100kW AC Inverter Project (Opensource)

Technical discussion on converting internal combustion to electric
bga
Senior Member
Posts: 492
Joined: Mon, 01 Sep 2008, 19:27
Real Name: Bruce Armstrong
Location: Perth WA

100kW AC Inverter Project (Opensource)

Post by bga » Fri, 10 Oct 2008, 05:33

Hi Johny,

Well spotted.

Yes, I'll call Mike at Fastron tomorrow and order one. It sounds about ideal for a smallish motor drive.

The PINout on the module suggests that it'll need a PCB for make the I/O bolt friendly. Better get the tables out and see how much copper I need for 100 Amps.

I am working on my own gate driver. It's not that hard after a couple of days research. Most parts sourced for a 2 Amp driver that can be adapted to 8 Amps easily. It wil use a HCPL-316J, a 2A isolated driver with desaturation feedback and should work just like a bought one.

Also have to get back to the main processor board and lay it out.

Cheers

BGA


Last edited by bga on Thu, 09 Oct 2008, 18:35, edited 1 time in total.

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

100kW AC Inverter Project (Opensource)

Post by Johny » Fri, 10 Oct 2008, 15:31

bga. Sounds good. If I compare their IGBT bridge Gate input capacitance with other IGBTs I've looked at:
Mitsubishi CM100DU-24F 39nF
Mitsubishi CM100DU-24H 15nF
Fastron BSM 100 GD 120 DLC 6.5nF

Which makes the Fastron way more sensitive so Gate drive can be lots less. The Gate charge figures bear this out as well. Your 2A driver should work as is.

During my VFD searching I've often wondered if some VFD manufacturers don't include desat protection in their Gate drivers. Based on the number of TECO app notes concerning IGBT blowouts I reckon they are suspects.

BTW I have a 24 pin DIP device called an MC3PHAC. It's basically a single chip 3 phase controller that does speed control only. It's VERY simple to set up (just caps and resisters) and provides gate drive at CMOS levels from a 0-5 volt pot input. I thought it would be a useful stepping stone to a DIY drive. It you want it - or details, just to prove the high power part of the system before commissioning your own DSP code, let me know. It will provide a sort of reference point at least.

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

100kW AC Inverter Project (Opensource)

Post by Johny » Fri, 10 Oct 2008, 16:12

I missed that Gate driver. The following isn't very useful to you but shows how much time I wasted looking for Gate drivers.

Manufacturer     Device        Price (USD ea)   Total Price     Via Powerex             VLA500               
Powerex        VLA504               
Mitsubishi      M57958L           35.24            211.44             email
Microsemi       APTRG8A120G   210               630             email
Omnirel        OM9401SD                    
On Semiconductor MC33153      2.45           14.7             octopart
IXYS            EVBD4400      148            444              digikey
IR             IR2114SS      15.01           45.03     
Powerex      BG2B-5015     114.43     343.29     newark

The HCPL_316 looks perfect (other than you have to provide the DC-DC convertor).

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

100kW AC Inverter Project (Opensource)

Post by Johny » Fri, 10 Oct 2008, 18:24

bga. Capacitors. The problem with using lots of small ones is that try to solder them to a solid copper bus is going to be difficult. I checked though my stuff and found the price I got from Richardson.
This is for 500 VDC 6800uF Nippon Con.
18.8 Amps ripple current at 105 degrees and 120 Hz.

----------------------------------------
Item   Qty Part Number                     Br Unit Price Extension Delivery
   1      5 ELXA501LGC682MF ELXA501LGC682MF UC      75.10     375.50 8 Weeks
            ELECTROLYTIC CAPS
            PRICE COMMENTS: United Chemi-Con
            ELXA501LGC682MFK0M
            5pcs moq/mults
            nc/nr
            DELIVERY COMMENTS: Lead time 6-8 weeks

----------------------------
At $75 each it's very good.
This was on 31st July 2008 so the lead time may be less now?
Do you want me to follow up and recheck lead time and whether the min. order is 5?
Last edited by Johny on Fri, 10 Oct 2008, 07:26, edited 1 time in total.

bga
Senior Member
Posts: 492
Joined: Mon, 01 Sep 2008, 19:27
Real Name: Bruce Armstrong
Location: Perth WA

100kW AC Inverter Project (Opensource)

Post by bga » Fri, 10 Oct 2008, 20:32

Johny,

I have contacted Mike Trubridge at Fastron, he is sending some info.

