A bigger motor

[TEG]

Some discussion of Roadster top spead, motor contraints and reasons started up here:

http://www.teslamotors.com/blog2/?p=51

My personal opinion:

#1: At speeds higher that 130mph the wind drag would cause the ESS to have to deliver too much current for prolonged periods. It could also overheat the air-cooled motor.
( Although the "Max Net Power" of the Roadster is rated at 185kW, I don't think it can maintain this power for extended periods of time. The Li-Ion cells Tesla are using are probably rated at 2C nominal which means two times pack capacity output = 112kW sustained. It would take about 112kW to maintain the roadster at 130mph, so going much faster could be harmful to the ESS or eMotor. )

#2: The gears were chosen based on the common measured speeds of 0-60 and 1/4 mile.
( 0-60 is accomplished entirely in first gear, then the shift to second should just get you through the 1/4 mile before hitting redline ).

#3: Since the roadster only has the energy equivalent of < 2 gallons of gas, your range at top speed would be very short.
( The Roadster gets by because it is highly efficient at lower speeds... But at 130mph+ wind resistance causes any vehicle to use a lot of energy fast ).

 

[TEG]

It looks to me like right around 6,000 rpm is where Back-EMF and battery internal IR drop, when added together, start to interfere with developing full torque output.

Is that really it? When I did a little research into AC motor controllers it appears that you have to optimize for a particular voltage and a particular frequency which results in max power at a certain RPM.
Above that speed (RPM/frequency) the voltage has to drop otherwise the transistors (IGBT's for Tesla) can't switch fast enough.
If they wanted the torque curve to stay more flat they could use faster IGBT's, but they would have to be lower voltage parts. They could get the flat torque curve at higher RPMs but max power would probably be less, and lower speeds would have less off the line torque.

For a 3 phase, 4 pole AC motor, syncronous RPMs at 60hz is 1800. If you create an inverter that creates 200hz AC then synchronous speed would be 6000rpms.
My guess is that the Tesla PEM IGBTs could create 200hz AC @ ~400v to get max power at 6000rpms. They vary the frequency and use the same voltage below 200hz to keep the same torque all the way down to ~0RPMs. As they go above 200hz, they have to reduce the voltage because the IGBTs (in my best Scotty voice) can't take any more.

Oh and I heard that motor windings need something like 4x inverter voltage capability, so the windings in the eMotor need to be able to hold onto ~1600v.

Some related references:
http://users.deec.ist.utl.pt/~ineit/imdrives/glossary.htm
(AC motor) "field weakening - Situation that occurs when over-rated speeds are achieved. Usually, over-rated frequency values imposes a flux reduction and peak torque decreases."

http://mag-net.ee.umist.ac.uk/reports/P1/p1.html
"In order to extend their speed range whilst keeping within an economic inverter rating variable-frequency synchronous motors are often operated with field weakening at speeds above a base speed. Applications such as electric vehicle drives require high torque capability from zero speed up to base speed and approximately constant power capability over a speed range of typically 3:1 above base speed."

So I think Tesla has engineered PEM & eMotor to have 6000RPM "base speed" where torque is constant from 0-6000RPMs. Then they may switch to "field weakening" mode which attempts to keep horsepower constant rather than torque.

 

[TEG]

Something slightly funny/ironic:

I have a Solar panel on my electric truck to keep the 12v accessory battery topped off and to avoid using power from the "traction pack" to drive accessories (like power steering, A/C, etc).

The Solar panel has a charge controller to keep the voltage going to the car at 14.2v max. (The panel can go a bit higher if unregulated).

I put a little LCD TV in the back of the truck so I could watch TV while recharging away from home.
I was watching TV at the middle of the day and the backlight on the TV went out (with a slight burnt smell following).

I pulled apart the TV and found that the backlight inverter had a blown transistor.

