Going Analog in a Digital World

You've read about the fancy digital dash app I've written that displays all sorts of information on a display in the hole that used to belong to the tachometer.  But that still leaves four other analog gauges in the instrument panel: The speedometer, clock, fuel, and temperature gauges.  These are old school VDO gauges that just look proper and cool so I want to make them all work with this new digital vehicle.

The instrument cluster with all analog VDO gauges in beautiful refinished real walnut

The rear of the original VDO gauges

The clock is no problem.  While young children may no longer be taught how to read an analog clock, I know how and mine has been serviced with a new quartz movement and works perfectly.  Done.

The original clock, modified with quartz movement

I have a plan for the speedometer but haven't executed it yet.  The idea is to buy a mechanical drive speed converter with a digitally controlled motor that spins a short speedometer cable.  I will send speed messages to the unit from my Raspberry Pi dash app via CANBus.  But that can wait.

The original speedometer and odometer

Speedhut's Speedbox digital to mechanical speedometer adapter

In the meantime, I've been focused on the fuel and temp gauges.  I'd like to show range on the fuel gauge and the "most relevant" (i.e. closest to critical) temperature on the temp gauge.  I have both of these values already calculated in my dash app but it would be nice to show them on the analog gauges also.

The original fuel and temperature gauges (upside down)

Plan A - Emulate the variable resistance sending units via digital potentiometers.

This seemed easy.  I would measure the range of resistance sent by the fuel float and temperature sending units, map my range and temp percentages to the corresponding resistance in my dash app, and then set a digital potentiometer to that value.  Easy peasy.

Measuring the resistance of the temperature sending unit from cold water to boiling

Charting the resistance of the fuel and temperature sending units

Chart of resistance curve for temp gauge

I verified the theory using a manually adjustable potentiometer and it worked great.  I can move the gauge needle by twisting the adjustment knob on the pot.  I saw a promising digital potentiometer on Adafruit that was already configured for I2C communication (supported by the Raspberry Pi) so I bought a couple and started programming.  All the sample code is in Python, which is fine but not what I'm using for the dash app (I use javascript) so it took a while to get it all working but I eventually was able to control the pot.

Testing gauge movement with adjustable potentiometer and a screwdriver

However, it turns out that pot has a range of 0 - 10k ohms with 256 steps (so roughly 40 ohms per step) while the range of ohms I need for the gauge is in a narrow range of about 0-220.  That means the gauge would jump between about half a dozen needle locations.  Not ideal.

So I found some 0-1k ohm digital potentiometer chips and got those.  Now I had steps of only 4 ohms so smooth needle movement was possible.  However, the gauge needs to share the electrical circuit with the pot but one is 12v and the other is a sensitive electronic 3.3v.  I couldn't get it to work.

Plan B - Build a circuit.

I'm not the first person to want to do this.  The Raspberry Pi and the Arduino both support Pulse Width Modulation (PWM) outputs which should be useful in creating variable voltage, resistance, or current to drive the gauges.  I found an article online that was promising so I bought some operational amplifier chips and combined one with a capacitor and some resistors to build a low pass filter and ... nothing.

Circuit on a proto board withe an Arduino, low pass filter, two operational amplifiers, and a MOSFET

I tried doing some math to figure out a proper circuit

And more math

But failed

One circuit

Another circuit

Another circuit

Some transistors, MOSFETs, and many circuits later and I eventually was able to program a circuit using an Arduino microprocessor to move the needle in the gauges!  Exciting.  However, when I added a second circuit for the other gauge, all the calibration was off.  Then, when I tried to illuminate the dash lights it was all off again.  The circuit is too sensitive to fluctuations in voltage and resistance in the car's electrical system.  This is because my circuit is running in "open loop" but I don't know how to fix it.

The videos above show how I can digitally control one gauge at a time, moving from empty (or cold), through 1/4 increments to full (or hot).

Plan C - Stepper motors.

Building a digital circuit to mimic the analog sending units proved beyond my meager electronics ability so I resorted to the modern approach -- replacing the guts of the gauges with stepper motors.  This is how modern car gauges operate.  The stepper motors are small, precise, low torque motors that can be easily controlled from a computer in increments as small as half a degree of rotation.  The X27.168 motors I am using have a total sweep of 315 degrees with 600 steps, or 1.9 steps per degree.  This is plenty of arc and precision for smooth gauge operation.  The motors have internal gearing but are extremely low torque -- only powerful enough to move something very light, such as a gauge needle.  The motors aren't super fast either - about 1 second for a full sweep - but they're fast enough for vehicle or engine speed and range and temps don't change quickly so, in this case, speed doesn't even matter.

The first step is to disassemble the original VDO gauges, removing the gauge itself, then removing the needle and gauge face from the gauge guts.

The original fuel gauge

The original gauge blown apart

The second step is to design a mount that will support the face, the stepper motor, and the needle and allow mounting back into the original gauge housing.  This is were Computer Aided Design (CAD) and 3D printing come in handy.  I designed and printed a two piece mount that has a recess to hold the stepper motor and screw holes for attaching the face.  This is covered by a spacer plate that allows for the electrical connections and has holes for attaching to the original gauge backing plate.  I printed it in clear PLA and left a space for light to come through from the original bulb location.

CAD model of stepper motor mount and spacer

The stepper motor installed in the 3d printed mount

Stepper motor mounted in clear 3d printed housing with room for light

Original gauge face mounted to stepper motor

Steppers all wired up nice and clean

The next step is attaching the needle.  The shafts on the VDO gauges are much smaller in diameter than than the shafts on the stepper motors so the needle won't press on.  The solution was to print a small disc that has an interference fit over the stepper motor shaft and then glue the needle to that disc.

Shaft on original gauge (left) is much smaller than on stepper motor (right) making needle installation a PITA

Time to build a controller circuit and program the motors.  There are a number of chips available for controlling stepper motors, such as AX1201728SG quad drivers, or L293D or TB6612 dual-channel H-Bridge motor drivers.  I opted for the TB6612 chip because all the chips are surface mount and that ship was available in a pre-mounted board configuration from Adafruit with polarity protection FET on the motor voltage input and a pullup on the "standby" enable pin. 

There are a few different Arduino libraries for working with steppers.  I used one called SwitecX25 because it offers good asynchronous control, which I need in order to move both gauges at the same time.  The basic code is silly simple.  It gets slightly more complex with CANbus integration but is still quite easy.

Simple circuit using an Arduino MKR Zero and two TB6612 driver chips

The hardest part of the whole process is attaching the original needles.  I'm still working on that.  Using paper needles in the meantime.  See it operational with a sample program that just moves between min/half/max values below.

Previous

Next