I am running a 480 volt 3 ph circ 120 ft to an 11 amp load.The voltage drop I come up with is 4.5 volts.I f we run this circ. through a vfd would that affect the voltage drop value I have calculated?Also is there any limit on the distance a vfd can be from the load?I'm new to working with vfd's.Thanks .
voltage drop with vfd?
- Thread starter tommyrice
- Start date
- Status
- Location
- San Francisco Bay Area, CA, USA
- Occupation
- Electrical Engineer
There are several issues with motor lead length and VFDs, but voltage drop is voltage drop, the VFD does not change that. Conductor sizing is the issue with that.
The longer the distance from the drive to the motor, the greater the risk of damage to the motor winding insulation. There is a phenomenon called "Standing Wave Generation" that is caused by the high speed switching of the transistors in the drive output interacting with the capacitance of the motor leads to create very high voltage pulses (i.e. 2000V) at the motor which eventually punches through the insulation of the windings and causes shorts. The longer the lead length, the more capacitance and the greater the problem. This can be minimized in several ways.
- Keep the distance as short as possible by locating the VFD as close to the motor as you can, under 25ft is preferable (not always practical however).
- Use motors that are "Inverter Rated" because they will have insulation that is designed to fight the effects of long motor lead length.
- Use "VFD Lead Wire" which is a shielded cable that is designed to help mitigate some of the effects, as well as protect other equipment from the RFI that the motor leads will generate (another problem). The RFI issue can be solved by simply using steel conduit, but that does little for the standing wave generation.
- Use "Long Lead Filters" on the load side of the VFD. Some drive have them built-in, most do not because they are expensive (to them) and unnecessary if the drive is close to the motor. But several companies like TCI and MTE make retrofit filters. Load Reactors are a low-end type of filter and are a minimum, but there are better options available.
Also, not directly related to lead length but NEVER put multiple VFD output conductors in the same conduit with each other or with any other conductors, unless you use the VFD Lead Wire where each set has its own shielding, otherwise you get mutual inductance.
Open Neutral
Senior Member
- Location
- Inside the Beltway
- Occupation
- Engineer
It took me quite a while to find any description of the issue. Recommendation after recommendations talked "what" (No long leads) about the problem vs. "why".... (Because.....)
I was still thinking telecom-ish: the longer the loop/line, the more capacitive, & the more HF rolloff, so good. I'm still not sure I understand the causes, but....
{I can't find a 3-ph 60 Hz slug for my Bird wattmeter, or I could measure the VSWR....}
- Location
- San Francisco Bay Area, CA, USA
- Occupation
- Electrical Engineer
The "why" is the subject of very long and boring white papers, many of which are available on-line if you wade through all the commercial stuff.
Super condensed version:
To make the magic that is modern VFDs, the transistors switch on and off at a very high rate, as in 3 - 16kHZ. This means the individual pulses have a very steep wave front, meaning they go from off to on as close to instantly as possible. When that steep wave front sees capacitance in the circuit, the capacitance interacts to over charge just a tiny bit. But that over charging in each pulse in each line can interact with each other to form higher voltages superimposed on the desired RMS signal. These added voltages form into what is called "Standing Waves" which have a wave length of their own. The longer the lead length, the more likely the wave length of the Standing Wave has a chance to couple with the Standing Waves in adjacent motor leads and interact again to increase the voltage. The peak can be 3-4X the peak voltage in the circuit so it can easily exceed 2000V. Older motor insulation was typically only 2.5X the RMS line voltage for the design, so for example a 480V motor would have insulation rated at 1200V. 2000V pulses, 1200V insulation = punch through. "Inverter Rated" motors now typically have much higher rated insulation, inverter lead wire has more shielding to prevent the inductive coupling between phase conductors, load filters lower the rise time of the PWM pulses, etc. etc. All are reasonably good strategies to lessen the likelihood of failure, using them all is the absolute best way.
mike_kilroy
Senior Member
- Location
- United States
open nuetral, u got it. same phenomenon as our ham radio SWR on xmission lines. I sold a 100hp vfd system to a customer a few years ago and it had clear pvc cover over the shielded twisted 4 conductor output wire; after a year of service it had burn marks at the peaks of the swr freq they had - IIRC it was around 4" apart, so say 16" wavelength maybe (I kept a piece of it if anyone wants to see a picture)? so about 150mhz band 
each newer generation vfd seems to get faster and faster turn on/off times of the IGBTs. I havent so they are down in sub 0.1 usec times so well over 1Mhz googled turn on time for 20-50amp igbts lately but bet they are in the .05usec range? so 20mhz-sh fundamental frequencies plus their harmonics....
Have a customer right now with 200hp high performance spindles with input and output reactors. Unfortunately they opted to use single conductor shielded cable for the motor leads due to space issues, and we believe the extra capacitance of of all those wires to shields is the cause of their motor failures - voltage spikes measured at a tad over 4000 volts peak at the vfd PWM switching freq of 6khz! at the motor side of the output reactor and "only" 3500V on the input side.... The drive mfgr had them change to an input choke with single turn windings around each phase (like a transformer) that are now tied to 2500watt regen style resistors; this dropped the voltages to 2500 & 3200v......
AC8V
each newer generation vfd seems to get faster and faster turn on/off times of the IGBTs. I havent so they are down in sub 0.1 usec times so well over 1Mhz googled turn on time for 20-50amp igbts lately but bet they are in the .05usec range? so 20mhz-sh fundamental frequencies plus their harmonics....Have a customer right now with 200hp high performance spindles with input and output reactors. Unfortunately they opted to use single conductor shielded cable for the motor leads due to space issues, and we believe the extra capacitance of of all those wires to shields is the cause of their motor failures - voltage spikes measured at a tad over 4000 volts peak at the vfd PWM switching freq of 6khz! at the motor side of the output reactor and "only" 3500V on the input side.... The drive mfgr had them change to an input choke with single turn windings around each phase (like a transformer) that are now tied to 2500watt regen style resistors; this dropped the voltages to 2500 & 3200v......
AC8V
Last edited:
Open Neutral
Senior Member
- Location
- Inside the Beltway
- Occupation
- Engineer
There's a solid reason for that, that many but not all here will surely grok. Switch transistors have 3 operating regions:
- Off
- Inbetween
- On
If off, the voltage across it is high, the current is zero, so watts dissipated is ~zero
If on, the voltage drop is low, the current is high, so watts is not as low as zero but...
If partially on, the voltage drop is medium, the current is too, and the power is *%^&*...until the smoke escapes.
So you want to get that IGBT or FET through the linear region ASAP.... or else. To that end, on the electric car three of use are building, we actually have multiple stages of gate drive for the FETS. We're getting bigger FETS and may need ZXGD3004E6TA's driving the Si8233 we now use by themselves. We're talking gates pulses in the amps {not micro or milli regions....} to get there.
Of course, those transitions then make other problems including RFI.... But our motor control leads are only a few feet long...
Jrafe:
I did understand it was a SWR issue but that's insufficient help to me re: fully comprehending why it's trouble in this case, at least without too much thinking to be fun...I'd think convention L-C filtering of the output would resolve things.
- Status