Mechanical speedometer drive solution

[Second Series]

How do I keep the mechanical speedometer if I replace the old transmission with a modern unit with reluctor ring?
Use an electric motor to drive the speedometer.
I have been working on this for a couple months now. I found a 17 page discussion that spanned several years. Someone had a working prototype, but failed to share any relevent information, and then vanished.
Break down what needs to be done:
1. Identify a motor that will function at a slow speed and have a range of speed wide enough.
2. Analyze the signals in a vehicle.
3. Build a motor controller to drive the motor from the vehicle signal.
4. Assemble the parts including speedometer cable.
I’m using an ’88 GMC K2500 for a frame swap for a ’47 panel truck. I needed a truck for work around my house, so I got a ’90 K1500 basically the same setup as the GMC. I’m using the K1500 to look at signals, and eventually to test the ’47 speedometer. A fair amount of math is involved, and I try to double check the convoluted equations. Please let me know if something is off!

The reluctor ring creates 40 pulses per driveshaft revolution. The DRAC converts the 40 pulses to 128,000 pulses per mile for the RWAL Brakes, 4,000 pulses per mile for cruise control, and 2,000 pulses per mile for the ECM.
Use the 2,000 ppm for the speedometer motor control.
Look at the signal on that wire:

[Second Series]

This is a screen shot of the oscilloscope, the truck is idling in 2nd gear at 10mph(on jackstands).
Channel 1 is the 2000ppm
Channel 3 is the 40 tooth Reluctor Ring that produces 40 pulses per driveshaft revolution.
Looking at the ch.1 2000ppm signal, one cycle is about 7 horizontal divisions. The time is set at 25ms.
The period of one cycle is 7 x 25ms= 175ms, Frequency = 1/period, so 1/175ms, or 1/.175 = 5.71Hz
This is close to the reading on another screenshot of 5.1Hz, so the math is correct.
60mph = 1 mile per minute, 10mph = 0.1666 miles per minute (10mph/60minutes = 0.1666) so 0.1666 x 60 = 10. And 1 mile in 6 minutes
2000ppm at 5.71Hz, or 5.71 cycles per second x 60 = 342.6 cycles per minute, or 342.6 pulses per minute 342.6 x 60 = 20,556 pulses per hour. At 10mph, 20,556/10= 2,055.6ppm. With some margin of error, close to 2000ppm. So for 10mph, the vehicle travels 1 mile in 6 minutes, and at (2000ppmile/6minutes) = 333.333ppminute. Thus at 10mph the vehicle travels for 1 minute a distance of 0.1666 miles with 333.333pulses. Let’s check this: 333.333ppminute/60seconds = 5.555cps, or 5.555Hz, that’s close to the measured 5.71Hz.
At 60mph the vehicle travels 1 mile in 1 minute and 2000ppmile. 2000ppm/60seconds = 33.333Hz
At 20mph take 60minutes/20mph = 3, the vehicle travels 1 mile in 3 minutes. At 2000ppm/3minutes = 666.666ppminute. 666.666ppminute/60seconds=11.111Hz So at 20mph, the vehicle travels .333 miles in 1 minute with a fequency of 11.11Hz on the 2000ppm signal.
With this math, a chart can be produced for the 2000ppm signal:

[Second Series]

“GM old standard was always 1000cableRPM = 60MPH. Start dividing by 2 and get down to 2MPH needs about 33RPM.”
(1000RPM=60MPH)/30= (33.3RPM=2MPH)
1000 Revolutions per minute = 60MPH, and 60MPH = 1 Mile per Minute so
1000 Revolutions per minute = 1000 Revolutions per mile
How to get 2000 pulses per mile 12 volt square wave to drive the motor 1000RPM?
At 60MPH, 2000 ppmile = 2000 ppminute. So (60MPH=2000ppMin)/30 = (2MPH=66.6ppMin), and 66.6ppMin/60seconds = 1.11ppSecond, or 1.11Hz
1000cable revolutions per minute = 2000 pulses per mile at 60 miles per hour, so
1.11Hz = 33.3RPM at 2MPH.
I’ll need to take the 1.11Hz signal from the DRAC and convert that to drive my motor 33.3RPM at 2MPH.
We need an electric motor capable of starting around 15RPM that can increase to something above 1000RPM.
I set up a test jig and tried several motors:

[Second Series]

The arduino supplies a variable drive signal that controls a darlington pair, the motor is connected to 12v and the darlington pair. The RPM sensor reads the reflective flag on the motor and sends a signal to the oscilloscope.

Makita Driver:
The driver can start and run slow:

[Second Series]

.5145*60= 30.87rpm
Makita fast:

[Second Series]

20.76*60= 1,245.6rpm
The Makita driver is the correct range for the speedometer driver.

