Tritium Wavesculptor

AC, DC, amps, volts and kilowatt. It's all discussed in here
Post Reply
User avatar
woody
Senior Member
Posts: 1716
Joined: Sat, 21 Jun 2008, 02:03
Real Name: Anthony Wood
Location: Mt Colah

Tritium Wavesculptor

Post by woody »

Anyone know anything about this 20kW AC Drive?
Planned EV: '63 Cortina using AC and LiFePO4 Battery Pack
a4x4kiwi
Senior Member
Posts: 772
Joined: Thu, 03 Jan 2008, 19:04
Real Name: Malcolm Faed
Location: Australia
Contact:

Tritium Wavesculptor

Post by a4x4kiwi »

Nice, I will be reading more about that one. Oz made as well.
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

Tritium Wavesculptor

Post by bga »

Hi All,

It's for driving the CSIRO hub motor, as used in solar challenge car race competitions.

The whole package is very efficient.
The controller is approx 97% efficient and the ~9kW peak CSIRO hub motor is about 98% efficient.

The motor is about $12,000 a piece. For serious solar racers only.

I had wondered what the teams did for a motor controller. The Tritium website says that this controller was used by 1/3 of the teams in the 2007 solar challenge with 6 out of the top 10 finishers using the controller. I suspect that the CSIRO motor website would say the same.

According to the documentation:
The controller is for a 160V battery system (165V absolute max) and will produce approximately 110V RMS at 100 AMPS. ~~11KW

There is also a 40KVA version of the controller. Given the paired motor, I would expect this to be similarly priced.


Cheers

BGA


Last edited by bga on Mon, 15 Dec 2008, 15:22, edited 1 time in total.
User avatar
acmotor
Senior Member
Posts: 3605
Joined: Thu, 26 Apr 2007, 03:30
Real Name: Tuarn
Location: Perth,Australia

Tritium Wavesculptor

Post by acmotor »

I looked at that motor and controller a few years ago and thought it would make a great ebike. Not certain people would believe the 200W sticker I had planned to fit !

I must say, I didn't get any answers to my emails at the time.

It is great to see home grown technology. (and AC !)
iMiEV MY12     110,230km in pure Electric and loving it !
User avatar
Electrocycle
Senior Member
Posts: 985
Joined: Sun, 19 Oct 2008, 20:23
Real Name: Andrew
Location: Sydney
MSN: dumhed@dumhed.com
Contact:

Tritium Wavesculptor

Post by Electrocycle »

actually, does anyone have an AC controller that'd do around 120v?
I have a t-flux motor (older solar car gear) sitting here that I'd like to use for something :)
The Engine Whisperer - fixer of things
Benonymous
Groupie
Posts: 92
Joined: Fri, 14 Mar 2008, 20:12
Real Name: Ben Roberts

Tritium Wavesculptor

Post by Benonymous »

I corresponded with one of the engineers at Tritium and they are developing a much larger controller (40kW I think) and will be pricing it around the $7K mark. No news on that for a while though.
EVLearner
Groupie
Posts: 51
Joined: Tue, 06 Jan 2009, 02:57
Real Name: Malcolm Moore
Location: Turramurra

Tritium Wavesculptor

Post by EVLearner »

I would love to see the circuit they are using, because I strongly suspect they are parallelling up power FETs to get their current with their rather limited voltage.

Not sure their approach is much beyond kiddie/university stuff, but what I do find interesting is that everybody seems to rush to a microcontroller to do "everything" - unfortunately, most of the younger engineers involved with this stuff don't have the analogue electronics experience - so the designs are not elegant - but have lots of software to make it happen!!

From my side I am looking for an FPLA that will have the 6 outputs in 120 deg phase, so that I can drive this State Machine by a much faster variable speed clock. Now if I can get PWM into the FPLA, then we have the makings of a very cheap to manufacture AC controller!! (probably around the $400 to $800 mark)
bga
Senior Member
Posts: 492
Joined: Mon, 01 Sep 2008, 19:27
Real Name: Bruce Armstrong
Location: Perth WA

Tritium Wavesculptor

Post by bga »

Hey, what's wrong with software!Image
I try to do as much as I can in software and keep the hardware simple.

I'm currently designing a three phase induction motor controller (600V/150Amps) and am using a Microchip DSPic30F6010A (30Mips, 80 pins and cheap). This will do what you need with all the other ADC SIO CAN etc stuff needed to make a complete and friendly motor controller. Other vendors (Atmel?) have similar products, but I chose the PIC because I have used them for years and there are sample applications in 'C'/Assembler for various types of motor control applications so the software development time should be reduced.

