Project: Build an On Board Diagnostics System for a carb'd car

[Buzo]

Nice work on the VI. Its all coming together very well. I'm really interested in how this will turn out. I love all the data your collecting and find it very interesting. Keep it up.

Hey, looks like you have some knowledge of Lab View, Great. I'm in the most exciting part of the code right now, making the pass fail criterias. Right now I am just asking If the signals are changing through the time. If all signals are found then I can start making smarter pass/fail decisions.
.

[Dr_Snooz]

I'm in the most exciting part of the code right now, making the pass fail criterias.

Oh yes, white-knuckled excitement that coding!

This project is awesome and really coming along quickly. It's nice to see how you think through each piece of it, gather data, develop a plan of attack and then execute. Awesome!

[Buzo]

Here the Final PCB.
I added one diode to each input to isolate them from each other. I found that was the cause of the failure in my Idle Solenoid.

Migrated to a 40 pins microprocessor.

I don't need a LED for each input, there will be some heat build up in this thing if I put 33 LEDs and they turn on at the same time. So the LEDs in the software are enough.

I will use the Software to tell the OBDZ if the car is standard/auto, with/whithout O2, AC/No AC, etc. 33 inputs were not enough to put switches in the PCB.

I will build the PCB tomorrow, test it and install it below the drivers seat.
Then will debugg the software and when done, just a matter of translating from LabView to Assembler

[Buzo]

This is the end of the project... from the hardware standpoint.
Software is the easiest part, I need a couple of days to finish it all.

This thing sends all the ECU signals, one at the time through the serial port. The software in the PC receives the signals, and writes them down to a file in the hard drive at the same time that turns led's ON and OFF in the main screen depending of the status of the inputs received.

[87roach]

You're doing some really great work here man, keep it up!

[Buzo]

Thanks for your comments, I have a few more pictures taken today as I installed it in the car.

Replaced the smaller microprocessor with a bigger one. For the first time I will be able to log everything that happens in the car during the same drive.

[Dr_Snooz]

You still have the stock stereo???? Does it work at all?

[Buzo]

You still have the stock stereo???? Does it work at all?

The radio works fine, I haven't tryed the cassette player, because I don't have a tape anymore! I usually use a FM transmitter with my ipod (that's why the ligther is not in place).

[Buzo]

After several trys I concluded that the best way to display results is through the graphs similar to what I have been posting in this thread.

And that was because the signals change through the time, so its hard to track or count how many times each signal changes.

And this is how it looks:


1.- Set up. Select your serial port, select if you want the data saved into a txt file. A 20 minutes log weights about 680 k.
Press start to begin. If you see the data received in the format shown, you are all set, if not, a 10 seconds timeout will occur.



2.-Inputs. Each signal has a excpected waveform to the left. This waveform was taken during a 20 minutes ride in my car, so it may not be perfect, but you will get the idea of what should happen in each input.



3.-Outputs. Same as inputs, my car's read signals as reference.





Next step is to talk to a neighbor of mine that actually has a couple of 3geez
to test the software and OBDZ in a car different than mine.

Then complete the assembler code to emulate the OBD system with the flashing LED the FI cars have.

The graphs shown in the screenshots above were taken in a 10 minutes ride around the neighborhood, max speed 40 MI, and you will notice there are some graphs missing. Well, I need to dig into that.

[Buzo]

I need to fix those inputs that look flat in below post, for example the solenoid responsible for the bubbling noise. I know that signal is there, but for some reason is not being detected, its a matter of rewiring those flat signals to known good inputs in the PCB.

Please stay tunned because during the weekend I plan to reproduce the failures I found in the car when I first bought it, just to see how this diagnostics tool could help me.

1.-Vacuum leak in the Air Intake door actuator
2.-Vacuum leak in the thermovalve actuator
3.-Frequency solenoid valve C hose pinched off
4.-General Idle system missadjustment
5.-Fuel flooding... But we have the fuel odor to detect this one, so I better don't mess with it for this time.
6.-Ignition coil signal to the tacho. Yeah, had no chance to replace the coil yet.

[2ndGenGuy]

That's so badass! Seriously mad skillz here. Your PCB looks awesome. What kind of processor is that that you use for something like this?

[Dr_Snooz]

After several trys I concluded that the best way to display results is through the graphs similar to what I have been posting in this thread.

And that was because the signals change through the time, so its hard to track or count how many times each signal changes.

And this is how it looks:


1.- Set up. Select your serial port, select if you want the data saved into a txt file. A 20 minutes log weights about 680 k.
Press start to begin. If you see the data received in the format shown, you are all set, if not, a 10 seconds timeout will occur.

