[PStechPaul]

I just found this thread and I plan to watch it with interest. So far I'm just working on a little utility vehicle and a garden tractor for electric conversion, and I am using a DC bus with a standard VFD and three phase motor. I'm also thinking about overclocking the motor and using a higher voltage for a HP boost. I even have a small motor that I rewound from a single phase 120V to three phase 8 VAC 12 pole, and overclocked it to 4x.

One thing I wanted to comment on is the large capacitors you are using on these VFDs. If you are running on pure DC from batteries, you shouldn't really need very much, unless you are looking for short-term surge power. I have about 1600 uF on my 320 VDC link, with a 2 HP VFD and motor, and I figure that will give me about 82 W-seconds, or only about 50 mSec at 2 HP (1600 W). They are critical for a VFD running on single phase AC, of course, but with three-phase input you only need to hold the voltage near peak for about 1.5 mSec.

I'm using a DC-DC converter which transforms 12 VDC to 320 VDC, using a square wave at 500 Hz, with deadtime of only 2 uSec, so a full wave bridge would not need much capacitance at all. But I'm using a doubler circuit for the 12 VDC input, so I need some capacitance. But at 500 Hz, it's only for 1 mSec.

Looking forward to hearing more about your conversion projects. Someday I hope to do an electric conversion for a highway vehicle, but for now I'm happy enough with my 1999 Saturn SL1 which averages 35 MPG and can get over 45 MPG highway.

[subcooledheatpump]

The capacitors aren't for smoothing input AC or for smoothing battery current/voltage.

They are for absorbing voltage spikes during IGBT turn off. The Capacitance needed is proportional to the current being switched. More current means more capacitors.

Just in my personal experience, I've noted that switching 100 amps will require 36000 microfarad, thats what I have in my own modified VFD controller. I've blown quite a few IGBTs using less than that.

[PStechPaul]

I've tried to find more information on the sizing of the DC link capacitors and the failure modes of IGBTs. The following article explains quite a few details of a large VFD and it only describes the capacitors as needed for smoothing the rectified AC. But it also mentions snubbers for the IGBTs and inductors and filters for smoothing the voltage supplied to the motor. It also shows the use of a dynamic brake, which is used to dissipate the kinetic energy stored in a rotating motor when stopping.
http://www.danfoss.com/NR/rdonlyres/...vfdlesson3.pdf

Here is the list of the lessons:
http://www.danfoss.com/North_America...ial+Market.htm

Has anyone looked at the waveforms of drives to compare the effects of link capacitance? I have a feeling that the failures may be due to high frequency switching transients, which are best handled by snubbers. And there also needs to be some smaller non-electrolytic capacitors across the DC supply of the bridge, with very low inductance and ESR at high frequencies. You might be able to use, say, a 0.47 uF polypropylene or metal film capacitor to reduce the high voltage transients.

Here is a detailed analysis of IGBT characteristics and failures, and snubbers:
http://www.fairchildsemi.com/an/AN/AN-9020.pdf

And other possibly helpful resources:
http://www.toshont.com/ag/vfddesign/ag_15_igbts.pdf
http://www.toshont.com/vfddes.htm
http://www.toshont.com/ag/vfddesign/AG40EM.pdf (specifically about DC link capacitors)
http://www.electro-tech-online.com/e...-inverter.html (discussion of DC link bypass and snubber capacitors)

Some of this information may be especially important for DIY modified drives. I have learned a lot just from an initial scan of some of these documents.

[subcooledheatpump]

http://www.ebay.com/itm/Lot-Of-10-Ae...item2ebc947134

These are your best bet as far as film snubber capacitors. I personally use 3 of them in my modded VFD

Also about sizing the capacitors; thats a little tricky but it usually comes down to this: higher currents and higher switching frequencies means you need more capacitance. Since V=L*DI/DT, that means if we are switching more current (DI) and we are switching it quickly (DT) we will get more voltage at the IGBTs terminals and therefore more capacitance is needed to absorb it.

