Wheel speed sensor noise and filter level

[Hoolio]

I'm connecting wheel speed sensors to the digital inputs of an M400. I'm using the OE toothed ring for the ABS system; 44 teeth with a diameter of 83 mm so the teeth are fairly small (3 mm high with a spacing of 4 mm). I'm not using the OE sensors because my particular car did not have ABS optioned and the mountings for the sensors are not present.

I thought I would use an industrial hall effect sensor because it is relatively cheap and easy to mount, so I have tried a Cherry GS100502. I'm getting a reliable and consistent speed reading, but there is a lot of noise on the logged speed reading. See the attached image; this was logged at 20 Hz and the units are minutes and seconds on the x-axis and km/h on the y-axis. The filter level in the M400 input setup was set to 2 (i.e. the minimum filtering other than off).

Is it normal to have this much noise in the signal? I would think not. What can I do to reduce it? Obviously I could increase the filter level, but I tried level 5 and this did not reduce the noise much but did significantly reduce the response to a deliberate brake lockup.

[Sean]

I suspect the teeth on the chopper wheel are too fine to work with that sensor. The Cherry data sheet for that sensor recommends 10mm between teeth.

http://www.mouser.com/ds/2/833/Gear-Too ... 075755.pdf

[Hoolio]

Yes, I was aware that the teeth are smaller than recommended for that sensor but I thought that I would try it anyway. Before driving the car I connected the sensor output the the M400 Ref input and used Ref/Sync Capture to look at the waveform while spinning the hub with a drill. This is the equivalent of c. 150 km/h and the waveform *looks* good by eye.

Is it the case with these sensors that if the teeth are too small the waveform will look okay but contain tiny timing variations which give a noisy speed reading?

[David Ferguson]

If you zoom in on the Ref/Sync capture, you will see that there is quite a bit of slope to those edges (possibly 50 or 60 uSec per edge), Have you tried both rising and falling edges to see which is the best? Have you tried adding an additional pull-up (or perhaps pull-down) resistor to speed up the edge?

I'm not sure the Ref input is the same as a digital input. What happens when you connect the drill motor (constant speed) with the sensor to a digital input, and view the Sensors screen (or log the data) -- is the signal a constant speed?

Other thoughts -- temperature -- does the signal look like this all the time, or just when the sensor get's hot? Many hall-effect sensors can be affected by temperature.
-- vibration -- how solid is your sensor mounting? If it is missing teeth occasionally, you might see this kind of noise/fluctuation in the input signal.

Is that logged data straight from the M400, or over the CAN bus to a dash/logger?

[Hoolio]

The logged data I showed was straight from the M400 while driving the car, with the sensor connected to a Digital input. The same sort of noise is evident looking at the speed trace in real time in a chart recorder in ECU Manager. The amplitude of the noise is less at lower speeds.

I have just run some more tests, turning the hub with a drill and the engine not running. The sensor was connected to the Digital input and I observed the speed trace in a chart recorder in ECU Manager. In all cases the noise was much the same. Everything was at ambient temperature, so that eliminates a hot sensor effect. The mounting I made for the sensor is reasonably stiff, so I don't expect vibration to be the cause. The sensor is powered from battery voltage and the Digital input has got a 4k7 pull-up to 5V. The data sheet for the sensor recommends a 1k0 pull-up to 5V, so I tried adding a pull-up of that value to 5V. I also tried it with the 1k0 pull-up to battery voltage instead. I tried both rising and falling trigger edges. I then powered the sensor from 5V instead of battery voltage, both with and without the additional 1k0 pull-up to 5V.

Any more suggestions?

[David Ferguson]

Sounds like you did all the tests. Have you tried playing with the sensor gap (or alignment) to see if that can improve things?

Other than that, I would be modifying the trigger wheel -- perhaps machine off half the teeth, or make one that fits the specs the sensor data sheet suggests.

Other than that I'm at a loss.

[Scott@FP]

Try a modern near zero speed capable HE sensor that is made for small teeth like this one-

http://www.mouser.com/ds/2/187/SNDH-T%2 ... 740268.pdf

If you do end up using this one it takes a pretty strong current (500-800 ohm pullup to 5v or 8v) to get a nice square wave output but it works well for small teeth at speed unlike the more popular older units.