The MC3PHAC, that's impressive. I think that the PIC is the best choice, the PIC looks to be a lot faster and is fully programmable, allowing phase currents and all of the strange EV things to be accomodated.

I think that I've had a similar experience with gate drivers. Powerex was OK value through richardson's. I was coming to the DIY position, though.
Then...
One of my associates here in WA is in the process of designing an IGBT driver for exciters on the 2MW gensets in K-class (1950s) locomotives. This will replace the mechanical governors that are becoming very diffcult to maintain.
We've been discussing the design of gate drivers.

This is how I got to the HCPL316J-500 (Farnell $5.00 maybe). The gate driver power supply is looking to be along the lines of a push-pull switchmode IC (TL494 RS $2.20) and a small transformer (MuRata 78613/1C or similar Farnewll $5.00). Add a handfull of small transistors and passives, an optocoupler for feedback and a cheap, but effective power supply should be possible. It'ss need some booster transistors to make 8 amps, enough for any of the IGBTs its likely to encounter.
I'll prototype this arrangement and confirm it's operation. I would expect that a 2 watt +15/-8V regulated output should be achievable.
This should make a fully optioned gate driver for $15-20.

A capacitor bank using little snap-in caps would need a number of them on a PCB, a finger labryinth to reach all of the caps. The price is probably somewhat cheaper to a set of 2 x 4700uF/500V. I've got Fastron on the job. I'm guessing 2 x 25, which should be about 50 Amps and 4000 uF. the caps have to be $5.00 each to make it worth while.
I haven't decided on the path of least resistance yet. Difficulty, that is.

Also, I haven't looked at snubbers for a bit, probably 0.5-1uF for 100 A IGBTs.

Cheers

BGA

Last edited by bga on Fri, 10 Oct 2008, 09:39, edited 1 time in total.

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

100kW AC Inverter Project (Opensource)

Post by Johny » Fri, 10 Oct 2008, 20:59

Hi bga. If you decide on the Nippon caps, I've uploaded the specs.
uploads/293/Electros_al-lxalg-e-070821.pdf

My concern is that a number of failed inverter project attempts I've read about involved less-than-optimum DC Bus impedance characteristics (actually they were woeful, using bits of wire)causing excessive dv/dt on the IGBTs. Big bolt-on stud caps. simplifies it a little bit.

I've also download and installed MPLAB and the PIC compiler on my work PC so I can at least track along with your project. I won't be much help because I haven't got any hardware (and my PC at home has a year 12 student attached to it) and I'm not as brave as you, and, and...

Is this the same version of the Circuit Cellar project that you have?
MotorController_CircuitCellar_MT2291.zip

Apparently the Open Source inverter scored a 4 to 5 page article in the August Circuit Cellar magazine. Pretty much the same as the doc.
http://www.circuitcellar.com/archives/v ... index.html

bga
Senior Member
Posts: 492
Joined: Mon, 01 Sep 2008, 19:27
Real Name: Bruce Armstrong
Location: Perth WA

100kW AC Inverter Project (Opensource)

Post by bga » Fri, 10 Oct 2008, 21:04

Agreed.

I was also considering film (the high current types) caps along the bus to reduce the bus inductance effects at the IGBTs.

1200V IGBTs offer some insurance.

I've seen the Circuit Cellar article and managed to download all of their documentation and some source code. It allowed me to compare the source code for their PIC controller with that from Microchip, only about 2% difference, easy to see their edits.

Cheers

BGA
Last edited by bga on Fri, 10 Oct 2008, 10:09, edited 1 time in total.

