openinverter forum

Uploaded my latest .bin with the inverter off and found fweakcalc showing as 12.5, which shifted back to fixed point via FP_FROMINT(fweakVal) put the expected 400 in fweakcalc. I think Im getting closer to something workable. Did some reading up on fixed point usage last night since I havent done much with them previously.

*update* fweak appears to be working as intended. By changing some Idle throttle values I was able to run through the full potnom range and observed desired behavior on fweak calc.

Now to read back through the throttle ramping stuff and make sure the data types agree. Will update github later today.

Alrightly folks....

Uploaded my latest changes and it seems to be working as intended. Nice smooth low throttle operation straight out of the firmware.

Below are my changes compared to the first draft I initially posted.
https://github.com/jonvolk/stm32-sine_L ... ...851f96c

https://github.com/jonvolk/stm32-sine_LDU

Unfortunately I will not get any prolonged seat time for a few days as its about to start raining for what appears to be an extended period.
Plot showing the reduction in fweakcalc with throttle application attached. Scaling is shit since I have raw pot on there. I have attached the compiled .bin if anyone wants to give it a try. Its based off the 4.97 release. The core of what this does is reduces throttle ramp to a very slow rate for lower throttle inputs. These eliminates any on/off throttle jerkiness. It also sets the default fweak value to 400 under 35% throttle. Over 35%, it will linearly ramp fweak down to the value you have in the parameter file at WOT. This gets rid of the low speed low throttle surging.

You should not have to change any of your standard parameters with these changes.

As always appreciate any insights or criticism.

I've been waiting for something like this:) Will give it a try in the E31.

Good deal. While I said you probably wont need to modify your existing parameters, fslipmin may need to be increased since fweak of 400 doesnt exactly generate world spinning torque at low throttle inputs..but thats the point

It dried out last night so I was able to give the car a proper beating this morning. Everything appears to work as advertised. Smooth low throttle/speed operation and I can stomp the pedal as violently as possible and the throttle ramping keeps all changes under control without oc.

After living with a static lower fweak value for a few days prior, I also noticed that with the updated firmware the low throttle 400 fweak value also happens to make for very smooth onset of regen as well.

jon volk wrote: ↑Tue Apr 13, 2021 11:28 am After living with a static lower fweak value for a few days prior, I also noticed that with the updated firmware the low throttle 400 fweak value also happens to make for very smooth onset of regen as well.

@Johannes, any chance this could propagate to ACIM side too? Id be happy to test .

EDIT: Sorry i saw there already is sine software posted

@ Jon Volk, can you also upload .hex file next to .bin? It is way easier for me to upload firmware via STlink and that uses hex file ...

Right, well I just uploaded the Jon firmware into the Panzer. Now, understand I am not a performance car guy but £$%^&& me that is some change

Low speed is nice and controllable, I did get one overcurrent when I got into the throttle but at that point the g was causing my organs to migrate to my spine so ....

Parameters attached if you see anything silly let me know. Thanks very much for doing this Jon and more importantly for sharing.

Jack Bauer wrote: ↑Tue Apr 13, 2021 12:47 pm
Parameters attached if you see anything silly let me know. Thanks very much for doing this Jon and more importantly for sharing.

fslipmax seems a bit high and probably the OC cause. I would say back it down to like 3 and creep back up to see what will avoid an OC. Youll probably end up around 3.8-4.2 if I had to guess.

arber333 wrote: ↑Tue Apr 13, 2021 12:34 pm
@ Jon Volk, can you also upload .hex file next to .bin? It is way easier for me to upload firmware via STlink and that uses hex file ...

hex file attached.

