Thread: Help designing gps speedo

Alright, short background to this project:

Original speedometer gear melted, no energy to diagnose or fix this ancient system. Tired of ECU not recieving speed signal, making my boost-by-gear stuck in first gear...

Time to crack the speedometer circuit, to allow any Nissan clusters to work with any gearboxes,wheel sizes, external factors, etc.: I want to build a GPS->Nissan cluster circuit.

Here is how I think the OEM works, correct me if I'm wrong:


Help needed: What type of input is sent from OEM speed sensors (red part)? Is it an alternating current, or pulsetrain? FSM talks about a reed switch?

Arduino nano+gps ordered, so this should be a <15$ project. Pretty nifty if it works, since then you could use for instance digidashes without worrying about matching speed sensors+clusters at all.

reed switch would be pulsetrain. there is no ac conversion anywhere on our vehicles afik.

already running a gps based speedo on my head unit's display via tunerview.

That's what I thought aswell..

Tomorrow, I'll be picking up a spare cluster from a friend, and I'll start generating waveforms into it and see if I can get the needle to move.

I imagine its sort of like pwm. either a longer pulse, or more rapid pulses should get it to move, and the way the speed sensor works, I would try modulating the frequency of pulses first as a longer pulse on time wouldn't make sense to me with a mechanical based sensor... so not really like pulse width modulation at all really.... but similar concept.

I believe that the speed sensor generates more pulses per second the faster it rotates, not pulse width modulation. For the mechanical speed sensor, this was converted into pulses at the dash, the electronic one does this in the sensor itself.

I'd be very concerned that the GPS wouldn't report an accurate speed regularly. I think it's a cool idea, but I'm not sure that in practice it would work well.

I have started looking into this.

I picked up a GA16DE 1994 cluster, should be easier to test with than a mechanical speedo cluster.


After doing some googling, I found out that the Nissan clusters use an odd way to signal the speed. Most do a digital pulse train, but our clusters need an AC sine wave. People have had succes with either +-5V or +-2.5V waves.

The manual talks about 1VAC:


So now I need to get the arduino uno to generate a suitable signal...

I believe it is 2 pulses per revolution of the sensor.

Correct.

Found a guy that made a similar circuit for an S13 cluster. Dakota speed converter->sine wave



Very heplful, he decided on -2.5 -> 2.5V, although it did work with -5->5V too.

I'll update this when I have the needle moving. Might need an OP-amp circuit...

Finally I can make some progress here

My picoscope showed up in the mail today. I went for a Hantek 6022BE, cost me only 50€, free shipping. The software that shipped with it is horrible (triggering not working), luckily some individuals have developed an open source version,

Here's a screencap. Notice the ripple, its around 10mV constant.


Pretty bad, but still damn good value for your money. Accuracy is well within the need for this project.

the mechanical speedos move the cluster needle by way of torque from an electric field generated by the ac signal, the electric ones take the ac signal and convert it somehow. I measured it once on my NX with digi dash and electric speedo sensor and it was an ac signal, but it varied with speed (larger a.c. voltage at higher speeds)