DIY Electric Car Forums banner

New controller prototype

122K views 537 replies 67 participants last post by  Qer 
#1 ·
The controller enclosure being machined:

http://tesseractcorp.com/******/Enclosure_Prototype_1.jpg

Next is the topside view (mostly-assembled, coolant passages not yet milled). We aren't 100% thrilled with the fans but the internal copper heat-spreader for the IGBT module(s) does increase the effectiveness of the heatsink, and the calculated volume of air required to limit the temperature rise of the heatsink to 25C with 1000W dissipated is only 45cfm; those two rinky-dink fans each move 44cfm so with an allowance for backpressure they will technically work fine (and 1000W is pretty much the worst case scenario).

http://tesseractcorp.com/******/Enclosure_Prototype_2.jpg


Next is a view of the laminated bus structure made from two pieces of 0.090" thick copper plate and 10mil mylar for insulation. Notice in the lower left hand corner there are two EV200 contactors installed (for a two-module controller - only one will be used for the one-module version). This allows the microcontroller to supervise pre-charging the film capacitor (hidden under the bus plates) as well as protect against hooking up the battery pack backwards, exceeding the allowed input voltage or catching on fire should the output section blow up... :rolleyes:

http://tesseractcorp.com/******/Enclosure_Prototype_3.jpg


Finally, a close-up of the interface section, showing the industrial ethernet jack (the matching housing for the plug can be installed "in the field" without any special tools), LEDs for power/status and error codes, and sealed barrier strips for 12V power, throttle, contactor interrupt ("E-Stop"), coolant pump, tachometer, etc...

http://tesseractcorp.com/******/Enclosure_Prototype_4.jpg


Now I just need to finish designing the new pc boards for this beast...
 
See less See more
4
#9 ·
Very cool! Can't wait to see what all the specs are on it and pricing.
Specs so far is 750 Ampere peak and 500 Ampere continuously (provided it's properly cooled) for the "small" one and 1500/1000 for the next step. Voltage is, uh, to quote Tesseract "North of 200 Volt" or however he put it. Not sure if he's finally made up his mind about HOW much north of 200 Volts...

Most stuff is controlled by a microcontroller (some panic stuff is made in hardware just for added safety/reliability) and all user settings will be done through an ordinary browser (yeah, the serial port got scrapped). Price is not decided yet, but expect it to be at Zilla levels rather than Nice Price.
 
#13 · (Edited)
I'm curious as to why you program your speed controller through a browser? An Ethernet port? Don't get me wrong it looks rock solid, and I'll definitely consider it when I look for a controller in a month or so. I'm just wondering what the interface will look like.

Why don't you call it the Titan? It looks like one!

In Greek mythology, the Titans (Greek: Τιτάν - Ti-tan; plural: Τιτᾶνες - Ti-tânes); were a race of powerful deities that ruled during the legendary Golden Age.

http://en.wikipedia.org/wiki/Titan_(mythology)
 
#14 ·
I'm curious as to why you program your speed controller through a browser? An Ethernet port? Don't get me wrong it looks rock solid, and I'll definitely consider it when I look for a controller in a month or so. I'm just wondering what the interface will look like.
That's not uncommon at all. The interface will be a webpage much like configuring a router for broadband. You'll have fields you can change the values of and then probably be able to save those settings. The other option being a serial interface with either a clunky text interface or the need to install software on the client machine which needs to be OS based. This allows you to connect a computer running any OS with a browser. Pretty cool, can't wait to hear and see more about it.
 
#15 ·
How about:

Power House 750
Big Block 750
The Flood Gate 750

"What kind of controller are you running in that EV?"
"Well I have a Warp 11 with a Power House 750 Controller!" ARR ARR ARRR!

"What kind of controller are you running in that EV?"
"It's a Big Block 750 coupled to a Warp 11" ARR ARR ARRR!

