEE Xplore if you have access through a library. No point in reinventing the wheel. This is if you try to “brute force” this. You can pull abstracts for free but accessing any IEEE papers means going through the pay wall. Nonmembers without a subscription pay typically around $30-50 per document.

Second you went the wrong way from the start and bought the wrong drive. So I’m assuming you started with a 6 pulse drive. When I’m training customers this is how I explain 6 pulse drives. Don’t think of the output as a “dirty” sine wave. Those pretty PWM pictures are totally wrong. With todays space vector modulation in particular those pictures are wrong. The reality is that on each phase a six pulse drive can only output one of two signals: +DC bus, or -DC bus. A true zero only exists in neutral point clamped drives because the back diodes continue to carry current. Sure we can run some hellacious 12 KHz switching frequencies and make very pretty sine waves that are trivially easy to filter with a proper filter because we are pushing all the harmonics to high frequencies. But we are making the harmony s from hell but that comes with two coats. IGBTs mostly only have losses during switching so drive efficiency will be terrible. A further cost is that if the output is unfiltered or relies on motor inductance we will have massive abc severe problems with reflected waves. High carrier frequencies was a thing in the late 1990s and we all learned to avoid it.

Instead what I teach in training is that the drive outputs a rectangular wave. It might kind of superficially look like a sine wave but nothing can be further from the truth. Don’t try to dig into the output pattern. Suffice to say that it’s a rectangular pulse pattern designed to operate a motor. We are well beyond the idea of simulating sine waves anymore. Todays drives use space vector modulation even in V/Hz mode. In PWM each phase is independent. We output positive or negative DC bus making decisions at a carrier frequency to match the average output of a sine wave. In space vector modulation we recognize that with all 3 phases there are just 7 possible output combinations. We choose the output pattern closest to the desired phase angle and voltage (vector) we want to send to the motor, plus some additional decisions where we try to do things like minimize audible noise or minimize cable reflections. Space vector switches a lot less and gives much cleaner output as a result. So it is a huge mistake except as a very basic “drives 101 theory” to use the PWM concept.


If you had started out wanting a clean power output then you would have gone in a completely different direction. If you stuck with a 6 pulse drive and if you use a DC link choke they are cheap and 1/4 of the size and weight of any sort of load side filter. In fact the relative merits of different filters are hotly debated. It is hard to argue for an active front end in a non-regenerative application over a DC link choke or additional front end filtering. This alone would have dramatically improved the output since you get approximately a 2 stage filter for very low cost since the marriage of the DC choke plus the DC bus capacitor gives you an LC filter with a very low 3 dB and it is already manufacturer tested so no wheel reinventing. You can get very close to the sinus wave filter output with just this for a much lower size, weight, and cost. Many VFD manufacturers bring the terminals out for this including LS.

If this isn’t enough, use a “clean power drive”. These use a phase shifting transformer followed by between typically 18 to 36 IGBTs where each step is a much smaller phase angle than the usual 60 degrees for a 6 pulse drive. The result is a much cleaner output. It is the same method used in “true sine wave” double conversion UPS systems. With one of these drives your project, whatever it is, would already be done. I will warn you ahead of time though that a clean power drive basically sums the outputs of multiple smaller six pulse phase shifted modules and requires the special multiple secondary transformer. The drives are surprisingly large and heavy for their size, and the price will be 2-3 times more. Not to say there is anything wrong. They just are not cheap for good reason. I would never recommend a clean power drive except for clean power.

Finally we haven’t discussed application. Someone will probably mention rotary phase converter or static phase converter. These work but output is not all that clean. It is mostly harmonic free but getting balanced voltages and phase angles is unlikely, even in a rotary phase converter. The ultimate in “clean power” output is an MG (motor generator) set. This is where you have a motor driving a generator. Generators can be asynchronous but most are alternators (aka synchronous generators). If it’s a DIY project you might be able to find a surplus generator and motor far easier and less costly than a big clean power VFD.

To give you an idea how good MG sets are most plants that have to test foreign voltages and frequencies such as US manufacturers use MG sets to produce say 50 Hz power from 60 Hz. All larger transformer plants use MG sets in their test cells to make 120, 180, or 240 Hz. They use the higher frequencies in testing transformers to prevent the core from saturating. This is part of standard testing requirements. None of them use VFDs.

LS would have steered you way away from the direction you took. I sell and install a LOT of LS drives. It sounds to me like you are attempting to fit a square peg in a round hole,
 

·
Registered
Joined
·
368 Posts
Discussion Starter · #15 ·
Thanks for all the information. It might be worthwhile looking into rewiring for three single phase loads if we can make the internal controller work.
Paul, I am leaning towards the idea that our small sized inductors are saturating. We should be able to see that happen while monitoring the capacitor current.
 

·
Registered
Power distribution and controls
Joined
·
1,790 Posts
Your going have a TON a problems with the heaters!

I did a single phase to 3 phase wafer machine and the VFD just crapped. So I called some old work mates at Eaton and I installed a VFD isolation transformer, backwards. Raised the min HZ up to about 66 and the project was off and running.

Wafer machine was used by the nuns to make money. My boss a Catholic kept telling me we were all going to hell if the project failed. With some help from some real electrical engineers the application worked flawlessly. Except for the extra 3 grand for the transformer.

Using a drive to make 3 phase works fine on rotating equipment when size correctly.


You have single phase loads and the drive is not going to like them. At least my understanding is single phase loads.
 

·
Chief Flunky
Field Service Engineer
Joined
·
3,305 Posts
Thanks for all the information. It might be worthwhile looking into rewiring for three single phase loads if we can make the internal controller work.
Paul, I am leaning towards the idea that our small sized inductors are saturating. We should be able to see that happen while monitoring the capacitor current.
If you really think so do this. Take extra cores. Run a single loop of fairly large wire, say 1/0, through the existing and extra cores to magnetically link them. If readings improve, there’s your result. Plus you should see the result with a scope on the output…flat topped waves.
 

·
Registered
Joined
·
368 Posts
Discussion Starter · #20 ·
Finally got back to this project. Using a Danfoss 130B2446 sine wave filter with the 2.2 uF capacitors worked very well. The VFD is quite happy driving the motor, compressor, resistive heater load we have in our equipment. The Danfoss inductor is heavier than the ferrites we started with and runs at 47C vs 90C for the ferrite inductors. So there is more core and the laminated steel is much harder to saturate than ferrite. The ferrite has no problem with low frequencies but it is easier to saturate which is what caused the problems we were having.

We also had the the SSRs driving the heater blow up because they had to deal with multiple zero crossings in the noisy poorly filter VFD output. With a good filter we longer kill SSRs!!


Thanks for all the advice, we learned a lot on the project and I would have no concern about driving an unbalanced load like this with a VFD.
 
1 - 20 of 22 Posts
Top