Can we get real 18A peak or ideally 21A peak current?

[rusefillc]

• at the moment GDI4 behavior changes at 18A commanded peak current: we have linear change of behavior up to and including 17.9A and then it falls off the cliff at 18A peak commanded
• in reality that 17.9A is more like 15.5A measured see Gdi4: amperage measurement precision  rusefi-hardware#256
• fancy flow benching is asking for 18A measured or ideally 21A measured.

[rusefillc]

Previous iteration rusefi/rusefi-hardware#221

[rusefillc]

@NOx-z a8091c4 mentions "our saturation is 17A" could that 17A be relevant here?

[rusefillc]

@NOx-z a8091c4 mentions "our saturation is 17A" could that 17A be relevant here?

I guess what I am really asking is "is the next step bumping Isense4_high further"?

[rusefillc]

@mck1117 poking DI :)

[NOx-z]

the 17A limit is due to the selection of gain (8.68) to the sense resistor (0.015 ohm last time I checked), which corresponds to a DAC value of 0xFC, nearly at the limit.. This conversion is bound to fall off in efficiency near the top ( & bottom?) which is likely why they are really only getting linear results up to 15.5A

Technically the DAC can currently accept up to 17.2A -> 0xFF

Luckily, there is headroom.. we can adjust the gain down to 5.79 (may be 5.8 in microcode) which hits the ceiling at 25.8 Amps..
and if we look at the 15.5 linearity from before (gain 8.7) that reads a value of 0xE8, with the updated gain (5.8) that would give 23.2 Amps

A microcode update would be required.

We utilize helper functions to determine this DAC value that we stick in the PT2001's ram, those will also need some updating to specify the new parameters.

The other alternative would be to swap out the sense resistors (currently 0.015 ohm) to something lower, like 0.010, which with the original gain of ~8.7 gives the same new ceiling of 25.8 Amps... this would still require that helper function to be updated to reflect the sense resistor. I don't like this method too much because the skill of individual soldering becomes a concern, whereas the other method is just software/firmware/microcode changes..

This seems like a one-off case, or at least rather niche.. I certainly wouldn't want a daily driver to have the gain set so low ( to have the microcode be set at this going forward)

We haven't even touched on the if the DC-DC converter on the board, the fets, and the board itself is/are capable of pulling this current for a sustained amount of time.. and some additional values may require tuning.

including images of PT2001 dev studio showing some of these values

[NOx-z]

https://github.com/NOx-z/libfirmware/tree/gain_8_7

not sure how to submit a PR to a brand new branch, since we do not want to merge this into main (but technically could)

[NOx-z]

Oh, and an update on the change.. originally the gain was ~12.5 .. not the 8.7 as assumed earlier.. the difference was actually that we are currently using 10mOhm resistors, not 15mOhm - exact similar change though, just lower the DAC gain number and update the helper function

[rusefillc]

@NOx-z good news

[rusefillc]

@NOx-z bad news - note how the rest of control profile is missing after peak

[rusefillc]

https://github.com/NOx-z/libfirmware/tree/gain_8_7

I see I've cherry picked that into https://github.com/rusefi/libfirmware/tree/gain_8_7
I also see https://github.com/rusefi/rusefi-hardware/tree/gain_8_7 and the original artifact seems to have already expired

[rusefillc]

GDI4-gain-8-7-may-13-2024.zip

official build just to have it