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

[Buzo]

Based on the succesful diagnostics I recently made in my car, I thought it would be good idea to build an interface that would automatically tell you which sensor is faulty, -similar to the one in the FI cars-.

So the ECU in a Fuel Injection car has a LED that blinks pre-determined number of times when the check engine signal comes on, to tell you there is something wrong in the system.

The project consists in adding a PCB with a PIC microprocessor and monitor the ECU inputs and outputs and make a LED blink if there is a missing input/output/start up secuence.

There are digital and analog inputs to the ECU in a carbureted car

DIGITAL INPUTS
Shift Position
Clutch
Vacuum Switches A and B
Ignition
Start, etc

ANALOG INPUTS
TA and TW sensors
O2 sensor, etc

OUTPUTS
The ECU is a solenoid valve controller, so I believe all the outputs are digital. Either they are at 12V or at 0V at any given time.

Let me start with the idea for the digital ones: Vacuum switches A and B are calibrated to switch at different vacuum levels, so the PIC will monitor the start signal, then Switch A will switch first and Switch B second as the engine "builds up" vacuum. If it doesn't happen in that secuence something is wrong and the LED will Blink "n" times, but will keep monitoring every other start in case the failure fixes itself (yeah).

The Analog inputs can be monitored with comparators to tell the PIC they are changing their value through the time and driving conditions. I.e. The O2 sensor produces a voltage of .4V when lean and .6V when rich, so compare against .5V. The comparator's output should be changing constantly. If it stays ON or OFF even after the car is warm, the LED should blink.

I have all I need right here at home to complete this project, I'll make one for me, but will share all the information with you guys (if interested) as I progress.

Let's wait if it works and then we'll figure out what else can be done with this basic OBD. It would be very helpful for those still with the stock carb I think.

[Dr_Snooz]

Very cool stuff.

[lostforawhile]

don't forget the engine senses rpm, and there's an electronic switch built into the cluster that tells the car when it's going over 5 miles an hour, this changes the status of the inputs and outputs, the problem with doing this, is the controller doesn't have sensors for everything, it's a Frankenstein system of vacuum controls and electrical controls that don't actually interface with each other in any way, in other words one system doesn't know what the other is doing, eg. the air jet controller, modifies mixture by vacuum controls depending on barometric pressure, it can be completely screwed up, and the control unit doesn't know. This is why it's so hard to get these systems right, two independently operating systems not in communication with each other but causing interaction between the two, depending on what they are doing. the ECU doesn't have all digital outputs either, the valves that control the mixture don't work on straight DC, it's a square wave, the control box changes the frequency of this square wave to modify the mixture, hence the name frequency control valves, you have other things also going on, such as the EFE heater, it has an independent control circuit and heats the mixture going through it on start up, also the system that controls air intake temp is completely vacuum controlled , it can influence the mixture, there is also an air intake sensor that can switch in a resistor to delay choke opening speed. If you look at the Preludes with the dual Keihin sidedraft carbs, they have a diagnostic system built in, they also have things like a mass airflow sensor, etc,that the three g doesn't have. it's basically twice the headache plus twice the vacuum lines. If one carb goes bad they both have to be replaced, as one controls the other. It didn't do much good though as those dual side drafts never worked right from day one. Thats why you don't see many of the cars with the dual carbs on the road, they died and couldn't be fixed. You may have heard people talk about Ferrari's having possessed carbs, they don't have anything on one of these systems when it goes bad.

[carotman]

It's a good thing Canadian 86-87 cars don't have an O2 or these frequency valves A and B. One less thing to go bad.

Good luck with your project. These cars would benefit from a certain diagnosis tool.