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

100kW AC Inverter Project (Opensource)

Post by Johny » Fri, 10 Oct 2008, 21:59

Snubbers here (standard postage is US 5.50):

http://cgi.ebay.com.au/1200V-2-0uF-Puls ... dZViewItem

These are leaded type but I reckon they could be mounted on the BUS +/- in such a way as to make the leads very short. I have a question for postage in for a USA seller with 2uF 1000VDC lug type.

http://cgi.ebay.com.au/4-Pcs-Aerovox-RB ... dZViewItem

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

100kW AC Inverter Project (Opensource)

Post by Johny » Fri, 10 Oct 2008, 22:03

bga. Are you basing your code on the Microchip or Circuit Cellar source?

The CC source I have is 48MB .zip.
Last edited by Johny on Fri, 10 Oct 2008, 11:04, edited 1 time in total.

bga
Senior Member
Posts: 492
Joined: Mon, 01 Sep 2008, 19:27
Real Name: Bruce Armstrong
Location: Perth WA

100kW AC Inverter Project (Opensource)

Post by bga » Sat, 11 Oct 2008, 07:35


That Zip file sounds about the correct size.

The CC code is substantially the same as the Microchip code in the motor
driver parts. It has additional CAN bus stuff, which I will attempt to
understand eventually. This may also be from Microchip, but I haven't looked at it.

I feel that the LabView stuff has to go for the sake of ease of management.
I'll use a hammer I'm familiar with : Visual C++.
There's an express free version from Microsoft.
I've built a numbrer of these sorts of control panel applications
so don't feel that this should be very difficult. Tedious, maybe.
Would like to get real-time parameter logging and display.

I dont like the way they've implemented the motor and dynamics constants,
which are compiled into the binary product. The PIC has a very functional
parameter flash memory for this purpose. (4K, which is plenty)
I would be unhappy having to alter the program binary for differing motors.

The motor control application uses approx 10% of the memory capacity
and 30% of the computrons (CPU speed).
Adding functionality should be no problem.

There is very little of the 'system' stuff that is needed to make a complete motor controller. This will have to be added.

A starting list:
-Bus management, including capacitor precharge and main contactor control.
-bus voltage monitoring
-Bus current monitoring
-Ground current monitoring (winding fault)
-Reversing, reversing lamp relay, reverse command input
-Braking, voltage and current limits, brake lamp relay.
-system Enable input
-ready lamp
-fault lamp (battery+motor)
-emergency stop, may be enable line going false.
-CAN bus
-safe recovery from a 12VDC outage while running.
-vehicle compatible I/O - not easily blown up.
-etc

The DSPic may run out of I/O in the above list. Haven't considered this problem yet.

cheers

bga

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

100kW AC Inverter Project (Opensource)

Post by Johny » Mon, 13 Oct 2008, 18:30

I agree that having changable parameters is a must. Two complete sets would be desirable. I have already hit a situation where 2 params sets with fast change between them would be helpful - and I haven't even got the controller in a car yet.

Is there serial I/O from the development board? A simple asynch serial command set would be good. That way custom PC apps for control (programming params) and logging would be easy.

Ground current (motor fault) monitoring you really get with the desaturation feedback via the Gate driver - assuming you ground the centre of the battery pack (a whole controversial thread by itself).

Is BUS or input current used by the controller? The commercial VFDs don't tend to monitor the BUS (or input) current. It guess it depends how much of the instrumentation you think that the unit should inevitably provide.


a4x4kiwi
Senior Member
Posts: 772
Joined: Thu, 03 Jan 2008, 19:04
Real Name: Malcolm Faed
Location: Australia
Contact:

100kW AC Inverter Project (Opensource)

Post by a4x4kiwi » Mon, 13 Oct 2008, 20:03

I plan on using the 4 profiles available on the Danfoss.

1. Forward economy
2. Forward power
3. Forward slow - for parking
4. Reverse slow

Multiple profiels or paramaters that can be altered by external inputs would be very helpful.

Regards, Mal

Last edited by a4x4kiwi on Mon, 13 Oct 2008, 09:04, edited 1 time in total.
Silicon is just sand with attitude.

