OSHW 100kW inverter for Electric Vehicles. What's your opinion?

[Alex_arg]

Hi everybody !
A few weeks ago i was wondering how power inverters used in cars work, so i made a search "100kw car inverter" and found
many sites with open source projects for this product, many of them are the DIY kind. Of course i've learn a lot about the insides
 of these devices, most of them designed to move a 3 phase AC induction motor, after this i'm enthusiastic about build one for
testing purposes.
Considering the fact that many countries adhere to low CO2 emission policy, do you think this could start developing a new industry?
In many countries it's difficult to buy a new electric car, so conversion of gas units to electric could be an increasing demand in near future?
Do we ,electronics engineers and technicians, be prepare for that?
I really like to hear your opinion, take a time do the search and tell me what you think. I heared Dave saying he dislike power electronics, but maybe
analize the principles of controlling AC Motors could be an interesing topic for a Fundamentals Friday video.

Thank You

Alex
 
   

[wraper]

If such a thing appears for real, there will be a lot of people electrocuted or harmed by some other way like fire of explosion.

[Ian.M]

If you want to get VFD and battery system management experience on the cheap, build a RC model with BLDC motors using open hardware ESCs.  A single screw boat is probably the easiest with a simple shrink-wrap over shaped polystyrene foam hull, and enough shaft down angle not to need a shaft gland, as all it needs is a two channel radio, a battery pack, a single motor ESC, the motor and  rudder servo.  + you get good exercise canoeing after it when it dies mid-pond.   RC cars etc. are mechanically complex and RC planes and quadcopters etc. have to be near perfect first time to avoid self-destruction by crashing.

I wouldn't even consider an electric vehicle for road transport use without a warranty on the pre-assembled battery pack.   Its just not cost-effective to invest so much money in a one-off - you could buy and run an 'eco' petrol car converted to LPG on the money you saved, and the income from working at a burger joint in the time you saved.

At least there is little risk of you getting such a vehicle on the road to be a hazard to other users - without considerable experience with component level design for related fields like grid tied solar systems and heavy electric traction (railways) the most likely outcome is an empty bank account and a burnt down garage, with possibly a large hospital bill for remove pieces of MOSFET shrapnel and patching up your LiO chemical burns.

[Jeroen3]

Converting a car from pertrol/gas to electric is a problem due to the weight of the batteries. All places this would easily fit are not an optimal place to put kilos of battery.
I've driven a Volkswagen Jetta Hybrid for a drifting course, the car is very difficult to control when it starts slipping due to the weight added in the back to make it a hybrid.
That is not going to happen unless some car proves to be suitable for this.

[awallin]

Here's someone spinning a lexus drivetrain (essentially the same but perhaps more powerful than a Prius):

Here's something on the inveter:


there are probably *many* more to be found by googling..

DIY ac-servordive builds for cnc-machinery might be worth a look too. Small pro-level cnc-mills will have ca 1kW motors on the axes and something like 6-12kW on the spindle. Prius MG2 is ca 50 kW at 500 VAC IIRC

[T3sl4co1l]

Doesn't sound like something you'd want to play with...  1kW is enough to kill or maim, 100kW is enough to explode and melt things significantly larger than a person.  And I wouldn't expect OSHW to be nearly safe enough, or well enough engineered, to deal with it responsibly and safely (not to mention efficiently and cheaply).

OS is something that works well when there's little or no incremental cost or barrier to entry associated with the subject.  Almost anyone can learn a programming language and sit down and write [bad] functions, or hopefully, rewrite someone else's drivel into something readable and good...  So OSS works well.  When it's big enough to be popular, also (like Firefox).

Droves of small OSS projects fester with poor development and support.  Often times very useful, but narrow in purpose, low in visibility, and never drawing the attention and time of those who would serve it well.

And that's just software.  Hardware these days is getting cheaper, but big hardware projects are largely unavailable to the average tinkerer:
1. You'll spend $100 just on surplus/eBay transistors for such a project, and without also having a comprehensive engineering background (whether formal or informal), you're guaranteed to be blowing those out on each. and. every. single. test.  And it will take hundreds of those cycles before you learn your lessons (pass/fail is a very, very low information-yielding criterion..).  And the rest of the hardware will run $2000 minimum, not to mention batteries (you can't run something like this off the AC line, not when the average household tops out at 30kW total for the whole drop), load (dyno??), motor, all the mechanical and electrical hardware, etc.
2. Development doesn't do much good unless every engineer working on the thing has a model to play with.  (See also #1.) The development/prototyping cost increases arithmetically with the number of developers.  That's insane.
3. There simply aren't very many engineers out there.  And most of them are busy working, and don't much care to do more of it during their off hours.  You'll still get a few hours here or there, but nothing focused.  And lack of focus and involvement is an absolute killer in something as comprehensive as this.