[Dr_Snooz]

don't forget the engine senses rpm, and there's an electronic switch built into the cluster that tells the car when it's going over 5 miles an hour, this changes the status of the inputs and outputs, the problem with doing this, is the controller doesn't have sensors for everything, it's a Frankenstein system of vacuum controls and electrical controls that don't actually interface with each other in any way, in other words one system doesn't know what the other is doing, eg. the air jet controller, modifies mixture by vacuum controls depending on barometric pressure, it can be completely screwed up, and the control unit doesn't know. This is why it's so hard to get these systems right, two independently operating systems not in communication with each other but causing interaction between the two, depending on what they are doing. the ECU doesn't have all digital outputs either, the valves that control the mixture don't work on straight DC, it's a square wave, the control box changes the frequency of this square wave to modify the mixture, hence the name frequency control valves, you have other things also going on, such as the EFE heater, it has an independent control circuit and heats the mixture going through it on start up, also the system that controls air intake temp is completely vacuum controlled , it can influence the mixture, there is also an air intake sensor that can switch in a resistor to delay choke opening speed. If you look at the Preludes with the dual Keihin sidedraft carbs, they have a diagnostic system built in, they also have things like a mass airflow sensor, etc,that the three g doesn't have. it's basically twice the headache plus twice the vacuum lines. If one carb goes bad they both have to be replaced, as one controls the other. It didn't do much good though as those dual side drafts never worked right from day one. Thats why you don't see many of the cars with the dual carbs on the road, they died and couldn't be fixed. You may have heard people talk about Ferrari's having possessed carbs, they don't have anything on one of these systems when it goes bad.

It doesn't have to be perfect. If he does nothing more than wire up the easy controllers it will be a million times better than nothing. He's spent the last couple weeks tracking down little electric carb controllers that have fritzed. Most of those will be pretty easy, I'd think. He can get more creative for the other stuff. A vacuum sensor would be relatively easy to create, as would a temp sensor. He could wire off the pulser in the speedo for speed, etc. I think it would be very do-able.

My only concern would be that by the time you finish working the bugs out of it, you would have spent about as much time as it would take to swap to FI.

Good luck and definitely keep us posted.

[lostforawhile]

It doesn't have to be perfect. If he does nothing more than wire up the easy controllers it will be a million times better than nothing. He's spent the last couple weeks tracking down little electric carb controllers that have fritzed. Most of those will be pretty easy, I'd think. He can get more creative for the other stuff. A vacuum sensor would be relatively easy to create, as would a temp sensor. He could wire off the pulser in the speedo for speed, etc. I think it would be very do-able.

My only concern would be that by the time you finish working the bugs out of it, you would have spent about as much time as it would take to swap to FI.

Good luck and definitely keep us posted.

I appreciate the effort, but these carbs are just not worth it, once any of these electronic keihins go bad, they are just about impossible to fix, I've got good diagnostics charts and i've been messing with them for over 25 years and I still end up throwing up my hands trying to fix the things. the older they get the harder they will be to fix, plus a lot of critical pieces are now discontinued for them

[Buzo]

I am new to honda's cars and all I know about them is what I can read in the service manual and what I have measured myself. I appreciate you guys telling me your experiences.

I don't intend to mess with the controllers and modify what the actuators actually do to the engine.

I just want a start point when you get a failure. I.e "the car started to stumble this morning, Is there a code in my OBD?" Yes - Good! No - Freakin project didn't work!

This is the PCB design so far. I added some LEDs to actually see if at least they are getting activated. As you pointed out there will be PWM and Analog signals that the sight will not be able to capture, but that's what the microprocessor is for.



In the other hand, monitoring signals is what I do for a living. So running wires to/from inputs/outputs is the easier part for me. The difficult part is learning what you guys have learned in all those years enjoying these cars.

There is no dead line to accomplish this project, I will be doing it slowly. I plan to first "learn" what a good O2 sensor does and then make it fail by disconnecting it or making a short. Then use that information in my PIC code.

