Gallery, Projects and General > Project Logs

Electronic Leadscrew for the New Lathe

<< < (27/32) > >>

rowbare:

--- Quote from: vtsteam on June 29, 2018, 09:39:27 PM ---And that means that simple integer division by the intended TPI will yield the proper pulse rate for the stepper to produce that TPI. Divide encoder pulses by 4, you get  a 4 TPI screw. Divide by 28, you get a 28 TPI screw. etc.


--- End quote ---
So what happens to the left over pulses?  3000/28 = 107 r 4

bob

vtsteam:
Well, rowbare, I know it's a little difficult to visualize at first, but think about it, and I think you'll get it...they just carry, as long as the thread continues. If you were doing only one turn, yes you'd have a remainder at the end of that turn, but this is a continuous process. In other words, the pulses don't have to line up with the end of each turn. It's just two different frequencies superimposed -- the spindle rotations, and the stepper pulses. Does that make sense?

rowbare:
If you look at it as superimposed frequencies, there has to be a harmonic relationship between them.

Given a 32 tpi thread (32 shows a bigger error than 28) and 3000 ppr, if the cutter has not moved exactly 1/32 of an inch after 3000 pulses, the thread pitch is off.

3000/32 = 93.75, so a divide by n will provide 93 pulses to the stepper per revolution. After 32 revolutions, the stepper will have gotten 32 * 93 = 2976 pulses and will have moved 0.992 inches cutting a 32.25 tpi thread. It isn't a huge error but an error just the same. It is certainly more significant in theory than in practice though...

bob

vtsteam:

--- Quote from: rowbare on July 04, 2018, 03:36:08 PM ---
3000/32 = 93.75, so a divide by n will provide 93 pulses to the stepper per revolution. .

--- End quote ---

Nope, it will provide 93 pulses, plus the .75 remainder as a "virtual" timing interval, because the spindle won't have to travel more that 1/4 of its normal inter-pulse distance before sending the next pulse, #94 to the stepper. The encoder has already moved that .75 amount.

See, the encoder doesn't reset once it gets to 93 which is what you're imagining. It's moved past that point just short of the next pulse position. The stepper just sees a continuous regular stream of pulses from the encoder. And those are timed exactly right.

Get it?

The only error is the fineness of resolution of the stepper. But there is no cumulative error from remaindering.

With a perfect leadscrew and a 3000 ppi stepper ratio the tolerance is a flat .00033" at any point along the full travel of the lathe.

beeshed:
How do you do metric? Since a Norton box is so easy the big value from an ELS is mixing imperial and metric. I think this is where a programmable device is an advantage.

Navigation

[0] Message Index

[#] Next page

[*] Previous page