Hi power inverter for Nissan leaf motor. Dyno's 302.3hp p15

Thanks everyone. And this was a Team effort. SO MANY people have helped me get here. I can't name everyone but its been amazing to see the EV circle I am in and how helpful almost every person is. So beneficial to man kind when we all come together like this.
 
fechter said:
Arlo1 said:
Secured the record for fastest Street conversion in the NEDRA for 1/4 mile and I think they can add the 1/8th mile as well lol.

http://www.nedra.com/records-sc-History.html

And by quite a margin. Way to go! You just raised the bar for the others. And with front wheel drive too, that's amazing.

Yeah I keep telling him to put a second motor in the rear... :mrgreen:
 
Lebowski said:
fechter said:
Arlo1 said:
Secured the record for fastest Street conversion in the NEDRA for 1/4 mile and I think they can add the 1/8th mile as well lol.

http://www.nedra.com/records-sc-History.html

And by quite a margin. Way to go! You just raised the bar for the others. And with front wheel drive too, that's amazing.

Yeah I keep telling him to put a second motor in the rear... :mrgreen:
That was the plan but with the batteries there is no room left
 
Hi Arlo

Have been following your project from a distance. Congrats on the very good results!
Got some questions about general ev-projects and the leafmotor, as you have been doing heavy diy and probably weighted most options.

The power you have gotten from yours is well enough, but can that be achieved with off the shelf parts as well?
If not, what would be your advice to someone jumping into a conversion? The little browsing on ev forums I have done, I see advice on 50kw brushed motors basically.
I am looking to convert my s2000 that I would like to be silent and electric (today an e85 turboconverted car dynoed to 420whp).
Less power is ok, as long as I can achieve north of say ~200kw, I would be happy. (Dont want less performance than the standard s2k, as that is much of the cars identity)

This has been in my mind for a few years now and with more and more ev batteries becoming available second hand, things are looking more promising in that regard.
The initial steps would be to get a grip on the size of the components that could be used, and then see if its a good idea or not.

What are your thoughts about a conversion like this? Could the leaf motor be a candidate or should I look elsewhere?
 
The s2k makes 200kw for a very small window the car will be just as fast with less power when you do it right. But on that note always build for the most power you can. My limits are the battery and controller but not for long in fact if I dynoed right now it would hit 330-350 at the wheels with the 800a parts which are maxing out the batteries.
 
So.... I broke down the other day... Turned out to be just a bad ground to the UVW encoder....

Non the less I came to a red light and stopped and then it would not go again.... I pushed the car up hill and around the corner into a gas station parking lot and looked at it... Realized the hall sensor was the problem some how.. So I pushed the car up hill across the parking lot and tried to push start it into sensorless speed with no luck.. So I walked to my shop a few blocks away... I got my bmx and peddled my ass home and got my computer then some tools and a spare brain chip as that had caused this 1 time on me when I accidentally fed more then 5.5v to the brain for a while... Non the less non of that solved it.

So I got my brother to tow me up the hill to my shop and I started looking at it. The encoder had minor rubbing on it so I adjusted the clearance and put a new on it then it worked! So I went to drive it home then it didn't work anymore... Put it back in the shop for the night. Then yesterday I got it running consistently well and spend the time to scope the hall signals to 100% perfect! Then It was so amazingly smooth! I was so happy... I decided I had not found the actual problem and put some CP70 on the connections and then it didn't work again this is when I found the ground had finally broke..
at this point I fixed it good real good even tried attaching the shield on each end of the sensor wires to maybe help the ground... I know everyone else says no only 1 end of the shield is hooked to ground... I tried it and it was horrible. As I give it more throttle and or more RPM it would get very rough. I made a video you can hear it. So I quickly snipped one end of the shield away from the ground and that did not help.

So after all that I realized what made it super awesome and buttery smooth... The ground wire was breaking and as it got weaker it gained resistance this resistance between the ground and the encoder helps make a filter... SO I pulled it back into the shop to excited to sleep at midnight and added a 15 ohm resistor in series between the negative and the encoder and BAM its perfect. Its now 100% OWM quality for starts!!!

I used my dash cam for a couple videos.
 
Here it is with the ground repaired!

[youtube]39J74cAv1bg[/youtube]
 
And with the resistor! Fixed and running awesome!

[youtube]AJyNw_UIRNQ[/youtube]
 
Basically you got it the way it was always intended to be :D

About being stuck at the gas station, there's a good chance setting it to sensorless would have gotten you home. On the flat with little load it will start from standstill, on my bike I can ride away sensorless under power with no help pedalling. Even the tinyest bit of movement and it will run in sensorless. About 'little load', this is load torque resistance relative to motor torque. With your 300hp a CRX would be 'little load' :D

For the transition hall-to-sensorless it looks at the hall speed, so with no valid hall signals you can push all you want, it'll not go to sensorless.

The UVW encoder chip has quite a large system on board with analog hall sensor, ADC converters etc etc. A stable supply is essential for its correct operation. Your 15 Ohm in the supply makes a low pass filter with the supply caps, meaning the UVW gets a more isolated and stable clean 5V power supply. I would move the 15Ohm to the 5V line... and my professor (who worked with sensative measurement equipment close to Nuclear Fusion tokamacs) tought me to always ground the isolation on both sides !