And being that real engineering is in short supply... they probably won't be taken seriously by the overall group/direction/management, leading to more festering of the project, and even less interest from real engineers (why would anyone want to join a project where they won't be taken seriously?).  Perhaps I'm overestimating the democratic hardships of OSHW projects, but I'm probably not wrong by direction, if by magnitude.



So in my opinion, I don't see OSHW being any more than a pipe dream for a product like this.

How can you get real engineers interested?  Draw up a comprehensive project plan, get it planned out by real engineers and managers.  Then get those steps executed.  Again, at least under the direction of real engineers, but preferably using real engineers to draw up the plans, do the layouts, you know.  You'll most likely need money along the way.  Kickstart it, or something?  That leaves it up to the donators: if it's simply not popular or vital or attractive enough, then obviously, no one wants it anyway, and that's as good a vote against going ahead as anything, regardless of how you feel about it.  If it does succeed, then soon enough (assuming it's well enough managed), it'll actually succeed.

But doing it by amateur hours, whenever, no, this simply isn't an Arduino weekend project.  There's almost no way to go from "here" to "there", when it's something this big.

Tim

[Ian.M]

Tim hasn't emphasised enough that there is no practical way that a part time hobby designer, even if they have *ALL* the skills required at expert level, can bring such a project to a successful conclusion (success being defined as a roadworthy electric vehicle + plans and full BOM for building more at a price less than the cost of purchasing one brand new from a country they are available in, importing it and going through all the expensive hoopla of getting it approved as roadworthy, taxed and insured).

The commercial competition is working 9-5, five days a week, 48 weeks a year on improving their electric vehicle designs.  Can you dedicate 40 hours a week to your project?

The competition has a team of engineers and technicians working on each project, with storekeeping/purchasing, IT, and general administrative support, because requiring busy people to work well outside their specialisation is a massive waste of resources.  You have to drop what you are working on to deal with administration just to keep the lights on and the suppliers off your back.  In one week even a small team with support, can put in hundreds of hours on the actual project, and you've just spent 80% of your free time for the week pleading with your bank manager, smoothing your suppliers who are sending progressively more colourful reminders for the overdue bills and updating your project page to try to stop your supporters deciding to diss your project on EEVblog, so have only got a couple of hours of 'real' project work done.

Lets assume you survive all the stress, and your wife doesn't divorce you for neglect taking the kids and 90% of everything.  5 years down the line, you *MAY* have a prototype ready for road trials.   You get it inspected and by the time the paperwork comes through and you've tested oit and tweaked the design another year is gone.   Now you are trying to put together the final production design and you find out that some of your key components are on E.O.L buy status or already discontinued because you bought the ones you could afford five years ago that were already half way through their product life cycle.

Meanwhile, the gear-head brother of someone from the government vehicle inspection department has gone out drinking with a guy he met on the competition's booth at a trade show, who's wife works for the competition's marketing department, your project gets mentioned, and the competition realises there is a market for electric vehicles in your country, splashes some cash entertaining government officials and sews up an exclusive inspection standard requirement for electric vehicles that their current models just happen to meet 100%, and places their latest model which is three generations ahead of yours, on show with dealers in your major cities with a national TV or Radio campaign.

At that point, you are totally screwed, with nothing to show for the past seven years except a broken marriage, a six figure overdraft and a vehicle that is no longer road legal.

How did you get yourself in that mess?  Well you cant put out a bushfire by pissing on it, and you just didn't have the resources in place at the beginning to do the job right.

[T3sl4co1l]

Well, I wouldn't say impossible (taking such a scenario as an example of 'impossible'), given that one could build their own drive, and mod their own car, and as far as I know, still be road legal (well.. depending on country too, for which I believe the US is much less stringent than most).

But the number of enthusiasts who might take this approach is quite small indeed, and existing drives are available for commercial applications (at reasonable prices!).  So even if you complete a design, it's unlikely to be of much benefit to anyone.