Slowing the IGBTs transistion from on to off works to reduce those DI/DT losses but that also means they will dissipate more heat

This is explained in the articles above, and also

www.mitsubishielectric.com/semiconductors/files/manuals/powermos4_0.pdf

Just to add to the list


If the bus inductance (L) is lowered that also means less capacitance is needed.

I always stick to the basics though, bigger is better. I'd rather play on the safe side and have too much capacitance rather than too little and end up with hundreds or thousands of dollars of destroyed IGBTs

[aeroscott]

Originally Posted by PStechPaul

I've tried to find more information on the sizing of the DC link capacitors and the failure modes of IGBTs. The following article explains quite a few details of a large VFD and it only describes the capacitors as needed for smoothing the rectified AC. But it also mentions snubbers for the IGBTs and inductors and filters for smoothing the voltage supplied to the motor. It also shows the use of a dynamic brake, which is used to dissipate the kinetic energy stored in a rotating motor when stopping.
http://www.danfoss.com/NR/rdonlyres/...vfdlesson3.pdf

Here is the list of the lessons:
http://www.danfoss.com/North_America...ial+Market.htm

Has anyone looked at the waveforms of drives to compare the effects of link capacitance? I have a feeling that the failures may be due to high frequency switching transients, which are best handled by snubbers. And there also needs to be some smaller non-electrolytic capacitors across the DC supply of the bridge, with very low inductance and ESR at high frequencies. You might be able to use, say, a 0.47 uF polypropylene or metal film capacitor to reduce the high voltage transients.

Here is a detailed analysis of IGBT characteristics and failures, and snubbers:
http://www.fairchildsemi.com/an/AN/AN-9020.pdf

And other possibly helpful resources:
http://www.toshont.com/ag/vfddesign/ag_15_igbts.pdf
http://www.toshont.com/vfddes.htm
http://www.toshont.com/ag/vfddesign/AG40EM.pdf (specifically about DC link capacitors)
http://www.electro-tech-online.com/e...-inverter.html (discussion of DC link bypass and snubber capacitors)

Some of this information may be especially important for DIY modified drives. I have learned a lot just from an initial scan of some of these documents.

A few hours to get threw the Danfoss lessons , thanks good find!

[PStechPaul]

Something else to consider is line and link reactors. They may not be as important for a battery powered DC link, but it would be worthwhile investigating. I searched again for information on link capacitors, and I did not find much more, but these came up:
http://emainc.net/newsletter/compari...dc-link-choke/
http://www.pge.com/includes/docs/pdf...e_reactors.pdf
http://igor.chudov.com/manuals/Sieme...Capacitors.pps

The last link is a PowerPoint presentation that, in page 10 and page 16, describes problems caused by higher link capacitors, and they use a 20 uF plastic capacitor rather than a more typical 1000 uF capacitor in their 15 HP drive. It shows how this lowers the peak switching current and therefore reducing high frequency transients. Here is another document that explains more. I think it is more about transients and harmonics caused by AC input switching, however:
http://www.us.sbt.siemens.com/HVP/Co...WhitePaper.pdf

Here is an application note from Microchip that explains how insufficient link capacitance may cause an overvoltage condition due to Regenerative Braking. With a battery bank, this might not occur, but if there is a diode between the batteries and the DC link, or if there is significant lead impedance, destructive high voltages could be generated.
http://ww1.microchip.com/downloads/e...tes/00887a.pdf

Here is an IEEE study of an alternate connection for regeneration ultracapacitors for greater efficiency. It's rather technical but might be helpful, especially since it directly addresses vehicular applications:
http://scholarsmine.mst.edu/post_pri...cc805e9646.pdf

This is very interesting:
http://www.nrel.gov/vehiclesandfuels...c_link_cap.pdf

I also found this youtube demonstration featuring a three-phase motor conversion:
http://youtu.be/2tFSVniO_Cc

Enough for now!