"What kind of controller are you running in that EV?"
"I just installed the latest Flood Gate 750 from TQ enterprises. It can cause a brown out in southern california if you floor it at a green light."


Yeah, they all sound about right. I like the look of this thing too. Takes that chinese crap out into a dark alley and kicks the crap out of them.

Good luck with the name! Can't wait for it to hit the markets.

Aaron
 
#17 ·
Power House 750
Big Block 750
The Flood Gate 750
Hmm. I kinda like Power House and Big Block makes me think of V8's and POWER. Dunno why... :cool:

Flood gate gives me a bad feeling, probably because of the image of water and electronics... :eek:
 
#23 ·
You might want to have those copper plates plated. I have used tin for copper bus bars...
For standard bus bars that are separated by inches I would totally agree with you, but the two plates here are only 0.01" apart and pure tin has a bad habit of forming whiskers while tin/lead plating (ie - solder), which doesn't whisker, is no longer permitted because of the RoHS directives in Europe... It does make for quite a dilemma...
 
#24 ·
Big and heavy? :rolleyes:

Without the mounting plate (size not yet determined) it is approx. 18.5" long by 7.75" wide by 4.5" tall (not including the fans, especially since they might get thicker). It weighs about 28 lbs as of now. I definitely wouldn't want to drop it on my foot!
 
#34 · (Edited)
electroless nickel is easy, we used to do it in kids swimming pools, like the ones you buy a K-mart. hard and soft anodize (mil-a-8625 ty 1 or 2), and chromate too. chrome was too hard because of the polishing required.

NASA style QC would be out, but you can get really close by being a control freak. some of the cooling pockets on top won't plate all that well on the bottom.

look up local planning rules. Some places like LA county get really strange about "hazardous metals" processes disposals.
 
#35 ·
electroless nickel is easy
...
look up local planning rules. Some places like LA county get really strange about "hazardous metals" processes disposals.
This is exactly what I was thinking when I wondered if it would be feasible or not for use to do it ourselves - that the regulatory hurdles might be more than we want to bother with.
 
#36 · (Edited)
the actual only hurdles are the chemical handling pre and post treatment.

elec nickel has fairly benign characteristics to the environment. I suggest a "visit" to a local processor on the pretext of doing a bid for your current object d' art. pick their mind. if they are a good shop, they will share their knowledge because they know what a PITA pollution controls are.

all the tables and equioment requirements are available online.

Yes it is cheaper to plate in house, but ONLY because your overhead costs are lower. The only other reason is wierd quality or timing issues that cannot tolerate a vendor's quirky schedule.
 
#38 ·
piotrsko - I did some more research and it appears there are RoHS-compliant EN processes that, unlike their soldering counterparts, are just as good as the old formulations that contained lead and cadmium. Thanks for nudging me along on this. :rolleyes:

Bowser330 - I expect 340VDC input will be the final safe rating as the IGBT and capacitor are both rated for 600VDC. I won't know for sure, though, until I've pounded the crap out of the controller with at least a 24 x 12V battery pack.
 
#42 · (Edited)
piotrsko - I did some more research and it appears there are RoHS-compliant EN processes that, unlike their soldering counterparts, are just as good as the old formulations that contained lead and cadmium. Thanks for nudging me along on this. :rolleyes:

Bowser330 - I expect 340VDC input will be the final safe rating as the IGBT and capacitor are both rated for 600VDC. I won't know for sure, though, until I've pounded the crap out of the controller with at least a 24 x 12V battery pack.
Can't wait to hear about the testing results...thats the only way to really know the limits...

About Regen...I really wonder about the calculable benefit, especially if you do more highway driving or driving during lighter traffic times...in some cases it could be a waste..
 
#40 ·
The controller hardware is capable of regen (half-bridge configuration with both legs pwm'ed) but it won't be implemented in the software until extensive testing of many motors on a dyno has been done.

You've been around EVs long enough to know that doing regen with a series dc motor - without blowing up the motor, battery pack or controller, that is - will take something just shy of a miracle to pull off.