Now... I do have the knowledge to design and build such a device, but, I really just don't feel all that interested in it.  (For a variety of reasons, one of which is having only one car to drive.  I like the idea of doing a mod, but I certainly don't have the resources to do that right now.)

And certainly, to build this as a commercial or OEM component, you'll have a long ways to go down the regulatory path.

Tim

[Ian.M]

Of course, if you are mad enough, dedicated enough and have a great ambition other than to get from point A to point B on the road network in some facsimile of comfort, a one off project build is possible.

E.g. http://failover-www.thesundaytimes.co.uk/ingear/cars/article1608045.html though that's a retrofit, to a '70s electric car, not a full conversion.

[rich]

The department where I worked had one of those Enfields cars for power electronics research projects. You didn't even need to drive it to know how heavy it was with all the lead acid batteries - when you looked at it, you would swear it was bending light around itself.

Considering the fact that many countries adhere to low CO2 emission policy, do you think this could start developing a new industry?
In many countries it's difficult to buy a new electric car, so conversion of gas units to electric could be an increasing demand in near future?

As new EVs are produced and sold, then more will appear 2nd hand at reasonable prices. Custom EV conversions will fail to compete on price, simplicity and support/maintenance/warranty for members of the public who only buy a car to get from A to B.  There are already EV conversion companies, but generally they do not design at the transistor level, instead they select commercial drives, batteries, motors etc, and integrate them. Great for car enthusiasts, but not suitable for the masses.

Battery EV conversions will always be niche given the current technology, but if you could find a popular, lightweight, chassis that would accept a cost efficient conversion kit then it could be a perfectly sustainable business model using commercial drives.

If you were looking to start an open source project it is not impossible but just very, very very challenging as others have stated. I believe an open source project like this is will happen in time when the right factors come together, although it might not be battery-EV as we know it today. Most likely financed by a university or a business.  I've thought about trying to get something like this started over the last year as I have a track car project I want to do an EV conversion on, but I can't invest enough time/money right now to stop it becoming yet another festering open source project.

[station240]

I did a quick look and what info I had handy on this area.

Start with the Nissan Leaf's battery pack (assume you got one from the wreckers).
Over 90Kw output, 364.8V, 294Kg
So first thing you need is a way to safely move this, and mount it one day into a car.

Then you need a suitable controller, salvaged from the wreckers also, or built.
For an off the shelf IGBT 3 phase brick, rated well beyond 274 Amps lets say 350A.
Now you need a heatsink, lets say the circuit is 98% efficient, well that gives you 2Kw of heat to get rid of, which means fans and bulky heatsinks (aluminium for less weight).

Safety concerns
1) Dropping any of this equipment on your foot/head, especially the battery.
2) Fault currents, you need things like a 300A rated DC breaker. I'd also recommend a heavy duty Contractor (relay), inductive current meter, etc. Better still test it with 12V low current first.
3) Explosions, despite all the warnings about lithium batteries,  the inverter is the real hazard.
4) Electrocution, rather obvious why.
5) Fire, this and the explosions is why you put the inverter somewhere bunker like. eg inside heavy metal box in a shed/shack, the operate from a safe distance with all your equipment and cables back to the equipment.

Despite all this, the worst that could realistically happen is you blow up a $200 used IGBT brick. As long as you can remotely cut power to the battery, there is no reason to get near it when power is applied.

Problem I see with this as an OSHW project, is people either don't document design criteria, or safety topics. Or the person building it leaves things out, eg a current meter. Proper documentation can take longer than the actual build, it's easy to forget something you did when building it. A parts list should also include technical details for critical parts, and why those parts were chosen.

[coppice]

The department where I worked had one of those Enfields cars for power electronics research projects. You didn't even need to drive it to know how heavy it was with all the lead acid batteries - when you looked at it, you would swear it was bending light around itself.

I was born and raised in Enfield, and it always felt like the Sun's rays were avoiding the place.

[pmbrunelle]

Tim hasn't emphasised enough that there is no practical way that a part time hobby designer, even if they have *ALL* the skills required at expert level, can bring such a project to a successful conclusion (success being defined as a roadworthy electric vehicle + plans and full BOM for building more at a price less than the cost of purchasing one brand new from a country they are available in, importing it and going through all the expensive hoopla of getting it approved as roadworthy, taxed and insured).

