Linear stage with 3.3" travel, stepper motor, and 0.5 micron encoder for about $100

Have questions about the equipment used for macro- or micro- photography? Post those questions in this forum.

Moderators: Pau, rjlittlefield, ChrisR, Chris S.

physicsmajor
Posts: 63
Joined: Sun May 10, 2020 12:56 pm

Linear stage with 3.3" travel, stepper motor, and 0.5 micron encoder for about $100

Post by physicsmajor »

In the grand tradition of re-purposing industrial equipment - especially things from biotech machines - at present there are quite a few OEM Parker-Hannifin stages on Ebay. These are extremely rigid and well made, and while you can get a (not as) nice linear stage for similar money from other manufacturers on Ebay these come with: a 0.488 um precision quadrature encoder(!!), a high quality 200 step/rev stepper motor, and an optical limit switch. This means you can do full closed-loop control - move, check position, correct to exactly where you want to be. They are specced at 3" of travel, but if you remove the hard stop you can get 3.3". The encoder goes the full length.

I am not aware of anything close to this price which has both the precision of this stage, plus the encoder, plus the stepper motor. I haven't seen anyone else posting about this here, and I thought it would be of interest around these parts.

One model is Parker 803-9288E, though some sellers just list it as Parker OEM. There are a couple versions of this thing, one with a Lin Engineering damped stepper motor and another with an Applied Motion stepper. I have the Lin Engineering variant. A couple links - they sell for under $100, depending on the seller you might end up total cost under $100 or a bit over.

https://www.ebay.com/itm/Parker-803-928 ... 3961637083
https://www.ebay.com/itm/Parker-803-928 ... 2993263894
https://www.ebay.com/itm/Parker-OEM-Lin ... 4272829268

I have successfully engineered this into a fully closed-loop controlled macro rail driven by a Raspberry Pi, a Polulu Tic stepper controller, and an Arduino which monitors the quadrature encoder and also fires the camera remotely by sending the correct IR LED pulse signal to my Nikon DSLR. The total BOM:

Parker OEM stage, shipped: $81.84 (may vary)
Rasperry Pi 4 model B: $35
Arduino Uno: $23
Polulu Tic (any which can do 1/16 microstepping): $31.95
IR LED, voltage drop resistors, random breadboards and hookup wires scavenged but let's say $10.

Total (close approximation): $181.79, definitely under $200.

This is under half the price of some commercial options, with far higher rigidity/precision and Stackshot/Wemacro do not offer anything with an encoder, never mind an encoder with submicron precision, so no closed-loop control. Each full step of the stepper motor travels 10 um, so with 1/16 microstepping you can actually achieve verifiable submicron positioning.

They are oddly shaped but easily adapted to photomacrography. On the Lin Engineering variant there is a tapped hole on the motor side of the stage which is photography-standard 1/4"-20, and another on the bottom which is also 1/4"-20. You can attach Arca-Swiss hardware to these and be off to the races!

The sled itself weighs a bit over 11 lb (not lightweight). Most would probably hard mount this for a horizontal setup, but I mount the whole thing to a tripod for mineral photography to good effect:
parker_macro_rail.JPG
If there is interest, I'd happily share more specs about the hardware as well as my Arduino sketch to monitor the quadrature encoder and fire the camera, and the Python code which interfaces with the Polulu Tic and implements full closed-loop control. It's a fantastic platform for photomacrography, at a fantastic price.

rjlittlefield
Site Admin
Posts: 21034
Joined: Tue Aug 01, 2006 8:34 am
Location: Richland, Washington State, USA
Contact:

Re: Linear stage with 3.3" travel, stepper motor, and 0.5 micron encoder for about $100

Post by rjlittlefield »

Interesting devices!

I notice that one of the eBay ads says "RLS model RLM2-12BCA RoLin™ linear incremental magnetic encoder which has a resolution of ≈ 0.488 μmm".

This is exactly what, 4096 counts in nominal 2 mm?

--Rik

physicsmajor
Posts: 63
Joined: Sun May 10, 2020 12:56 pm

Re: Linear stage with 3.3" travel, stepper motor, and 0.5 micron encoder for about $100

Post by physicsmajor »

Yes - that seller (Starlight Photonics) knows the most about these, and they're correct. The hardware is relatively standardized, they all have that encoder.

As you say, it's exactly 4096 counts per 2 mm. Or if you prefer, 2048 per 1 mm, etc. Each encoder step is 0.488 um, or 488 nm. Kind of boggles the mind that it's accurately reading position at intervals equal to the wavelength of teal-blue light.

