I installed a hall effect sensor on the spindle pully.  At 900 RPM, I get a pulse width of approximately 300 uS.  Up to about 900 RPM the DRO reads correctly, but at any speed much above this, the readout reads erratically.  The Index Debounce is set at 2, but that made little or no improvement from when it was set at 10.  I bumped up the kernel speed (to 45khz), and that improved things (previously, the rpm reading maxed out at something over 600).  

The numbers seem to indicate that this should work up to a much higher speed.  Any thoughts as to why it does not?

I suspect that widening the pulse width would improve results, and if I don't hear of some other solution, I will install a circuit to do that.  

SK

I use one and reads very accurately out to 3600 rpm which is my max spindle speed and use a kernel speed of 25000.
Could be a number of things, here are some thoughts
- the sensor must be capable of switching at higher rates ( see your spec )
- Sensor position and magnet used ( i use a rare earth magnet 1/8" diameter )
- Quality of the signal ( see if there is any noise or spikes ) using a o'scope

RICH

Also the width of the tigger point defines the time domain of the signal.  I alway try to use 50/50, 50% time on 50% time off.

Just a thought, (;-) TP

Is this problem solved? i have the same problem and cant seem to fix it.

RPM going erratic at high speed usually means that the Index pulse coming into whatever hardware you are using (even if it is just LPT) is inadequate. You can get this from a slow sensor, too much HW filtering, or too much SW filtering. Yes, been there.

Cheers
Roger