Blog: http://malfunction.faed.name

bga
Senior Member
Posts: 492
Joined: Mon, 01 Sep 2008, 19:27
Real Name: Bruce Armstrong
Location: Perth WA

100kW AC Inverter Project (Opensource)

Post by bga » Thu, 16 Oct 2008, 23:33

Hi All,

Multiple parameter sets should be easy(ish). I would expect thet these will be stored in block of the flash memory. probably about 128 bytes each of the 4K. May need a command/logic line for loading these. My expectation would be that there will be a serial command set for manipulating these via a serial link. My design has a 'reverse' input, so this may allow a set for reversing. Mostly the parameters will not be the bas motor parameters, but a set of control parameters that determine the interaction of throttle, brake etc.

I haven't thought of what to do with the communications links, except that because they are easy implement, they should be. Four of these can provide an almost limitless expansion capbility. (Impressive feature of the controller, I wonder if they can all run at the same time?)

I would expect something like the following:
ICD : The in-circuit debugging in real-time. The ICD port allows single stepping, program loading and a few other operations.
UART1 : Diagnostic terminal for operational debuging on a laptop.
UART2 : Unallocated, possibly for third party hardware. eg battery managers or chargers.
CAN1 : Proprietary related components, such as an I/O module or a dashboard controller. I haven't thought of this much yet except to use these devices to make wiring up the controller easy and convenient.
CAN2: third party I/O expansion. Allows interfaces to various CAN devices or external equipment.

I have been doing an allocation for the I/O, but this is likely to be partly placed on a separate PCB with a logic level flat cable between the processor and the I/O PCBs. Makes the I/O easier to locate and keeps the boards small for prototyping.

Desaturation of IGBTs
Mostly this is used to protect the devices from a short circuit or an IGBT fault on the inverse side. Good drivers have this built in so that it's quick and sure. A motor stall could also cause this if the IGBTs have less capacity than the windings, although the controller should be on the job to prevent high current from this source.


Cheers
BGA

   
Last edited by bga on Thu, 16 Oct 2008, 14:19, edited 1 time in total.

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

100kW AC Inverter Project (Opensource)

Post by Johny » Mon, 20 Oct 2008, 16:53

FYI
First photo. The Lenze re-assembled with cover removed. Compare with the bare power board (earlier post) photo to see how the high power DC routing fits together.

Image


My test setup. The little red box on top of the controller has Enable, Forward/Reverse and a 5K pot for accelerator. The motor is a 2.2kW TECO.
The two transformers in the foreground provide an extra 90 VAC which I have in series with neutral to give around 330 VAC - just enough to be above the low-voltage DC limit of 430 VDC.

Image

Edit: Added CR here to make picture visible???
Last edited by Johny on Wed, 22 Oct 2008, 17:49, edited 1 time in total.

bga
Senior Member
Posts: 492
Joined: Mon, 01 Sep 2008, 19:27
Real Name: Bruce Armstrong
Location: Perth WA

100kW AC Inverter Project (Opensource)

Post by bga » Sat, 05 Sep 2009, 20:55

DC Bus capacitors have been one of the biggeest problems for the controller.

Conventional Alum. Electros fall short in the voltage, ripple current and ESR.

I estimate : needs 0.5 milliohm ESR and at least 2500uF so that a 10Khz PWM controller has low ripple:
   0.5 volt ripple from the ESR
   5 Volt ripple from capacitance (1% at 500 volts)

My hunting to date has yielded:

1) Cornell Dubilier type 947
   800V, 600uF, 70 Amps RMS, 2 milliohm ESR
   4 or 5 for a 300 Amp Bus
   They are available at Farnell (ex USA) for AUD$315 each
   Hmmm.