I studied this little bitty (3"x1") inverter circuit and I can see that it has one major chip (a PWM chip to turn DC into AC) and some "switching power supply" transformer type components to boost 12v up to the voltage needed by the backlight. Well I am fairly sure that I blew the power transistor because I asked it to switch at the same speed, but asked it to use a voltage that was a bit too high. This TV was intended to be used with a 12v wall transformer, and wasn't likely engineered to handle auto voltages that could go to alternator (or solar panel) recharge voltages around 14v. I try to teach myself electronics by goofing around with things, and it struck me a bit funny that I had an "inverter failure" just as I was researching how the Tesla PEM inverter works, and I was just researching about how switching transistors have to have carefully controlled voltages. I guess I got to see first hand (but on a very small scale) what would happen to PEM IGBTs if you ask them to go beyond rated voltage for a particular frequency.

 

[W8MM]

It looks to me like right around 6,000 rpm is where Back-EMF and battery internal IR drop, when added together, start to interfere with developing full torque output.

Is that really it?  When I did a little research into AC motor controllers it appears that you have to optimize for a particular voltage and a particular frequency which results in max power at a certain RPM.
Above that speed (RPM/frequency) the voltage has to drop otherwise the transistors (IGBT's for Tesla) can't switch fast enough.
If they wanted the torque curve to stay more flat they could use faster IGBT's, but they would have to be lower voltage parts.  They could get the flat torque curve at higher RPMs but max power would probably be less, and lower speeds would have less off the line  torque.

I only paraphrased what was explained to me by a Tesla employee while I was taking an HQ tour and asking "engineering" type questions.  They found somebody that could talk "geek" better than the marketing people for me to ask my questions.

IIRC, I intuited from the conversation that the newest motor controller design added more IGBTs in parallel to reduce the voltage drop at maximum motor current.  It seemed related to R-on as opposed to VceSat since paralleling would reduce the former, but have no effect on the latter.  I can't figure out how more parallel IGBTs would help with switching speed, but I can deduce how it would lower R-on and reduce voltage drop across the controller at max motor current.

AFAIK, all electric motors have more back EMF as rotor speed increases, irrespective of their being AC or DC motors.  Do I have this wrong?

Mike

 

[TEG]

This may be the blind leading the blind at this point as I "don't know what I don't know" in this area. (I am just dabbling.)

With that said, I too was wondering how the extra IGBTs helped.
I had thought perhaps they have a way to feed each IGBT less voltage (so it can switch faster) then pool the voltage back together again.
On the other hand it would take a lot of voltage step up circuits to do that, so I wonder if I am wrong.

The other thing I wonder is how close we are to Tesla "secrets" when talking about this. I don't know if there is any "secret sauce" in this area that we have to avoid.

They have showed photos of the innards of the PEM. They have said clearly that they added more IGBTs to help the torque curve.

 

[TEG]

Some related comments from Anatoly and Roy over on the Teslamotors blog:

Anatoly:

"IBGT switching frequency surely is a limitation more important than winding isolation voltage limit. But if someone set several IGBT in parallel, add small inductances for temporal energy buffering and allow IGBT swith one after another with some phase shift it would reduce IGBT swithicng limit impact considerably. I believe hard IGBT switching related limit comes at higher frequency than needed for EV so it is hardly a show stopper limitation to get much higher eMotor power than what exist in Tesla roadster."

Roy:

"All this discussion about limit of motor power. Simple fact: If too much power is fed to an electric motor it overheats and burns up. The insulation will then fail and the motor windings short. This is almost catastrophic."

So based on these comments, my current thinking is as follows:

#1: They started by making the best (air cooled) motor they could with the idea that it needs to produce 250hp in bursts (150hp for extended periods).

#2: They designed an ESS that could ouput 112kW 2C (150hp) continuous, and 185kW 3.5C (250hp) in bursts.

#3: They engineered the PEM to be able to have a fairly good amount of low end torque (200ft-lb) to get nice off the line and get to maximum HP by middle of the rev range (6000 RPMs) before they have to switch from "flat torque program" to "flat HP program". Torque drops off after 6000 RPMs because they need it to so that you don't ask the motor to create too much horsepower.

