Accurate homing switch and sensor for stepper

[snoopy27]

Stepper motors without the added complexity of encoders still need axis homing that is accurate and repeatable. I am wondering how accurate a homing switch without encoders could be. I am talking about limiting the cost for expensive mechanical switches and proximity sensors to something reasonable for small HF, Taig, and Sherline type machines.

I would like to design or develop something that is accurate and repeatable to .00005" and with a resolution of .0001" or greater.

I have some thoughts, but wonder what others might suggest.

[Karl_T]

I used a slot sensor on my hardinge CHNC for the spindle index mark. Just a few bucks from digikey. You'd want it on the ballscrew shaft. The larger the disk, the more accuracy you have. Do some math to find out what you'd need.

[snoopy27]

Originally Posted by Karl_T I used a slot sensor on my hardinge CHNC for the spindle index mark. Just a few bucks from digikey. You'd want it on the ballscrew shaft. The larger the disk, the more accuracy you have. Do some math to find out what you'd need.

Thanks for the thought.

Let me explain what I am seeing as the problem with homing an axis without a precision switch and encoder index pulse:

1. Mechanical switches that are reasonable in price have low repeatability and can only be used to slow an axis to start looking for an index or marker signal if you want your home position to be precise.

2. I would like to design a pulse or marker signal with a resolution of .0001" or greater not to induce an initial positioning error larger than the pulse or marker signal without an encoder for stepper axis homing.

3. To be very accurate once the home mechanical switch is made and slows down an axis it should look for a marker or index pulse from an encoder, then it must zero the axis precisely on the position of the marker or pulse before the homing switch and marker signal or index pulse transitions from a high or low state.

The homing of axis and the repeatability of homing accuracy is important if you were to lose power, or have a system crash or restart or recover from a cutting file and not have to manually find a starting posiotion for your cutter.

A precise homing setup on an axis can also alert you when there are certain mechanical issues that are present on an axis, like backlash or loose gibs.

The homing resolution on a stepper axis can never be greater than the resolution of the stepper itself, and may have larger initial errors even when micro-stepping because of the nature of all stepper motors.

I think that putting encoders on steppers has limited value, and with the cost of servo motors and amplifiers being very reasonable, if you have to add this cost you might as well go to a servo system.

Having said that I feel there are advantages with steppers on small machines and having a precise homing system on an axis might be a benefit.

Again, I have some thoughts I am willing to share, but would be interested in first hearng what others have done, and what suggestions they might have.

[LeeWay]

This is kinda why I just use the soft limits for now. I rapid around 300 IPM on both my mill and the router. Since zero position on the mill is usually center table, the soft limits work fine there. It also has hard stops. The only switch other than Estop on the mill is the upper limit on Z. Standar push button switch, but when I am rapiding up, it almost over runs that switch. It is moving so fast. I have to pay attention. Momentum is the issue.

This is even more true on the router. Probably have near 150 pound gantry. When X is rapiding @ 300 IMP and I hit a limit, like a car, it takes more time to stop than instant halting. It would run way over a switch. I have been thinking about a double switch setup myself. One for the slow zone. Typical cheap switch like you mentioned. Then a more accurate home switch that it would search for in slow speed. Not as worried about Z on this one, but X and Y are what takes awhile to zero in at home.
I have a hole drilled at 1" in both X and Y on all my jigs. Right now though, it is still by eye.
This is fine if there isn't any problems and I leave the same jig on for a long period of time, but not near as accurate as the double nut ball screws. Trial and error then gets it dead on, but that costs a part or two sometimes.

I was considering using an index pulse card on each axis like these.

Not real sure what the best way to accomplish this would be. Possibly having a long strip with a hole in it mounted to the axis. This metal strip would be long enough to be in this index pickup when the first switch is hit. Then it looks for the hole.
Only two card s would be needed I think. Home X and Y.

Thoughts?

[snoopy27]

Originally Posted by Karl_T I used a slot sensor on my hardinge CHNC for the spindle index mark. Just a few bucks from digikey. You'd want it on the ballscrew shaft. The larger the disk, the more accuracy you have. Do some math to find out what you'd need.

Originally Posted by LeeWay This is kinda why I just use the soft limits for now. I rapid around 300 IPM on both my mill and the router. Since zero position on the mill is usually center table, the soft limits work fine there. It also has hard stops. The only switch other than Estop on the mill is the upper limit on Z. Standar push button switch, but when I am rapiding up, it almost over runs that switch. It is moving so fast. I have to pay attention. Momentum is the issue. This is even more true on the router. Probably have near 150 pound gantry. When X is rapiding @ 300 IMP and I hit a limit, like a car, it takes more time to stop than instant halting. It would run way over a switch. I have been thinking about a double switch setup myself. One for the slow zone. Typical cheap switch like you mentioned. Then a more accurate home switch that it would search for in slow speed. Not as worried about Z on this one, but X and Y are what takes awhile to zero in at home. I have a hole drilled at 1" in both X and Y on all my jigs. Right now though, it is still by eye. This is fine if there isn't any problems and I leave the same jig on for a long period of time, but not near as accurate as the double nut ball screws. Trial and error then gets it dead on, but that costs a part or two sometimes. I was considering using an index pulse card on each axis like these. Not real sure what the best way to accomplish this would be. Possibly having a long strip with a hole in it mounted to the axis. This metal strip would be long enough to be in this index pickup when the first switch is hit. Then it looks for the hole. Only two card s would be needed I think. Home X and Y. Thoughts?

Thanks for the Link Leeway,

I don't know what software you use, but for your router if cutting only in wood, a double switch setup for homing would help with your overshooting problem. If you only need to home your machine once during a cutting operation or part then for you the problems you are talking about would be solved.

I looked at the link for this index pulse detecting board but couldn't obtain enough information from the site. It might solve half the problem, although from the picture of the photo transistor that creates the index pulse I am not sure that it would produce the accuracy needed for a precise repeatable homing of an axis that I would like to acheive.

Has an additional thought I would like the marker for the homing signal to be non-rotary induced, and not subject to gearing , belt drive or pulley tolerances and issues. That is why I am entertaining the idea of an integrated solution that can be mounted to detect home along a machined surface of an axis.

Your idea isn't bad at all, but again a small hole if not protected can easily be contaminated.
If nobody has any more thoughts, I will try and get some drawings and parts list together for what might be used as a universal precision homing system for small machines with mounting brakets, sensor and detector.

[ger21]

There was a company called CNC Building Blocks that was making boards for homing to an encoder pulse for use with Mach3. Ed Gilbert was the guy's name. But he wasn't selling enough to pay for his website, so he shut it down. There's still a Yahoo group (filled with spam) that has the manuals in the files section.
http://tech.groups.yahoo.com/group/CNC_Building_Blocks/
I emailed Ed a few months ago and he still had boards available if people were interested.