ratiometric hall effect sensors outputs at high RPM

Im using an allegro A1308 linear ratiometric hall sensor to measure the rpm and angular position in a shaft rotating at 20K rpm. My magnet is a 2mm dia 1120 mT magnet placed at around 15mm from the shaft. The TEAG(total effective air gap) of the sensor and the magnet is 1.5 mm. I notice that the output voltage of the sensor falls dramatically as the RPM increases, and by 10000rpm falls to such a level its indistinguishable from the sensors no-flux output. Im trying to figure out why exactly this is -

1. is this because the magnet flux experienced by the sensor at such high speed is actually less,

2. or is this because the sensors output cant rise fast enough?

any thoughts will be welcome thanks harsha

Edit After Sunnysky and Andy s answers below :

3.If I used 2 latching hall effect sensors -the ones that go high when it sees a N pole and stay high till it sees a S pole, with two magnets 180 degrees apart and with opposite poles facing the sensors - would that work? The frequency then should be just double the rps and well within the sensors bandwidth. Thanks

• You have an explanation of the bandwidth and sensitivity problem but no specs on flux and speed to validate your concept. In theory maybe, but it comes down to more engineering specs. or test measurements for you to understand what you are doing and what the sensors are capable and all the variables that influence a reduction in signal. Commented Jun 12, 2019 at 6:44
• Maybe all you need is a more sensitive , nonlinear Hall sensor rather than a ratiometric linear one with a nyquist filter sensing.honeywell.com/hallbook.pdf Commented Jun 12, 2019 at 15:06
• Personally, I would use something else like the motor BEMF or an IR optical reflector or rotary encoder Commented Jun 12, 2019 at 15:14

You have a sensor with a 20kHz BW ( typical) and a flux signal from a 3mm magnet at a 30 mm radius (?) with loss of signal at 10kRPM or 167 Hz.

The duty cycle of the pulse is less than 1.6% = 3mm/(30mm*2π) * 100% .

This small duty cycle reduces the fundamental f signal by -33dB and spreads the bandwidth of the signal up to 1/1.6% or 62 times the fundamental , where there is a null in the spectrum.

That is over 10kHz, so it means yes due to the small width of the magnet you will lose amplitude with RPM from the harmonics exceeding the sensor BW.

There may be some GBW sensitivity improvements that could be made, if you provide more details, but the fundamental design problem is the magnet width at high RPM.

• Hi - have edited the original question in response. Also your right, the magnets at 15mm from the shaft not 30mm Commented Jun 12, 2019 at 5:56

any thoughts will be welcome

10,000 rpm is a shaft rotation frequency of 167 Hz. The sensor starts to "give-up" at 20 kHz so, on the face of it, there should be no problem but, you need to consider that the magnet only has it's flux modulated over a few angular degrees of the shaft.

So, if the signal pulses over (say) 5 degrees of shaft movement, it has an effective signal frequency that is equivalent to 167 Hz * 360 / 5 = 12 kHz and this is getting close to the 3 dB limit. Remember this is just a worked example with me throwing some numbers at the problem.

is this because the magnet flux experienced by the sensor at such high speed is actually less

No, that probably isn't relevant

Of course, without knowing further details about the shaft (is it 30 mm diameter?) and what circuit you are using (and how noisy it is and how noisy your power rails are) it's difficult to be more helpful.

• Hi - have edited the original question in response. Also your right, the magnets at 15mm from the shaft not 30mm Commented Jun 12, 2019 at 5:56