The IGBT switching limits, and back-EMFs that I went on about are probably things that would come into play eventually if you didn't need to worry about eMotor heat first. Or maybe they were a factor, but the PEM engineers had a bar under which they could design so they only had to dip so deep into their bag of tricks.

(I once asked one of the PEM engineers about "biggest challenges", and there was nothing about EMFs or switching limits mentioned, rather it was concern about making everything fit, and having to keep making design changes to adjust to different requirements)

 

[TEG]

Some other things I wrote in the tesla blog:

# Anatoly Moskalev wrote on June 10th, 2007 at 9:25 am
## To NiMH EV:
## Tesla Motors seems make their choices to keep 200+ range versus to have outrageous 1/4 mile timing and other racing capabilities.

Yes, the more I learn about the inner workings of the Roadster, the more I see how they crafted a "balanced" set of parameters to meet a number of design goals. To me these goals all make sense, and I would NOT want to suggest changing anything. Now, Tesla started with other designs (e.g.; ACP TZero) as a basis, so they had a good starting point, and many of the differences appear to be "incremental improvements", although the ACP people seem brilliant so improving their designs takes some serious skill!

## But if someone set several IGBT in parallel,
## add small inductances for temporal energy buffering and allow IGBT swith one after another with some phase shift it
## would reduce IGBT swithicng limit impact considerably.

Thanks for that information. I know Tesla added more IGBTs to improve their torque curve on their PEM, but I didn't have any idea how adding more would make it possible. Your description gives me an idea of how they may be doing it.

IGBT 101:
http://www.powerdesigners.com/InfoWeb/design_center/articles/IGBTs/igbts.shtm

## I believe hard IGBT switching related limit comes at higher frequency than needed for EV
## so it is hardly a show stopper limitation to get much higher eMotor power than what exist in Tesla roadster.

Really? Well maybe I am wrong that IGBT limits cause them to have to drop torque starting at 6000RPMs, but some AC motor controller design articles I read suggested that IGBT switching limits would cause a torque dropoff just like what I see for the Roadster.

## Choice of voltage versus current for eMotor and ESS most likely was driven by trade off to avoid too much
## electrical isolation complexity on one side and too heavy wiring on another.

Yes, I considered that. Yet another area where it is an "art" to balance all the constraints and still meet the design goals.

## Just the safety issues for mass market will be severely difficult but for some racing it could be done.

Yes, I think they are already pushing the limit of what they would like to do safety wise. There would be safety benefits and easier design to use a lower voltage, but they need some of the benefits of going with higher voltages. If you look at the history of EVs, the voltages keep going up. Year 2000 EVs (like EV1, Rav4EV, EV1) used ~300V systems. Now Tesla is going closer to 400V. Many older home brew EVs use much lower voltages like 96V. Toyota hybrids tend to use a lower voltage battery pack, but then use a circuit to boost the voltage right before it goes into the eMotor.

Some comparisons:
Year Bat Motor kW Model
1997 288 288 30 Prius gen1
2000 273 273 33 Prius gen2
2003 202 500 50 Prius gen3
2005 288 650 123 Highlander Hybrid AWD
2006 288 650 147 Lexus GS450h
2006 245 650 105 Camry Hybrid
2007 400 400 185 Tesla Roadster

Notice how Toyota has been using a more and more aggressive "boost converter" to raise the battery voltage before they run the eMotor. So does that mean that Toyota uses IGBTs rated for use at 650volts in their latest hybrids?

Given that Toyota is limited in NiMH capacity by Cobasys, it is amazing the power increases they are getting on their eMotors even though they can't increase the traction pack size. Their P/E (Power-to-Energy) ratio keeps going up. The GS450h has only a 2kWh pack, yet produces 147kW from the eMotor. (can they really get ~70C power from their batteries?!)

http://www.lexus.com/pdf/models/GSh_driving_performance_guide.pdf
http://www.greencarcongress.com/2007/02/the_lexus_gs_45.html
http://www1.eere.energy.gov/vehiclesandfuels/pdfs/program/phev_rd_plan_02-28-07.pdf
http://www.ipieca.org/activities/climate_change/downloads/workshops/27sept_06/Session_4/Wimmer.pdf
http://www.ricardo.com/download/pdf/R119361S.pdf

It appears that Toyota is using a strategy of higher voltage, lower RPMs as compared to Tesla.

