Gallery, Projects and General > Project Logs
Electronic Leadscrew for the New Lathe
RussellT:
Steve
I think I understand the spreadsheet, but if in your spreadsheet I increase the spindle pulses per revolution, then I think that should reduce the error. For example if I was trying to take a step every 1.7 pulses and I increased the pulse count by a factor of 10 then I would need to take a step every 17 pulses - doing away with the error. I think the error will be much smaller than your calculations suggest.
For example for a 36tpi thread there will be 3000 pulses per spindle rev and you need 166.67 steps to advance the leadscrew 1/36". That works out you need to advance the stepper one step every 18 pulses exactly. No error apart from the stepping nature of the advance.
In a worse scenario (27tpi) you need to advance the stepper once for every 13.5 pulses. As long as your software takes care of the half pulses then the maximum error at any point on the tread will be one step or 0.0001666 inches.
Russell
Edited to remove my percentage error calculation which was wrong.
vtsteam:
Oops....almost killed your last post, Russell -- I hit edit insted of reply -- luckily the cache had a copy......phew!
Checking the spreadsheet Russell -- i'll post a corrected version......
RussellT:
Steve
I realised that my percentage error calculation was wrong so I removed it. The difficulty with the error calculation is that if the software keeps track of the accumulated errors then over a long thread the error will be eliminated. However it will correct by adding a step when the error accumulates and it should do that when it has fallen one step behind. That means that the line of the thread will show waviness a maximum of one step from peak to trough or oscillate by a maximum of half a step either side of a mean line.
Where I am having trouble is working out over what length the maximum deviation from the mean line will occur to get a percentage error.
To take the 27tpi example the stepper should step at 14, 27, 41, 54, 68, 81, 95, 108 etc pulses. The error is zero over two steps but the error in the middle is only by mistiming the step by half a pulse.
I need to think about how to calculate the error some more.
Russell
awemawson:
--- Quote from: vtsteam on April 01, 2015, 08:50:50 AM ---I was thinking about something like this:
http://www.ebay.com/itm/Encoder-600P-R-Incremental-Rotary-Encoder-AB-2-phase-6mm-Shaft-5V-24V-coupling-/321160456974
--- End quote ---
Amazing encoder that Steve, according to the description it has an 'output triode' - now valves, that's my vintage of electronics :lol:
vtsteam:
I think I've got it-- please check!
If correct, one thing that looks doable in it is dropping the stepper/leadscrew reduction to 1.5 to 1, and running the 600 line encoder at 5 to 1 reduction. What this does with my particular leadscrew, is match the encoder pulses per revolution with the steps per inch. That's 3000 each.
It looks like that means that the number of pulses per encoder revolution equals the thread pitch. That means no errors percentage. And the reduction on leadscrew is easy to do (small gear ratio) and the encoder can run at spindle motor speed (also easy).
Unless I'm wrong in any (or all of the above) -- which is a distinct possibility......
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version