The IM driver consumes about 10K instruction words and 10Mips, so there's lots left to do the throttle, displays, lights etc. I think that the BLDC applications are similar.

I think that you'd be quite correct in assuming that the tritium controller uses parallel FETs to get the conduction losses down.
They may also be using the FETs to do active flyback control to reduce diode losses.

Ian Hooper's motor contoller page is a good read and very informative. Please see "Motor Controller Design: A Photo Journal"

Back to the LCD interface, this doesn't seen to be as easy as it usually is Image).

[edit:typos]
Last edited by bga on Sun, 08 Feb 2009, 16:47, edited 1 time in total.
EVLearner
Groupie
Posts: 51
Joined: Tue, 06 Jan 2009, 02:57
Real Name: Malcolm Moore
Location: Turramurra

Tritium Wavesculptor

Post by EVLearner »

Hi bga

Nothing is wrong with software - providing it is used in an eligant manner, and when it comes to a 3 phase (120 deg apart) firing system, with a variable frequency drive, I see strong uses for both digital to work the angles and analogue to manage the PWM - but I am bending towards doing the PWM using digital too!!

BTW that is one nice processor, and good to hear that they have done most of the software to date. Lucky that you have played with micros recently, I have not touched them since about 1983 (basically the same, wider and much faster now)

I can see that the PWM can be clocked from the 30 MHz in small enough incriments, so that this can be a stepped triangular (digital) waveform for comparison with the 3 pseudo sine (flattened) waves and that should not should not be a problem, but what I am still having mental tussle about is the Throttle frequency oscillator to produce the variable frequency flattened sine waves at 120 deg apart.

The chip has a 10 bit A/D converter, so that could be used to convert a voltage (current / resistance), into a digital word that could then be used as a timer for an inverter speed (with 3 phases), so that in turn could be used as a comparison to produce PWM to drive the IGBTs. Is that getting close to the agreed design??

They may have used FETS, but when you look at the overall resistance and the currents, the drops will be about the same and I think the IGBTs probably have as good an on resistance as the 30 FETs - and far easier to mount than FETs!! Comments appreciated..


It sounds like what you are designing is very close to what I was looking to do so I will leave the micro part to you.


I am looking at using IGBTs as the switches and my thoughts were to use optical isolators to interface to connect to the drivers for the the IGBTs, and to have a small (open loop regulated) switched supply to power the isolated gate drive circuits. This should make the all the drive circuits reliable, simple and well isolated. Comments appreciated.

Think I read Ian's journey last night (is that the 47 Mbyte one?)

Good luck with your LCD interface.

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

Tritium Wavesculptor

Post by bga »

Hi EVlearner,

Ian's page is big, but also very interesting. I'm glad that he's seen fit to publish a warts & all account.

I don't have a problem with using IGBTs or FETs. I like IGBTs because they're a lot more abuse tolerant that FETs, particularly for over-voltage punch through.

I think that there may be some efficiency gains from using FETs in lower voltage systems (less than 200V) because the motors have a typical cruise current far less than the maximum, so the conduction losses in FETs should be significantly lower than with IGBTs.
If the design is using discrete parts, many of the FETs are interchangeable with IGBTs, alowing options for either.

Above about 200 volts, FET choice is limited and IGBTs become a better solution. Even in low voltage systems, the BIG 600Amp IGBT modules could make them a better chioce from a complexity viewpoint.

My favorites are:
FET   Fairchild FDA75N28 (280V, 75Amp, RDSon=41mOhm)
      IXYS IXFX120N20 (200V, 120Amp, RDSon=17mOhm) (ZEVA uses this)
IGBT Fairchild FGA25N120ANTDTU (1200V, 50Amp)

I think that these are essentially interchangeable without too much component lead origami.
Both Fairchild devices are available for low$$ from farnell.

Also, a good isolated gate driver:
Avago HCPL-316J (2.5Amp) from RS, moderate $ for the ROHS part (they list two).
These will need a current boost for high current operation above about 150Amps.
I like this device because it has undervoltage and desaturation detection built in. This should save a lot of grief in the real world.

After the LCD capitulates, the next task is to get the HV isolated power supplies sorted out so that I can lay out and make the gate drivers.

