S2000 Rotrex + Electric Motor

Do you think a single injector will provide good cylinder to cylinder fuel matching? I've been thinking an FMU may be better at this, as it hits all 4 injectors the same. I asked ChapGPT and she(?) says the single injector will be fine. I'll probably start with the combination and then consider getting rid of the FMU.
 
Do you think a single injector will provide good cylinder to cylinder fuel matching? I've been thinking an FMU may be better at this, as it hits all 4 injectors the same. I asked ChapGPT and she(?) says the single injector will be fine. I'll probably start with the combination and then consider getting rid of the FMU.
You're not running high boost so it should work fine however you could also check to see if there is a fuel injector replacement that flows a little bit more fuel than your current stock injectors and swap them in. The stock computer can adjust fuel within a certain range ~10% to counter the extra fuel, or if you can use one of the adjustable fuel pressure regulators I've seen available for the S2000 you can adjust the pressure downward a little to achieve the equivalent flow to the stock injectors and still maintain the increased flow potential higher up.

I started with this stuff some years before direct injection and had lots of injector flow specs where many injectors were interchangeable with little to no mechanical mods necessary to fit them on domestic vehicles, Honda may be very different.
 
I think I finally got the VESC reliable. It took a lot of trial and error and fooling around, but in the end, I believe my setup is ending up better because of the VESC. I wanted something that I could just get in the car and drive, and not worry about the supercharger or battery (or fluid levels, etc). It would have been more difficult getting a flat boost control, controlling the battery, cooling, etc, without it, but it has been a lot of work. If anyone wants to use a VESC, I can help, but be prepared to struggle some for a while.

It has a wonky On and Off input that took me 4 different circuits before I got it to shut down reliably, but I finally got there. I'm a retired analog circuit designer, so I tried my best to use a transistor, but had to give in and use a micro relay. I don't have a separate switch between the battery and ESC, so I rely on the ESC shutting down when the engine stops, to not drain the battery. You can't simply switch off 12V. I switched over to the micro relay when I realized that if the 12V battery ran down, the VESC would wake up and eventually drain the blower battery. I have a design and the parts to build a solid state power switch for the battery, but as Elon says "No part is the best part". So far I don't need it.

Then there is FOC motor control, which is very sensitive to some motors (like mine) and took days of trial and error before it worked well with my motor. Now that I know how it works, I could get a motor optimized in a hour or so. A basic BLDC controller just works straight out of the box (or so I'm told).

It took me a while before I realized there is a setting that gradually limits motor current at high eRPM. The default was something like 80k eRPM, which I didn't worry about until I realized that is only 5.7k rotor rpm for a 14 pole motor. The motor really woke up when I changed that to 120k.

It had settings that should have allowed it shut down gently with a low cell voltage in the battery, but after fooling around for a few days, I realized those settings didn't do anything. Apparently, it is hard coded for LiPo, and ignores the settings I put in for LTO. I was able to find a work-around, by moving the battery management to the Arduino, but there goes another week of learning about how to move data across CAN. I'm a hacker, not a programmer, so it takes me a while.

These are just some examples of my VESC trials. There are others..

But there is a lot of good as well. Changing motor control over to a CAN bus interface was painful, but it allows me to command motor duty cycle when under boost, and command motor current when shutting it down. DC spools the motor up at max motor current, then the current drops back when the motor rpm gets to where it should be. That gives fastest spool up. A Motor Current command of 0 amps allows it to coast down gently and stay spooled up between shifts. This worked great on the bench, but I haven't tested it much on the car. I have a table in the Arduino of DC vs RPM that it uses to keep the boost constant. I can shape the boost curve anyway I'd like, or have multiple boost curves that I can set from the driver's seat. (for now, I have the Cruise Control switch sent to the Arduino). The CAN bus also gives good noise immunity, which fixed some of the problems I had during bench testing.

The VESC sends back motor temperature acoss CAN which I use to control the fans, and also to slowly shut down the motor if it gets too hot. Battery status is on the CAN bus, so the Arduino gradually reduces power when the cell with the lowest voltage in the pack gets low. I used a VESC compatible BMS that talks over CAN. But it only charges the battery and won't maintain the battery (as I found out late in the design) ie it shuts the charger down after the battery is charged. So I used the Arduino over the CAN bus to control the battery charger, so now I can do whatever I want with it. I have a dumb $30 dc upconverter for the charger, but I can get away with that because the Arduino is in control of it.

I have to admit, my setup is more complicated than most people like, but I have enjoyed this so far, and hopefully it will end up working seemlessly and hands-off in the car. I spent a lot of time testing everything in my spare bedroom (with ear protection and a very understanding wife) and I think that paid off. Everything seems to be working well after 3 or 4 test drives. The main problem is getting the fueling and tune right.

If anyone is interested, I have a complete set of schematics and documentation, which is probably about 95% finalized. The main thing that is changing is control of the engine tuning, which will mostly be done by the Arduino plus some hardware that is still tbd.

Below is a link to the schematics and Arduino code.

 
Top