My understanding is that if your motor is not neutrally timed with interpoles then you can pretty much forget it, no matter how fast and well-compensated the feedback loop inside the controller is. I intend to find out, though, as this seems to be one of those things that if you ask 5 different engineers about you get 6 different answers... :rolleyes:
 
#43 · (Edited)
Their H2 model has regen as an option, but from what I've read it seems to blow up with alarming regularity if it is actually enabled. That may be more the result of attempting regen with an advanced timing motor rather than a design fault with the controller (to be fair, though, the controller still ought to be able to protect itself).

Unfortunately, Zapi's website is just a step above useless and no one over here seems to have the H2 in stock... Ah well, we'll go it alone as usual ;)


Bowser330 said:
About Regen...I really wonder about the calculable benefit,
From a recovered energy standpoint it makes very little sense, especially with lead-acid batteries because the Peukert Effect affects charging just as much as it does discharging. Yep, the more amps you try to cram back into a Pb battery the more get converted into heat. If the charge rate is kept at 1/5C or less then charging efficiency can be as high as 90% efficient, but at higher rates the charging efficiency can drop to 50%, and is awfully dangerous to boot (raise your hand if you want a battery chock full of sulfuric acid exploding inside your vehicle!).

And keep in mind that regen requires twice the conversion, so you have to multiply the efficiencies of every part in this chain twice! For the sake of example, let's assign: 85% for the motor, 95% for the controller, and 50%-90% for the battery. Motoring efficiency, ie - driving, will range from 40% on the low end to 73% on the high end (neglecting drive train losses). Efficiency during regenerative braking will then range from an unremarkable best of 53% (0.73 x 0.73) to a truly awful 16% :eek: ...

That said, there is one benefit to regen that makes it a compelling feature despite the abuse it heaps on batteries, controllers and motors: it enables an electric motor to simulate "engine braking" just like an ICE. Since pretty much everyone learned to drive on a vehicle that behaves the same way, regen will help make the transition to electric seamless. Also, psychologically speaking, it really seems to annoy the hell out of EV drivers when they have to convert their forward momentum into heat via the brakes (rather than convert it into heat in their batteries, I know, I know... not rational, but we all have our moments... :rolleyes: )
 
#50 · (Edited)
Are you going to be casting that case?

If so you should really rethink placement of the fans. They're currently not blowing across the sink's fins making it nearly worthless to have the fans (unless there's another sink underneath that sink, but even then). Likewise even if they were it wouldn't be much heat removal (fins are not very deep)

The sink will effectively act like it had no fans at all in that set up, even if you have another fan on the side to blow air through the circuit ... the sink's design (with turns in the sink fins I assume to increase convection pick up) isn't being utilized.

You should also look into mounting the IGBTs on a circular fanned sink like I attached.

If you built a decently tight case around that (and forced air at high speeds though the sink) it would be significantly more efficient and keep the IGBTS cooler (which of course helps efficiency).

Also current copper prices are extremely low... copper sinks are beyond reasonably priced now

http://www.metalprices.com/FreeSite/metals/cu/cu.asp

it's down to what aluminum prices USED to be $1.25/lb... Not bad for twice the thermal conductivity.

If you're casting that case in china (which you should definitely look into first since mold costs would be a mere $1000 or so), expect to get copper for about $0.60/lb in ingots
 

Attachments

#55 ·
Are you going to be casting that case?
Only if annual volume exceeds 500 or so units. The aluminum alloys suitable for casting are very different, thermally and mechanically, from machined 6061-T6 and it would pretty much require a total redesign of the housing.

That said, the goal with this enclosure was to require as little human labor as required to make it. Same with the mechanical design of everything else.