2) EACO, a chinese company, but has an office in Canada.
   www.eaco.com
   The caps look to be a good spec, but I haven't managed to contact them yet
   eg:
     SHD-700-600: 700V, 600uF, RMS 54 Amp, 2.7mOhm
     Would need 6 for a 300 Amp bus.
     (3600uF / 324 Amp RMS / 0.45 milliohm ESR)
    or
     SHD-700-470: 700V, 470uF, RMS 47 Amp, 3.1mOhm
     Would need 6 for a 300 Amp bus.
     (2800uF / 282 Amp RMS / 0.5 milliohm ESR)


Has anybody contacted eaco?



antiscab
Senior Member
Posts: 2490
Joined: Mon, 26 Nov 2007, 05:39
Real Name: Matthew Lacey
Location: Perth, WA

100kW AC Inverter Project (Opensource)

Post by antiscab » Sat, 05 Sep 2009, 22:38

you could pull the same trick as for the DC controllers, use lotsa little ones.

This also helps with heat dissipation, though increases assembly time a bit.

do the 1v due to ESR and 5v due to cap take into account the relative capacitance of the battery?

Matt
Matt
2017 Renault zoe - 25'000km
2007 vectrix - 156'000km
1998 prius - needs Batt
1999 Prius - needs batt
2000 prius - has 200 x headway 38120 cells

User avatar
acmotor
Senior Member
Posts: 3603
Joined: Thu, 26 Apr 2007, 03:30
Real Name: Tuarn
Location: Perth,Australia

100kW AC Inverter Project (Opensource)

Post by acmotor » Sat, 05 Sep 2009, 23:35

Ditto on Matt's Q re including battery pack in calculation ?
Batteries claim lower impedance at 1Khz than DC.
Ah, 0.5mohm you say ? Mmmm.. don't worry about battery in the calc. then. A pack is going to be in the order of 500mohm before contactor/wiring/fuses ?

Is this another reason for wanting to work with higher frequencies say 100Hz or 200Hz for the 100kW waveform and 20kHz for the PWM ?

Explain why you need to limit DC bus ripple to 5V ?
I know the VLT5042 30kW runs 4500uF and switching frequencies from 3 to 14kHz. Wouldn't bus ripple of 5 to 10% be acceptable ? Image
iMiEV MY12     110,230km in pure Electric and loving it !

User avatar
Johny
Senior Member
Posts: 3729
Joined: Mon, 23 Jun 2008, 16:26
Real Name: John Wright
Location: Melbourne
Contact:

100kW AC Inverter Project (Opensource)

Post by Johny » Mon, 07 Sep 2009, 18:52

bga. I got a price of Richardson mid last year for 500V 6800uF electros.
Basically $74 each.

Richardson Electronics Pty Ltd
Price Quotation 2913483                      110/33 Lexington Drive
                                               Bella Vista, NSW 2153 Australia
31 JUL 2008 Page 1 of 1                      Phone: +61 1300882819
                                               Fax : +61 2 8883 4671
Valid for 30 days                            
                                               
      Customer 106645                           
                                               
Your Reference        Terms               Currency          Customer Fax
                        NET 30 EM            Australian Dollar 0397303968

Item   Qty Part Number                     Br Unit Price Extension Delivery
   1      5 ELXA501LGC682MF ELXA501LGC682MF UC      75.10     375.50 8 Weeks
            ELECTROLYTIC CAPS
            PRICE COMMENTS: United Chemi-Con
            ELXA501LGC682MFK0M
            5pcs moq/mults
            nc/nr
            DELIVERY COMMENTS: Lead time 6-8 weeks
            aro.

                                                  Total      375.50




bga
Senior Member
Posts: 492
Joined: Mon, 01 Sep 2008, 19:27
Real Name: Bruce Armstrong
Location: Perth WA

100kW AC Inverter Project (Opensource)

Post by bga » Mon, 07 Sep 2009, 21:44

I started with the electro option, bit think that film capacitors offer improved performance over electros. The low ESR (10x lower than similar electros) and higher voltage (up to 1700V) allow a simpler and more robust filter. The very low ESR should result in cooler running capacitors with less requirement for forced air cooling, a problem in closed enclosures.
A 300Amp/700V filter should be achievable with 6 x 76mm dia x 120mm long devices. A total capacity of 2820uf and ESR of 0.5mR
A simlar 300A/700V EPCOS filter would require
12 x B43564 450V/4700uf/25mR for 900V/14000uf/8.3mR as a series-parallel network. and roughly twice the volume of the film caps.