[lostforawhile]

if it helps, Dr Snooze scanned the entire electrical troubleshooting manual, it's on the 3geez wiki, it's for 89 so most of it is right for the earlier years, one difference is all the cars, cruise or not have a clutch switch, on later models it's connected to an interlock, on earlier ones, it's not connected to the interlock, but I believe it's an input to the controller, I'll need to do a continuity test between the pink wire on the switch and the input plug of the control module

[Dr_Snooz]

Subscribed.

[Buzo]

I appreciate the effort, but these carbs are just not worth it, once any of these electronic keihins go bad, they are just about impossible to fix, I've got good diagnostics charts and i've been messing with them for over 25 years and I still end up throwing up my hands trying to fix the things. the older they get the harder they will be to fix, plus a lot of critical pieces are now discontinued for them

I fully understand what you are saying here. And it may change the whole scope of my project. I just would like to do these final comments.

To be honest with you guys, when I see the check engine on in my other cars, I take them to the dealer.

But as far as I know, you can't do the same with these cars because they will say it is not worth it. So that was the main Idea behind this project. No need to take it to the dealer, a way to run the diagnostic yourself and based on real data, decide what to do. (The cause of white smoke in the muffler is a blown head gasket? Or a coolant leak in the carburetor's base? With this question I mean that there may be a hard fix (head gasket), and may be an Easy fix (Carb base) for the same sympthom. So a right diagnostic may be a huge difference). I learned it in this forum

Of course there will be no magic in a basic OBD, you still would need some troubleshooting skills to find the root cause of the blinking code.

Thanks again for your input. That was the purpose of posting the idea first.

[carotman]

This is an awesome project!

I know you'll need to monitor the idle solenoid, air temp sensor, EGR.

[Buzo]

With this PCB I can monitor 4 analog signals at the same time.
The board communicates directly to my laptop's serial port, so I will be plotting whatever data i get in excel.

I'm thinking in the Air temperature, Coolant temperature and oxigen sensor, since these signals are used by the ECU to make decisions.
With this information I can set working limits and later, rise alarms if out of such limits.

Here is the PCB in theory:



And here is the same PCB, but in the real word:



Tomorrow will hook it up to the car and see what happens.

[lostforawhile]

With this PCB I can monitor 4 analog signals at the same time.
The board communicates directly to my laptop's serial port, so I will be plotting whatever data i get in excel.

I'm thinking in the Air temperature, Coolant temperature and oxigen sensor, since these signals are used by the ECU to make decisions.
With this information I can set working limits and later, rise alarms if out of such limits.

Here is the PCB in theory:



And here is the same PCB, but in the real word:



Tomorrow will hook it up to the car and see what happens.

the air temp sensor isn't an analog signal,it's just on and off, and it doesn't even communicate with the control box, it switches a resistor in and out to control the choke opening speed, the temp sensor only connects to the gauge, the o2 sensor will be a narrow band signal. i'll have to look at the wiring diagram, but I believe the other temp sensor on the block only is a switch for some of the emissions valves

[Buzo]

the air temp sensor isn't an analog signal,it's just on and off, and it doesn't even communicate with the control box, it switches a resistor in and out to control the choke opening speed, the temp sensor only connects to the gauge, the o2 sensor will be a narrow band signal. i'll have to look at the wiring diagram, but I believe the other temp sensor on the block only is a switch for some of the emissions valves

Of course they are connected to the control box.
Look at this page (11-87) of the service manual.
At least these are the ones I am referring to.



The ECU makes lots of desicions based on the coolant and air temperatures.
Here is just one example.



I spent too much time watching the footbal games, so very little progress today

[lostforawhile]

Of course they are connected to the control box.
Look at this page (11-87) of the service manual.
At least these are the ones I am referring to.



The ECU makes lots of desicions based on the coolant and air temperatures.
Here is just one example.



I spent too much time watching the footbal games, so very little progress today

that must be the 88-89 the one i have shows no temp input to the ecu

[carotman]

I know the ECU also checks something to activate the Idle solenoid valve.

