[DRIVING] Renault Clio Electrique -97

[bexander]

Finaly got a more permanent inverter installation done. Still need to sort out the motor cable as it didn't fit as I planed, hence the blue tape. Best option is to source a original MG2 cable for the Yaris inverter and adopt it on the motor side.
Mounted a water cooler on the inverter coolant as well. Test and calcualtion showed that the temp rise would be to large. With the Leaf setup I never needed a radiatior as there where so much mass in the motor to absorb the heat.

When the mtpv SW is ready, I'm ready to test!

[Pete9008]

Looking good, must be about the emptiest engine bay I've seen (and the smallest radiator) - little surprised at the size of your coolant header tank though, there must be almost a kg of spare liquid in there!

How much power do you think you are having to dissipate in the cooling loop?

Done a little more on the pseudo-MTPV algorithm in that it's now been tested on a low L motor simulation too. It needed a few tweaks but now seems to work pretty well, need to check the changes haven't affected high L motor use though. The main blocker at the moment is that I would like to test it on a F103 board before releasing it, just to check the execution times are OK, and my bench and test kit are currently occupied with F405 hardware. Hopefully have a look at it towards the end of the weekend though.

Have you tried the over-modulation yet?

[bexander]

Well I did a test drive without the radiator and with 840g of coolant (50/50 water glycol, with specific heat of 3,3kJ/Kkg) I calculated a power loss of approx 60W based on a temp rise of 20degC and a duration of 15min.

With the radiator the total amount of coolant is closer to 1200g. Before I added the radiator I considered just adding in a larger header tankt to provide a buffer but if I want to drive longer distances it might be a bad idea. There is a 120mm fan in front of the radiator as well which in turn should help circulate some air over the motor as well (air cooled only). Always feels good to get double use out of thing...

Did a drive today and the inverter temp never got above 10degC with ambient temp close to 0degC so seem to make a difference.

The radiator is a heater core from a daewoo matiz. A 140mm fan pretty much covers the whole heater core. Weights less then 800g incl the fan.

Not tested over-modulation yet but it's getting closer. Made me a bit nervous when I relized I have to change the code an compile it, don't want to mess something up...

EDIT: Corrected fan size to 140mm.

[johu]

You can grab a binary here: viewtopic.php?p=48964&hilit=modulation#p48964

[Bigpie]

Awesome. I need to try over modulation too

[mjc506]

Square waves are fun Who needs a nice smooth sine?

[bexander]

Yeah, my thinking is that for high inductance motors, like the Toyota motors, it should smooth out that square wave anyway. Might introduce more interference, specially for low inductance motors, but who knows...
Waiting for daylight to give it a test.

[bexander]

Alright, the over-modulation do indeed increase power with approx 10%.
I got, from battery, 45A @ 387V @ 70km/h compared without over-modulation 40A for the same case.
That SW stays in the car! Would be a good idea as Johu already suggested to make modmax a parameter in the future.

The log from my very short test drive, if anyone is interested.

log 14.csv

(83.84 KiB) Downloaded 327 times

[johu]

Nice Same finding on Audi.
I wonder where the upper limit for this is and whether it's diminishing return (or eventuell an int overflow). In the extreme case you'd end up with 6-step modulation but which modmax would this correspond to?

edit: thought I'd plot your data

[bexander]

Did a longer test drive and the oscilations at high speed and low current demand is with over-modulation barely noticable so a good improvment here as well!

A quick serch on the topic of over modulation lead me to this article,
https://microchipdeveloper.com/motor:overmodulation
Seem like 1.15 and above no longer increase Uq, but I'm not shure I understand if 1.0 is what is run in the original code or not?

[Pete9008]

A quick serch on the topic of over modulation lead me to this article,
https://microchipdeveloper.com/motor:overmodulation

Good find, particularly interesting that over-modulation can start to link Uq into Ud at high levels.

Wondering whether the best way to do over-modulation is to have two parameters, modmax and modmaxfrq? The later is the frequency by which the modmax value should be fully used. Below that frequency it gradually increases from the standard SVM value. It means that we don't get the waveform distortion at low frequencies but still get the benefits where it's needed (see SciroccoEV's post viewtopic.php?p=48646#p48646 for another reference)

[bexander]

I have now reached 70Wh/km as a new record while driving to work. Best I ever got with the Leaf setup on the same distance is 77kWh/km.

Some of that improvement is due to 40kg lower weight but most of it is due to the MGR motor setup itself. My guess is that the MGR do have lower peak efficiency compared to the leaf setup but for my use case, at cruise, there seem to be quite some improvement.

[Ev8]

So about 14km or 9miles per kWh? That’s extremely impressive with most production evs hitting 3-4.5 miles per kWh

[bexander]

So about 14km or 9miles per kWh?

