[rusEfi] 2003 Dodge Neon (#9)

[AndreyB]

I suspect that my pulses are not perfect. Self-stimulation does not look perfect, I need to try with one stm stimulating another and get to the bottom of it

[AndreyB]

I suspect that my pulses are not perfect.

These pictures show show troublesome stuff, I would report once I trouble shoot this further:

[AndreyB]

On a more positive note: I've got vehicle speed sensor working.

My sensor is a Hall type, the wire coming into ECU is constantly switched between ground and floating. At first I've tried feeding this right into a digital input with just a 1K pull-up, but this did not work - all kinds of noises - I was getting lot of false period measurements.

For second attempt I've routed VSS sensor wire, with same 1K pull-up, into the op-amp - this has helped me before with trigger input. Success! Got a nice clean VSS signal.

[abecedarian]

1K pull-up is kind of strong; maybe 4.7-10K is more apropos. You need a fast transition from high to low and 1K will adversely affect the slew rate, if the chip is even really able to reliably overcome that at all.

[kb1gtt]

Most halls have a max output around to 20mA to 30mA. If you have 1k pull up and 12V for your VCC supply, you'll have about 12mA of current. If you are using 5V, then your only at 5mA of current in the hall sensor.

See page 26 of this http://cherryswitches.com/cherrydownloads/cherry_sensors.pdf which suggests resistors for that particular MFG's sensors. Here are some snippets.

hall_resistors_location.PNG (31.53 KiB) Viewed 24310 times

hall_resistors.PNG (7.62 KiB) Viewed 24310 times

[AndreyB]

With the bugfix by @ hip9011 stuff is ready to be tested on a real car. I've added all the knock detection stuff to my in-the-car board and I was really looking to run the test, but something went wrong.

My car starts on rusEfi but it idles lower then usual (currently 800 vs usual 1200) and it would not rev above 2k. Something is wrong but I have no idea what Do not remember changing much in the code recently. The most suspicious thing I see is MAP showing sometimes 50kPa, sometimes 70kPa on idle - I believe it's a bit too much. I remember and I have the logs showing MAP about 30kPa on idle.

Checked the input voltage - the voltage is in fact above 2.5 volts, for my sensor that's in fact abouve 50kPa which is wrong.

Leaving town tomorrow, will only look into this next weekend

Oh, and stock ECU acts funny too, but maybe it's due to limping mode due to disconnected o2. Will start by swapping back the OEM o2 on Saturday.

[kb1gtt]

It's possible a broken MAP can also put your OEM ECU in limp home mode. I've seen a suby that only needed an RPM sensor to technically work. It did not produce much for power, but it still ran which surprised me. Can we verify the MAP voltage on the stock ECU?

[AndreyB]

[video][/video]

P0501 is vehicle speed sensor, how critical would it be for a manual trans? yet the car barely moves in 1st gear, I am unable to drive out of the underground garage since I need to go up-hill.

[kb1gtt]

Sounds like your sensors have gone wonky. Do you have an OBDII scan tool? Can you check TPS, O2, RPM, IAT, MAP/MAF with an OBDII scan tool? Your vehicle speed code could have originally been a variety of other codes before this one became dominate. I would say first things first, check the other devices, also check the O2 sensor. If that's disconnected it will go into limp-home mode. Your lack of power is because your ECU has gone into limp home mode. This has probably happened because the ECU has identified the signals to the ECU are not correct. The key is figuring out what signal is wrong and fixing it.

[AndreyB]

Here's how it looks like with disconnected TPS, I believe that's how limping mode looks like:

[video][/video]

Now, TPS is reconnected, new MAP sensor, after code reset. The car does not want to rev with WOT, while it would rev to some extent with 70% open throttle:

[video][/video]

I guess I will go get new spark plugs, do not have other ideas

[kb1gtt]

plugs would miss-fire. See if they can let you use a scan tool.

[AndreyB]

Talked to my buddy Keith and he had a brilliant suggestion to remove the O2 sensor and see if that would make the difference, to check if the cat or resonator is clogged. The fact that the engine gets really hot really fast plays well with this theory.

Well, I took the o2 sensor out and suddenly I can rev to 6k & hold it! Still no power, but I think I am on the right track

[kb1gtt]

What type of cat do you have? Is it honey comb platinum, ceramic beads, or other? Is it OEM or aftermarket? I'm guessing you'll only be able to guess if it's aftermarket or OEM.