So far in my project, I've made some PCBs.
1) I have a processor card based on the Circuit cellar project and the Microchip development kit. I've added I/O to make it mre useful for EV use. It's currently having firmare witten for it.
2) I have 3+3 discrete IGBT PCB that should be good for about 100-150 Amps at 600Volts.
3) I have a set of link capacitor boards using 400V/330uF capacitors in series-parallel. This should result in 4500uF at maximum 800 volts.
4)Current sensor PCBs for the blue LEM HAIS current sensor modules.

To do:
1) Gate driver PCBs (two channel 1200V isloated, suitable for my IGBTs
2) System PSU PCB (12V/3Amps, 5V/1Amp) I should move the 5V to the CPU PCB so it'll work off 8 to 40V.
3) The EV I/O PCB. to contain driver transistors, relays, switch and analog sensor interface for an EV.
4) Metakwork for a box and busbars, water cooling plumbing etc.
5) Fans, Contactors (Tyco Kilovac LEV200-A4NAA) terminals etc.

The recent pace has been slow due to over-work commitments.

Last edited by bga on Sun, 15 Feb 2009, 11:29, edited 1 time in total.
EVLearner
Groupie
Posts: 51
Joined: Tue, 06 Jan 2009, 02:57
Real Name: Malcolm Moore
Location: Turramurra

Tritium Wavesculptor

Post by EVLearner »

Hi bga

Sounds like you are on the same wavelength as I am on (and that is good)! Your reasoning for the IGBTs is exactly what I had considered - and I can see the voltage going up - not down (I started thinking in terms of 144 V then moved to 200 V DC. Now, depending on the batteries, 5 groups of 4 batteries does not sit comfortably, so I am not thinking in terms of 4 or 6 blocks of batteries 48 * 4 = 192 V, or 48 * 6 = 288 V (which somewhat aligns with the 240 V AC), or four blocks of five batteries: 72 V * 4 = 288 V, just another twist; you see the voltage is rising!

Had a look at the opto isolator (HCPL-316J) data and personally I'm not too keen on that approach - although it might work very well - it just looks like a weak link to me (I just don't like the way the opto is powered from the high potential side). I will be further investigating a small converter with well isolated +/-12 V feeds to connect ot the opto isolator. (Will keep you posted)

You are way ahead of me WRT the microchip development - the nearest that I got to that was with an Apple II, but then my career changed dramatically - so I am going to have to work from well behind the 8 ball there!!

WRT the IGBTs, I have seen some layouts that put these side by side and run a common bus over the three module/pairs. Something is worrying me with this layout and it has a lot to do with close proximity high voltage parallel bus collisions (incidents).

Some time ago I made a current shunt with a very thick wire, and another thin wire as the parallel connection for a lower current current probe. It struck me that the thick wires that connect to each IGBT can be used as current sense resistors and the differential voltage will be proportional to the current flow. The high voltage side of the IGBTs modules can connect to a common (heatshrink insulated) busbar - or disk, if the IGBTs are positioned 120 deg to each other!
Tritium_James
Senior Member
Posts: 683
Joined: Wed, 04 Mar 2009, 17:15
Real Name: James Kennedy
Contact:

Tritium Wavesculptor

Post by Tritium_James »

Hey guys, I am probably qualified to comment on this one   Image

Yes, our 20kW drive is designed pretty much for solar cars. The design priorities are optimised for efficiency, reliability, weight, size, cost, in that order. For teams spending 500k+ on solarcells, $7k for a controller is a bargain...

Electrocycle, this controller will run a T-flux perfectly, we have one here on our bench dyno, along with a CSIRO hub motor.

EVLearner, yes, this is a bit beyond kiddie level - and it's all software! The sort of design you're talking about is known as a "six-step" controller, where you drive the phases 120° offset. I'm afraid it's quite a primitive way to go. Ours runs a sinewave output (matches the back-EMF = more efficient, quieter), and uses sensorless control algorithms to keep the drive in sync with the motor.

The key thing with these sorts of drives is the phase current regulation, and for a BLDC motor you've got almost no hope of managing it successfully with an analog control solution, especially if you're running 6-step. The inductance is really low, and you need a very rapid response. The huge current step waveform as you transition from one phase to the next means it's practically impossible to keep your current in regulation. It can probably be made to work if your silicon can tolerate the resulting current overshoot, but then your efficiency drops, and you're carrying around more silicon than you need to.

The controller is using paralleled FETs. Exactly as bga pointed out, above 200V the choice starts to become limited (and crappy, too!). We're using 200V rated parts in this design (IXFR120N20, for those interested), three in parallel on each phase. For anything higher power, IGBTs are the way to go.