Correct, but this is with my own driving, keeping speed at or below 70km/h in a car with just the leagally required stuff. Don't know how it would preform in WLTP.
Got more improvments comming as well...

[celeron55]

Soon you'll have to get a big hammer and get the Clio more into the shape shown in your profile picture, for the true peak efficiency.

[bexander]

Soon you'll have to get a big hammer and get the Clio more into the shape shown in your profile picture, for the true peak efficiency.

Yes, that thought has crossed my mind more than once but my panel beating skills is not up to the task.

[Pete9008]

Nice, some impressive results there

[bexander]

Been driving almost 4000km now and it works well except when the battery voltage drops down below 360V. At lower voltages it starts to oscillate at high speed with low torque requested, i.e. at cruise. This isn't to bad but I don't like it so I want to rectifiy it.

On idea is to raise the battery voltage. I could go up to 100S (96S at the moment), any higher and I need a new charger as well. Should delay the phenomena but It would still be there.

Another idea is to to change the motor. I do have a Kia HSG which I'm keen on testing. I expect this motor to be on the week side but might be ok. In order to give the HSG every chance of success I think I need to have a gear ratio at approx. 10:1.
I have found the Renault Twizy (80km/h version) has 9,23:1 in a neat package but it is hard to find used.
Smart Fortwo 450 0,6 do have 10,31:1 in 2nd gear. These gearboxes are easy to find but a bit bulky for use as a single speed reduction, could be ok though. Don't know how this would hold up long term with 7-8000rpm continious at the input shaft...

Anyone have any other suggestions for a gear reduction at approx. 10:1 including diff in a small package capable of 50Nm and 8-10000rpm??

[johu]

Audi also has slight oscillation at high speed and low torque. I think it can be fixed in software but I don't know how. Have you played perhaps with vlimmargin?

Unfortunately Audi is gone now so I can't test myself.

[bexander]

The issue can probably be fixed with SW but I don't know how either. These Toyota motors love high voltage so they become tricky to handle at high rpm if you don't have that high voltage.
I have considered rewinding the motor for higher speed/lower torque but I'm not shure I can do it successfully.
It would be very interesting to find out if a HSG can power a small car,

[celeron55]

What would also be interesting is trying to drive the motor with an OEM Toyota control board. There's a high chance that it will be compatible with the MGR, and that it will be smooth at high rpm. However, conclusive tests haven't been made with this combo. I've only seen some questionable test bench videos.

I want to do it on my test vehicle, but annoyingly I'm having to spend my time on other things and can't get to it in another half a year probably.

Would love to see this tested on the Clio - just swap an OEM control board into the inverter and get a ZombieVerter or a GS450H VCU to command it.

(Or get another inverter... if you need one sponsored, please ask - it's not out of the question for me at least to cover some of the cost. And you could ask Damien for a "reserach"-priced VCU, as everyone wants to know if the inverter will work in this way.)

[bexander]

What would also be interesting is trying to drive the motor with an OEM Toyota control board. There's a high chance that it will be compatible with the MGR, and that it will be smooth at high rpm. However, conclusive tests haven't been made with this combo. I've only seen some questionable test bench videos.

I want to do it on my test vehicle, but annoyingly I'm having to spend my time on other things and can't get to it in another half a year probably.

Would love to see this tested on the Clio - just swap an OEM control board into the inverter and get a ZombieVerter or a GS450H VCU to command it.

(Or get another inverter... if you need one sponsored, please ask - it's not out of the question for me at least to cover some of the cost. And you could ask Damien for a "reserach"-priced VCU, as everyone wants to know if the inverter will work in this way.)

That is an excellent idea!!
My thinking was that I need a RX450h with the extra MGR output and then need to hack that inverter etc.
Quickly looked in to it and at least Johannes uses similar parameters for the MG2 in the Audi project, specially the missmatching resolver/motor pole pairs.
I will investigate this further. I do still have the OEM control board for the inverter.

[bexander]

I was a bit concerned about resolver position with a random motor/inverter setup and also my inverter is heavily modified on the inside to only supply voltage to motor side and using that same voltage to supply the DC/DC but after watching this video posted by Damien I'm thinking that this might actually work! Very Exciting!!

Video will start at the interesting bits.


Also, I have a Arduino Due and the V1 of GS450h inverter controller uses that same ARM processor so I could probably get away with a simple home made shield on top and some adoptions of the code, at least for testing.

[bexander]

I've created a shield for my DUE to be able to test communicating with a Toyota logic board using Clocked Serial Interface Protocol.

DUE shield

Sourced suitable connectors and cables as well. I do have the resolver part of the Toyota connector as well.

All connected

I have created a rough SW based on the Lexus GS450H VCU SW and ZombiVerter VCU SW.
First step is to establish communication with the logic board and then step by step find out if I can control the MGR with this setup.

[Ev8]

Cool will be an interesting comparison,