The commercial competition is working 9-5, five days a week, 48 weeks a year on improving their electric vehicle designs.  Can you dedicate 40 hours a week to your project?

The commercial competition is also wasting their time doing FMEAs, sending emails, doing supplier management, and generally spending their time doing anything but design. Management probably thinks that that productivity goes linearly with the number of designers, but in reality project advancement goes up logarithmically with the number of staff; as team size increases, so does the amount of communications overhead between the nodes.

I know of a team comprising maybe 3 or 4 guys that designed a proof-of-concept prototype (nothing production-quality) of a regenerative motor drive. It took them two weeks, and they didn't leave work, they slept (not much) at work, and they called order-in food. And no, this was not a modification of an existing product. Some sort of miracle, but apparently true.

I believe in the power of the single designer. However, you must be versed in a variety of subjects; software, analog hardware, digital hardware, power, heat transfer, machining/fabrication, PCB layout. You will have to do this all yourself; motivating other people (without paying them) who do not have your burning desire will be difficult.

Probably a decent hobbyist should be able to come up with a motor drive in about two years, assuming 10~15 hours/week of hobby time (that's more or less the time I spend on hobbies). In this scenario, there would be zero regulatory compliance, but look at something like MegaSquirt. It is said to be "for offroad use only", but guess what, it's used on plenty of road cars anyway.

As for the extent of open-source, I think you can open-source the project in the sense that you complete most of the work, then you start documenting and then publish the design to the benefit of the public.

[Ian.M]

I think your estimate for the hobbyist fits quite well with the scenario I described.  2 Years for the drive,  then there's the battery management systems, preparing the donor vehicle and all the mechanical work to be done.   If all goes well you might beat my 5 year estimate by 18 months or so, but if you have any setbacks and break/burn-out/blow-up anything expensive, I strongly suspect 5 years will be closer for the total project design and build phases. 

Life has a nasty habit of dumping time-consuming problems in one's lap and in any 5 year period, I would expect a significant risk of having to take a 6 month break from the project.

The basic argument and scenario remain the same whether it takes a year or two less or more.

[pmbrunelle]

I just don't see the life-ruining doomsday scenario...

You can spend tens of thousands of dollars to buy a brand-new Corvette, or you can spend the same money over some period of years on a project car.

But that is key... it needs to be disposable "fun spending money"... you mustn't be expecting any sort of financial return.

So yes, I believe that a DIY vehicle electrification project is possible without divorce/losing sanity with the condition of having enough of a stable income.

And so what if it takes longer than expected? Heck, I've had my project car for 7 years now, and it's not yet done. I basically stopped all activity on the car while I went to school for 4 years. It's for fun. I work on my car when I feel like it, and I set it aside when I want to do something else instead.

Besides, if you finish, you'll need to find something else to focus your energy on...

[free_electron]

the biggest risk is : powering it up for the first time , under full operating voltage and current...
you better make sure your insurance policies are up to snuff and you are wearing your asbestos underwear ...

There are safety procedures to be followed.

- make sure that drive train is properly bolted down. and i mean BOLTED DOWN. screwing it to the table is not properly bolted down. it will flip table and all... i'm talking concrete blocks , buried deep enough so it can't rip em out.
- you will need to load the drivetrain... good luck finding a used dyno ... and don't muck with flywheels , that alone is a nomination for a darwin award ... especially at those power levels.
- make sure your fuses can handle the breaking current without just sitting there fizzing popping and arcing... cutting a 300 amp flow is not easy.
- that battery is extremely dangerous. the battery may be rated for 300 amps continous. short it and it will give a tenfold or more of that .... no wearing anything metallic around these circuits. one crossconduction error in driving the igbt bridge and you have full-on china syndrome...
- safetyglasses ALL times ! ( better : full face shield and leather apron. )

- don't attempt to write code to run the inverter using printf to spit out debug information.... hard time deterministic code is needed and proper debugging techniques

-beware of eddy currents ! chopping such large currents at high frequencies can create nasty side-effects ! stray magnetic fields, electrical fields and before you know it everything is alive with the sound of fizzing and popping !

[Ian.M]

I agree its a fun hobby project, if you have the resources, skills and the time, but the O.P. was hoping for more:

... after this i'm enthusiastic about build one for testing purposes.
Considering the fact that many countries adhere to low CO2 emission policy, do you think this could start developing a new industry?
In many countries it's difficult to buy a new electric car, so conversion of gas units to electric could be an increasing demand in near future?
Do we ,electronics engineers and technicians, be prepare for that?