I feel that most (all?) of the DC controllers are grossly under specified in the filter capacitor department. The 4700uf/200V electros are typically only rated for about 5 amps of ripple current. So 10 will be about 50 amps, but the controller has a 750 amp circulating current.
Something doesn't add up.

On the ESR side, the battery ESR is ignored. As you say, it's a lot more than the filter ESR, or about the same if electros are used

The 5 Volt ripple was so that the battery current doesn't vary much, 10 Amps or so with the TS LFP40AHA's ESR. No particular criteria except that large ripple will make the battery system less efficient due to I2R and peukert losses, so it's good to minimise it if possible. Mostly, I considered this at low speed, where the average battery current will be low compared to the motor current, so ripple will be more significant.

I was using a 50% duty cycle on the ripple current, so that for a 10KHz PWM frequency, 1/2 is discharge and 1/2 is charge, the ramp time for the ripple is effectively 50us instead of the 100us period. Yes, the PWM frequecy will likely vary between 3 and 15kHz. I don't yet know what the best frequency will be, but expect it to be variable, depending on conditions. Min/max off and on times are factor and the IGBTs won't be happy much above 15khz.

The 200Hz is more than should be needed, as this will be ~6000RPM and 180km/h ... maybe in a vacuum. Perhaps 150hz is a better design limit. The system is likely to run at 120Hz as the high speed operating limit for 100km/h. According to ABB docs, the 15kw motors are operable to 4500 RPM**. CP, Teco and others are probably the same.

**Needs an operation to re-arrange the motor's windngs for lower voltage so that 150Hz is useful from the limited battery voltage.
I'm thinking 450 volts battery because of weight and possibly 1/2 415V initially or a rewind to the calculated optimum - currently guessing 150-175 volts, in which case 300 amps won't seem a lot.

Battery and motor limits will probably not permit 100kw. The idea was to prevent the controller being the weak point.

bga
Senior Member
Posts: 492
Joined: Mon, 01 Sep 2008, 19:27
Real Name: Bruce Armstrong
Location: Perth WA

100kW AC Inverter Project (Opensource)

Post by bga » Mon, 07 Sep 2009, 21:48

Hi Johny,

That's not too scary a price.

I like the spec on the film caps, but Farnell was quoting near $300 for Cornell Dubilier type 947, which is a bit over the top.

I have been looking for other suppliers to get a more realistic option.

Cheers
Last edited by bga on Mon, 07 Sep 2009, 12:07, edited 1 time in total.

HeadsUp
Groupie
Posts: 265
Joined: Sat, 28 Feb 2009, 09:13
Real Name: Mark W
Location: Sydney

100kW AC Inverter Project (Opensource)

Post by HeadsUp » Sun, 01 Nov 2009, 06:37


If anyone is interested in reading basic specs on current generation Fuji dedicated EV/hybrid IGBT's

spec sheet below

Fuji dedicated EV IGBT

what interested me the most was mention on most of those links for an " evaluation board " to suit each module

( as i had been looking for gate drivers for some Fuji 6MB1300U-120 IGBT's )


Download page for code and user guide for the Microchip controller here

just posting the link here coz the original site was no longer posting the information and it was a b*stid to find

Microchip developer board and code guide download page

.
edited 10 zillion times to fix corrupted links
.
Last edited by HeadsUp on Sat, 31 Oct 2009, 19:44, edited 1 time in total.

