[michaelhigh]

All the manuals I've read show using tools to read the backlash amount. I've a 7ish year old dsls micromill, with rotary encoders. Is there a reason I can't split the signals off to a separate reader that can keep track of the absolute position of the machine at all times (dont worry about the electronics, I design digital electronics for a living), its the mechanical / pulses that I'm curious about. Shouldn't that give me the exact count of pulses on a backlash?, as well as allowing me to reset the system by entering the absolute co-ordinates back into
mach3, (after jogging the machine in a standardized direction to account for backlash). I've just found that when something sounds this simple,, I'm usually missing something.

[Algirdas]

Not sure if I got your question. Do you want to find out the amount backlash without mechanical measurements?

[dazp1976]

If you're talking about rotary encoders on the back of steppers, don't they just read bits of information to tell you of position errors?. Don't do anything with it other than fault?.
DRO scales might help you read physical positions but again won't do anything with it.

You sound like you need servo's and the like that have indexing and high speed pulse feedback etc.

[michaelhigh]

What they do is provide pulses which provides speed and direction. They are currently used for feedback into the servo motor controller, and will trigger an e-stop on position misalignment. That said, the question is why can't I split the signal off, (think high impedence oscilloscope probe eavesdropping on the line), and using a small micro, keep track of the absolute position of the screw at all times, its resolution would the distance between two of the pulses, and then display it, on something as simple as seven segment displays. So if for example, I set "HOME" as dead center of travel of all three axis, then jog left about a quarter of the travel availability, then give the machine a command to go back to "HOME". I record the #s on the display for the x-axis. Repeat with jog to the right. The difference in absolute position displayed now for the x-axis from the value recorded earlier should = backlash. I should also be able to do limit switches this way, without needing mechanical switches. Am I missing something? As long as the micro doesn't crash or loose power,, it should always know the absolute position of each axis.

[dazp1976]

Am I missing something? .

Yep.
1. Seems like a total waste of power and it can crash & fail anyway.
2. you can check backlash with a £20 dial indicator.
What will all the scopes etc cost?.

[michaelhigh]

Yep.
1. Seems like a total waste of power and it can crash & fail anyway.
2. you can check backlash with a £20 dial indicator.
What will all the scopes etc cost?.

Probably about $40 in parts (the scope example was to simplify explaining hi-impedence). In addition to backlash, it could also be used for more accurate limit switches, and home positioning. 1 pulse from it should equal 0.000125" worth of travel, and ive yet to see a mechanical "home" switch have that accuracy. Also, it would simplify realigning after an e-stop, for example if the servos trip during a high speed jog, or power failure it would allow you to resume from the exact position where it tripped. Also, in addition, if it works, would be able to provide realtime info on the backlash over time, ie in its simplest form, a led lights up when it gets over "x". Electrically I know its sound,, the point of the post was to ask what I dont know, for example is the mill always 8000 pulses per inch, or can that change over time, (also, the pulses on the encoders match the steps in the motor, as I bought the all in package from taig). That said, it did just occur to me, it wouldn't work for backlash, as I thought about if the locknut wasn't tightened down.

[Algirdas]

Reading your ideas makes me think you are missing something very important.

and using a small micro, keep track of the absolute position of the screw at all times,

What is "micro" you mention as a device?
Sorry being nasty: what is your objective once again?

[Al_The_Man]

, and ive yet to see a mechanical "home" switch have that accuracy.

On high end and professional systems, the home switch itself is not used to register "home" , The typical routine is the machine would rapid to the home switch, where it would stop, then reverse and Slowly move until it saw the one-per-rev encoder Z marker pulse. It would then register this as home into the system.
.

[samco]

Linuxcnc...

[dazp1976]

On high end and professional systems, the home switch itself is not used to register "home" , The typical routine is the machine would rapid to the home switch, where it would stop, then reverse and Slowly move until it saw the one-per-rev encoder Z marker pulse. It would then register this as home into the system.
.

Is that what the indexing is on a servo?.
Just thought I'd ask.

[Jim Dawson]

Probably about $40 in parts (the scope example was to simplify explaining hi-impedence). In addition to backlash, it could also be used for more accurate limit switches, and home positioning. 1 pulse from it should equal 0.000125" worth of travel, and ive yet to see a mechanical "home" switch have that accuracy. Also, it would simplify realigning after an e-stop, for example if the servos trip during a high speed jog, or power failure it would allow you to resume from the exact position where it tripped. Also, in addition, if it works, would be able to provide realtime info on the backlash over time, ie in its simplest form, a led lights up when it gets over "x". Electrically I know its sound,, the point of the post was to ask what I dont know, for example is the mill always 8000 pulses per inch, or can that change over time, (also, the pulses on the encoders match the steps in the motor, as I bought the all in package from taig). That said, it did just occur to me, it wouldn't work for backlash, as I thought about if the locknut wasn't tightened down.

Backlash is the mechanical difference between the leadscrew (ballscrew) and the actual table position when measured in both directions. In other words, there might be enough mechanical looseness in the system where reversing the direction of the leadscrew does not instantly move the table the expected amount. Normally you would measure this with a dial indicator, but it can be done electronically also.

A combination of linear scales and the rotary encoder on the motor shaft would be needed to read this difference. This allows you to compare the actual table position to the leadscrew position.

Without some serious mechanical issues, if 8000 pulses per inch is correct today, then I would expect it to also be 8000 PPI 10 years from now.

There are some high precision mechanical switches available that are supposed to be repeatable to 0.0005mm (0.00002'') https://www.automationdirect.com/adc...eat_accuracy)# But this kind of repeatability would be way overkill in a normal home shop.

[Al_The_Man]

Is that what the indexing is on a servo?.
Just thought I'd ask.

The Z pulse is known as the marker or zero pulse, one per rev.

[Stigoe]

You're talking about using rotary encoders connected to the motors or screws and tapping into those pulses? If so, you wouldn't be able to detect backlash, as those encoders wouldn't detect movements of the table. You would need linear encoders connected to the table to be able to detect the backlash and know the correct position of the table.

[desktoplathes]

All the manuals I've read show using tools to read the backlash amount. I've a 7ish year old dsls micromill, with rotary encoders. Is there a reason I can't split the signals off to a separate reader that can keep track of the absolute position of the machine at all times (dont worry about the electronics, I design digital electronics for a living), its the mechanical / pulses that I'm curious about. Shouldn't that give me the exact count of pulses on a backlash?, as well as allowing me to reset the system by entering the absolute co-ordinates back into
mach3, (after jogging the machine in a standardized direction to account for backlash). I've just found that when something sounds this simple,, I'm usually missing something.

Backlash doesn’t accumulate…. If I’m understanding your intentions, this might be what you are missing. A certain amount exists on each axis (except generally not Z due to weight keeping things in one side of the thread gap). Every time direction is reversed, it occurs once in each direction opposite the previous, canceling itself out. Mach3 has a backlash compensation capability but I’ve never tested it… effectively adding a backlash value to travel at each axis direction change I assume. This is the best you could probably hope for accomplishing no?