My experience with the MachTach, Tachulator and most the cheaper ones on eBay is that they either work, or do not work. The Tach Hall sensors you buy on eBay are rated to something like 9999 RPM with a single magnet. If you take the barrel off you will find a small IC hall sensor at the tip, a resistor and a LED. Spacing of magnets or on optical sensor reflective material can cause errors. The MachTach and Tachulator both can accommodate multiple marks/pulses per revolution, this only effects the lowest RPM and not the accuracy.
On the MachTach when I use shielded wire and grounded one end, it picked up all the VFD noise. I have not had this issue with the cheaper tachs. Both the MachTach and Tachulator you can set the pulse sensitivity which looks at on/off voltages, this is sometimes an issue when switching types of sensors. In the case of noise contamination/sensor sensitivity, typically the numbers jump all over the place. I have done a few custom tach/hall sensor installs, and I do occasionally have issue with decreased RPM reading at the higher speeds (typically more erratic readings). I find that the sensor sensitivity goes down when the sensor is embedded in metal and some hall sensors are more axis directional.
Unlikely to be the belt(s), you would see a drop in RPM when turning if they were slipping. If you are not the original machine owner it is possible that the motor pulley was changed. Many people have used the linked belts on the G4003 series to diminished the vibration issues with the surface finish, I have mixed feelings on them vs. getting a better belts. Single or double belt should make no difference if not loaded.
) about adjusting/calibrating the Hall sensor via the pot on the Machtach board. I didn't do that but wonder if it'll make that much a difference. I'm more worried about the Hall sensor I got from Amazon not being fast enough for the higher speeds to read correctly. Gotta look that up later.