Suggestions for a machine

I need to make a test fixture, that's going to look a whole lot like either part of an ink-jet printer or a really lightly loaded CNC axis.

Basically, I'm making a measuring instrument for a client that -- from the user's perspective -- involves putting a magnet on a board and getting a length measurement. The test fixture is to verify the design and later, to calibrate units in the manufacturing process.

For testing and calibration, I'd like to make a fixture that moves a carriage up and down, horizontally, over about a meter's stroke, and drops or lifts the magnet. The magnet is 10mm x 6mm; that, an RC servo (or whatever) and a carriage to hold it all is the extent of the "load" the machine needs to carry.

It needs to be accurate to about +/- 0.1mm. Repeatability is less important -- if I could reliably stop the carriage at as much as +/- 0.5mm that would be OK, but more accuracy is always better.

I'm looking for suggestions for how to do this using parts that are available in quantities of 1ea -- I know how to do it if I was going to be making 100 a year for ten years, but not for just one or two.

I started with a vague plan of using bits and bobs from the hobby CNC market, but those are all pretty heavy-duty for what I want to do. I had the notion of trying to get a 1 meter long, 6mm lead screw, but it looks like long precision lead screws are hard to find. Some sort of motor driving a long toothed belt attached to the carriage would probably work, if I could measure the position of the carriage along the stroke.

So -- comments and suggestions are welcome, particularly if they come with part numbers and company names from whom to order.

TIA.

Reply to
Tim Wescott
Loading thread data ...

String pot?

formatting link

-jsw

Reply to
Jim Wilkins

That's on my short list, although things that can just plug into a PC, or do so with little fuss, are desirable.

Reply to
Tim Wescott

+/- .1 millimeter? right about .004 inches? I believe that's going to be the defining factor, and probably mean the CNC stuff is what you will have to use.
Reply to
Richard

Before I got the four serial-port multimeters running simultaneously I was looking for a just-plug-into-the-PC data acquisition system.

formatting link

Radio Shack's 4-port passive / powered USB3 hub and this USB2 power meter have been useful:

formatting link

A USB3 bus-powered external hard drive operates through the meter, at USB2 speed.

Reply to
Jim Wilkins

For a meter long travel you are going to need to use some form spring loaded anti-backlash nut AND map the screw over its entire length of travel. You might get good repeatability with the screw unmapped, but you will not get good accuracy over its entire range.

You "might" be able to get away with a smashed on /cut in place acetal nut and an acme screw for backlash for a short useable life, but you still need to map the screw.

1/4 16 acme screws are available cheaply enough.

Here is a 6' precision acme screw from MacMaster for $50.

formatting link

I have good luck using these with Dumpster CNC spring loaded antibacklash nuts, but they wear out and loosen up under any significant load. Usually the mounting flanges wear first, so they can be tightened up a couple times before they need to be replaced.

You will need to make arbors on the ends (atleast one end) for bearings, and make bearing blocks, but that is doable. Even a pair modestly inexpensive v-groove bearings will work with a shim in between the inner races, and modest clamping pressure on the outer races for a while. If you want to make it last a little longer use a wave washer between the inner races. As the bearings wear the inner races will be pushed out by the washer.

JMO

Reply to
Bob La Londe

P.S. Mach 3 supports screw mapping, and it will run upto 300 lines of code unregistered. Use any motion control hardware you like, but I favor the G540 behind a Smoothstepper for stepper motors, and G320X servo controllers behind a smooth stepper and a decent breakout board for servos. Smoothsteppers are available in USB or Ethernet. If this is going to be a slow no load application you might even use a UC100 instead of a Smoothstepper. Or even just a parallal port to your breakout board. I still have one machine running a G540 directly off an LPT port.

Reply to
Bob La Londe

Check out Kerk, they make special rolled thread screws with anti-backlash nuts. I used one in my laser photoplotter, and it has worked great. I had them select out one with better than average accuracy for a few $ extra.

Jon

Reply to
Jon Elson

How about a large format sign printer. Replace the normal print head with a simple electromagnet to lift/drop the magnet. Run it using the cutting software that goes with it. That can be set up to cut down to .5 mm or better normally.

Reply to
Steve W.

To me the difficulty would seem to be the accuracy over 1 meter of length. Not sure what your budget is but perhaps a linear encoder DRO scale. Another possibility could be to map a lead screw using a dro scale, store calibration values, and correct for lead screw error per calibration data. Of course if you use the DRO scale approach, it should work for a timing belt drive, rack & pinion, whatever.

I wonder if the mechanism from an old pen plotter could be of any use?

RogerN

Reply to
RogerN

"RogerN" wrote in news:QpqdnWU46cQgHIbJnZ2dnUVZ snipped-for-privacy@earthlink.com:

How about a far out there idea for the position sensing? Friction drive to a rotary encoder!

Use a Dremel 7122 Diamond Wheel Point 2.4mm which is a shaft with bonded diamond dust round it at one end running on a soft brass strip the full length of the work area. It will need a fair bit of pressure, but once the random diamond grains have indented the surface they run on, it will be zero slip.

Add a photo-interruptor for the home position.

If the difference in coefficient of thermal expansion between brass and steel is a problem (and a 10 deg C change would give you an error of about

0.1 mm in 1 m), you'd need to go to a copper or brass plated steel strip for the track or add a temperature sensor and a photo-interruptor at the far limit and compensate in software.

With a 1000 pulse/rev encoder, you'd get 0.01 mm resolution. A good mechanical setup would have the strip on edge rigidly mounted at one end and spring tensioned at the other, with the 7122 diamond coated shaft in a custom made collet to fit the bore of an Omron E6H-C series hollow shaft encoder and a small rubber coated roller (e.g. VHS VCR pinch roller) running on the back of the strip opposite the shaft to maintain the contact pressure. Add some felt wipers to keep dirt out. The E6H-C series is rated for 29.4 N side load so should be OK with a couple of Newtons pressure and approx 4 cm overhang.

If the E6H-C hollow shaft encoder is too steeply priced, you'd need two bearings on the 7122 shaft (possibly on the outside of a custom made collet assembly) and a normal encoder coupled to it.

HTH

Reply to
Ian Malcolm

PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.