If so you should really rethink placement of the fans. They're currently not blowing across the sink's fins making it nearly worthless to have the fans (unless there's another sink underneath that sink, but even then). Likewise even if they were it wouldn't be much heat removal (fins are not very deep)
Impingement cooling - in which the fans blow down onto the fins - is actually more effective than the traditional method of directing the airflow across the fins. Every CPU and high end graphics card uses impingement cooling, btw.

All of this is mostly irrelevant, though, because forced air convection is only the backup cooling method - the primary means is liquid cooling.


[Copper is] down to what aluminum prices USED to be $1.25/lb... Not bad for twice the thermal conductivity.
Copper machines very poorly (it galls and work hardens), it is still more expensive than aluminum and it is a lot denser. The thing already weighs 28 lbs, you know.

That said, there is a copper heat spreader in between the module(s) and the heat sink which effectively multiplies the baseplate area of a two-module controller by 53% and a single module controller by a ridiculous amount (~3x). This is mainly to let us get away with drilling fewer liquid cooling passages.


If you're casting that case in china (which you should definitely look into first since mold costs would be a mere $1000 or so), expect to get copper for about $0.60/lb in ingots
I am not nearly the Sinophile that you are, but even if I was more favorably disposed to outsouring manufacturing to that country I doubt I would anyway because the expected annual sales volume won't justify it. And then there is the "slow boat from China" problem - a lean manufacturing setup can not afford either to wait around 10-12 weeks for a shipment to come in, nor hold a massive amount of inventory to cushion against demand.

As of now the machining time per controller is... rather long, I'll admit, but after optimizing the tool paths we expect to be able to crank one out every 6-8 hours. If it looks as if sales volumes are even approaching that limit then we will next look into having one of the local places that does custom aluminum cylinder heads and such into making sand castings for us.
 
#52 ·
IGBTs should be directly mounted to the sink similar to output MOSFETS in an amplifier... that'd be the best way to rig up the setup which is why I suggested another heatsink type as well.

You definitely would want fairly thin fins but lots of them, copper and probably an axial or cylindrical fin design for efficiency. Likewise controllers are bound to be a tad noisy... so I say just dump as much velocity from a fan system as you can get... (ie. 2000-4000rpm fans through a small area of case surrounding the sink).

Might as well make it 25C all the time :cool:

I'm not sure how much heat is actually produced though... any sort of wattage figure would be helpful in figuring that out. I've seen 2000-3000w of heat be dissipated without fans before (class A amps)...
 
#54 · (Edited)
#57 ·
lets see: 5 a week, 250 a year, need more just add 'nother machine, don't even change code.......

6061 can be cast, but not there in FLA unless there is huge aerospace facilities using same process. still at least 3 hrs machining, just less chips to recycle.

Looks to me that you have covered multitudes of really intricate and obscure details, and done so in an impressively sophisticated manner not readily apparent to the non-production-manufacturing oriented.

Is it done yet??
 
#58 · (Edited)
Is it done yet??
Let's just say there's still a few details *cough* that has to be *coff* ironed out. :D

Here's a tidbit about the software as a contrast to all this hardware talk, software controlled soft start:



Something is acting up slightly right now, it shouldn't take 60 ms for the PWM to start running. It probably has something to do with the recently added www-server that has had some unforeseen effect on the code. Gonna look into that tonight, really. Anyway, here's a close up of the soft start:



As you see it dithers the PWM to compensate for the fact that the pulse has to be a few us minimum to not blow the IGBT to Kingdom Come. Hopefully that will mean that the controller can start the car without a jerk.

Oh, and here's a test web page that the controller serves while running the pwm:



And no, the teddy bear won't stay. It's just there for testing that the server can handle both pictures and html. Anyway, this does not mean that the software is finished, but I'm getting there. Sloooowly... :cool:
 

Attachments

#60 ·
Being an IT project manager and programmer myself I should have known that last 20% of work usually take 80% of time...
But that would also mean you know how important it is to not rush like mad to hit an impossible deadline to guarantee the quality. After all, you don't want to drive a version 0.1 alpha pre-pre-release, do you? :D
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top