Hello Guest it is January 29, 2020, 03:18:25 AM

Author Topic: Hall effect home switches - Nonsense and bother  (Read 3745 times)

0 Members and 1 Guest are viewing this topic.

Hall effect home switches - Nonsense and bother
« on: February 25, 2008, 05:33:39 PM »
Well, I'm annoyed again.

My machine has three hall-effect proximity switches for homing. These have inbuilt magnets and react to plain steel vanes on the slides, or the leadscrew nut in the case of Z. They trigger a logic  pin on my second port with an arrangement that uses a 2k2 resistor in series with the sensor, and the whole lot across a 9V supply (these sensors originally had a 9v line). Across the resistor is the input side of a darlington-output optoisolator in series with a common LED to give a combined junction voltage of around 2.2 volts (plus a current limiting resistor). Effectively when the sensor is triggered its resistance rises from around 300 ohms to 13K; the voltage across the 2k2 resistor drops sharply to below 2.2 volts and the opto shuts off, allowing the logic pin on its output side to float high.

All this is very fine and well, but when I reference the axes, zero all, turn off the automatic zeroing and move all the axes -10mm, then re-home them all, they end up in slightly different places to where they started.. not usually much, 1-2 microns usually, with the exception of Y which can be much further out.

A few questions:

 - Do hall proximity sensors work best if the vane passes very close by it, or does a bit of separation help?
 - Can i expect day-to-day drift due to temperature and age affecting the interface circuit?
 - Is it really reasonable to expect a machine to home accurately to EXACTLY the same position 24 hours after it last did it?
 - if so, how?

I can find precision switches giving 1 micron repeatability on RS, but if I'm to have three of them they'll be around £500, and I won't be popular with the bean-countress.

Suggestions? I did consider that of "buy a new one from Haas" as very favourable, but... see above  ;D Thanks in advance.

Offline Hood

*
  •  25,856 25,856
  • Carnoustie, Scotland
    • View Profile
Re: Hall effect home switches - Nonsense and bother
« Reply #1 on: February 25, 2008, 06:00:30 PM »
I use optical switches on my Bridgeport retro, they are housed in Alu boxes and a rod protrudes from either side and is held central by a spring either side. The rod has a flag on it which blocks the LED until the rod comes against the stop. I did tests on these with glass scales accurate to 0.005mm and out of 100 runs I only had 1 (possibly 2, it was a long time ago) where the DRO did not register 0.000mm and the 1 time it read 0.005mm, so its safe to say they are accurate. Oh and another thing was this was on my first retrofit and it was a well worn Bridgeport so the 1 time it was off could well have been due to the saddle pivoting.
 Another option you have if you have encoders is the board from CNC Building blocks, it looks for the index pulse then counts the preset amount of encoder lines (set by dip switches) then tells Mach it is homed. I have this setup on my Lathe and it too is deadly accurate, probably more so than the optical switches.

The opticals are the cheapest option, about £10 for the 3 if I remember correctly but obviously you need to account for your time in making up suitable waterproof housings.
Below is a pic of the one I made for the Z Axis so you get the idea of what my ramblings are about ;)
Hood
Re: Hall effect home switches - Nonsense and bother
« Reply #2 on: February 25, 2008, 07:19:03 PM »
thanks Hood - making waterproof boxes will be no problem. Have a look at RSwww, part number 304-560, a nice optical gate with a built in schmidt triger and logic output. Is this the sort of thing you're using? It''s only £5 and very neat - and would run off my 9V setup.

no encoders, sadly, I'm running three hefty steppers. Encoders would be nice.

Offline Hood

*
  •  25,856 25,856
  • Carnoustie, Scotland
    • View Profile
Re: Hall effect home switches - Nonsense and bother
« Reply #3 on: February 25, 2008, 07:30:18 PM »
The ones I used were 455-0896 also use them for the spindle speed sensors.
Hood
Re: Hall effect home switches - Nonsense and bother
« Reply #4 on: February 25, 2008, 08:00:21 PM »
Grand. Will look into trying one of these out. Thanks :)

Offline jimpinder

*
  •  1,233 1,233
  • Wakefield, West Yorks, UK
    • View Profile
Re: Hall effect home switches - Nonsense and bother
« Reply #5 on: February 26, 2008, 04:23:45 AM »
Reading the initial post, I think you have got very good results if you are accurate to a few microns, using Hall effect proximity switches. Bearing in mind these work on induced magnetism, then I think it is likely that things change on each pass. You might not be able to see, but even a slight movement in the table, or bit of metal about could affect it.

There was an old saying - don't bring a bucket near the ships compass, for exactly that reason.

I think the optical solution would be far more positive
Not me driving the engine - I'm better looking.
Re: Hall effect home switches - Nonsense and bother
« Reply #6 on: February 26, 2008, 04:12:38 PM »
I'm still very interested by the optical solution. I took a scientific approach to the problem today and plotted all the values of the three axes on 20 re-homing moves from a zeroed homing move. X and Z deviated by +/- 3 or 4 microns  while Y was sometimes as much as 90 microns out. I tried winding Y's sensor out a bit until the vane only just missed the front of the sensor, doing the same for Z as it was a bit further out than I liked the look of.
The second test run showed a marked improvement on Y with a slight improvement on Z, at this time I had around +/-3 microns on each axis according to the DRO. Certainly good enough for most work but there's room for improvement.

Ideally if I can get dead-zero rehoming I should have it - I am, after all, worth it. I'll prototype one with some RS clobber, screw it to the bed and wire it into a home switch input to test its repeatability. It sounds like a safer bet than the old hall effect sensors anyway.

The next step is to test the positioning accuracy of the machine and its steppers, drivers and software, and try out the leadscrews for backlash. more as it arrives.

Oddly enough, i asked our workshop manager about this lot and he reckoned that Y was always a bit dicky... Took five minutes to fix. That, and the brake contactor with the fried coil. and, of course, the control system that gave up its ghost, well that took a bit longer...   time aside, it's fun, isn't it? 

Offline jimpinder

*
  •  1,233 1,233
  • Wakefield, West Yorks, UK
    • View Profile
Re: Hall effect home switches - Nonsense and bother
« Reply #7 on: February 29, 2008, 03:52:27 AM »
I am just trying a laser screwed onto the cross slide, shining at a detector at the end of the lathe - I will let you know how accurate this is. I am not too hopeful, the laser seems a bit "wide-beam" but if I mount it, and the detector behind a lemgth of narrow tube, I should be able to get reasonable accuracy - the parts were £10 for the laser and £1.25 for the detector(s) - so not too expensive.

At least this way, there is nothing for the machine to crash into.
Not me driving the engine - I'm better looking.
Re: Hall effect home switches - Nonsense and bother
« Reply #8 on: February 29, 2008, 05:37:07 PM »
The  laser with its well-collimated beam should provide the potential for great  accuracy. I did have an idea for using a laser for homing - if a laser mounted at 90 degrees to (say) X shines upward  onto a black pad to prevent is  going into anyone's eyes. As X approaches it brings with it a highly polished peg of round bar which interrupts the beam. Reflecting off the round bar causes the beam to sweep downwards  rapidly  and  hit a sensor.  It's the  high sweep speed  that should allow very rapid triggering, especially if the sensor is mounted a distance from the peg.