LCD Instrumentation for Cougar Controller

[ianbartie]

It seems as if there is something to do with the dash panel stopping the tacho. It didn't respond to the 200Hz square wave. It could be a missing signal of some sort so I am going back over the wiring diagrams & going to rerun the DTC scanner & see what faults are logged.
Failing that, I may resort to fitting a regular aftermarket tacho into the dash instead.

[ianbartie]

I have been playing around with the fuel tank sender & a couple of little trim pots.
Even though the sender wire connects via the sender to a dash panel return wire, it works equally well connected via vehicle ground.
This is what I have found.
Empty needs 71 ohms & measures 35mV
Quarter needs 119 ohms & measures 48mV
Half needs 166 ohms & measures 58mV
Three quarters needs 227 ohms & measures 64mV
Full needs 283 ohms & measures 70 mV

Is this wonderful little LCD display able to drive this gauge somehow?

[Freyguy]

Ian,
I believe we can make this work, but you'll need to change one line of code to invert PWM vs SOC direction.
1. Referencing version 1.3, change line 1755 from "subfwb RES2,f" to "addwf RES2,f ". This will make PWM increase as SOC decreases.
2. Make R20 on LCD board a 71 Ohm resistor (or something slightly less). A 100ohm in parallel with 220ohm will give 68.75ohms which is close enough.
3. Change constant "PWM_MAX" from 255 to 100~130 (decimal not hex)

This is going to be a trial and error process with PWM_MAX. Start with 100 and see where the needle goes to at 100%SOC. If it's less than full, increment to 110, program and observe again. Once needle is at full, then repower while holding down SW2 to force 20%SOC. Adjust VR2 until the needle reads empty. Cycle power to bring 100%SOC back again. The needle should still be at full. If not, adjust it, then recheck needle at 20% adjusting VR2 again if necessary.

[ianbartie]

Thanks very much Chris,
I'll get some bits & start experimenting. I hope I can make it work because it will look so good to have the original fuel gauge as a proper SOC meter.

My last little trick will then be to have the original temperature gauge respond to the battery amps being pulled. Not quite sure if this will work OK though.
I am sure I can come up with the right input but the needle may be too slow to respond reasonably.

[ianbartie]

I have made the change to line 1755, put in a 68 ohm resistor & have done some experimenting.
I have tried various PWM_MAX values between to 100 & 125 range to see what happens.

The car measures +0.030mV @ 71 ohms = empty & +0.70mV @ 283 ohms = full between the sender wire & car earth. The sender wire by itself measures 106mV to earth.

So far, I haven't been able to get the needle to move via the PWM yet but it will move with a variable resistor between the sender wire & earth.

By my reckoning, the PWM should be about 38% for full & about 80% for empty. That equates to a PWM_MAX of 0x61.

Reading some stuff, it seems like I may need a low pass filter. That may give a signal that the computer likes better. I could also try a digital pot perhaps.

[Freyguy]

Ian,
PWM_Max values of 100 (0x64) to 125 (0x7D) equates to a pwm range of 39~49%. I’m assuming needle just remained pegged at full? You could try higher values, but if you forced SOC to 20% (SW2), VR2 should allow adjustment of PWM over the full range up to 100%. With Q1 “on” 100% through the 68ohm resistor, it would be equivalent to 68 ohm from sender wire to earth. 95% duty would be closer to average current produced by 71Ohms. I would think the needle would move to empty, similar to the 71ohm resistor test you did.
Did you confirm PWM pulses on TB1 position 10 when connected to sender wire? If so, did duty change with VR2?
Does the attached look like your setup? It shows either LCD brd or your test trim pots as they would connect to sender wire going to instrument cluster.

The voltage drop across the 71~283 ohm indicates current in the 2~5mA range…definitely not driving a coil! I think this interface is current driven (similar to 4mA current loop).
If everything above checks out but still no needle movement from full, then try adding a large cap from TB1-10 to gnd (100~470uf) .

[DJBecker]

Gauges are an unpredictable load. They have an inductive component, but it's often not enough to filter the PWM signal to an average value.

When testing with a 25KHz PWM signal and MOSFETs with negligible on resistance, we found that most Miata gauges required 50-95% pulse width to operate. Each gauge type reacted differently, with some generating enough "kick" to require a freewheel diode.

The point is that you aren't going to be able to guess at the appropriate pulse width for a specific gauge reading. You really have to sit down and try it out, building a calibration table.

[ianbartie]

Hi Chris
My setup is the same as your drawing. There was no movement of the needle at all. Stayed on empty even adjusting VR2.
I can see the pulses on TB1-10 on the scope when I bench test it but I haven't scoped it in the car. The duty changes between 100% or 20% on the bench.

I'll do that this morning & try the capacitor too. I'll also measure the current between full & empty using the variable resistors.
The sender wire definately is not driving a coil. The VW uses stepper motors in the gauges. The wiring diagram just shows the sender wire going into the mysterious 'dash panel'.

I have found during this conversion that if there was a simple way or a complicated way to do things, VW always choose the most complicated way.

[ianbartie]

Put the scope on the PWM output & hooked up the display. The wave form went really strange so I hooked the scope straight to the sender wire without the LCD display.
With the ignition on, there is a square wave present on the sender wire.
It has an amplitude of 3V at full and down to an amplitude of 1.8V at empty.
The frequency is approximately 230Hz.
I think this is the reason the gauge is not working with the current PWM.
The gauge still works fine with the trim pots though. Maybe a digital potentiometer driven by the PWM?

I am starting to think that using my custom LED gauge may be easier. It is a series of LEDS driven by an LM3914. It was originally going to reference the pack voltage via a voltage divider down to 3.2V.
I can probably drive it with the PWM output & an RC filter instead.

[harlequin2]

I assume the sender is the tank unit and you have it on the bench so you can vary it? Also, I guess it has a "ground" to the frame rather than 2 wires? Have you measured its range of resistance from full to empty and what is it?
I wonder if the computer is simply sending a square wave through the fuel gauge and the sender in series although there must be some damping there too. How quickly does the gauge follow the changes in sender?
Also, what's on the sender wire with the sender disconnected?