HeadsUp
Groupie
Posts: 265
Joined: Sat, 28 Feb 2009, 09:13
Real Name: Mark W
Location: Sydney

100kW AC Inverter Project (Opensource)

Post by HeadsUp » Sun, 01 Nov 2009, 07:17


Free application from Fuji for simulating heat loss , efficiency , and heatsink requirements in their IGBT.

you will need to register before downloading it


Free Fuji calculator for heat loss , heat sink , efficiency

sample below as based on 6MB1300U4-120

Image

Image
Last edited by HeadsUp on Sat, 31 Oct 2009, 20:40, edited 1 time in total.

bga
Senior Member
Posts: 492
Joined: Mon, 01 Sep 2008, 19:27
Real Name: Bruce Armstrong
Location: Perth WA

100kW AC Inverter Project (Opensource)

Post by bga » Sun, 01 Nov 2009, 17:39

Hi All,
An update on the film capacitor front:

I have ordered some EACO power film caps for myself and Richo:
Capacitance: 390uF
Rated Voltage: 900V (+/- 10%) (Un)
Life expectancy: 100,000 hours at Un and 70C
RMS Current : 51 Amps
ESR : 3mR
Dimensions: 122mm long x 86mm diameter
I will probably use 6 in the controller for 300A RMS.

I also ordered some 2uF/1000V snubbers for the IGBT legs.

Hopefully, this will arrive in a couple of weeks.
Last edited by bga on Sun, 01 Nov 2009, 06:41, edited 1 time in total.

HeadsUp
Groupie
Posts: 265
Joined: Sat, 28 Feb 2009, 09:13
Real Name: Mark W
Location: Sydney

100kW AC Inverter Project (Opensource)

Post by HeadsUp » Mon, 02 Nov 2009, 01:15

whats peoples thoughts on bus bar connector design to eliminate spikes from destroying components , this thread suggests there is a bit more care required in the science rather than just making insulated copper bars on the high current ( DC bus ) side.

I think a common mistake is to size the copper bars but not allow for the drill hole weakening the cross-section , in fact i would prefer to punch the holes in them with a tapered spear punch rather than drill them so no material is lost , then you end up with a pregnant bulge around each hole , other way is have your bus bars or bus plates profile cut with a rounded bulge around each hole.


thread link regarding busbar design from DIY forum

Quoting here

I think someone here pointed out that the DC bus really needed plates, not just bars. I looked up a company that makes them, but they do it on a "custom" basis. I could have them provide bus plates, and it's possible they have a boiler plate design, but they don't have a real product line that you can select from. I talked to a guy from this company (guy and co. name I forgot) and he told me a lot about how they make the plates. He calculated some things with me on the phone, and said that if they were doing it, they'd use .062" thick plate. Between the layers, they'd put .010" nomex film. They punch the appropriate holes, and emboss the terminals on the different layers in such a way so that the terminals all end up in the same plane so they'll mate to the IGBTs. Then they'd apply an adhesive, put the assembly in a press, and bake it at around 350 F for several hours. Anyway, the plates are supposed to allow for very low resistance and inductance, and provide a little bit of distributed capacitance. The problem with power spikes feeding back to the IGBTs is an alleged killer of expensive semiconductors. When you're switching hundreds of amps at tens of kilohertz, these things matter a lot. That's also why the driver boards have their own DC-DC converters. With all the voltage fluctuations caused by all those amps being switched, it's best to have a separate and robust supply driving the gates of the IGBTs. In order to control switching times, you also want a sturdy driver, which is provided with modules on the driver boars. Like many things in the world of electronics, going to the extremes causes things to get a bit weird. Ask anyone working in very high power, very low power, or very high frequency. I'm not really one of those people, so I'm trying to absorb as much info as I can. The other problem I think I'll eventually be faced with is a safe source of power for testing. 100 kW at 240 Vac exceeds the power company feed into my "laboratory" (home), so the only way to do it will probably be a massive bank of batteries. That's a long way off, though. I first need to get to the stage where I can even turn a (rather large) unloaded ACIM. -Mark

User avatar
Electrocycle
Senior Member
Posts: 985
Joined: Sun, 19 Oct 2008, 20:23
Real Name: Andrew
Location: Sydney
MSN: dumhed@dumhed.com
Contact:

100kW AC Inverter Project (Opensource)

Post by Electrocycle » Mon, 02 Nov 2009, 01:21

just make the bars bigger than you "need" :)
The Engine Whisperer - fixer of things

Post Reply