Question and suggestion about ampmin [SOLVED]

[catphish]

Last week while tuning my car, I attempted to use ampmin to set a minimum amplitude to reduce throttle dead zone. With the car stationary, I increased ampmin to approx ampmin=35, not enough to move the vehicle, but enough to eliminate the throttle dead zone.

Unfortunately this has a side-effect that when the car is moving, you get positive torque at zero throttle. This isn't surprising. The wiki states that ampmin sets a *relative* minimum amplitude, and hence when the motor is moving, it applies a much larger minimum voltage.

It seems to me that the purpose of ampmin is to apply a minimum voltage to the motor, but in reality it seems the result is just a minimum throttle position.

My question is: could we have "absampmin" in addition to "ampmin" that specifies an absolute minimum amplitude? This would allow the throttle deadzone to be eliminated when stationary by setting a more aggressive absolute minimum amplitude, but without applying so much minimum throttle at higher speeds.

[johu]

This brings me to an open task.

It would be really cool if someone with a matching motor could take over software maintenance for the sine variant.

[catphish]

It would be really cool if someone with a matching motor could take over software maintenance for the sine variant.

It would be great if someone could keep working on this. I'd be happy to help where I can with this, but I don't currently have any hardware that's not embedded in a car. I'm happy to write code, but I'm not really able to test it. I realize that's not very helpful

[johu]

So it's good that you have hardware embedded in a car because that's what I lack (in terms of asynchronous motors)

If it helps I can send you some old dev board.

[catphish]

So it's good that you have hardware embedded in a car because that's what I lack (in terms of asynchronous motors)

If it helps I can send you some old dev board.

Do you not have your original Polo conversion any more? That's very sad!

I guess in order to be able to bench test the stm32-sine, I would need the following:

* An STM32 dev board with IGBT drivers and current sensors
* A hex IGBT and DC bus caps
* A 230V induction motor
* Rotary encoder
* 300V of batteries and charger
* A current limiting device like a heating element
* A couple of contactors

None of these parts need to be large. When I developed my VFD a few years ago I used a 250W motor and HobbyKing batteries. Unfortunately I didn't keep any of that hardware (apart from the rotary encoder).

Perhaps I'll price up a dev kit.

[johu]

When working with a Toyota inverter (e.g. Gen2) you can run the synchronous buck/boost converter to boost, say, a 48V battery to 300V. Since it is synchronous it will buck down any regen current to 48V and boost battery voltage to 300V to have a realistic motor behaviour.

The Gen2 control board can constantly run the converter with a fixed duty cycle that can be configured on the web interface.

EDIT:
In addition you could even run two motors with two controller kits. One with positive, one with negative torque.

[catphish]

Going back to the question of ampmin:

In my vehicle, I found that increasing ampmin reduces throttle deadzone when statuonary, but by setting it high enough to eliminate the deadzone entirely resulted in the car continuing to accelerate with zero throttle applied when in motion.

In other words, with an example, setting ampmin=40 (for example), the value is not enough to accelerate the car when stationary, but is enough to accelerate the car when moving.

After experimenting on the bench it is now clear that this is not a problem with ampmin itself, but is indicative of an incorrect boost value.

Specifically, the fact that 40% of V/Hz is enough to accelerate the car when already in motion but not when stationary means that boost is set too low. I have always been a bit cautious with boost because I felt that setting it too high could easily cause damage, however I now realize that ampmin is a tool that can be used to optimize the value of boost.

It should be possible to use the following procedure to tune this, with the vehicle safely off the ground:
1) Reduce boost to something that is definitely a bit too low and disable all regen.
2) Gradually increase ampmin until when accelerating and then releasing the throttle, the wheels keep spinning.
3) Reduce ampmin just a little so that the wheels *do* stop when the throttle is released.
4) Increase boost until the wheels start spinning with no throttle applied.
5) Reduce boost just a little until the wheels no longer spin without throttle applied.
6) Reduce ampmin just a little more for safety.

The vehicle should then be in a state where boost is correctly calibrated and regardless of vehicle speed, the applied torque at zero throttle (ie ampmin) is just below what is required to accelerate. I believe this should eliminate throttle dead zones, as well as create more consistent torque from 0RPM up to fweak.

Unfortunately it occurs to me that this procedure could result in boost ending up way too high (and pulling too much current at 100% throttle), but I also think that ideally the same throttle position should ideally provide the same torque at all speeds, so I definitely think it's worth experimenting with this approach to try to correctly balance boost with fweak.

I will do this procedure in my own car and update my SDU tuning thread accordingly.