Using digital caliper for positioning?

The other day I bought a caliper and was astounded to to find perfectly fine digital calipers for only $10 or so. These even have digital readouts and data ports! Then I found that similar digital slides are actually used with lathes. Naturally I'm wondering if these would be usable for reprap. The idea is to get accuracy limited by these devices instead of by accuracy of mechanical construction.

The page at [www.shumatech.com] explains the innards of these linear slides. They can be programmed to give output at a rate of 50 Hz, which seems enough for real-time control of speed. So my question: wouldn't two digital calipers for measuring and cheap DC motors for driving threaded rods make for a fine positioning system?

We add new somewhat expensive components of course, but on the other hand can do with cheaper and easier to drive DC motors instead of steppers.

-Geert

Hi Geert,

it's possible, of course - and it's accurate until 0.01 mm too ...

But a feedback with 50 Hz isn't so very fast, this would slowdown the fabbing seriously.

An example: - my CNC-mill is capable of 10000 interpolations per second or virtually positioning in 3D pixel-wise with max. 10kHz - and this is slow, when milling with high accuracy!

If you want to control your fabber (or mill) with a 50Hz-rate-feedback, then either you have to reduce your accuracy drastically or you will have a maximum moving speed of 5mm/s (or even much slower with positioning-error-reduction!) at 0,1mm accuracy in respect to 50mm/s at 0,01mm accuracy at my CNC-mill ...

Viktor

Sorry Viktor
but although the scale can only update every 20ms it can be moved alot faster than that it simply doesn't display every 0.01mm
If it displays every 0.1mm (what should be fairly enough for the darwin) it can be moved 50mm/s much faster that nophead is printing at the moment.

And those scales would help alot realigning different layers, it doesn't matter if the head takes half a second more to perfectly align the last 0.09 mm

So, adding digital scales is a pretty good idea.

'sid

Edited 1 time(s). Last edit at 10/21/2008 07:20PM by sid.

It seems frequency of updates only needs to be sufficient to accomodate desired changes in speed. Basically, if we're zooming along at 50mm/sec, and know we're at 5.00mm from our destination, we know that's exactly 100ms from now. So then we're going to have to do one of two things: just keep our speed and cut off the extruder at exactly the right time, or, more likely, slow down, use one or two measurements to determine our new velocity and compute a new time-to-arrival.

In the end, the speed and especially maximum acceleration/deceleration is mostly dependent on how good a model of the motor and moving parts we have. Ideally, we can use relatively infrequent position measurements to calibrate and adjust the model of the drive system. It is of utmost importance however, to have accurate clocks and little jitter in the measurements. It's not particularly interesting to now that we have been at position P, unless we know exactly when we were there. The challenge is to read the scale, update the model and control the motors in real time.

-Geert

"Ain't no need to watch where I'm goin'; just need to know where I've been." - Tow Mater

... right for straigt lines in one axis only ...

As i understand the servo-feedback you have to wait after every changing of your moving-angle (e.g. when drawing arcs or such) until the sensor prove true the last target-point.

When driving inclinated lines you need an extremely precise speed-controlled motors or your line would be a sinusoidal wave around the desired path with the frequenzy of your measuring/validating.

While you can use Bresenham-calculations for stepper-driving, it's not so easy with servos - here you specify two speeds for the two driving motors (see drawing vector-lines) and have to recalibrate every n-th position - ye faster, ye preciser is the line.

And while most stepper-driven mills run smooth and accurate with 10kHz and microstepping, the servo-driven mills i know needs much higher feedback- and steering-frequencies (some above 200kHz?).

Viktor

While they may not be suitable for closed loop control it seems worth having a quick think about other applications. A couple I can think of are:

- Non-linearity mapping, finding places where a one step is shorter or larger than usual
This information could then be used to add or remove compensating steps to keep the absolute error to within half a step. This may not be needed, I don't know how far from linear Darwins tend to be. I would imagine that if the belt gears are slightly off center that would cause irregularly sized steps.

- Reducing or removing the movement between layers for homing
As the home position can be anywhere on the table via absolute positioning it can be placed nearer the work, which should be faster and more accurate.

Homing could even be removed entirely by measuring drift and compensating between layers.

I'll take your word for it that 50hz is too low to be useful as a real-time positioning system, but...

this could be used for homing, like peteredworthy said.

Imagine you are printing 8 corner cubes (forgot the real name) - you could use the caliper to measure the precise location of each corner cube, basically offering improved repeatability and accuracy, without compromising speed.

-Leav

Much as the idea is a good one, I don't think cheap calipers would be very good. All of mine drift slowly, by about 0.01mm every few changes in direction. Since the average build contains hundreads of such changes, the drift might reach into the mms, which would really defeat the point. Some form of quadrature encoder on a motor shaft might work better, but is there much need? My steppers don't drift, and I'm not sure dc-motors are any cheaper or better if you need to have the expense of a caliper and addition electronics for each axis.

Closed loop control is better if--apparently--harder. Whether that is a shaft encoder or whatever...and, yes, I think there is a need. You start trying to push the steppers faster--faster ramping and such--and they will eventually lose a step here and there. Over a large build that adds up.

Demented

Quote

You start trying to push the steppers faster--faster ramping and such--and they will eventually lose a step here and there.

Yes they will skip if you exceed the pull in or pull out torque, but then you just back off and they run reliably. There is no need for feedback with steppers if you use them properly.

---

[www.hydraraptor.blogspot.com]

True, if you use them properly, they don't need feedback. If you want to go faster, however, you need to change to a closed loop system and some servos or something.

Demented

Hi Demented,

... how fast do you want to go?

My CNC-mill (as many other stepper-driven systems) drives with half-step until 6kHz and 12.5 microns per step = 75mm/s or 4.5 meters per minute.

When switching to 1/4 microstepping with 10kHz and 6.25 microns per step it's 62.5mm/s or 3.75 meters per minute ... this is more then enough for most tasks

Viktor

The best use of the Digital calliper is to maintain the accuracy of your rep strap or Rep Rap build I could not do without mine I don't really use my 0-25mm micrometer any more the calliper is easier & quicker. If you are in the UK buy it from Maplin when it goes on special offer

---

Bodge It [reprap.org]
=======================================

---

BIQ Sanguinololu SD LCD board	BIQ Stepcon	BIQ Opto Endstop
BIQ Heater Block PCB	BIQ Extruder Peek clamp replacement	BIQ Huxley Seedling
BIQ Sanguinololu mounting	BIQ standalone Sanguinololu or Ramps mounting	Print It Stick It Cut it

My rep strap: [repstrapbertha.blogspot.com]

Buy the bits from B&Q pipestrap [diyrepstrap.blogspot.com]
How to Build a Darwin without any Rep Rap Parts [repstrapdarwin.blogspot.com]
Web Site [www.takeaway3dtech.com]

Have just seen that Maplin has these calipers on sale again -