The Toyota hybrids have eMotor horspower relatively flat from around 1200 RPMS through to a modest redline.
(no 13500 RPMs like Tesla)
Their flat torque curve (as HPs rise) goes from 0 up to a modest / low RPM.
(no 0-6000RPM flat torque like Telsa)
Peaks torque/power RPMs:
MaxTorqueRPMs : Model
0-0400 RPM Prius Gen 1
0-1200 RPM Prius Gen 2
0-1500 RPM Highlander & Camry

I wonder how things would have worked out if Tesla had picked different targets:
600volts
Torque peak up to 4800RPMs.
375hp @ 5000 RPMs.
9000 RPM redline
Would they have been able to get the same performance and range if they configured the ESS, PEM and eMotor to do that? From a marketing standpoint, higher torque and HP numbers would have been attractive.
Maybe they couldn't do that because of battery power characteristics, so they had to meet HP numbers with wide power band rather than higher HP.

I think Telsa is really "pushing the envelope" with motor RPMs and high torque peak for an EV.
I wonder why they chose such high revs rather than higher voltage...

========

# Roy wrote on June 10th, 2007 at 7:43 pm
## As many have noted the torque remains constant from 0 to some rpm and then falls linearly.
## At the same inflection point the horsepower stops going up.
## The horsepower does not drop at this point, it is just that it has reached its maximum design power output.
## If excess voltage were to be applied, the motor would burn up.

OK, so I can see them using the "flat torque" algorithm from 0-6000 RPMs then they get to "target HP" and switch to "flat horsepower" algorithm. Yet eMotor HP/Torque graph shows horsepower falling past about 8000 RPMs. Is this intentional due to eMotor heat too? Can eMotor only produce ~250hp for a short time then needs to drop off?

So what happens if you did something awful like pull a heavy trailer with the Roadster such that you were using all of your 200ft-lbs of torque and you got up to ~7000 RPMs and torque started to drop off so you couldn't accelerate any faster. Now you are stuck at max HP for an extended period of time. Would the batteries or eMotor die at that point because you failed to accelerate into the HP dropoff zone above 8000 RPMs? What about someone with a load of gold bricks in the trunk trying to drive up a very steep hill at sustained 8000 RPMs? I wonder if the PEM will limit voltage / horsepower intentionally even if you stay at 8000 RPMs. Perhaps the eMotor graph published by Tesla is only if you accelerate through the range without staying at max horsepower for too long?

 

[TEG]

Related discussion here:

http://www.eng-tips.com/viewthread.cfm?qid=183175&page=1

 

[TEG]

I can't seem to give it a rest...

Some comments from AC propulsion (some of which may apply to Tesla):
http://www.acpropulsion.com/reports/tzero_EVS17_Paper.pdf

"The traction inverter power stage is based on three bipolar IGBT-based smart poles switches.
Unlike most other traction inverters, the switches are made up of paralleled discrete IGBTs in TO220 packages.
This approach allows the heat load to be spread over a larger area, making air cooling possible.
Additionally, the cost per kVA for discrete IGBTs is currently about half that for packaged IGBT modules.
Control of the phase current is through discrete analog circuits, with a lookup table for optimum slip speed.
Patented techniques are employed for high speed stability and for maximizing the motor output in the voltage-limited (high speed) portion of the operating region."

I think Tesla may have used their design as inspiration for what they did.
I think they added more IGBTs to "beef up" the system to make it more powerful.
Notice how AC propulsion says that the higher RPMs are "voltage limited". I think that means they intentionally reduce voltage to avoid overheating the eMotor.