How to derive vehicle speed from motor speed:
(1000rpm/30)=33.3rpm, so (1000rpm/33.3rpm)=30 take this result and divide into 60mph:
(60mph/30)=2mph
Let’s look at our test results:
The Makita Slow:
(1000rpm/30.87rpm)=32.39 and (60mph/32.39)=1.85mph
The Makita Fast:
(1000rpm/1,245.6rpm)=0.80 and (60mph/0.80)=75mph

The Makita driver is a work tool, so I went to the second hand store to find a good deal on a cordless driver.
I found a Chicago Electric driver. I extracted the motor, but it was too fast. I put the planetary gear back on and that got it in the ball park.
Chicago Electric w/Planetary Gear:

[Second Series]

2.23Hz * 60s = 133.8rpm
(1000rpm/133.8rpm)=7.47 and (60mph/7.47)=8.03mph

[Second Series]

22.51Hz * 60s = 1,350.6rpm
(1000rpm/1,350rpm)=0.740 and (60mph/0.740)=81.08mph

I may be able to get the motor spinning and then slow it down to drive the speedometer less than 8mph.
Let’s look at RPM of the electric motor vs. Hz. of the DRAC:
The Chicago Electric slow speed is 133.8rpm.
The formula to find MPH is (1000rpm/133.8rpm)=7.47 and (60mph/7.47)=8.03mph
Let’s rearrange this equation to find rpm:
(60mph/8.03mph) = 7.47 and (1000rpm/7.47) = 133.8rpm
Put this formula into a spreadsheet with the mph and Hz to find the range of RPM for the speedo motor, and create a chart:

[Second Series]

Now we need to write some code for the Arduino to convert the range of frequencies to the range of motor speed.
The Arduino uses a scale of 0-255 to drive a motor. I could use 0-50Hz as my drive signal.
255/50 = 5.1, so my code will have a line that looks something like this:
analogWrite(motor, freq*5.1);
where analogWrite is the command, motor has been designated as the output pin in another line, freq is the frequency calculated in another line, and freq times the multiplier gives me the variable output between 0 and 255 from the variable input between 0 and 50Hz.
There are limitations here that will need to be addressed. The range 0-50Hz is good to about 90MPH, if I go faster than that, the signal will be out of range of the code. I can add a line that treats anything above 50Hz as 50Hz. The motor drive signal range 0-255 is a pwm with 255 being dc, 255 is the fastest the motor will spin. I’ll set the maximum RPM on the speedometer face to be driven by the 255 setting. The motor has a minimum speed that it will start at, and a minimum speed that it will run at. The minimum speed that it will run at may be lower than the starting speed. The Chicago Electric motor with planetary gear starts at what would be 8MPH. If it can run slower, I’ll have to goose it to get it moving, and then immediately back it down to under 8MPH. The pwm range 0 to 255 won’t start at 0, it will start at what it takes to keep the motor running at it’s slowest speed.
I’ve played around with a few examples of Arduino code that I found on-line. I have reconfigured the Arduino to accept a variable frequency from a fuction generator. I discovered that the Arduino does not recognize a 4volt frequency, it must be 5volts.
The motor will run with the Arduino signal 28, but it will reliably start to run with signal 35.
The code is adjusted to add 35 to frequency x 5.66. The upper end is limited to 255, so any input above 254(60Hz.) is converted to 255. The low end. The code named “freqCount” works well. But because 35 is added to the equation, the top speed is achieved with 39Hz. I’ll have to re-code so the motor starts with 35 at 1Hz, and runs through to 60Hz for 255. 39Hz x 5.66 = 220.74, and 220.74 + 35 = 255. 220/60Hz = 3.66, so I’ll change the multiplier in the code to 3.66.
The code seems easy enough to modify, and still needs improvement. I’ll check this setup with the RPM sensor when it becomes available.
The vehicle has 12volt signals, the Arduino runs on 5volt signals. Use a 7805 or similar to change 12v 2000ppm to 5v for Ardunio input. I’ll have to look at datasheets to verify this component reacts at the appropriate speed.
My transmission is from an ‘88GMC. The DRAC is integrated in the gauge cluster so I need to find a DRAC from ’92 to ’95 Truck, or ’90 to ’95 Van. I went to a couple wrecking yards last week, all the electronics were gone…
I may be able to use the Arduino for a DRAC. That would make it useful on a wider range of applications. The signal will require conditioning to transition between vehicle and Arduino. The VSS, reluctor signal looks like a .5v ripple on 12v. The 12v can be blocked with a capacitor, and the .5v ripple will need to be boosted to 5v so the Arduino can see it. That should be simple enough to achieve. The 128,000ppm, 4,000ppm, and 2,000ppm signals will need to be boosted from 5v to 12v.
The circuit in my truck takes the signal from the VSS/Reluctor sensor, runs it through the Ratio Adapter and produces a 128,000ppmile signal, I read about a Schmitt trigger used to convert the signal to a 12v square wave, from my o-scope results, it appears to happens with the divide by 32 circuit to produce the 4,000ppmile signal, finally a divide by 2 circuit produces the 2,000ppmile signal.

