I am building a 3-dimensional positioning system for my university bachelor thesis. You can think of it as a 3D printer, except the nozzle is replaced with a holder for an optic fibre. We have received some demanding specs from our client (listed at the bottom) and a budget of roughly 2300 dollars. I am thinking of using Arduino and a shield called RAMPS 1.4 to control the actuator motors. My struggle is finding affordable actuators meet the requirements. NEMA stepper motors with a lead screw or ball screw seems like a good option, but I am uncertain on how to assemble the railing, and what railing parts I need. Some complete actuators seem decent, but I fear the accuracy will plummet when the axis is mounted together. Delivery time is also an issue, so parts need to be shipped from Europe (or have fast delivery to Norway). I have little to no experience with linear actuators before this project. Other tips you think I might need also very much appreciated
Thank you!
10 µm is 0.01 millimeter. Most mechanical constructions will flex even more than that when a hand is placed on the construction.
This precision is not a suitable subject for hobby people.
Perhaps you better research for what precision is reachable with your budget. 8x8x20mm allows for massive parts for strong guidance. Linear actuators usually are built with screw thread drives, find out which precision devices are available, including repeatability, at what cost. Also keep in mind thermal expansion/stability of the device.
DrDiettrich:
Perhaps you better research for what precision is reachable with your budget. 8x8x20mm allows for massive parts for strong guidance. Linear actuators usually are built with screw thread drives, find out which precision devices are available, including repeatability, at what cost. Also keep in mind thermal expansion/stability of the device.
If you go with actuators with ball screw followers, be sure to get lead screws with adjustable thrust bearings so you have control when changing directions. Otherwise you will need to quantify the lead screw movement and add or subtract that from your actuator turns.
Paul
I've done a lot of optomechanics in my past life. Freedom is never free Have you looked at thorlabs or newport, the typical places for optomechanics? If memory serves, each motorized degree of freedom is worth of 1000USD in that accuracy range so your client is bullshitting. The cost is combined actuator, stage with railing, and controller. Positioners like the ones you want already exist for fibers. They just don't cost that low. The cost comes from the actuator (500usd), motion stage with railing (250usd), and controller (250usd), ball park. Now with your skill level, which one or two of the three can you make at a lower cost than purchasing something off the shelf? Manual degrees of freedom costs about 500usd per degree of freedom because the actuator is just a micrometer and the stage with railing is the same as used by motorized ones.
SKF works in the nanometer range. Programming and running such equipment gave a unique experience.....
Just thought like You, the budget ought to be some 100 times higher or more, and done by really experienced and skilled people.
Some thoughts I have:
It should be easy to accomplish desired resolution with a stepper motor and leadscrew. A screw with small pitch can be used in addition to half/quarter stepping. Will it then possible to bypass the accuracy/repeatability problem using feedback loops? I was thinking distance sensors measuring actual movement for each actuator or some camera solution. That said, I am struggling to find sensors with desired accuracy. A third idea is to install photoelectric sensors at the desired x and y position (z-position don’t need to be exactly accurate, the make a simple algorithm to guide the optic fibre to its position. The time requirements are not very relevant. If the system finds the desired position within 20-30 sec it should be ok.
Will it then possible to bypass the accuracy/repeatability problem using feedback loops?
henningfrestad, Magnescale(was Sony) offers gauging probes with .1um resolution , 1um accuracy and .1um repeatability. but I think 3 of them would be over your current budget(maybe ebay).
henningfrestad:
Some thoughts I have:
It should be easy to accomplish desired resolution with a stepper motor and leadscrew.
I think you missed the part where OP is talking about 10 µm and repeatability of half that. if that are the requirements of the experiment to be performed, the actual system target tolerance should be better than that.
The thermal expansion issue as mentioned in #2 is no joke, nor is the flex by touching it (maybe even from looking at it too hard) from #1. That's the kind of precision OP is asking for. Regular precision lead screws, which have no discernable play when pushed and pulled by hand, are quite certainly not good enough for this.
In the final build the parts will also have to be aligned with such precision. A 90.01° corner instead of the desired 90° could be too much already.
Agreed, get an off the shelf system. Unless your thesis IS building such a system, and then I hope they have allowed you sufficient time for it (which is probably several times that of a typical thesis project).
Railroader:
10 µm is 0.01 millimeter. Most mechanical constructions will flex even more than that when a hand is placed on the construction.
This precision is not a suitable subject for hobby people.
Then don't place your hand on it!
This demands careful construction, and precision measurements to align the axes at right angles,
but its not out of the realms of hobby construction by any means, especially for such a small
active volume. You can certainly get precision ballscrews that are well capable of this and stick
some NEMA14 or NEMA11 steppers on them.
Anti-backlash acme nuts are also available and will be cheaper and acme is easier to source in
miniature sizes too I think.
henningfrestad:
Some thoughts I have:
It should be easy to accomplish desired resolution with a stepper motor and leadscrew. A screw with small pitch can be used in addition to half/quarter stepping. Will it then possible to bypass the accuracy/repeatability problem using feedback loops? I was thinking distance sensors measuring actual movement for each actuator or some camera solution. That said, I am struggling to find sensors with desired accuracy. A third idea is to install photoelectric sensors at the desired x and y position (z-position don’t need to be exactly accurate, the make a simple algorithm to guide the optic fibre to its position. The time requirements are not very relevant. If the system finds the desired position within 20-30 sec it should be ok.
Again, thank you
You should be using much more than 1/2 or 1/4 stepping in a motion control system,
x16 microstepping is a good starting place. Vibration and noise are terrible without
microstepping and with a leadscrew system you will have resonance issues without
microstepping.
I usually demonstrate a Michelson interferometer by setting one up on an optical breadboard (good teaching ones for $1K) and pressing one finger on the breadboard and then talking to the mirror (you could light up a lighter as well) to show how sensitive the interferometer is. It also shows how a rigid metal board flexes like a water bed, in the quarter wavelength range, i.e. 0.1-0.2um. At 10um you don't get a large effect unless you move long distances. I think a student should start with thorlabs and newport to research the cost of off-the-shelf solutions and understand their capabilities and technologies to have a base line, not to just take the project and do it with little to no experience.
Have you looked at thorlabs or newport, the typical places for optomechanics? If memory serves, each motorized degree of freedom is worth of 1000USD in that accuracy range so your client is bullshitting. The cost is combined actuator, stage with railing, and controller. Positioners like the ones you want already exist for fibers. They just don't cost that low. The cost comes from the actuator (500usd), motion stage with railing (250usd), and controller (250usd), ball park. Now with your skill level, which one or two of the three can you make at a lower cost than purchasing something off the shelf? Manual degrees of freedom costs about 500usd per degree of freedom because the actuator is just a micrometer and the stage with railing is the same as used by motorized ones.