See this thesis for the concept: chapter 2 page 13 (19 of the pdf file).... this gets very complex in a hurry
View attachment 448622.pdf
 
I know you keep telling me that...
But everyone else says ground only one end of a shield and its because you can create ground loops if you ground more then one end.
 
Thanks for posting that paper Lebowski - constant debate about this for CAN cable too.
 
encoder manufacturers will say how they want their equipment grounded and i see it in their document as grounded at both ends. ya you definitely getting serious ground loop current if a resistor in this ground cable helps. this resistor with the current flowing through it will make your resolver's reference not "zero". thankfully its 5V so you got some room. you can also try a ferrite bead instead of resistor. you get the impedance but only for high frequency, DC or low frequency will see no resistance.

why you get so much on the ground wire? well... i know part of the story. recall our recent emails on subject of dV/dt and "common mode noise" from switching IGBT traveling through windings, through motor bearing to shaft. once on the shaft it hits the resolver, crawls over case and onto cable shield going back to inverter.. this noise stuff is creepy, it goes everywhere and causes strange results. noise... so freak'n hard to debug. its impressive you were able to find a solution so fast!

with your super large IGBTs coming those things are going to have lots of thermal capacity unused. they won't run hot (my thinking). so you might have opportunity to decrease your switching rise/fall time to reduce system noise. and adjust deadtime to suit new rise/fall times. don't let deadtime represent more than 5% of your switching period. if you had to meet a design spec for EMI this wouldn't be enough.
 
In practice, you drain the shield to the controller first and see what happens. If it's noisy, drain shield to the motor side. If thats noisy, then try the mid point of the shield and drain to chassis, then try motor, then try controller, then try draining shields to random points on the vehicle using star terminations.

Whatever direction it starts to get better in, you keep doing that thing more until it gets worse doing more.

To date I've not yet seen an EV application where draining from both ends was the lowest noise solution, but after weeks of my life spent in EMI chambers working out solutions to pass emissions tests and em immunity testing, I feel like I know less about it.

Shielding and noise reduction at EV power levels is vastly easier to not make the noise when possible vs trying to convert it into heat with external bits. Anytime a ground or shield isn't single ended and/or star terminated it will create potential for recirculating current induced noise in my limited experience.
 
Good information guys.

For the record. I tried not grounding the shield at all tried grounding both ends and tried grounding one end then the other only 1 at a time.
But I did not try grounding to chassis. I grounded to the common for the brain board which is connected to chassis ground. Reason for the common to be grounded from the brain to the chassis is for the gauges to work its a long storey but there is a method to my madness.

PS I love doing experiments like this!

I would love to spend time in an EMI chamber.

-Arlin
 
I would love to spend time in an EMI chamber.
so long as you don't get claustrophobic it can be very interesting. and i agree with luke, its hard to ascertain "the rules" of EMI and noise.. the subject seems to have a chaotic life of its own. what i learned the hard way is to do absolutely everything you can in hardware to mitigate conducted/radiated emissions at the design stage because debugging after its been built always seemed like endless trial and error.
 
no because we want to know where the frequency will be exactly at high energy (not spread out with lower) so we can taylor a single highly tuned filter to suit. that's an LC hardware filter. hopefully i understood your question. if you could meet DO-160 or MIL-STD 704 with spread spectrum and NO hardware filter, then that's something else.. but so far as i know its not possible. you need the filter and therefor you need to know exactly where in the spectrum the noise is.
 
The idea behind a spread spectrum clock is to not have a single strong (interference causing) frequency but to have many much smaller ones.

3503Fig03.gif

The clock source driving the PWM (or the entire controller processor), standard you would have a single tone for your clock like the blue curve above (coming from a Xtal oscillator for instance), this makes that the interference etc caused is also a (or many) single high amplitude peaks.

With a spread spectrum clock you make a clock signal having the yellow spectrum. The interference energy from this is spread out over a larger area, and with a lower amplitude...

There are chips available that generate a clock signal with the yellow spectrum, these can be used to drive the entire controller and reduce the local peaks in the EMI, spreading the EMI energy over a larger area instead... maybe, as I have never tried this.
 
yes i know what spread spectrum is. maybe in the digital world where noise is low it's good strategy. in power world its not good becuase if you spread out noise over larger spectrum all you get is a larger spectrum of noise ALL of which exceeds your allowable limits (DO-160, MIL-STD 704) because the spread spectrum will not lower the noise so low as to pass a real EMI spec. so.. how are you going to filter that? keep in mind we're talking power.. filtering 1000 amps... 600Vots... if you have a weight/volume/cost method of filtering a huge range of spectrum let me know so i can patent it.
 
Skip ahead to the 20 min mark.


https://www.discovery.ca/Shows/Daily-Planet/video?vid=1247385

EDIT: if you can't get the discovery link to work try this one it has an echo on the narrators voice but its almost as good.
https://review.bellmedia.ca/view/468036009
 
Lebowski said:
'Not avaliable for viewing in your region'.... :(
Will try to get it on YouTube. But maybe use a vpn
 
Congrats Arlo! Thats awesome that they featured your project. I'm sure that will inspire more people to build performance EVs.
 
Back
Top