One thing that occured to me is implementing into other applicaitons, is that the low throttle fweak is hard coded at 400. It would be more flexible to add that a a parameter input. I can try working on that later this week if it seems this addition is functional. This would also allow function to be bypassed by setting the values equal.

jon volk wrote: ↑Tue Apr 13, 2021 1:01 pm
One thing that occured to me is implementing into other applicaitons, is that the low throttle fweak is hard coded at 400. It would be more flexible to add that a a parameter input. I can try working on that later this week if it seems this addition is functional. This would also allow function to be bypassed by setting the values equal.

Today i tested it in Mazda and the first thing i noticed was a lack of vibration with slow driveoff or any kind of driveoff really. Nicely done!
There is however an odd thing which is probably tied to your solution. I noticed quite some time shift from the time i order throttle and the time it actually gives any effect on start. Maybe some 1/2s worth? I should have tested this on the road at high speed, but i only have a small 360Vdc battery in Mazda. My 2nd BMS isnt ready yet so i cant use Kokam yet for full power.

Whenever you implement the FW shifting code please post bin and hex files. It is just easier for me to update code with my phone.

tnx

Yeah I get that little throttle delay also. That said its a massive improvement to the driveability of the E31. Now I gotta find new front discs ....

Great feedback.
I have a firmware draft that adds an fweakstrt value so you can adjust that.

The very slow throttle ramping that was done to prevent initial on/off throttle blips has a pretty high threshold. I was translating raw pot values to the new throttle percent value and erred on the slower side to be safe.

I need to think about how to logically do this in a manner that's universally configurable with maybe one additional parameter. Alternatively I could just remove that for the next "public" iteration as it's a pretty mild nuisance fix in comparison to the other low speed vibration/oscillation/jerkiness what have you.

Glad to hear it's also an improvement in other applications as well. Ill take a look at the code tonight to see what I can come up with.

Here's an updated one that allows you to tweak the "fweakstrt" value. I have the default to 400. If you want to turn this feature off, set fweakstrt to the same value as fweak.

For the throttle ramping, I found Im an idiot and didnt change over all the code from using my raw pot values so it was basically stuck at a .45 throttle ramp value if you were below 80% throttle. This was updated to be .45 under 10%, ramping to your full normal throtramp thereafter. I still need to work on making this user configurable,

Code is pushed to my github, compiled files below.

EDIT: Spent some time on the road and things seem to work well. Throttle ramping can still be more aggressive but it's improved over previous iteration.

jon volk wrote: ↑Thu Apr 15, 2021 11:17 pm Here's an updated one that allows you to tweak the "fweakstrt" value. I have the default to 400. If you want to turn this feature off, set fweakstrt to the same value as fweak.

For the throttle ramping, I found Im an idiot and didnt change over all the code from using my raw pot values so it was basically stuck at a .45 throttle ramp value if you were below 80% throttle. This was updated to be .45 under 10%, ramping to your full normal throtramp thereafter. I still need to work on making this user configurable,

Code is pushed to my github, compiled files below.

EDIT: Spent some time on the road and things seem to work well. Throttle ramping can still be more aggressive but it's improved over previous iteration.

I tested the second iteration and i must say it is even better. I can get throttle response to comfortable level and system resonance is virtually gone. There is still barely noticable lag in throttle pedal response, but ii is not uncomfortable. I still have to test this at speeds.
I use fweakstrt at 300.

And i figured my LH rear brake locked up... service required.

Also while Mazda was static for a long period and i drove newer Peugeot i thought how much better a single gear gearbox is to end user VS transmission shifting. It is just so much better i want to change mazda to Outlander motor and gearbox...

Well this is great news! Will look into integrating those changes in the main firmware.

Now Jon should have a look at FOC firmware and eliminate low speed oscillation

johu wrote: ↑Sat Apr 17, 2021 6:57 pm Well this is great news! Will look into integrating those changes in the main firmware.

Now Jon should have a look at FOC firmware and eliminate low speed oscillation

Sounds like you’re offering to whip up FOC for asynchronous motors

Having put a bunch of miles on it I can’t say I’ve had any issues with the fweak mods. For an official release you can probably omit the throttle ramping for now. I’m still messing with that to find a sweet spot. Transition from regen to heavy throttle has a noticeable dead spot that isn’t there if just transitioning from coasting to heavy throttle. Need to look a little deeper at what’s going on.