We have a new design in the prototype stage at the moment, aimed at the EV market. It's rated to 200kW, and is about the same size as the 20kW solarcar drive. It's based on an IGBT module, and the software is almost identical to the solarcar drive, and it's already up and running on the new processor (a TMS320F2809 32-bit DSP @ 100MHz). Specs are:
- up to 450V DC bus
- 400Arms output (phase) current
- 500x170x80mm
- 8kg
- ~97-98% efficient at cruise
- Watercooled
- IP67 sealed
- CAN bus control/telemetry
- 3-phase BLDC or induction motor drive

We expect to have it ready for sale by the end of October, possibly earlier, depending on some of the on-road testing. It will be running in a Honda Civic conversion in the World Solar Challenge mid-october, that's expected to be our final testing before general sale. Target price is around the $7k mark in low volumes, but that (unlike the solarcar 20kW design) has the possibility of dropping significantly as the volumes pick up. This one has been designed with ease of manufacture and lower cost a bit higher on the priority list!

Any questions (about the solarcar drive, or the new one) please ask!
HeadsUp
Groupie
Posts: 265
Joined: Sat, 28 Feb 2009, 09:13
Real Name: Mark W
Location: Sydney

Tritium Wavesculptor

Post by HeadsUp »

thanks for the update James , keep us posted when release date and price is known.

will the interface have ability to connect to a pushbutton cruise control system as i mentioned in another post ?
i see value in having a dashboard mounted pushbutton system to control speed accurately as people drive between different speed limit zones , 40,50,60,70,80,90,100,110 km/hr at the push of a button

can you pop a beta test unit in the mail for me guv ? Image

thanks . and best wishes
Last edited by HeadsUp on Mon, 09 Mar 2009, 08:00, edited 1 time in total.
Tritium_James
Senior Member
Posts: 683
Joined: Wed, 04 Mar 2009, 17:15
Real Name: James Kennedy
Contact:

Tritium Wavesculptor

Post by Tritium_James »

The interface to the controller is purely digital via messages on the CAN bus. You tell it a target torque and speed and it does the rest. Both values can be positive or negative, this covers drive/regen and fwd/rev. You can also set limits on battery voltage (upper/lower), battery current, controller temp, and motor temp. It will dial the motor current back as appropriate to keep everything else within limits.

We sell a driver controls box that takes pots, encoders, switches, etc as inputs and sends the appropriate commands on the CAN bus to the motor controller (or controllers). See here:http://www.tritium.com.au/products/TRI63/index.html This has open-source software running on it (and schematics are there too), so you can make your user interface do whatever you like. It's using an MSP430 micro, the source is in C, the dev tools are free. So if you can understand some fairly basic software, your pushbutton cruise controls is certainly possible...

(edit - fixed link)
Last edited by Tritium_James on Tue, 10 Mar 2009, 05:55, edited 1 time in total.
Tritium_James
Senior Member
Posts: 683
Joined: Wed, 04 Mar 2009, 17:15
Real Name: James Kennedy
Contact:

Tritium Wavesculptor

Post by Tritium_James »

I hope that didn't sound too much like an ad! Admins, please let me know if that wasn't OK.

James.
User avatar
acmotor
Senior Member
Posts: 3605
Joined: Thu, 26 Apr 2007, 03:30
Real Name: Tuarn
Location: Perth,Australia

Tritium Wavesculptor

Post by acmotor »

... Maybe the fine line between designed and made in Oz and just sold in Oz ? Image
I'm all ears ( ... Will probably need those green lithiums to power your beast though !) Image   
iMiEV MY12     110,230km in pure Electric and loving it !
Tritium_James
Senior Member
Posts: 683
Joined: Wed, 04 Mar 2009, 17:15
Real Name: James Kennedy
Contact:

Tritium Wavesculptor

Post by Tritium_James »

acmotor wrote: ... Maybe the fine line between designed and made in Oz and just sold in Oz ? Image
I'm all ears ( ... Will probably need those green lithiums to power your beast though !) Image   


I cannot wait myself! Just got to do all the grunt work between here and a commercial product, documentation - ugh.

... and they're silver, actually.
User avatar
acmotor
Senior Member
Posts: 3605
Joined: Thu, 26 Apr 2007, 03:30
Real Name: Tuarn
Location: Perth,Australia

Tritium Wavesculptor

Post by acmotor »

Hey, silver is the new black ! Image
Were you referring to the KoKams ? I meant the round LiFePO4 but the LiPo may be better.
iMiEV MY12     110,230km in pure Electric and loving it !
Tritium_James
Senior Member
Posts: 683
Joined: Wed, 04 Mar 2009, 17:15
Real Name: James Kennedy
Contact:

Tritium Wavesculptor

Post by Tritium_James »

Yes, I meant the Kokams. They're slightly scary, and we haven't worked out how to package them properly, and it's going to require testing for transport, etc, but they're sure hard to beat for power density!

