Hall Effect Sensor (latch) measuring RPM problem

[waswart]

Hi,

I am having problems with a Melexis US1881 Hall Effect sensor (latch type) which is used to measure rpm. It is used with an Arduino Uno R3. I have attached some images showing the setup and extracts of the datasheet.

From a cold start, it works correctly initially but then starts degrading (not latching correctly) after typically twenty minutes and this gets worse as time goes on until it almost doesn't latch at all (except at very low rpm). When started again soon after it would degrade sooner (say after 8 minutes) then again after 5 minutes etc. It would still latch correctly at very low rpm, even when in degraded state.

In typically use the rpm would be about 60Hz and the sensor is rated for 10KHz.

From a layman's point of view, it seems it is a problem of a component degrading over time (or being saturated over time) to the point where it no longer functions correctly.

The setup is as per the images.


Things I have tried:
1. Magnets: I have replaced the magnets (rare-earth magnets which are quite strong) as the fly-wheel gets quite hot (it can get up to 70-80 degrees Celsius) as it could be possible that heat affects the magnetic strength. Have also mounted the magnets on wood to reduce heating up of magnets which has not solved the issue.
2. Replaced the Hall Effect sensor several times in case it was damaged.

Possibilities:
1. Magnetic environment: there are other magnets in the housing behind the flywheel (it is a magneto trainer) mounted on the inside of the housing with the Hall Effect sensor magnets mounted on the outside of the flywheel.
2. I am not too clear on the internal workings of the sensor and if the problem is related to output voltage saturation or the magnetic specifications for the operating, release and hysteris. Perhaps the magnets are too strong or too close, and perhaps could have a cummulative effect?

Links to datasheets and application notes:
https://www.melexis.com/Hall-Effect-Sensor-ICs/Hall-Effect-Latches/US1881-140.aspx

I am very much a hobbyist and any comments/suggestions as how to proceed would be appreciated.
Thank you

 

[chuckey]

I would suspect the magnets. the hall device is good for 85 degs C (at least). Have you tried to effect of moving the magnets further away? just to see how much "field" you have in hand?
Frank

 

[jiripolivka]

Many magnets have some temperature dependence which could affect the system function. I would test a set of rotating magnets with your sensor while blowing hot air on them by a heat gun. You can monitor their temperature with an IR non-contact thermometer. This way you can find what affects your system.

 

[FvM]

I'm missing a clear description of the "not-latching" fault. Where do I see it in your waveforms respectively how does the faulty waveform look like?

I don't understand your schematic. The sensor has an open drain output. Where's the pull-up resistor that assures a correct high level?

 

[waswart]

I would suspect the magnets. the hall device is good for 85 degs C (at least). Have you tried to effect of moving the magnets further away? just to see how much "field" you have in hand?
Frank

Thanks, I was using 62 ND magnets which affected the sensor from about 15mm with an airgap of 4 to 5mm.
I have swapped those out for 32 ND magnets which affect the sensor from about 8mm now with an airgap of 6 to 7mm. This seems to have solved the problem, I have tested about three hours now and have not been able to reproduce the problem.

Thanks very much to all for the replies.

 

[waswart]

I'm missing a clear description of the "not-latching" fault. Where do I see it in your waveforms respectively how does the faulty waveform look like?

I don't understand your schematic. The sensor has an open drain output. Where's the pull-up resistor that assures a correct high level?

Thanks for the reply, unfortunately I don't have a scope, the wave forms are just readings on an analogue pin, but the second waveform shows the irregular latching.

I do have a 10K pull up resistor as well as 100nF and 4.7nF capacitors as per the little schematic (a bit obscured on the setup.jpg bottom left corner).

 

[D.A.(Tony)Stewart]

1) For optimal detection the gap should be less than the sensor width, which means any runout (wobble) must be smaller to avoid collision.

2) heat wont bother the sensor, but will reduce magnet strength but smaller gap will improve sensitivity. All you need is 10mT pp to exceed hysteresis which is slightly more than a fridge magnet

3) circuit resonance will degrade results. use short paths, lower impedance, to make resonance less than hysteresis

4) although sensor may 30pps the small size of magnet or circumference/magnet ratio increases the bandwidth required by this ratio or est. >100x = 3kHz which is approaching limit of sensor so 3) and 1) becomes important.