Each full step of the motor is calibrated to be 10 um - and with microstepping I have had no problem demanding precision better than ±2 um (this is sufficient for 10x and I don't have a need for more, but I do believe it can achieve reliable ±1 um). There are some minor nonlinear factors in the leadscrew or system at large, as you expect when you measure to this fine of a level. At first I trusted the stepper more than I trusted the encoder, so I did open loop control. Upon experimentation, the encoder is just dead on correct and with closed loop control you just occasionally nudge the system back into cal. Closed loop has the advantage of being innately calibrated (for relative movement), so you just look up the appropriate step given effective f-number and magnification and plug it in as the interval.

In case anyone wonders if they could do without the Arduino - not without a dedicated quadrature encoder readout chip, and these cost about as much as the Arduino. The Pi just cannot reliably poll the encoder fast and regularly enough for an accurate measurement. Even going relatively slowly, you rack up a high quadrature rate when you get a new state every 0.488 um!

physicsmajor
Posts: 63
Joined: Sun May 10, 2020 12:56 pm

Re: Linear stage with 3.3" travel, stepper motor, and 0.5 micron encoder for about $100

Post by physicsmajor »

Expanding a bit on the encoder:

Properly mounted, aligned, and wired up is a RLS model "RLM2-12BCA" encoder (a Renishaw company, one of the big names in industrial high precision scales). I believe it's actually a RLM2-12BAC, as the final letter per the company is either C or nothing and it definitely comes with this option (the 2 mm reference). The encoder itself is mounted for more travel than the system is capable of, close to 100mm. The scale and encoder are mounted on the undersurface of the moving stage, where it is unlikely to get dirty or even significantly dusty. The site with technical data is located here: https://www.rls.si/eng/rlc2hd-miniature ... ic-encoder

This particular model of encoder is specced for travel up to 1.82 m/s, with a maximum encoder frequency of 8 MHz. You won't get close to that speed over a 3.3" actuator, though; depending the speed of your quadrature encoder you might start skipping states well before the encoder's capability.

Encoder pinout:
- Red = +5V
- Black = ground
- White = quad out
- Blue = quad out
- Brown = reference out

Use is simple. Hook up power (from the Arduino) and monitor the quadrature outputs with digital interrupt pins. The reference output is available as a convenience and provides a check every 2 mm of travel (4096 steps). I currently only use the quadrature outputs.

You can buy this scale and encoder from RLS for $89.09. With this one piece, you've broken even with the cost of the whole unit on Ebay - never mind the need for painstaking precision in mounting the scale and reader head.

physicsmajor
Posts: 63
Joined: Sun May 10, 2020 12:56 pm

Re: Linear stage with 3.3" travel, stepper motor, and 0.5 micron encoder for about $100

Post by physicsmajor »

The stepper motor:

Mine came with a Lin Engineering 4118M-03S stepper motor. This is a custom model not available on Lin Engineering's site, but seems extremely similar to a standard 4118M-06S except it's specced for marginally more current (0.85A instead of 0.7A) and possibly higher precision. It is a 200 step/rev or 1.8 °/step motor. Each full step moves the stage nominally 10 um. With 1/16 microstepping you can easily have ±2 um precision and likely ±1 um precision. My motor has a through-shaft with a damper mounted on the reverse.

It's trivial to figure out how to wire a stepper motor. There are four leads. Use a DMM to determine which leads are connected with a low resistance. Those are one pair, the other leads... are the other pair. Hook it up to your controller like this. If it moves in the opposite direction expected, reverse one of the sets of leads.

A Lin 4118M-06S stepper motor with rear shaft runs $50.87 for a prototype, sans damper or hardware to mate the shaft to the leadscrew.

physicsmajor
Posts: 63
Joined: Sun May 10, 2020 12:56 pm

Re: Linear stage with 3.3" travel, stepper motor, and 0.5 micron encoder for about $100

Post by physicsmajor »

The physical hardware:

The stage itself is overbuilt and highly constrained to only linear motion. It is CNC machined from a solid block of either aluminum or steel (quite heavy). It is mounted on two Hiwin MGN9HH miniature linear guideways per each of two rails attached to the base. The rails also extend well beyond the actual travel, they run the entire length of the stage at 210mm long. A small amount of what may be white lithium grease is applied. I can't find an exact quote for the Hiwin hardware, but it likely is in excess of $100 by itself. Hiwin's info about their guideways is here: https://www.hiwin.com/pdf/linear_guideways.pdf - these are the miniature MG series.

The leadscrew has 12mm outer diameter with a 2mm pitch. The nut attached to the stage is heavily preloaded to virtually eliminate backlash. For each full step, the stage travels 10 um. Both ends of the leadscrew are mounted to sealed roller bearings.


