I got the speedo working with a PWM signal generated by a generic output from the Megasquirt.  I then send this signal through a PWM converter with a 32 multiplier.  This generates a 499Hz signal that indicates a speed of 18km/u, what corresponds with what I calculated.  It was kinda cool to see the odo ticking forward.  I let the car run for 8.8km .  The speedo should be spot on .
Unfortunately, I didn’t get it working in the car yet.  My transmission generates an AC signal that I convert to a square wave and send to the PWM converter.  Although it works with my self generated signal, it doesn’t work with the signal coming out of my VR converter.  It looks ok on the oscilloscope though.  I’m a bit puzzled, but I’m sure I’ll figure it out eventually.

Only thing left is to program the PIC16F628 to correct the signal.

Decided to oder a LM1815 and build a new VR circuit like this.

Very simple VR circuit.

Just after ordering I found this circuit.  Needs only a couple of resistors and 2 transistors.  Gonna stick with the LM1815 based circuit though, now that I’ve ordered the chip …

Here’s another one with extremely low parts count.  I actually have the Maxim chip, but it’s surface mount so very difficult to solder

Code for the pic is here.

Put the Westfield with its wheels in the air to test the speedo.  Nothing .
Don’t know exactly why it isn’t working, my oscilloscope gets a reading so it should indicate something.

At 50 km/h (13.889 m/s) I have 41.5 Hz, iow 41.5 pulses for 13.889 m or 4800 pulses per mile.  This is weird as it’s supposed to be 4000 ppm.  Either way, it should read something, even at 4800 ppm.
The speedo expects 156000 ppm (32.5 times faster).  I drove the car up to 170 kph, so the speedo should at least indicate 5 kph, but I get nothing.
Difficult to diagnose if it’s the speedo or a problem with my signal.  Will do some more testing.  Worst case, I make a VR conditioner to directly condition the signal of the transmission to 0-5V and feed that into the speedo.

The wire coming from the sensor no longer goes to the instrument cluster but to the MS3X on port ADC13 instead.  Since the oil pressure sender outputs no voltage (it’s a simple lineair resistor), a bias resistor is added on the MS3X card in the R67 position.  The oil pressure sender has a fairly low resistance so I used 330ohm to get good accuracy.  Worst case, this pulls about 15mA through the sensor so it shouldn’t be a problem.

The factory manual mentions these resistor values:
0psi => 52 ohms
90psi => 16 ohms

With a 330 ohm bias resistor we can calculate the corresponding input voltage on ADC13:
0psi => 5*52/(52+330) = 0.68V
90psi => 5*16/(16+330) = 0.23V

With this we can calculate the 0V and 5V values in Tunerstudio:
0V => (90-0)/(0.23-0.68)*(0-0.68) =  136
5V => (90-0)/(0.23-0.68)*(5-0.68) = -864

Enter these values in Generic Sensor Input 1
sensor – source: Spare ADC (ADC13)
Transformation: Linear mode

Now you can use a spare output port to trigger the oil warning light when oil pressure goes below 15psi.  I use these values:
Output port IAC2
Power on value: 1
Trigger value: 0
Output channel: sensor01 < 15  Hysteresis 2

If you have a real oil pressure gauge, you probably want keep using it (no point in disconnecting a perfectly good oil pressure gauge!).  You’d obviously need no bias resistor and you need to make sure you do NOT send over 5V into the ADC port.  The gauge works with 12V, so it’s probable that you measure more then 5V over the oil pressure sender (I never checked though).  If so, you’ll have to use a voltage divider to bring it down to 5V.
Obviously, you’ll have to do some math since you’re now using an unknown bias resistor (easy to calculate its value).

The Honda connectors are removed and new wires soldered in.  I always use the oem Mazda color codes (an old wiring loom is super handy for this kind of thing) as it’s easier to spot mistakes.  I also made sure to have a stress relief for the wires should I ever accidently rip out the connectors.  It’s not pretty, but it does the job just fine.
As always, more pictures in the S2000 album.

The NB miata gearbox has a VR sensor which outputs an 8200 ppm AC waveform.  The miata instrument cluster has a built in VR conditioner to convert that signal to a square wave signal which drives the miata speedo.  This square wave is also used as an input for the megasquirt.  After reading the Megasquirt doc, it appears the megasquirt also has an VSS output so I’m going to (ry and) use it to drive the S2000 speedo.

