After staring at these charts for a while, I realise I'm running into a fairly simple problem that's limiting my performance significantly. My desire not to exceed 450A means 2 things:
1) I limit AC current at 0RPM to 450A by setting boost. This works perfectly.
2) I limit DC current to 450A by setting fweak. This also works perfectly. By setting this to 250, we reach 450A at exactly the same time we reach 100% battery voltage, meaning this current is never exceeded. The fact that this occurs at 60mph is a handy bonus.
The resulting AC and DC current looks like this:
In the top left we have 450A of AC current, and in the top-right we have 450A of DC current. The problem is that in the middle, there is a big area where both AC and DC current are less than 450A. For example at 80 seconds, the AC current has reduced to 400A, but the DC current has only risen to 290A. This means that for a significant portion of the acceleration, I am pushing only 90% of my target current.
When I tune more aggressively (boost=4000) to bring AC current up to 500A, the problem is even more apparent:
Here we see that the initial 500A created by increasing the boost falls off quite quickly, with fweak being the factor that determines current for the majority of the acceleration curve. This means that most of the time, the currents are well below the 500A AC and 450A DC limits.
There would seem to be 2 solutions here:
1) Add a third point in the V/Hz cure. So in addition to 10% voltage at 0Hz and 100% voltage at 250Hz, I could add some more "boost" at 125Hz which would allow current to be increased at around 30mph without changing the 0mph peak and 60mph peak that are already tuned.
2) Reduce fweak. This would allow me to push a consistent 500A AC, yay, but would cause me to hit the DC current limiter. I had hoped to avoid the limiters (derates) because right now hitting this reduces the throttle to near enough zero, making for a fairly unpleasant driving experience, however with some improvement, we could make a smoother current limiter.
In a way, this brings me back to a more fundamental question. Should we be using an open loop, as I have favoured with this tuning, or a closed loop, where we measure current and try to maintain it within idcmax. From what I've experienced so far, the closed loop DC current control in openinverter isn't fast or accurate enough to provide a good driving experience, however I haven't yet spent enough time tuning idckp to see if this can be improved upon with configuration alone. I suspect that my suggested changes to the AC current measurement maths would help here, but that's beyond the scope of this thread.
What is implemented is the P-part of PID. What is lacking is some low pass filtering to dampen oscillation