2.-Inputs. Each signal has a excpected waveform to the left. This waveform was taken during a 20 minutes ride in my car, so it may not be perfect, but you will get the idea of what should happen in each input.

3.-Outputs. Same as inputs, my car's read signals as reference.

Next step is to talk to a neighbor of mine that actually has a couple of 3geez
to test the software and OBDZ in a car different than mine.

Then complete the assembler code to emulate the OBD system with the flashing LED the FI cars have.

The graphs shown in the screenshots above were taken in a 10 minutes ride around the neighborhood, max speed 40 MI, and you will notice there are some graphs missing. Well, I need to dig into that.

Now I have a headache...

[Buzo]

Now I have a headache...

Yeah, I know. But I tried different things like add a PASS/FAIL word to each signal, and none seems to work better than the waveforms.

Actually I want to thank you for scanning and uploading the service manual, without it this work could not be possible and the interaction of the carb'd fuel system would remain un-comprehended.

[Buzo]

That's so badass! Seriously mad skillz here. Your PCB looks awesome. What kind of processor is that that you use for something like this?

It is called PIC 16F877. 33 inputs, Analog to digital converter, serial communication port, perfect for this application.

This project itself could be considered out of the scope of this forum, that's why I didn't want to add anything related to the build of the PCB to this thread, I would bored you guys to death.

I am thinking seriously in aborting the blinking led capability. I mean, a laptop with a USB to serial port adapter is available almost anywhere. So I think I can close this project after I run my failures simulation this weekend.

[Buzo]

One good new and one bad new

Good one:
I was able to induce failures to my car, and detect them with my system.


Bad one:
I don't recal exactly which action produced which failure.

For example in the chart below I was able to make my system too rich, but I disconnected several things during the experiment that I am unsure which one produced the failure.

So I need some more time to come up with a relationship between a disconnected hose, and its effect in the car.

[Buzo]

I just found that its a pain in the butt to connect the laptop to the OBDZ everytime, so I am re-considering putting the Diagnostics LED back.

I have made good modifications latelly, I am able to measure speed and engine rpms and send that information to the Laptop.

Will update my post soon.

[2ndGenGuy]

It is called PIC 16F877. 33 inputs, Analog to digital converter, serial communication port, perfect for this application.

This project itself could be considered out of the scope of this forum, that's why I didn't want to add anything related to the build of the PCB to this thread, I would bored you guys to death.

I am thinking seriously in aborting the blinking led capability. I mean, a laptop with a USB to serial port adapter is available almost anywhere. So I think I can close this project after I run my failures simulation this weekend.

You have a link to another forum where all the "boring" stuff is? I love that stuff! Feel free to post it here! I think it would be great!

[Buzo]

I can say I finished my project.

As its older brothers the OBDI and OBDII, my OBD0 only detects failures related to the emissions control. It autodiagnoses the ECU and reports out if one output is bad because there are missing inputs, or due to an internal ECU failure (just like the FI one).

It is not able to find vacuum leaks, or why your engine is stalling.

However, I learned how all the sensors and actuators interact with the engine, I whish I have met one of these cars when new. The concept itself is, as I said before, a dream come true. In the old days of the carbed cars, we all wanted an automatic throttle control, or an idle boost to avoid adjusting the idle too high so the engine doesn't die when in "D", and a lot of other nice features.

Of course I am not done with monitoring signals in the car, I'll keep adding external sensors in strategic places to cover more failures than the ECU-related ones.

This is the last screen I added, it is really nice to see your speed/rpms in the screen of the computer. Someday we could put a monitor inside the instrument panel and simulate all the needle indicators, costomize them, change them to digital, interchange positions, play music, etc.

Thanks for your comments and opinions and thanks for reading this thread. More to come yet!

[A20A1]

This is an incredible tool

Do you plan to hook up a vacuum/pressure sensor?


I wonder if Honda engineers ever made something like this to test their own system when producing the 3G

One good new and one bad new

Good one:
I was able to induce failures to my car, and detect them with my system.


Bad one:
I don't recal exactly which action produced which failure.

For example in the chart below I was able to make my system too rich, but I disconnected several things during the experiment that I am unsure which one produced the failure.

So I need some more time to come up with a relationship between a disconnected hose, and its effect in the car.

Which hoses? Was it one of the A,B,C lines?
Is there any difference between cars sold in Mexico and ones sold in the US? Not all 3G's out there came with the same heap of emission controls the US cars did.