My understanding of a honey comb platinum is that it can only fail if it either falls off, or gets antifreeze in it. Carbon will burn out, which will make it hot, but the platinum will degrade long after the metal enclosure that holds it has turned to a liquid. If you blow a head gasket, that can get antifreeze in them, which won't burn out. AKA they typically can't get plugged. However ceramic after market things can fuse the beads together, so those are sub par quality. I think some USA vehicles tried using the ceramic beads for a while, but quickly found the failures / warranty issues, then switched to the platinum honey comb style. I guess it's possible your low cost Neon could potentially have used a ceramic technology, but I don't recall seeing any OEM ceramic stuff after about 1999.

I once had a 1980 Renault station wagon. First car in high school. The muffler didn't have a drain hole, and the muffler filled with water. My dad and I had it on the lift trying to figure out why it had a very significant lack of power. It sounded just fine, mild exhaust note. But then we noticed there was no flow out the tail pipe. It turned out the air pump hose in the cat had failed, and was muffling the exhaust. The entire exhaust was coming out a perforated pipe the size of my pinky finger. So I might also suggest you check to see if you have flow out the exhaust. If you don't have flow, you might want to try drilling a small hole in the bottom of you muffler, or what ever low point you have in the system. The muffler is often a good place to put the hole, as it will typically stay muffled.

[AndreyB]

My old 0.1 board:

my_0.1.jpg (1.04 MiB) Viewed 24719 times

My new 0.4 board is mostly ready:

my_0.4.jpg (1.3 MiB) Viewed 24719 times

Damn it good grounds! If a spark is sparking, how comes you need BETTER ground to actually start the car?! Note the cool 3D plug covers by @ and vertical coil drivers in the proto area. Now need smaller high side power switch board.

[AndreyB]

[video][/video]

[AndreyB]

this signal is supposed to be 50% duty cycle with all tooth the same width, but that's the tragedy of VR type of sensor: only the rises are reliable, the falls are pretty much useless for precise angle measurement.

[abecedarian]

this signal is supposed to be 50% duty cycle with all tooth the same width, but that's the tragedy of VR type of sensor: only the rises are reliable, the falls are pretty much useless for precise angle measurement.

That's going to depend mostly on the correlation between tooth face and VR sensor. If the sensor is isn't matched to the trigger wheel, the voltage will rise as the tooth enters, then fall to zero, then drop below zero as the tooth leaves the sensor, or vice versa if wired in reverse... which might be something worth checking if you're triggering isn't working as expected.

Code: Select all

tooth: \   /      versus: \       /   or: \   /
        \_/                \_____/         \_/
         _                    _            ___
sensor: | |                  | |          |   |

... for instance ... the first would be correct, the 2nd and 3rd are mismatched and will give erroneous voltages leaving you stuck trying to catch either the zero going positive edge or negative going zero edge.

MAX9924/6 set up as mode A2, I believe, is supposed to catch the positive-going-negative zero crossing, so VR polarity is important, as is matching the sensor to the trigger as mentioned. If the trigger's top (face) and bottom (valley between teeth) lands are different widths you won't get 50% duty no matter what you try. Also, if they're different and the VR polarity is reversed, the chip will be trying to detect a tooth when the sensor is actually looking at the space between teeth, because it's looking for the positive-going-negative zero crossing, creating a situation much like what I ASCII-art drew above.

Also, the MAX chips use the positive edge to set hysteresis levels making polarity even more important to proper function.

[kb1gtt]

From this page, http://www.diycardoctor.com/cam_sensor.htm see below signal. Note the typical OEM wave form has one edge that is clearly a valid edge, but the other edge kind of hangs around 0V. As well that will really hang around 0V when you have a skipped tooth. Most detectors are a 0V crossing detector with a hysteresis window to prevent nuisance trips. That nuisance tripping is a fair part of what makes the 9926 special. It automatically scales they 0V crossing sensitivity based on the peak detected. In the below picture you can see the raising edge 0V crossing's are very much blah, and the actual crossings could vary by large tolerances. While the falling edge is very clear when it crosses. In your case you appear to have the wires flipped which flips the signal, and flips which edge is accurately detected.

[AndreyB]

IMAG1721.jpg (1.29 MiB) Viewed 24270 times

More VR fun.
The green board is the older one - it has rusty water spots and it used to be a Honda board and you can see it's a mess of wires. Problem is this boars gives me a nice signal on both channels:

screenshot_ok.png (45.11 KiB) Viewed 24270 times

While the brand new super clean red board to which I want to upgrade shows me some crap on the second channel:

screenshot_broken.png (32.68 KiB) Viewed 24270 times

And so far I am failing to find the issue - all the connectivities are where they should be. Will re-heat the whole thing, hopefully it's just a bad solder joint somewhere, Hate to be stuck on such trivial things

[kb1gtt]

This was VR? I thinks your VR wires are flipped. The falling edge vs rising edge thing.

