Need to find a compatible magnetic speed sensor for speedometer

Thread Starter

MB107

Joined Jul 24, 2016
345
It appears that what you have is a fairly standard magnetic impulse sensor, generating an AC signal. As already mentioned I wouldn't be terribly worried about using a 1KΩ sensor, you can simply add an 820Ω resistor in series if absolutely necessary. The important consideration is the number of pulses per mile, and the signal level. You will need to determine what is the lowest signal voltage that the speedometer will work at and the peak voltage at max road speed.
A magnetic impulse crankshaft sensor or abs sensor will do the job, or if you are fussy about the voltage level, a 5V Hall type sensor (often used for camshaft sensors) should work with a voltage level spot in the middle of your desired range. It will have the added complication of requiring a 5V supply , which can be easily obtained, and it will have a square wave output. I would be surprised if the speedometer is fussy whether it is a square wave or AC pulse.
Before you start, you will need to accurately determine the number of pulses per mile. By far the best option is to use a signal generator to drive the speedometer and graph frequency V's speedometer reading. If you don't have access to one you can easily make one with an arduino.
. Your Tyres are 205/65-15 which turn 792 times per mile. Your top speed is 140MPH = so your tyre turns at max speed are 30.8 times per second.
Going by the max figure you quoted in your original post of 400HZ, you will need to generate 13 pulses per wheel rotation.(30.8x 13 = 400.4) which is accurate to 0.1%
This is very easily achieved by making a new split collar in two halves,the same diameter and approx weight as the existing one on the OE driveshaft, with 13 evenly spaced teeth attached to it. It is then a simple matter of mounting the sensor on a bracket so that it is pulsed as each tooth passes.
Thanks for the reply.

Most of that work is already figured by the manufacture. Mercedes already uses a 4 tooth wheel on the transmission output and so will I. Consequently this produces ~ 13 pulses per wheel rev depending on the differential ratio. My toothed wheel design is shown in some of the first few posts. Speeds of operation have been calculated to be between 30 and 1000 in/sec at the diameter wheel proposed (5MPH to 180MPH). Also running between 10Hz to 500Hz. Challenge seems to be the low surface speeds, the point where the speedometer just starts to sense motion. At the other end of the spectrum is at high speeds not producing too much voltage that I fry something.

In this car the signal does not go through an ECE it goes directly to the speedometer which I assume must have some kind of frequency to dc voltage converter built into it.

Since these sensors are very linear I believe that matching the threshold voltage at a low driving speed is the key to getting this right. For example if the stock vehicle produces 5V pk-pk at 10MPH then I need to find a sensor and gap combination that will also produce 5V pk-pk at 10MPH.
 

Thread Starter

MB107

Joined Jul 24, 2016
345
Well I believe I am there. Monday I will be placing an order for a model 0010M-52954 sensor from Spectec. Unfortunately because it is custom made I had to by 3 for a total of $550.00. Not to bad considering retail over the shelf sensors are about $150 in the 1.8K Ohm range. Unfortunately finding one off the shelf with 1.8K Ohm and under 1/2" diameter was not happening. It will also take about 1 month to deliver.
 

Attachments

Thread Starter

MB107

Joined Jul 24, 2016
345
OK so it all works pretty good except the speedometer needle has a slight bounce to it. Not bad but not like Mercedes Benz would make it. I believe I have found the problem in that the tone wheel is difficult to get concentric to the shaft. There is nothing to pilot on so it can be off by the tolerance of the bolt holes.

So was wondering if there might be a way to clip the signal if it exceeds 10V. That would make all the peak voltages the same for the usable part of the operating range.
 

Attachments

Alec_t

Joined Sep 17, 2013
14,280
A 10V Zener diode would clip the sensor signal. The diode power rating would depend on the value of any series resistor between sensor and diode, and the value of the input impedance of whatever reads the signal.
 

Thread Starter

MB107

Joined Jul 24, 2016
345
Thanks for the reply. So based on that I'm coming up with a proposed circuit that looks like the attached picture. And with that I should be able to get the desired output. The only thin I can do is measure the resistance in ohms of the circuit the system feeds, which I believe I already did and it was 1K ohm. I can check again, based on that I getting a 10m watt Zerer diode. I have only found 1 10V, 10m watt zener diode on Digikey P/N CDLL5240B-ND. Unfortunately the minimum order is 329 pieces at $1000. I'm sure I can do better but does this all look right. Also which
way does the zener diode go?