For some reason, my ECU doesn't activate it anymore . I had to ghetto rig something to bypass it.

[Buzo]

Here is the first data logged.

You can see how the TA changes resistance accordingly to the door position. If its open, lets hot air enter and the resistance and voltage increases.
If its closed, lets cold air enter, the resistnce and voltage decreases.


The 5V at the end is because I disconnected the white connector from the TA sensor. If I make a short, of course it will be 0V. But I don't want to do that, he-he

For my diagnostics purposes, I can monitor that the sensor changes above and below 2 Volts. Or I can put wider open limits, "as long as it is not at +5V or 0V then the sensor is OK".





Does it make more sense now?

[Buzo]

O2 sensor data log

Need to elaborate more in this one, like logging the signal during a trip.
Just fired up my car and waited until warmed up then started acquiring data.
The las three peaks are some 3000 RPMS-5 seconds-lenght accelerations and the 0V at the end is when shut the car off.

[AccordB20A]

thats neat as! good work

[88Accord-DX]

Oscilloscope readings are pretty neat on diagnostics. Looks like you have a interesting project going on. Good luck with it, I'll stay tuned in.

[forrest89sei]

Cool Project Carbs And Computers Together could be good?? WOW!

[2ndGenGuy]

This is super badass! I've been poking around this thread every once in a while since you started it. It's looking really awesome so far! I love the flow from concept to prototype board! This is by far the coolest DIY stuff, IMO. Something about this is more exciting to me than turbos and horsepower and all that stuff!!

One thought that popped into my head was to T into vacuum lines at certain points with some MAP sensors (if they would really even be MAP sensors anymore), and monitoring vacuum at thermovalves to see if they are opening at proper coolant temps. You could use sensors to check vacuum at any of the crazy 130483948 lines in the carb system on the fly to monitor if things are getting vacuum when they should be... Just an idea, anyways.

What do you do? Must be some kind of electrical engineer? Or just a hobbyist?

[Buzo]

This is super badass! I've been poking around this thread every once in a while since you started it. It's looking really awesome so far! I love the flow from concept to prototype board! This is by far the coolest DIY stuff, IMO. Something about this is more exciting to me than turbos and horsepower and all that stuff!!

One thought that popped into my head was to T into vacuum lines at certain points with some MAP sensors (if they would really even be MAP sensors anymore), and monitoring vacuum at thermovalves to see if they are opening at proper coolant temps. You could use sensors to check vacuum at any of the crazy 130483948 lines in the carb system on the fly to monitor if things are getting vacuum when they should be... Just an idea, anyways.

What do you do? Must be some kind of electrical engineer? Or just a hobbyist?

Oh thanks for your comments & ideas. I only hope it helps to track down failures in this complex system.

I am electrical engineer and I have been always doing electronic stuff as a hobby, like transforming a simple $20 USD remote controlled car into a robot that can follow a line drawn in the floor, etc. So now that I got this real car full of sensors, its time to do something more beneficial.

But going back to the project, tomorrow will be an exciting day, I will permanently hook up the Microprocessor to the ECU and log all digital signals at the same time during different driving conditions.

After I understand better how things interact, I can start writing the code that will let me know when "something should be activated at this time and it was not".

This is how the Acquisition board has evoluted in these two weeks:
I added the screw blocks to make the connections to the ECU and some resistors below to be able to reduce the 12V signals to 5V and protect the micro.

[Buzo]

...thats neat as! good work

...Oscilloscope readings are pretty neat on diagnostics. Looks like you have a interesting project going on. Good luck with it, I'll stay tuned in.

...Cool Project Carbs And Computers Together could be good?? WOW!

Not sure how to do multiple quotes here, but I thank you for your comments, This thing is working better than I expected.

We will know by tomorrow the secuence of the star up, from the time the key is open, until when the car is fully warmed up.

[Dr_Snooz]

Off the hook!