Exclusive to the Lin Engineering motor variant of this stage (identifiable by the damper on the motor), there is a tapped hole in the stage and another on bottom of the base with 1/4"-20 threading. This is how my Arca-Swiss hardware is mounted in the image on the original post. The Applied Motion stepper variant does not have these, but has other mounting holes I cannot be sure about.

ray_parkhurst
Posts: 2750
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Re: Linear stage with 3.3" travel, stepper motor, and 0.5 micron encoder for about $100

Post by ray_parkhurst »

The leadscrew appears to be a very high quality teflon-coated type. My first stacking stage used a teflon leadscrew and had excellent performance.

mawyatt
Posts: 2479
Joined: Thu Aug 22, 2013 6:54 pm
Location: Clearwater

Re: Linear stage with 3.3" travel, stepper motor, and 0.5 micron encoder for about $100

Post by mawyatt »

Very interesting device and nice setup!!

Would be interesting to see how this measures up to the popular THK KR types folks are utilizing.

A number of years ago we started off with the excellent Pololu Tic devices on the RPi, but later moved up to the Trinamic devices because of the Motor & Velocity Profiles which allow extremely smooth and precise control, yet also allow very fast movements without any jerky type behavior. The Trinamic TMC5130, TMC5160, & TMC5072 also support encoder feedback, although we haven't employed this and elected to go with the Continuous Closed Loop Piezo Stages for very precise nanometer levels of performance.

I see no reason the Trinamic devices wouldn't work with these Parker type rails.

Best,
Research is like a treasure hunt, you don't know where to look or what you'll find!
~Mike

ray_parkhurst
Posts: 2750
Joined: Sat Nov 20, 2010 10:40 am
Location: Santa Clara, CA, USA
Contact:

Re: Linear stage with 3.3" travel, stepper motor, and 0.5 micron encoder for about $100

Post by ray_parkhurst »

mawyatt wrote:
Wed Jul 01, 2020 10:25 am
Would be interesting to see how this measures up to the popular THK KR types folks are utilizing.
The stage uses a dual-rail, quad-carriage arrangement, so from a 3-axis torque perspective this stage probably has 10x the capability of a KR26. The off-axis drive arrangement introduces static loading torque for vertical arrangement, or dynamic loading for horizontal, so this will limit load capabilities somewhat, but with the kind of loads we use in our setups it should be fine, similar to my preferred idler rail arrangements.

Although the ball screw is in a "zero backlash" nut arrangement, these nuts have some amount of inherent rotational play, so there is always a bit of backlash that must be dealt with. The teflon coating helps with this but more important is the rigidity of the ball screw mounting. THK does a superb job here and their backlash numbers are very low.

Biggest limit I see with these stages is the 200-step motor, which is easily replaced with 400-step type, and the 2mm pitch screw. These are of course mitigated by the positional feedback. It's also not clear what load capability the ball screw is capable of, but again for our systems this should not be an issue.

physicsmajor
Posts: 63
Joined: Sun May 10, 2020 12:56 pm

Re: Linear stage with 3.3" travel, stepper motor, and 0.5 micron encoder for about $100

Post by physicsmajor »

Indeed there is a minimal amount of backlash, though I have adaptive code in my closed-loop driving algorithm which functionally allows one to ignore system nonlinearities.

Regarding load tolerance - early on when I wasn't sure how much load the stepper and gearing would support, I put my entire strength against the stage braced on the undercarriage and was unable to stop, significantly slow its movement, or even cause the stepper to skip. I do not think any amount of photographic gear one could mount on the stage would create a problem. I use this from horizontal to probably around 60 degrees on the tripod, depending on specimen, and it doesn't skip a beat.

I did initially consider replacing the stepper with a 400 step type. This would improve the precision, potentially beyond the encoder's ability to measure at 1/16 or higher microstepping. I decided not to, because the inbuilt stepper matches the encoder resolution fairly well at 1/16 microstepping and I am satisfied with its performance at 10x magnification. 400 step motors also have somewhat lower torque and holding force.

I should have mentioned this earlier, but I have no relation to any of the sellers of this item. I stumbled across it in late April in quarantine and decided it was attractive as an entry point into this hobby; it's exceeded my expectations and I wanted to share this in case anyone else finds it useful.

mawyatt
Posts: 2479
Joined: Thu Aug 22, 2013 6:54 pm
Location: Clearwater

Re: Linear stage with 3.3" travel, stepper motor, and 0.5 micron encoder for about $100

Post by mawyatt »

The THK KR types we have show no measurable backlash.

