[rusEfi] 1991 Honda CRX B16A Swap #32

[ZHoob2004]

Rebooted several times. Goes away when I reset the STM, comes back when I rewrite settings.

[AndreyB]

<constant name="vbattAdcChannel">"PC4"</constant>

<constant name="map_sensor_hwChannel">"PA0"</constant>

this looks good. so if reboot helps but write breaks you could be stuck in some loop of valid and invalid settings. If you think it's a bug can you find a way to reproduce it? what does analoginfo command says - which one is using the port you think it should not be using?

[ZHoob2004]

Both are set to what they're supposed to be. I can reproduce it just by using that msq file. One thing I noticed is in tunerstudio under map settings the setting is labeled "battery voltage adc" or something like that.

I see this in the file, is this normal?

Code: Select all

<constant name="console_rx_pin">"PA0"</constant>
<constant name="console_tx_pin">"PA0"</constant>

<constant name="vRefAdcChannel">"PA0"</constant>

[AndreyB]

console_rx_pin & console_tx_pin are still not used in the code - this is an unfinished change, should not be a problem for you.

vRefAdcChannel is a bit worse - that's unused but semi-active code - firmware has already started to use it (for no reason) while TunerStudio does not have UI to change it. I've just disabled the un-needed code to make things safer. Thank you for reporting the issue!

[ZHoob2004]

So with that change, a new firmware build should fix this? I suppose I could always remove that pin specification if I needed to, I'm just guessing the new tunerstudio file defaulted it to that.

[AndreyB]

I think we had a confusion "MAP ADC pin being used by VBAT ADC" between vBat and vRef. Yes latest firmware would not be using vRef which you cannot adjust via TS.

[ZHoob2004]

Cool. Should have more cranking (and maybe starting!) tonight. If this works, I might put in an order for an extra VR board and see if I can get the 1 tooth working with the MAX or if I need to go the comparator route.

The Frankenso VR's are set to mode A2 and per page 3 of the below PDF, the VMIN-THRESH for the 9926 is 15mV, but could be as low as 4mV. Per tWD the threshold will be set to this minimum in about 85mS, but could happen in as little as 45mS. So it does not take much noise to trigger a false reading. The 2.2k ohm is aiming to decrease your normal VR signal to be closer to a 150mV peak to peak signal, such that noise is below 4mV.

https://datasheets.maximintegrated.com/en/ds/MAX9924-MAX9927.pdf

What is the current of the MIL? Can you use your 10A meter to check the current draw? I expect the hi/lo drivers can only drive about 0.125A before they hit a long term thermal limit. You can drive them at the higher current for shorter periods of time, but eventually if the average is above 0.125A, I expect they will over heat and enter thermal shutdown.

I don't know, TBH. It's an incandescent. Google says 1.4W, but I'm not sure if that's the same bulb, so ~.12A? MIL bulb might be a lower wattage than that, because that might be one of the gauge backlight bulbs.

I'm hoping the 2.2k work out here, but I feel like they should. Should a pad be added to the board to allow for such a resistor in the future? I can see this being useful for a lot of applications, and I think I read somewhere that impedance matching is helpful for VR circuits.

[kb1gtt]

If you can, wire the MIL light to one hi/lo driver chip and use other chips for spark. The thermal limit was based on both channels in one chip. So if you only have one channel for the MIL, that will help with the thermal issues.

[ZHoob2004]

Spark has been moved to a low side driver because the Honda ignition module wants to be pulled low to trigger spark and left floating otherwise. I have all the drivers not doing anything right now.

[ZHoob2004]

IT'S RUNNING!

4 tooth wheel, ended up using a trigger angle of 102 (I'm still not sure how that number works, my distributor is set to somewhere between 16 and 18 base timing)

Engine seems to be rather smooth, but I'm getting some very regular skipped spark errors, here's a capture of it running and I'll see what I can do about a datalog.

IT RUNS

2017-02-15_20_35_28rpm_1234_maf_NaN.png (41.5 KiB) Viewed 30093 times

I have no idea how to deal with a datalog yet, it's only for a few seconds but there should be a few skipped spark events in there still.

[AndreyB]

VIDEO?!

Now comes tuning

[ZHoob2004]

[video][/video]

I didn't know my phone was recording in 360p until it was done uploading...

never mind, youtube seems to be upping the quality as it processes.

Also, do you have one of these starter solenoid buttons? Harbor Freight - $10. So worth it for this project (I know it's like $0.10 of parts)

[AndreyB]

Love your remote start button! Mine on the Neon was a much smaller pink-red button

[ZHoob2004]

Doing idle tuning now. I'd get in the car for a gentle cruise around the neighborhood but my laptop battery can't be trusted so I'm tethered to the garage until I grab an inverter.

[sleepingAwake]

Excellent work!!! It runs very smooth.

[kb1gtt]

Sweet, and it seemed to start very nice. Happy tuning.

[ZHoob2004]

I'm still getting missed spark events and I can't figure out if it's a trigger thing or a software bug of some sort. There's no real pattern to it, I just get a message of "looks like missed spark" every 20 or so cycles (not a consistent number). Afr seems to drop slightly when these happen, so I'm pretty sure the sparks are being missed but it doesn't show on the sniffer.

Sniffer with engine running and acting weird.