[subcooledheatpump]

Some good info.

That video you posted is actually etischers conversion, he's posted it on this forum too.

There are also videos on youtube of my own working 3 phase VFD conversion

http://www.youtube.com/subcooledheatpump

[PStechPaul]

Interesting. I posted a comment on your video. I'm working on an electric conversion for a riding mower / utility vehicle, and once I am comfortable with the design, I may scale up to a road legal vehicle. I am convinced that the induction motor is the best choice overall for EVs. I'd be interested in seeing your complete schematic and other technical details. The 600 volts of batteries would scare me. And I bet they are expensive, probably about $1500 for 10 kWH. I'd use 10 90 A-H deep cycles for about $800.

There seem to be many different approaches to electric conversions, and I'm trying to look at the big picture to see where improvements might be made and where the existing technology is more or less efficient. There are always tradeoffs and it's hard to say what is really "best". If you haven't seen my own little three-phase conversion, here's my video:

http://youtu.be/y0qWY4bVnEA

[aeroscott]

Originally Posted by PStechPaul

Something else to consider is line and link reactors. They may not be as important for a battery powered DC link, but it would be worthwhile investigating. I searched again for information on link capacitors, and I did not find much more, but these came up:
http://emainc.net/newsletter/compari...dc-link-choke/
http://www.pge.com/includes/docs/pdf...e_reactors.pdf
http://igor.chudov.com/manuals/Sieme...Capacitors.pps

The last link is a PowerPoint presentation that, in page 10 and page 16, describes problems caused by higher link capacitors, and they use a 20 uF plastic capacitor rather than a more typical 1000 uF capacitor in their 15 HP drive. It shows how this lowers the peak switching current and therefore reducing high frequency transients. Here is another document that explains more. I think it is more about transients and harmonics caused by AC input switching, however:
http://www.us.sbt.siemens.com/HVP/Co...WhitePaper.pdf

Here is an application note from Microchip that explains how insufficient link capacitance may cause an overvoltage condition due to regenerative braking. With a battery bank, this might not occur, but if there is a diode between the batteries and the DC link, or if there is significant lead impedance, destructive high voltages could be generated.
http://ww1.microchip.com/downloads/e...tes/00887a.pdf

Here is an IEEE study of an alternate connection for regeneration ultracapacitors for greater efficiency. It's rather technical but might be helpful, especially since it directly addresses vehicular applications:
http://scholarsmine.mst.edu/post_pri...cc805e9646.pdf

This is very interesting:
http://www.nrel.gov/vehiclesandfuels...c_link_cap.pdf

I also found this youtube demonstration featuring a three-phase motor conversion:
http://youtu.be/2tFSVniO_Cc

Enough for now!

very good stuff. Thanks , I would like to see what happens to the harmonics with the 12 and 18 diode system and how they are implemented . I'm just having a hard time seeing how more diodes would change things , but it is obviously what happens , just more expensive seams to be the reason for not using them . Funny it's called a 6,12,18 step inverter just because of the diode count .

[PStechPaul]

Originally Posted by aeroscott

very good stuff. Thanks , I would like to see what happens to the harmonics with the 12 and 18 diode system and how they are implemented . I'm just having a hard time seeing how more diodes would change things , but it is obviously what happens , just more expensive seams to be the reason for not using them . Funny it's called a 6,12,18 step inverter just because of the diode count .

My understanding is that they use additional transformers to obtain 6 phase or 12 phase power, with 60 or 30 degrees per phase. The magnitude of ripple drops and the ripple frequency goes up, making it easier to filter. But I don't think it makes that much difference, especially if you are generating an output voltage lower than the input.

http://www.allaboutcircuits.com/vol_3/chpt_3/4.html
http://www05.abb.com/global/scot/sco...ce%20sheet.pdf
http://www.joliettech.com/abb_evalua...-ac-drives.htm