Clipping Circute.jpg output file on Samson.jpg
 

Alec_t

Joined Sep 17, 2013
14,280
The zener diode would (a) need to connect in parallel with the F-to-V circuit, not in series as shown above, and (b) be normally reverse biased. Here's a pessimistic simulated setup with and without a zener diode present:
SpeedoInputClamp.PNG
This setup cuts off negative peaks, but I doubt that will worry the speedo circuit. One concern with this simple circuit is whether the action of the diode in effectively shorting the sensor at high revs could cause undue heating of the sensor coil.
The green trace shows the power dissipation in the zener diode. Average power is 165mW with peaks of 1.5W. Your mileage may vary. I'd be looking for a 2W diode (plenty listed in 1-off quantities at Digikey etc) and expect it to get moderately warm at high revs. If your sensor output under load is less than in the sim then a lower rated diode will fit the bill and sensor over-heating is less of a concern.
 
Last edited:

Thread Starter

MB107

Joined Jul 24, 2016
345
Many thanks Alec_t.

Looks like I may have a solution if I need one but I'm not sure if I will need one. Yesterday driving home from work I noticed there was no pulsing of the speedometer. Thought maybe it broke in and fixed itself. Then this morning pulsing again. On the way home there was pulsing but it cleared up after a short burst of speed to 80MPH. Morning temperatures have been ~ 50's here with the afternoons in the 80's. I'm thinking this may be a mechanical issue with a speedometer that has been sitting dormant for the last 10 years. I will be taking a trip to Athens GA in two weeks, about 500 miles round trip, so that should be a good workout for it. So until then I wont be doing anything with it.
 

Thread Starter

MB107

Joined Jul 24, 2016
345
The zener diode would (a) need to connect in parallel with the F-to-V circuit, not in series as shown above, and (b) be normally reverse biased. Here's a pessimistic simulated setup with and without a zener diode present:
View attachment 175530
This setup cuts off negative peaks, but I doubt that will worry the speedo circuit. One concern with this simple circuit is whether the action of the diode in effectively shorting the sensor at high revs could cause undue heating of the sensor coil.
The green trace shows the power dissipation in the zener diode. Average power is 165mW with peaks of 1.5W. Your mileage may vary. I'd be looking for a 2W diode (plenty listed in 1-off quantities at Digikey etc) and expect it to get moderately warm at high revs. If your sensor output under load is less than in the sim then a lower rated diode will fit the bill and sensor over-heating is less of a concern.
I do have one question about this circuit. The input signal shown is a constant amplitude of +/- 300 V the output is a constant amplitude of +10V. If the input amplitude were to periodically change as mine does, would the output remain the same?

Also I'm not sure what you consider high frequency but the most this car could possibly see would be about 350 Hz. at 180MPH during a Bonneville speed trial lasting about 20 seconds plus about 60 sec to get there.
 

Thread Starter

MB107

Joined Jul 24, 2016
345
Well I just got back from Atlanta 900 mile round trip. Still not out of the woods yet although the issues seam to be limited to the range of 60 to 80 mph now. Its looking more and more like mechanical issues in the speedometer head. Wort comes to worst I can swap speedometers with another car. But right now I'm just thinking of seeing how it reacts to a pure sine wave input in the same frequency range.
 

Thread Starter

MB107

Joined Jul 24, 2016
345
OK so I am off to this solution. A pure sine input makes a very smooth speedometer. S I will order the diodes. One thing I noted was that there is a 1.5W diode and a 3W diode. Both were under 50 cents. A 2W that was $13.00 +. I'm going to try several different voltages Should I halve the watt rating if I double the voltage from 10 to 20.
 

Alec_t

Joined Sep 17, 2013
14,280
Don't pay anything like $13 !
My previous sim was deliberately pessimistic.
Assuming your sensor output increases linearly with speed and gives 10V pp at 10mph, then at 180mph the output would be 180V pp. Plugging that value into my sim gives, for a 10V zener, a peak dissipation of ~ 340mW and an average of 34mW. For a 22V zener the peak is ~360mW and the average is 26mW. Those figures are for a sustained 180mph, so in practice the average zener dissipation would be a good bit lower. On that (possibly over-optimistic) basis, a 500mW-rated zener in the 10V-22V range should be adequate for the job, although it wouldn't hurt to go for 1W if the price difference is negligible, since we don't have an actual figure for the sensor output at high speeds.
 
Last edited:

Thread Starter

MB107

Joined Jul 24, 2016
345
Thanks

I just ordered a set of zener diodes 12, 20 and 30 volts in both 1W and 2W. I got 2 of each size, 12 pieces total for under $5.00 + about $8.00 shipping. I'm also going to try a few techniques at centering the tone wheel. I also found that the speedometer will respond to a sine wave way down at 0.4V. Not sure if I could go that low with a pulse but I may have some wiggle room to increase the gap and reduce overall voltage especially if I can center the tone wheel.
 

Thread Starter

MB107

Joined Jul 24, 2016
345
OK I just received the diodes tried a 12V 1W diode in the lathe with the tone wheel offset to simulate the problem and the results seem positive. The sensor does not appear to even get warm, however I really cant get the lathe to simulate over about 50 MPH. The only way to really tell will be to get it in the car. Stay tuned.

Thank You Alex

Without Diode.jpg With Diode.jpg
 
Top