2017-02-15_22_51_49rpm_1228_maf_NaN.png (42.4 KiB) Viewed 30072 times

Also on the topic of bugs, with the trigger offset and fixed advance both set to 0, the coil cannot fire. The sniffer shows actual dwell as 0.1ms despite it being set hard to 3ms. I'll switch computers and upload a sniffer capture of that. It almost looks like the trigger is cutting off the spark.

2017-02-15_20_08_46rpm_0_maf_NaN.png (37.25 KiB) Viewed 30072 times

And finally, I need some clarification on injection settings. I realized that batch injection could only be an upgrade from simultaneous (since I have 4 injectors anyway), but I don't understand what the "two wire batch injection" option does. Since I have individual wiring for all my injectors, I'm going to guess I don't need this?

[stefanst]

I had a problem with dwell that I was able to fix by changing the lowest rpm setting in the dwell table to 0, while still having a dwell value in there that makes sense.

[AndreyB]

Please attach currenttune.msq, some .csv console log and .msl logs

This way we can see your error codes

[ZHoob2004]

Here's my most recent msq (it's set a bit rich atm) and a tunerstudio log. I can't find anywhere in the console to record a log, and I can't imagine a copy-paste being much more useful than me saying there's no apparent pattern in the missed spark events. I was trying to do some tuning earlier with my new basemap I tried to make from a dump of the honda ecu, but I noticed some afterfiring on deceleration (only did idle tuning) and realized I had liquid fuel coming out of the exhaust, so I'm missing a lot more sparks than I thought. Engine runs fine on the stock ecu, so it's not a mechanical issue.

Attached is a log of the engine idling for a bit. Starting about 30 seconds in, I started the auto tuner (I was running like 12:1 afr) until I was up near 14:1. There is a slight blip of the throttle at 260s marking where I turned auto tuning off.

If there's a better log I can get you, such as from the rusefi console itself tell me how and I'll try to get it uploaded tomorrow. I tried SD card logging, but I can't get more than a few seconds of logging, so I might have to check my solder on the socket.

[AndreyB]

Your .msl is full of
CUSTOM_OBD_MMC_ERROR = 6043,
and
CUSTOM_OBD_SKIPPED_SPARK = 6045,

for 6043 let'd disable SD card for now?

To see what's up with skipped spark console .csv logs would help, those are in your 'out' folder. Sounds like you were not aware that console is writing logs? Just added a text mentioning this on the bottom of couple of console tabs.

[stefanst]

Interesting: It looks like there's essentially not a single record in your log without a skipped spark. Many, if not most of them, have multiple skipped sparks even. I spot checked and you have 16 skipped sparks in 500ms. Running at 900rpm, you get 30 sparks per second. (900 x 4 / 2 / 60). 16 missed sparks in 500ms means there is no spark. That simply can't be.
And even with much fewer missed sparks, you would still see spikes in AFR, caused by un-burned oxygen. Since your AFRs look pretty spot-on, I would look for some other problem.

[ZHoob2004]

I feel like the AFR should be smoother, but the motor is running fairly smooth if it is missing 100% of the sparks...

Here are some of the most recent logs, maybe they'll shed more light. Can I get an explanation of how the ECU is determining that spark events have been missed? there should be 1x spark event per trigger event, which might be throwing the ecu for a loop unless it's also counting map pulses and that's where it's finding a discrepancy. Maybe with a 16x4 trigger I might have all my problems disappear...

[stefanst]

That last msl doesn't seem to include error codes and counts. Different type of log? The names are all different too....

[AndreyB]

That last msl doesn't seem to include error codes and counts. Different type of log? The names are all different too....

that''s rusEfi console log file. Unfortunately inconsistent naming
lastErrorCode and warningCounter are there

[AndreyB]

0) simultaneous injection means short squirts means too close to injector lag which is a huge unknown probably means hard to tune fuel?

1) trigger setting is asking to use "rise only" for VR sensor. I am tempted to rename this setting to "Do you have VR sensor?" Should I? Maybe as a consequence some of your dwells are not 2ms but only 0.68ms because imprecise trigger shape definition (do you have 50% duty cycle?) throws scheduling off a bit?

[ZHoob2004]

Under "Custom Trigger" I have "Use Rise Edge" as true, is that not the same thing? I have tried using the "use only rising edge" setting, but it's locked me out of the serial console in the past. I've had to take the discovery board out of the car back to my desktop to reset and reflash (because my laptop runs linux and I can't flash firmware)

[AndreyB]

Under "Custom Trigger" I have "Use Rise Edge" as true, is that not the same thing?

Sorry this is an unused unneeded property. Just removed from project definition.

I have tried using the "use only rising edge" setting, but it's locked me out of the serial console in the past. I've had to take the discovery board out of the car back to my desktop to reset and reflash (because my laptop runs linux and I can't flash firmware)

Can you try this one more time? Let me know if you have exact S2R (steps to reproduce) if that looks like an issue.

[ZHoob2004]

I suppose that explains why nothing really changed when I tried turning that setting off... I'll run out to the car and give it a shot with that enabled again. As far as steps to reproduce, I'm just going to open up tunerstudio, change the setting, flash it, restart, then start the car.

[ZHoob2004]

Disabled SD and restarted, worked fine. Changed trigger settings (set 'only rising edge' to true and the unused one to false), locked. LCD on frankesno is completely blank and the fuel pump runs continually. Solid red? led on discovery right by reset button. I'm going to try a completely fresh tunerstudio project and just export my tables I need, we'll see how that works. Is the latest bundle what I should be using?