Connecting encoders

I need to connect encoders to moving slides. I wanted to do this with a 5 to 1 increase in encoder revolutions so that .0001 resolution can
be attained with a 500 line encoder using quadrature counting. But that would mean that the encoder shaft would need to be about .063 dia. Much more realistic is an encoder shaft that is about .318 dia. Then, 1 rev. would equal 1 inch and a direct drive can be used from the slide with a 2500 line encoder. Still, the encoder needs to be driven by some kind of belt or wire. I'm thinking that timing belts would work because they are used in CNC machines and seem to be backlash free. Granted, they usually drive a screw with a pitch of about .200 or 5mm. This means that .0004 belt backlash translates to .0001 table backlash. Right? Looking at a WM Berg catalog some time ago I saw belts made from cable with teeth molded to the cable. I wonder if these would be better. Any thoughts appreciated. Cheers, Eric
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Look at how the head is moved by band positioners in floppy drives. If physical sizes are not too far off, tearing apart an old floppy drive is a very good start.

And at the same time remember that floppy drives (with much much coarser stepper motors) have roughly 0.001 inch spacing (96 or 135 or 192 TPI). They have to hold this positioning to a fraction of the track spacing so for 0.0001 inch resolution the principle ought to be OK if not the implementation.
Tim.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
wrote:

Thanks Tim. That indeed might work. And I have seen the band positioners in 'em. Eric
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

But note that I misplaced a decimal place at least once in my math! Of course if I did it twice and the errors cancel out, then it's not so bad :-).
Tim.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Wouldn't it be easier and better to just get new encoder disks? Anyway, it's not at all clear what you're trying to. It sounds like you want more resolution out of the encoder on a screw.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Mon, 26 Sep 2005 16:53:41 GMT, "Mike Young"

That's what I said above. I ordered the encoder with 2500 lines. But that doesn't address the whole problem. I will not be beasuring screw rotations. Because there is backlash in the screw that cannot be removed easily. So, the thing to do is to actually measure the linear travel. This is then translated into rotary motion by a belt or similar wrapped around the encoder shaft. What I want to avoid is slipping belts, backlash in cogged belts, and errors from temperature. I hope that explains better what I want. Cheers, Eric
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
snipped-for-privacy@whidbey.com says...

Why not the obvious solution - a linear encoder? I think you'll have a very tough time getting your encoder to repeat anywhere near to .0001 with any of the suggestions to date, unless the stroke is very short, and I can't think of anything better. How long is the linear travel?
Ned Simmons
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
wrote:

Greetings Ned, The travel in one axis is only about 1.5 inches. But the other is just over 8 inches. Also, I have not found any linear encoders that have .0001 resolution that I can afford. The rotary encoders from US Digital, which have a Hewlett Packard HEDS reader at the heart, are indeed accurate enough. I had to replace an encoder on a machine tool and it moves .0001 at a time and repeats within the .0001. I don't see why, if the proper cable, metal strip, or belt is used, won't be accurate enough. If the drum is within diametral tolerance is OK, and the belt doesn't slip, and the encoder is accurate enough, why won't the system be accurate enough? If I sound like a wise ass please know I don't mean to. If it won't work, I want to know why. I know that temperature will affect the accuracy. So the measurement would only be relative, not absolute. So when moving the axis I expect at the time of movement the display will show .0001 when the axis moves .0001. Thanks, Eric R Snow
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Eric R Snow writes:

My hunch is you're asking for more precision than the mechanism itself can maintain. Timing belts are surprisingly free of backlash, but 0.0001 is not reasonable. The stretch alone would exceed that. Same for metal bands. Assuming you have inches or more of motion involved here.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Mon, 26 Sep 2005 12:25:23 -0500, Richard J Kinch

Actually...1 tenth is reasonable. Many "C axis"cnc lathes use a timing belt.
Gunner
"Pax Americana is a philosophy. Hardly an empire. Making sure other people play nice and dont kill each other (and us) off in job lots is hardly empire building, particularly when you give them self determination under "play nice" rules.
Think of it as having your older brother knock the shit out of you for torturing the cat." Gunner
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Gunner writes:

No way. The tool position may be 0.0001", but the belt is before the gearing and is not that precise.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Tue, 27 Sep 2005 11:54:16 -0500, Richard J Kinch

Way dude. Not all..but many
Gunner
"Pax Americana is a philosophy. Hardly an empire. Making sure other people play nice and dont kill each other (and us) off in job lots is hardly empire building, particularly when you give them self determination under "play nice" rules.
Think of it as having your older brother knock the shit out of you for torturing the cat." Gunner
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Eric R Snow wrote:

Look at the tooth to tooth accuracy spec on the belts. It accumulates over distances to a sizable number in not very much distance. I think that ten thousandths of an inch accuracy would not be doable with timing belts. The teethmarks in my hindquarters from this problem have long since healed, but I still remember it pretty well.
Bob
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
ya know, if you really want superb accuracy, why not use a laser interferrometer - use an IR laser (cheap) - all you need is a couple of mirrors and a cheap CCD one line sensor

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Mon, 26 Sep 2005 23:14:02 -0700, "william_b_noble"
Can I use you method to measure long travels? If so, tell us more. Eric

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
On Tue, 27 Sep 2005 02:45:26 GMT, MetalHead

Bob, My CNC lathes and mill all use timing belts to position within .0001. And these belts have been in the machines a long time. And the encoders are on the motors, not the screws. Of course, the positioning accuracy is not accurate to .0001. Just the repeatability. So the machine is programmed to cut 1.0000, but cuts 1.0003, so an offset of -.0003 is used to make the cut on size. And this is what I'm looking to do. Get close, measure, then correct. So the system has to be really accurate only for a short distance. I should have put this in my original post. Eric Eric
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.