At the end of such a project, if you are lucky, you may be able to get a technical job with an electric car company, but if you are expecting to turn it into a business opportunity, the outcome I described is not unrealistic.

Also, even as a hobby project, with very few EVs in Argentina, key drivetrain components will have to either be custom machined or imported, and the batteries will probably have to be bought new.  Even with Argentinian drivers being amongst the worst in the world, (2010 highest ratio of fatal traffic accidents in Latin America) there just aren't going to be anything like the numbers of scrapped EVs you will find in North America or Western Europe to source parts from.

Also Renault is actively moving into the market, so if Alex wants to be the EV king of Argentina, he's about five years too late . . . .
http://www.buenosairesherald.com/article/163947/renault-tests-100-electric-cars-in-country

[T3sl4co1l]

- make sure your fuses can handle the breaking current without just sitting there fizzing popping and arcing... cutting a 300 amp flow is not easy.
- that battery is extremely dangerous. the battery may be rated for 300 amps continous. short it and it will give a tenfold or more of that .... no wearing anything metallic around these circuits. one crossconduction error in driving the igbt bridge and you have full-on china syndrome...
- safetyglasses ALL times ! ( better : full face shield and leather apron. )

The three of these go hand-in-hand:

- There are "semiconductor" fuses available, which are simply fuses which blow within a cycle (1-2 ms), so it doesn't matter if it's AC or DC.
- They're fast enough to protect the most robust (minority carrier) devices (diodes and SCRs), though hopeless on IGBTs and such (which need electronic protection).
- However, when IGBTs (and such) fail, the arc flash formed at the die and bondwires propels the case and/or encapsulation at explosive speeds.  Short circuit conditions give megawatts of transient power, and when protected by conventional fuses (or none at all), this power draw will persist for tens of milliseconds or longer!
-  While semiconductor fuses aren't enough to protect IGBTs (and such), they do better to prevent damage to capacitors, wiring and battery, and prevent injury to personnel due to propelled shrapnel (less oomph behind the arc flash).

You should still have arc flash / explosion / finger guards in place over the transistors, and really, any place high voltage is available.  Not to mention PPE and OSHA approved safety procedures (safety ropes, blinkenlights, lockout-tagout, verified zero energy state, etc.).

This is more pertinent to industrial 480VAC equipment, but with batteries pushing hundreds of volts, it's quite applicable here, too.  Biggest downside is, the battery doesn't know when to stop, but even an industrial installation has a breaker *somewhere*, though it might be a rather large one (e.g., maybe a 1600A main entry, or some hundreds-A disconnect hanging off a busbar or wiring panel).

Tim

[Alex_arg]

Well well ! Thank you very much to all of you for your contributions, and your advise.
You make me think about the inviability of this project, at least for me, and recall me many things i've no take into account.
Although i have experience in 3 phase 400V equipment ,switching and distribution, i've never use electronics components for
these level of power ( just contactors, thermal relays and the like), you were right about the complex process of build, debug,
and test such device.
Anyway i like to share with you one the pages that led me  to think i can do that, one of them is:

http://www.instructables.com/id/200kW-AC-Motor-Controller-for-Electric-Car/ 

just for your interest and curiosity.

I'll surely find another electronics project to work on, i've build many simple intruments for my work, and thats the direction i should follow.

Again ,thank you very much for your professional advice, and for caring about my post.

A big hug from Argentina ,and read you soon!

Alejandro
 
 

[NiHaoMike]

- make sure that drive train is properly bolted down. and i mean BOLTED DOWN. screwing it to the table is not properly bolted down. it will flip table and all... i'm talking concrete blocks , buried deep enough so it can't rip em out.
- you will need to load the drivetrain... good luck finding a used dyno ... and don't muck with flywheels , that alone is a nomination for a darwin award ... especially at those power levels.

That's the "easy" part. The most obvious place to mount the motor is in the vehicle itself and for a test load, hack the old engine (assuming it's not seized) into a makeshift air compressor.

That said, a much better beginner's EV project would be a bicycle. A kilowatt or two is easily done with low voltage and will probably go faster than you want to. Actually, I know someone who did just that for his senior design project.