Loaded this yesterday and adjusted my parameters to suit no pot2 input and removed CAN mapping.

It's good, very good. Like above, I get the throttle delay with a slight elastic sensation that power builds quickly once things start to move. I also get the same situation when hard on the throttle, off, and then back on again. It's all fairly minor stuff and great progress overall. Thanks Jon.

Ill pop up another version without the throttle stuff or turn the throttle stuff completely off after say 1500rpm. Once at speed the throttle ramping proves more of a hindrance. From a stop it does help quite a bit with a softer initial turn on to brake off idle speeds. The elastic feel you describe is quite prevalent for me when Im hard on the brakes/regen into a turn and jump back on to power out. Not really noticeable driving like a sane person.

jon volk wrote: ↑Fri Apr 30, 2021 8:16 pm Ill pop up another version without the throttle stuff or turn the throttle stuff completely off after say 1500rpm. Once at speed the throttle ramping proves more of a hindrance. From a stop it does help quite a bit with a softer initial turn on to brake off idle speeds. The elastic feel you describe is quite prevalent for me when Im hard on the brakes/regen into a turn and jump back on to power out. Not really noticeable driving like a sane person.

I also noticed the stretching feeling on quick driveoff from roundabout. Also that throttle time delay is really annoying from the start and after some RPM. Usually HMI (human machine interface) needs to have less than 0.5s delay so it is not annoying.
Could you make this delay or throttle stretch effect tunable parameter from the interface? Or variable to the rate of throttle movement?

EDIT: Oh yes... i tried to set fweakstrt at 240 and motor seems to have sharper response but still being managable.
Throttle delay was very annoying to DMV person making a car test on track today . Nontheless i passed another annual inspection .

tnx

The people have spoken. I will omit the "feature" and hopefully post something later this evening.

arber333 wrote: ↑Fri Apr 30, 2021 8:28 pm

jon volk wrote: ↑Fri Apr 30, 2021 8:16 pm Ill pop up another version without the throttle stuff or turn the throttle stuff completely off after say 1500rpm. Once at speed the throttle ramping proves more of a hindrance. From a stop it does help quite a bit with a softer initial turn on to brake off idle speeds. The elastic feel you describe is quite prevalent for me when Im hard on the brakes/regen into a turn and jump back on to power out. Not really noticeable driving like a sane person.

I also noticed the stretching feeling on quick driveoff from roundabout. Also that throttle time delay is really annoying from the start and after some RPM. Usually HMI (human machine interface) needs to have less than 0.5s delay so it is not annoying.
Could you make this delay or throttle stretch effect tunable parameter from the interface? Or variable to the rate of throttle movement?

EDIT: Oh yes... i tried to set fweakstrt at 240 and motor seems to have sharper response but still being managable.
Throttle delay was very annoying to DMV person making a car test on track today . Nontheless i passed another annual inspection .

tnx

I have my annual MOT inspection tomorrow. Always a fun time when the tester gets in the car and expects an engine to fire up.

By later this evening I apparently meant now. Removed throttle ramping mods. Just fweak stuff remains. I can probably drop the LDU part since it's now universally configurable with the fweakstrt parameter.

jon volk wrote: ↑Fri Apr 30, 2021 8:57 pm By later this evening I apparently meant now. Removed throttle ramping mods. Just fweak stuff remains.

Thanks! Will test tomorrow after the inspection.

jon volk wrote: ↑Sat Apr 17, 2021 10:00 pm Sounds like you’re offering to whip up FOC for asynchronous motors

Been asked for many times. I suggest someone with a Toyota inverter tries it first
Maybe I can try to keep the basic slip calculation and just replace V/f by FOC? Hmm.