No announcement yet.

Vehicle Diagnostic Unit

  • Filter
  • Time
  • Show
Clear All
new posts

  • Vehicle Diagnostic Unit

    Regrettably due to a number of expensive upcoming major repairs that will be needed I have decided to retire my '92 Geo Tracker after ten years of reliable service sometime this spring. In it's place it makes sense I bought something older. A 1982 AMC Eagle.

    This car existed in the VERY very early days of computer emissions control and at the tail end of emissions systems of the 70's that really strangled engines to keep smog down. The 6-cylinder engine in the Eagle uses a number of digital sensors and a stepper motor attached to the Carter BBD dual barrel carburetor which on its own seems to be the scorn of most old farts because how dare a computer do the lord's work in a carburetor!
    Anyways the major issue with these configurations is that when something fails the computer does not tell you. The 4 and 6 cylinder Computerized Engine Control (CeC) does not utilize a Check Engine light or any method to store codes on the 6-cylinder model. It just assumes everything works, otherwise the car runs like crap. The computer itself is encapsulated in potting compound. If I can find another of these computers I would like to take a shot at stripping the potting off to reveal exactly what is going on inside this.

    To diagnose the system when something does eventually go wrong AMC for many years used a pair of Molex plugs under the hood for the home user or garage to diagnose the inputs and outputs of the CeC. Diagnosis of the expensive computer itself is otherwise impossible. AMC at one point sold a special tool that plugged into these Molex connectors to give you realtime sensor information. I have never seen one of these for sale.

    The obvious solution here is to just make my own. The great thing about the CeC is that when I mean "digital" I mean almost all the inputs and outputs are either at 0v or +12v. On or off. With the exception of the O2 sensor every input and output for the CeC is accessible from the diagnostic connectors.

    Only the O2 sensor exists as an analog input. Once heated it will vary between 0 and 1v to inform the CeC if the engine is burning too rich or too lean and that in turn allows the CeC to step pins inside the carburator to adjust the air/fuel ratio.

    This project consist of three phases, each progressively more complicated:

    -Add test pin for O2 sensor to the diagnostic harness, which is easy as the 6-pin Molex has four unused pins
    -Assemble wiring harness to connect vehicle to the Diagnostic Unit. I know we have the parts needed in the Hack Room
    -Assemble the Diagnostic Unit

    The Diagnostic Unit requires no logic. If a solenoid is on, an LED lights. If it's not, the LED goes out. As an extra feature for switches I can use a relay driven by the sensor to toggle indicator lights to show if something is opened or closed. Battery and O2 sensor readings can be read using regular analog meters. A 1 and 15V meter respectively are perfect. Engine RPM can also be monitored using a digital tachometer. Additionally the Molex Plugs can be represented with a group of banana plug sockets to aid in diagnostics where signals must be forced by grounding or adding voltage to a line. The whole assembly fits into a sloped enclosure, presumably of plastic. I have two ideas in mind for what to buy but it comes down to finding one at the right price. Hobby enclosures seem to get really expensive once you want more than just a box. Nobody in town sells enclosures this complicated that I have found and Lee's Electronics in Vancouver does not have boxes large enough in plastic.

  • #2

    This is the wire harness I spent the last two Hack Nights assembling and repinning. The Molex plugs and pins were purchased at RTS electronics and cost about $7.

    This is most of the components for the Diagnostic Unit. The 12v relays are still in the mail.
    The items purchased can be (or could be) found below:

    The fuse holder was something random I found in the Hack Room as well. The Gould WindoGraf machine will make an excellent piece of secondary equipment for monitoring circuits over a prolonged period of time however requires sufficient transient supression and can only monitor four channels at a time.
    Last edited by MIPS; 12-18-2018, 05:39 PM.


    • #3
      After some procrastinating I decided to buy one of the double sloped enclosures seen above, albeit it is coming from the UK for $50 shipped.

      It is not however coming with the aluminum panels that you mount your components to. I know we have some sheet at the makerspace still, would anyone be able to help me by cutting out two pieces in the following dimensions?

      248 x 99mm
      248 x 49mm

      Failing that it shouldn't be too hard to quickly cut out two pieces of acrylic.


      • #4
        Sat down tonight and drafted the layout for the panels.

        Meters mount on the upper panel with the digital tachometer in the middle. The lower panel has the red, green and yellow indicator lights on the left and right sides (plus the power switch) with an electrically identical arrangement of banana plugs in the middle for jumpering or additional metering of the pins directly from the diagnostic connector.
        Remember that all the indicators operate either on or off. Each red and green LED operates with a relay between the lamps and the pin on the diagnostic harness. The two large connectors that connect the box to the harness stick out the back and the entire unit is protected by a fuse on the +12v pin.

        Some highlighting will be done...somehow. Either paint or cut vinyl. Lettering...I dunno but I'm not doing it by hand.
        Last edited by MIPS; 01-04-2019, 09:22 PM.


        • #5
          Very nice, Cant wait to see the completed build.


          • #6
            The box arrived. It's tight. A little bit smaller that what I wanted but there were no other affordable options available.