Comparison of tZero motor to Tesla Roadster motor:

TZero : Tesla
181 : 200 ft-lb max torque
177 : 185 max power (kW)
238 : 248 max power (hp)
687 : 6?? max amps
330 : 400 max voltage
326 : 375 voltage at horsepower peak
5000 : 6000 "base speed" RPMs (torque starts to drop off at higher RPMs)
12000 : 13500 redline

As you can probably see the Tesla "PEM" outputs a lot like the TZero "PEU", only with a "little bit more" of everything.
Crunching the numbers above it seems they made it about 10% better.

Tesla's eMotor has to take the abuse of extra volts and RPMs, and still managed to drop about 20lbs or so.

 

[vfx]

This approach allows the heat load to be spread over a larger area, making air cooling possible.

It's worth pointing out here that Pheonix makes a big deal about how their batteries do not need liquid cooling but avoid saying that thier PEM is liquid cooled.

 

[TEG]

I was looking at some AC motor specs here:

http://www.proev.com/P1Motor.htm

They show some RPM vs Torque vs Horsepower numbers.
If you graph their #s, the general shape and trajectory of the HP & Torque look a lot like those from Tesla (although their torque peak is at a lower RPM).
It is interesting to note that at 340VDC they have flat torque up to 3500RPMS, then switch to flat horsepower for a while then it slowly drops off... Then you look at 380VDC and the (same amount of) torque stays flat up to 4000RPMS before the HP plateau.

Those charts don't say anything about the inverter. It is just on a page talking about the motor, so I don't think IGBT switching comes into play at all there. Assuming that motor heat is a limiting factor then I would have to assume that at very high RPMs you have to limit torque & HP to prevent overheating. (Something must turn more of the electrical current into heat)

I don't know if their chart holds true for other AC motors, but it does suggest the following:

#1: Higher voltage doesn't provide any more peak torque. ( I guess they just use more AMPs at lower voltages )
#2: Higher voltage appears to push the torque peak to a higher RPM. ( Why? )
#3: Higher voltage can provide higher horsepower ( because you can keep the same torque at a higher RPM )

 

[TEG]

More research suggests that the Tesla PEM could deliver more battery power to the eMotor at high RPMs and keep the HP up there, but it would cause issues for the batteries and/or the eMotor.

So I feel fairly confident now that the PEM or IGBTs aren't really the limiting factor here.

So if you made a "better" eMotor and got better batteries you could probably reprogram the PEM to make it take advantage of the enhanced capabilities.

If you wanted to go racing you could probably change the PEM software to get some extra performance in exchange for reduced reliability and durability.

 

[W8MM]

Notice how AC propulsion says that the higher RPMs are "voltage limited".  I think that means they intentionally reduce voltage to avoid overheating the eMotor.

I don't think that is the case.

One should think that back EMF caused by RPMs limits the maximum speed garnered from a given battery voltage.  Once the back EMF exceeds (and opposes) the battery voltage, then no more current can flow through the controller, thereby setting a current-starved maximum  motor speed.

What's so hard about that?

 

[TEG]

The confusing part to me is just what limit comes into play.

Some people tell me that motor HP is limited by the ability of the eMotor to handle heat (otherwise the windings melt), and once you get to max possible HP, you have to start reducing voltage at higher RPMs to avoid making too much power and heat.

Others have said that IGBT switching limits prevent you from creating proper field frequency at high RPMs if the voltage is too high.

Others have said that the Motor Controller and eMotor have a max "base frequency" and you have to drop voltage as you go into "field weakening mode" beyond the base frequency.

And (as you said) another camp is that back EMFs from the motor are what causes a limit to how much voltage you can apply.

I have basically given up on asking, because I don't know who is a real source of authority to say for sure.

At this point I now suspect that the IGBT switching limits aren't the gating factor.

I think Motor heat and/or back-EMFs (or some combination of the two) may be what causes the (intentionally designed in) HP dropoff (and likely voltage drop), but I really can't say for sure. Sorry if I am missing something obvious. I am not an EE or eMotor expert even though I play one on TV.