[Second Series]

This is a screen shot of the oscilloscope, the truck is idling in 2nd gear at 10mph.
ppm = pulse per mile
Channel 1 is the 2000ppm at 5.1Hz
Channel 2 is the 4000ppm at 10.45Hz
Channel 3 is the 128,000ppm at 332Hz
The Gauge Cluster diagram shows a divide by 32 circuit, so 332Hz/32= 10.375
10.375 is close to 10.45Hz, so the frequency counter in the o-scope is acurate and the readings are valid.

[Second Series]

A close look at the 128,000ppm signal:
This is a screen shot of the oscilloscope, the truck is idling in 1st gear.
The 128,000ppm at 158Hz.

[Second Series]

I’ll have to convert the VSS to something the Arduino can detect, Multiply that to get the 128,000ppm and use some external circuit to get this kind of waveform, the 4000ppm and 2000ppm can be produced with code and boosted to 12v with discrete components. I could add a potentiometer to adjust the multiplier for the 128,000 signal to accommodate a range or gear ratios and tire sizes.

I’ll need to take some more o-scope readings to double check the math, let’s look at it from the other end.
Stock tire size is 225/75R16, or 29.3” so 3.14 x 29.3 = 92.002” circumference of tire. 63,630 inches in a mile. 63,630/92.002 = 691.615 tire rotation per mile x 3.73 gear ratio = 2,579.725 driveshaft revs per mile. 2,579.725 x 40 teeth = 103,189.018 pulses per mile. At 60mph(1 mile a minute), that’s 103,189 ppminute And 103,189/60s = 1,719.81667pps, or 1,719.81667Hz @ 60MPH
At 20mph take 60minutes/20mph = 3, the vehicle travels 1 mile in 3 minutes. At 103,189 ppm/3minutes = 34,396.333ppminute. 34,396.333ppminute/60seconds=573.27Hz So at 20mph, the vehicle travels .333 miles in 1 minute with 573.27Hz from the VSS.
Take this equation, plug it into a spreadsheet, and come up with a chart for MPH vs. VSS frequency:

[Second Series]

This is good for any transmission with 40 tooth reluctor, range will be slightly different with other gear ratios and tire sizes.

So, this is what I have done for now. I will continue to look for a suitable motor, probably a cordless driver from a second hand shop. They are in the right range, and include gears. I tried looking on-line for motors, and gears with poor results. The Mabuchi RS-555PH was suggested in that other thread, but it needs to be geared down.
I need to build the box, get a connector for the speedometer cable. Where can I get the connector like on a transmission that the cable attaches to? 5/8” thread for GM?
I’ll take some more o-scope readings of the VSS.
Breadboard the discrete components to modify the VSS signal for Arduino, etc.

[mongocanfly]

looks like you know your stuff......waaaayyyy over my head...
me..i just get a Dakota digital convertor box and move on...

[skyphix]

I love this at home hacking stuff. I haven't messed a bunch with the arduino's yet but keep wanting to pick up a basic kit and try it out.

I just ended up using the OBDII port, a bluetooth sender, and will be using a tablet for my gauges.

[nsb29]

Originally Posted by mongocanfly

looks like you know your stuff......waaaayyyy over my head...
me..i just get a Dakota digital convertor box and move on...

I’ll second that

[prodjay10]

I’ve been wanting to do something like this with a stepper motor. That’s is supposed to be used in the speedos anyway. I just lost interest.

You can get arduino clones for cheap.

[dayj1]

Originally Posted by Second Series

I will continue to look for a suitable motor, probably a cordless driver from a second hand shop. They are in the right range, and include gears. I tried looking on-line for motors, and gears with poor results.

If you'll PM me your address, I'll send you a stepper motor with a built in optical tachometer for feedback to your controller. It's RPM range is good up to about 120 MPH (2000 RPM). I was going to do this a few years back and sourced a couple of motors. In my case, I decided it was just easier to use a mid-90s GM speedometer and re-face it for my application. But, you obviously know what you're doing. If you can put the motor to good use, you can have it.