[AndreyB]

And so far I am failing to find the issue - all the connectivities are where they should be.

cursing on the forum has worked! I've located the bad solder joint - it was that jumper spot on the output side. Since output side is so much simpler I was not even looking there. Lesson learned: poor joint would somewhat work even on a digital line.

[AndreyB]

Got back from Michigan, temperature here dropped to about 30F/0C. Cold dodge would not start on 75% diesel/25% gas - sometimes it catches but then it dies, spark plugs are wet.

Bigger issue: discovery which was inserted into Frankenso 0.4 inside a case under the hood in a warm garage has died (this one now has "1" drawn on it): a bunch of system LEDs blink randomly as if there is a power issue. Took another discovery which was in the garage as well and somehow 10 minutes later this one would not recognize ST-LINK as USB device while rusEfi seemed to be running fine (this one has "2" drawn on it). Had to grab another one from upstairs to use OEM fuel pump to remove the diesel mixture.

Now the Neon is back on the road on pure gas but today two discovery boards have left for a better place This makes me upset since I have no clue why.

[kb1gtt]

My condolences and sorry for you loss

Hmm, I wonder why that could happen. You only replaced the discovery board, correct? So the 5V seems to be generating 5V? Does he discovery match any of the situations noted here http://rusefi.com/wiki/index.php?title=Hardware:stm32f4discovery_repair

[AndreyB]

Looks like Neon uses Bosch 0280155976 injectors, I will start by using http://www.boschdealer.com/specsheets/0280158117cs.jpg since that's better then nothing. Unless somebody can point me at a more relevant calibration summary?

[AndreyB]

I like how the end result of this looks like but I really want move on for now, hopefully my 0.333 alpha would server me good enough.

Today auto-tune session: https://svn.code.sf.net/p/rusefi/code/trunk/misc/logs/2003_dodge_neon/2016-01-30_autotune.zip

[AndreyB]

Now that I am done with that warm-up PID and resolved the no-start-due-to-no-gas issue, I need to get back to acceleration enrichment. Short test drive today just for a record:

https://svn.code.sf.net/p/rusefi/code/trunk/misc/logs/2003_dodge_neon/DataLogs/2016-02-15_21.09.49_drive2.msl

[kb1gtt]

I see CLT jumps from 88C to 103C. That's quite a swing, your thermostat might be a bit bad. Is that hard to change? I'd bet you get better CLT regulation if you change your thermostat.

Thermostats have a “rated” temperature such as 180F (82.2C) or 195F (90.6C)
This is the temperature the thermostat will start to open, give or take 3 degreesF (1.7C)
The thermostat fully open about 15(8.3C)-20(11.1C) degrees above its rated temperature

Found here http://www.stant.com/index.php/english/products/consumer-products/thermostats/abcs-thermostats/

Your thermostat appears to be going full open, which I do not expect at light engine loads with cold temperatures. This change in CLT is likely one source of control jitter that's causing some problems for tuning.


I see the AFR still spikes at times, but seems much better than it was. The AFR spikes on accell appear to be less than 1 second, which seems most common at low TPS changes, like an 8% TPS change seems to bring the most AFR spike. If you tromp on the long pedal a bit more TPS of 25%, it seems your AFR is better. So you seem to be tuning in the correct direction. I think you are currently hitting limits with your tuning because of system jitter.

I thought I remembered you showing me a injector compensation for voltage. I'm not seeing a change in the fuel:pulse based on voltage. Perhaps that's because the compensation would have a minimal change, perhaps something else is happening. I can't find the table right now, so I can't tell you if that compensation is happening or not.

[AndreyB]

With life in the way it has been about a month since I've started my test mule on rusEfi. Good news current firmware is still functional, not so good news is poor start with weak battery: note how trigger synchronization is lost after one cycle of ignition/injection.

Once the battery spent 3 minutes on 50A charge the car has started right away-ish.

Defect ticket: https://sourceforge.net/p/rusefi/tickets/279/

[AndreyB]

Summer is great time to test A/C logic. Looks like ECU reads the switch state and turns the relay accordingly? Probably no A/C while cranking and no A/C while wide open throttle? Also radiator fan always on in A/C is on?

http://rusefi.com/wiki/index.php?title=Vehicle:Dodge_Neon_2003#A.2FC

[kb1gtt]

Also use a pull up resistor on the AC sense switch, and if you sense 0 volts, turn off AC, if you see +V then it's OK to use the AC. That's a safety that prevents over pressurizing which results in bursting, as well it prevents operating when your compressor is in suction which prevents the oil from getting around the system and can damage your compressor.