GDI4 runs out of juice Isense4_high

[rusefillc]

user report:
boost voltage setting 40v
user has one proteus with two GDI4 boxes (8 injectors sequential) for a GDI flow bench

at 3000RPM 2.5ms injector pulses 39v volts are observed at peak voltage as EXPECTED
at 4000rpm voltage drops to 30v, that's an ISSUE

L2
10uH

[Qwerty-OFF]

Это измерение напряжения на форсунке или на тестовой точке высокого напряжения на плате? Хочу узнать что в этой точке:

[rusefillc]

@Qwerty-OFF GDI4 case is closed that's on injector wires

[rusefillc]

rev D

no activity
vboost=64.7v

low RPM bench
vboost=64.7v

1200 RPM
vboost=64.7v

[rusefillc]

there is a very sharp drop of vboost around 5200 rpm

4900 everything pretty fine

5100

5300 rpm

[rusefillc]

just attempted bumping max boost time from 400 to 600 us, did not help. tried bumping max current targer from 13A to 15A I think it made matters worse

reducing boostCurrent to 10A has helped - there is no more voltage drop

[mck1117]

reducing boostCurrent to 10A has helped - there is no more voltage drop

this suggests that the inductor was saturating at 13A

[rusefillc]

this suggests that the inductor was saturating at 13A

what is the right way to improve things? do I change inductor or just change default boostCurrent to 10A and live happily ever after?

[mck1117]

Wait - I read that wrong, I thought you were adjusting the SMPS parameters, not the injector parameters. Changing the boost current will change the injector's opening time and potentially other things.

max boost time from 400 to 600 us

This just allows the injector to take that amount of time for the boost threshold to be reached, it doesn't change behavior at all so long as the injector hits the current limit faster than that time.

[rusefillc]

@mck1117 I was changing the below I assume I was not changing injector parameters?

[mck1117]

Those are absolutely injector parameters. They control the behavior during the boost phase of the injection (high current "slam it open" part).

[rusefillc]

Are "boost current target" and "peak phase current target" both injector parameters? I was assuming that only the latter is directly related to individual injector control while "boost current target" is something to do with DCDC core same as boost voltage is a DCDC setting?

[rusefillc]

current plan is to try https://www.mouser.com/ProductDetail/EPCOS-TDK/B82559A5103A033?qs=amGC7iS6iy9mEhDlIgUEKA%3D%3D Power Inductors - SMD 10uH 12% 1.2mOhm ERU33 AEC-Q200

@kb1gtt @NOx-z do you guys have opinions on what could be going on here?

[NOx-z]

I would like to see the vboost (pin) on an oscilloscope as multimeters are slow and don't give the full story.. (and I only have two test injectors, which may not be enough to replicate)

we may be able to setup vboost_mon_en which can set an IRQ (a certain pin, maybe) for when undervoltage of VBoost occurs, so we can detect/flag/log on this issue.

We have two additional settings, at least in the microcode, (I do not know if they are accessed elsewhere ) for the current sensing of the dc-dc converter:
#define Isense4_high 2;
#define Isense4_low 3;

We may be able to tweak those to allow more current, but we should do more research into this before just setting new values.

I am reviewing some of the microcode now.

[rusefillc]

@NOx-z current L2 is https://jlcpcb.com/partdetail/bc_bao_Cheng_elec-BCIHP1770100M/C237980 see https://github.com/rusefi/rusefi-hardware/blob/main/GDI-4ch/gerber/rev.%20d/bom.csv

[rusefillc]

happy 3000 rpm

[rusefillc]

unhappy 5300

[rusefillc]

sad 5300 video - note how thing are slowly falling down

https://youtu.be/iUY8sWQcSHo

[rusefillc]

[rusefillc]

with magic byte adjustment

[rusefillc]

See rusefi/libfirmware#8 for solution

[NOx-z]

This magic byte was the Isense4_high in the microcode dram, which is not exposed to the GDI configuration as it is specific to the DI board and not someone's car. The value was changed from ~3.6Amps to ~6.5Amps.. we should spend a little time in the future really tweaking this setting, but this quick change resolved the issue

[rusefillc]

rusefi/libfirmware#9