Here's some scope grabs from our initial testing. They're a bit wide so I'll link them:
http://www.tritium.com.au/james/kokam.gif
http://www.tritium.com.au/james/thundersky.gif

Blue trace is cell voltage, Red trace is current (1V = 100A). So the Kokam 100Ah cell dropped 280mV under a 500A load (5C) whereas the TS 90Ah dropped 520mV under a 450A load (also 5C). Both cells were at about 80% charge. The reason the current drops away during the test is our resistive load heating up and changing resistance.

This drop makes quite a difference to the power able to be put through the motor controller.
Last edited by Tritium_James on Tue, 10 Mar 2009, 05:54, edited 1 time in total.
User avatar
acmotor
Senior Member
Posts: 3605
Joined: Thu, 26 Apr 2007, 03:30
Real Name: Tuarn
Location: Perth,Australia

Tritium Wavesculptor

Post by acmotor »

TJ, the links only worked by cut and paste not click. (I did go there)
Did you use post reply or quick reply ? Need post reply for links to enter correctly.
Yep, LiPo is good. So are BMI, then so are ultra caps !

The 280mV is from 3.7V whereas the 520mV is from 3.2V so in comparison it is 520mV x 3.7/3.2 = 601mV for TS. I guess we all know that though.
Have you tested BMI from lithbattboss ? Do you plan on any other brand tests ?

Are you expecting heat problems with LiPo ?


iMiEV MY12     110,230km in pure Electric and loving it !
Tritium_James
Senior Member
Posts: 683
Joined: Wed, 04 Mar 2009, 17:15
Real Name: James Kennedy
Contact:

Tritium Wavesculptor

Post by Tritium_James »

acmotor wrote: TJ, the links only worked by cut and paste not click. (I did go there)
Did you use post reply or quick reply ? Need post reply for links to enter correctly.
Yep, LiPo is good. So are BMI, then so are ultra caps !

The 280mV is from 3.7V whereas the 520mV is from 3.2V so in comparison it is 520mV x 3.7/3.2 = 601mV for TS. I guess we all know that though.
Have you tested BMI from lithbattboss ? Do you plan on any other brand tests ?

Are you expecting heat problems with LiPo ?


Ah, OK, I just hit the quick reply button. Guess those broken links will save our webserver some traffic!

Haven't thought about BMI up until this point. The two options we've been considering for different vehicles are either A) Cheap, a bit nasty (TS) or B) Expensive, no compromise performance (Kokam). I think there's probably room in the middle for something else.

No heat problems so far with LiPo, but we've hardly done comprehensive pack testing. I don't expect major issues though, the internal resistance is low (as this test shows) and there's a massive surface area to volume ratio to help dissipate heat quickly.

And at this point I'm not convinced ultracaps are good. By the time you've spent the required amount of money, you'd get better performance by spending it on a bigger (lithium) battery pack!
User avatar
Richo
Senior Member
Posts: 3737
Joined: Mon, 16 Jun 2008, 00:19
Real Name: Richard
Location: Perth, WA

Tritium Wavesculptor

Post by Richo »

acmotor wrote: TJ, the links only worked by cut and paste not click. (I did go there)
Did you use post reply or quick reply ? Need post reply for links to enter correctly.


Actually the problem is when it inserts the URL text it puts in a space.
So if you don't take the space out it wont work.
[ U R L =   ] [ / U R L ]
Do you see the space afetr the = and before the ]
Take it out and problem goes away.
I did tell Andrew about it...

http://www.tritium.com.au/james/kokam.gif
http://www.tritium.com.au/james/thundersky.gif
Last edited by Richo on Mon, 09 Mar 2009, 19:32, edited 1 time in total.
So the short answer is NO but the long answer is YES.
Help prevent road rage - get outta my way!
Tritium_James
Senior Member
Posts: 683
Joined: Wed, 04 Mar 2009, 17:15
Real Name: James Kennedy
Contact:

Tritium Wavesculptor

Post by Tritium_James »

Thanks Richo, actually to make them work I had to take out the = sign as well. I'm guessing that's for when you want to have the link text be different from the link destination...
Post Reply