A test awhile back at an effective magnification of 800X showed no signs of backlash nor serious induced vibration, suspect the Parker rails would also be good. Too many projects keep me from evaluating these rails with the Trinamic controller tho, like designing and evaluating sub-ppm drift voltage references!

https://drive.google.com/file/d/19wE-B- ... sp=sharing

Watch out for the TI and Toshiba stepper motor driver chips, they are used in some of the Tic devices. The TI chips have shown a horrible result when micro-stepping, a few posts about these for 3D printing is online. The Toshiba chips have a dead zone in the motor current zero crossing, causing a torque "jerk" when transitioning zero (learned this from a Pololu engineer), supposedly you can feel this on the shaft if you monitor the motor currents and know where this occurs. We don't use these chips so haven't had to deal with this.

Best,
Research is like a treasure hunt, you don't know where to look or what you'll find!
~Mike

physicsmajor
Posts: 63
Joined: Sun May 10, 2020 12:56 pm

Re: Linear stage with 3.3" travel, stepper motor, and 0.5 micron encoder for about $100

Post by physicsmajor »

Good to know about the TI chips, that could be relevant for anyone looking to replicate this. I am currently using their Toshiba controller offering, the Tic T249, because I have a few other steppers including some higher power options I wanted some headroom for. It's working fine. I can't claim to have experienced the zero crossing issue, but perhaps I should swap it out for the discrete Tic 36v4 I have lying around and see if things seem smoother.

I haven't used any of the Trinamic offerings. The Polulu is easy to control via USB and has command-line scripting.

Again, backlash is mostly academic when you have an encoder and closed loop control.

mawyatt
Posts: 2479
Joined: Thu Aug 22, 2013 6:54 pm
Location: Clearwater

Re: Linear stage with 3.3" travel, stepper motor, and 0.5 micron encoder for about $100

Post by mawyatt »

We don't use the Toshiba devices but was told about them, were using some of the TI devices which were all replaced.

Feedback doesn't "cure" backlash, it only masks the effects, sometimes making it worse. The dead zone caused by backlash creates "hunting" in closed loop systems, this can cause considerable issues when trying to find the exact location. What happens is the system approaches a position, may slightly overshoot, then attempts to reverse and experiences the backlash induced dead zone. When a dead zone is encountered in a negative feedback system the system effectively becomes "open loop" because no feedback is present and the controller attempts to close the loop by driving hard. This may overshoot in the opposite direction and the whole process repeats in reverse. If it's a large dead zone the system may continuously bounce back and forth across the dead zone "hunting" for the final position and never settle into a steady state condition. We've seen this effect in many closed loop systems, however this may not affect your results depending on how close you view and how the feedback is implemented (continuous, discrete, resolution and such).

Interestingly this is the very feature of a system called a "Bang Bang Servo System", where the dead zone is purposely introduced to cause the system to bang into each of the rails, or limit cycle. The desired result is then integrated to average the limit cycles, similar to Pulse Width Modulation. In another case the Ring Laser Gyro had to deal with an optical dead zone caused by the two counter rotating optical HeNe laser beams locking together at low gyro rates, the solution was too "dither" the optical effective path length to bounce across the dead zone and then integrate for the result.

An effective but simple means to mitigate backlash in open loop and closed loop rail systems is the popular unidirectional loading of the rail with gravity, springs, or rubber bands.

Anyway to see this type of feedback control hunting, look at output of the control system that drives the motor. If this is slightly varying under steady state conditions then the likely cause is hunting, which may look like excessive noise (common result).

Best,
Research is like a treasure hunt, you don't know where to look or what you'll find!
~Mike

physicsmajor
Posts: 63
Joined: Sun May 10, 2020 12:56 pm

Re: Linear stage with 3.3" travel, stepper motor, and 0.5 micron encoder for about $100

Post by physicsmajor »

This system's dead zone is significantly less than 2 um which is the lowest I've bothered to test. There are no problems with closed-loop convergence using this hardware and current algorithms. If you tried to do something silly like ask for 0.5 um precision, with 1/16 microstepping corresponding to 0.625 um per step and 0.488 um encoder increments, then you would have a problem.

Online
lothman
Posts: 480
Joined: Sat Feb 14, 2009 7:00 am
Location: Stuttgart/Germany

Re: Linear stage with 3.3" travel, stepper motor, and 0.5 micron encoder for about $100

Post by lothman »

I think in the sub µm range the problem is no longer the motor but stick-slip phenomena of the spindel/nut. May be the ball screw of the THK KR series has an edge over all the thread screw spindels. But probably such small movements with high accuracy and repeatability are the domain of piezo positioners or thermal positioners.

Post Reply Previous topicNext topic