Originally Posted by Second Series

I need to build the box, get a connector for the speedometer cable. Where can I get the connector like on a transmission that the cable attaches to? 5/8” thread for GM?.

You're pretty much going to have to machine a fitting for this. I've never seen anything off the shelf that would work without some modification. I was going to machine a coupler to attach the motor directly to the gauge (and eliminate the cable and the resistance that comes along with it). I have access to the tools to make such a fitting, but, again, I just said forget it and went the route of least resistance.

[Second Series]

Originally Posted by dayj1

If you'll PM me your address, I'll send you a stepper motor with a built in optical tachometer for feedback to your controller. It's RPM range is good up to about 120 MPH (2000 RPM). I was going to do this a few years back and sourced a couple of motors. In my case, I decided it was just easier to use a mid-90s GM speedometer and re-face it for my application. But, you obviously know what you're doing. If you can put the motor to good use, you can have it.



You're pretty much going to have to machine a fitting for this. I've never seen anything off the shelf that would work without some modification. I was going to machine a coupler to attach the motor directly to the gauge (and eliminate the cable and the resistance that comes along with it). I have access to the tools to make such a fitting, but, again, I just said forget it and went the route of least resistance.

Very generous offer, Thanks! PM sent. I'm interested in comparing the different motors and how the speedo reacts. Once I get the circuit down, I can take the IC off the arduino, and mount it to a new pcb.

[pilotXC]

Dakota Digital pulse sensor works just fine in mine as Mongo stated.

[solidaxel]

D D after you get past the price it works great and easy to install!

[pilotXC]

The pulse sensor is $40-$50, I didn't do the digital gauges that are $1000.

[Second Series]

My initial plan was to use an off the shelf solution, I found a couple available. In my search I also found discussions of building your own. I also discovered that I can replace the tail housing and use a mechanical speedometer cable, there are several signal sensors with pass through for mechanical cable available. It’s good to have options. I’m having fun playing with electronics.

I received the motor from Dayj1, Thanks again! It is a DC motor with optical disk, not a stepper. I found it is used in a Lexmark printer for paper feed. It does run at 2,136rpm for 128mph, but I couldn’t get it to run slower than 160rpm that works out to 9.6mph. I could gear it down, but 1.5:1 would get 6.4mph to 85mph. I like that it has the optical disk, I’ll play with that.

Part of the issue could be my test setup, I’m using a BJT transistor for the motor supply, a FET might be better. The PWM is 12v, I could try different frequencies, or voltages. I may fine tune it at some point, but it works for now.

Someone tore apart a broken vacuum at my work, I found the motor in the scrap bin. It’s from a Dyson cordless, unfortunately the shaft was bent. It kind of worked, I tried to unbend the shaft, and it has a good range, but too fast at 100rpm to 8,000rpm . The shaft is 2.3mm and I have some gears from the hobby shop for 2mm. I set it up with 5.5:1 gears and it’s good for 3mph to 100mph. It is a Johnson motor, but I was unable to cross reference the numbers on their website. I did locate a similar motor by looking at size and supply voltage. There is also the spec. of Stall Torque, that may be related to how slow it will run. I’ll order a few different motors to test. I’m still curious about using a stepper motor, so I’ll keep searching.

The code has some bugs and needs to be cleaned up, but it works for 2Hz and up. I’m ready to connect my speedometer to these motors to see how it effects the motors.

It looks simple enough to build a DRAC. I can use an op-amp as a Schmitt trigger to clean up the VSS signal and produce a nice square wave.

[Second Series]

A 5volt regulator to feed the square wave into the Arduino. The signal can be converted from 40 pulse per driveshaft rev to 128,000ppm, 4,000ppm and 2,000ppm in the software. A potentiometer can be used to adjust for tire size and gear ratio. The 128,000ppm looks like a modified square wave on my truck, A differentiator will produce that signal. It looks like the 4,000ppm is 5v, and the 2,000ppm is 12v. The Arduino operates at 5v so a transistor switch can bump the signal up to 12v.
I’ll need to order some components and breadboard this. I’d like to set up a reluctor ring and sensor on my bench so I don’t need to jack up the truck when I’m trying to figure this out.

[Jason Banks]

Just found this thread, I actually started working on this also. I got sidetracked and haven't come back to it yet, but I'll post some pictures of my setup. DC Motor with Timing belt and gears that drive another shaft that has a rotary encoder. I used a Teensy development board instead of a real arduino. The rotatry encoded which Is not installed in my pictures and I can't seem to find right now is a CUI AMT103-V. The whole thing is designed to thread right on to the back of the speedometer head.

I kept smoking regular transistors, or they were not working right(can't remember) so that other little thing plugged into the board is a better surface mount transistor or mosfet.