            The meters need to be trimmed slightly so that they will mount properly to the top half. There's a circular indent in the back which serves no purpose and couldn't remain or else the aluminum plate would be too thin at the top and bottom and prone to breaking. Removing the indent gives me 50% of the material for strength.

            I will be in on Wednesday to try and get the drill holes for the plates marked out. I'm hoping I only need to do this ONCE.

            I also marked out where the harness comes into the box on the back. A metal plate lives on the inside to reinforce both connectors as one assembly.
            Last edited by MIPS; 01-14-2019, 10:16 PM.


            • #7
              Man I love Eagles. There was a guy in 100 Mile House in the neighborhood I grew up in that had dozens of those things in his yard.

              This is a really interesting look at how to communicate with old barely computerized vehicles. Keep us updated!


              • #8
                Transferred out the holes and openings form the layout to aluminum sheet and then after center punching and pilot drilling we got the metalwork finished.

                Up next is widening all the holes to fit their designated lights and switches. Still unsure how to apply the labels.


                • #9
                  While I work on the enclosure I am seeking a bit of external help on how to make all the lamps work.
                  the original design for an aftermarket diagnostic unit is here however is in comparison a far simpler device where a series wired resistor and LED connect to each computer input/output. My box as seen above has two lights to represent each sensor. That way I can better identify when something is opened or closed. To do so requires a dual position relay for each lamp. That however means that the computer then needs to deal with a 33ma current as opposed to something like 3ma from an LED and you have a connection to ground through the coil which may also interfere with the computer. The solution is to use a MOSFET or a darlington pair set of transistors to isolate the CeC control circuit from the lamp switch circuit.
                  Initially I was working on salvaging parts from a shift register controlled 16 relay security module.

                  This however did not work when breadboarded, partially because I reverse engineered the original circuit wrong. When I referred to a friend his suggestion was to scrap the darlington pair entirely and use something like a 2N7000 MOSFET.

                  I will have to purchase the parts and wait for them to arrive before proceeding with additional testing.
                  Last edited by MIPS; 02-02-2019, 08:38 AM.


                  • #10
                    Aaaaand the exterior is all done up. Looks sharp!

                    Now of course that leaves us now with the internal wiring to complete. That will wait until the MOSFETs arrive. In the meantime I still do not know how to put down the labels for each indicator. It's too small for vinyl and I don't remotely have the supplies to silkscreen.


                    • Garret H
                      Garret H commented
                      Editing a comment
                      Rob had talked about using some kind of overlay on stainless steel that you can then use the laser cutter to etch. I don't know more details then that...

                  • #11
                    Another circuit idea is the use of relatively inexpensive 4N33 optoisolators, as demonstrated below:

                    Edited: Because I am using 12V relays with exceptionally low coil current (30ma) you can actually rather safely drive the relay directly from the optoisolator. An even simpler circuit is the following:

                    Where the transistor is the output side of the 4N33. The reason I would suggest this way is that it closely reflects an earlier diagnostic unit someone else made in that the computer inputs and outputs do not mind the very light load of an LED but we can still isolate the entire diagnostics system from the computer in case a catastrophic failure occurs.

                    Edited: Edited: And here is the above breadboarded. The only selective component is the 680 ohm resistor which will let me not blow out the optocoupler in the unwieldy 11v-14v DC range a car will run on. The orange wire is what you would have tap into the diagnostic connector.

                    Last edited by MIPS; 02-07-2019, 11:48 AM.


                    • #12
                      Last night consisted almost entirely of preparing for the one thing I was not looking forward to: Making the connections.

                      This PCB is what controls all the lights and distributes EVERYTHING. The ten relays operate the red/green lamps for the more important sensors. Each relay has a diode for the flywheel and the board will consist of ten of the above circuit, plus potentially a bit more space in case I need to add tweaks for the tach-to-tachometer signal. While the two rear connectors are hardwired to the plugboard everything else passes through three 26 pin connectors. Just preparing everything for the soldering took a significant portion of the night.

                      I'm still not done. There's another 26 wires left and in the process I found out I screwed up the pinning on the rear connectors, so that harness will have to be dismantled and de-pinned form the molex plugs and done again. I'm exhausted. I'm done for today.


                      • #13
                        Nice work!


                        • #14
                          Work resumes.

                          The cabling harness had to be rebuilt due to pin errors discovered during assembly.

                          The cable was made as a separate piece due to AMC and later Chrysler/Jeep continuing to use the same diagnostic connector, but with a different pinout. This way if you need to use the box with another unit it only requires a different harness.


                          • #15
                            A few people might of seen it on Saturday.
                            The unit required some additional research to verify it would work. This required as part of another project to reverse engineer the computer completely for conversion to an arduino device to source another computer, depot it and do a little reverse engineering to see how exactly the signal ground worked when the computer itself is well noted in documentation to NEVER be put in contact with chassis ground.
                            I won't go totally into detail as to what I've done and how but here's pretty pictures for the moment.

                            Edited: Nifty. The depotting process was documented by me on another forum however I submitted it to Hackaday and it was posted on the 27th.
                            Last edited by MIPS; 06-03-2019, 09:11 PM.