MX5 circuit board installed with some stick on spacers to prevent it touching the S2000 pcb.

The problem is that when I remove the miata cluster, I’ll no longer have the VR conditioner.  Since the megasquirt cannot directly accept the AC waveform, I’d have to build a new VR conditioner in the proto area.  This seemed a bit too much trouble (couldn’t find a LM1815 locally), so instead I removed the small pcb/display (with VR conditioner) from the miata cluster and installed that inside the S2000 cluster.  Just like before, the output of the miata pcb goes to the megasquirt and the megasquirt output will be used to drive the S2000 speedo.

Modifry ECT module glued to the MS3X module

I installed the Modifry ECT module inside the Megasquirt.  This module takes the signal from the coolant temp sensor and converts it to a format that the S2000 cluster understands.  The input of the module taps into the sensor wire, the output goes to the S2000 cluster.  The module is fully programmable and even has an adjustable alarm output (as a warning for overheating for example).

For my own reference:
ECU pin 1G is Modifry CLT out to S2000 cluster 2A (V/W) via an additional wire
ECU pin 2D is MS3 VSS input from MX5 module 2L (W/B)
ECU pin 2A is MS3 VSS output to S2000 cluster 2O (P/L) via an additional wire

The S2000 cluster is a lot smaller than the miata cluster (that was the reason of doing the swap all along), so I’ll have to make a template out of plexi to make it fit.  Shouldn’t be too hard to do though.  A bit of plexi and some paint or vinyl sticky stuff should do it.  I haven’t yet decided which color, probably matte black.

I don’t have the S2000 connectors and I don’t want to cut the wiring of the car, so I decided to remove the S2000 connectors from the pcb and replace them with the connectors from the miata cluster.  Removing the S2000 connectors proved to be a bit challenging and in the process I managed to let my dremel slip and touch the pcb in one or 2 places.  It took me quite a while to repair the damage.  As Murphy would want it, I cut right in the middle of a bunch of very thin traces.  Luckely i managed to avoid cutting into components.

The problem with the miata connectors is that they are part of the instrument cluster, so I had to cut them out.  It looks worse then it is really.

S2000 cluster with cutouts for the miata connectors (bottom).I cut these out of the miata cluster.

Miata connectors glued in and circuit ready to be soldered.It looks messy, but everything is pretty firmly glued in.

I cut some holes in the S2000 cluster and glued the miata connectors in.  The left one was a straight fit, the right one needed enlarging and the middle one is an all new hole.  I’ll probably add a couple of screws to make sure they don’t come loose.
I used the corresponding bits of the circuit board layer and will solder the wires straight on them.

As I can’t really see all my dials, I decided to buy a Honda S2000 instrument cluster today.  It’s a lot smaller and should be entirely visible through the steering wheel.
The cluster is a bit of a pain to connect because the S2000 has a 156000 ppm (pulses per mile) speed sensor, while the miata has a 4000 ppm.  I’ll have to come up with a way to multiply the stock signal by 39.

Also, the coolant sensor does not just measure the resistance (voltage) of the coolant sensor.  Instead, the ecu measures it and sends a 5V signal that has a 30ms high pulse and a 150ms to 1500ms low pulse, relavite to the temperature, so that’ll require some modifications as well.
Oh, and I’ll have to reprogram the km reading, because I only have 12K kms, not 142K!

I have no idea yet if it’ll fit inside the Westfield dash as the upper part of the cluster is rather high.  I’ll make it fit if necessary.

the color filter can be easily replaced with a different color

I’m also planning on changing the color of the display because I don’t like the yellow/orange backlight.  It’s just a piece of plastic, so it shouldn’t be too hard to swap it with a different color.  The bulbs are a bit yellow-ish as well, so I’m going to look into replacing them with bright white leds.

More pics in the S2000 album.

Built the megasquirt this weekend.  The MS3 build is a lot easier than my previous MS2 builds.  With the MS3X, most of the extra circuits and flyback diodes can be omitted, greatly facilitating the build.  Again, I made the MS3 fully plug and play.  I really don’t like using extra wiring looms.  The 64pin connector has more than enough room to be put inside the case, so why make a wiring loom!
Build instructions here.