Accuracy of a print head sensor in a DIY linear drive
ฝัง
- เผยแพร่เมื่อ 15 ก.ค. 2016
- Table Of Contents:
00:05 Introduction
00:48 Measurement setup
02:29 Controlling the drive with a microcontroller
04:05 Backlash
06:46 Limits of the linear sensor
The project page:
www.homofaciens.de/technics-ma... - วิทยาศาสตร์และเทคโนโลยี
You really do give the best formation to learn about this . I wondered why your set ups are so "home made" but then I realized that even with a basic set up the fundamentals are excellent so the project works fine. Well done for mastering all this and thanks again for sharing.
Great video... thanks for such a precious sharing of knowledge and all of your hardwork with us.
He sound like Arnie! Nice experiment, nicely explained too. I learned lots, thank you.
9:00 note to self: put temperature readout on my diy 3dprinter ! And the springy end of the encoder strip can have a larger error than the fixed end! Didn't think of that! Great video!
i like your videos, you have a fan in México, thanks for share
Now that's a scientific approach!
Excellent work :)
Great Work!
Great demo! The oscillations at 7:43 could be eliminated without slowing down the firmware-constants by adding hardware-damping by means of a drag-ring (ideal viscous damper) by using an aluminium disc dragging on a magnet-airgap :)
The fixed end of the threaded rod is linked with the base plate through two ball bearings that are pressed on a square tube (see the previous video for details: th-cam.com/video/ULJE4MsJpR4/w-d-xo.htmlm15s ). When adjusting the ball bearings tightly, you get more friction that damps oscillations.
Thanks for sharing your thoughts. I'll keep that in mind for future constructions.
Cool! The bearings may give coulomb-friction or stiction more than viscous friction (and that would undermine the counteracting of oscillations). A drag-ring, on the other hand, is near-ideal damping without any stiction. But I have a hunch you know that already ;)
Ball bearings used in that "wrong" way are close to drag rings. At least I could not feel stiction when turning the plywood disc by hand, but of course I can't say if there is viscous friction at higher revolution speeds.
so cool!!! thank you.
Thanks for sharing! As i understand this kind of linear encoder will compensate any backlash in the thread/nut but will oscilate if it's too prescice, can be improved by using a PID control for the motor so the oscilation will be dampened?
You'll never get any backlash compensated, but yes, PID will give better results.
+HomoFaciens Interesting!
Great work! So, my question is: can we use this motors with encoders for a 3D printer? Thanx 🙏
Yes, but not for a really good one with the sensor strip from old printers as the results of this video show.
Another nice video! Will you be trying it with an optical motion sensor next? :D
Maybe a laser module like the ADNS-9800, hehe ye its a bit pricey at the moment, maybe they will come down in price some day.
There will be a video about using an optical mouse as motion sensor for this linear drive, soon. Stay tuned!
+HomoFaciens Looking forward to it - my attempts with mouse sensor failed due to non-linearity of its output. Can't wait how you tackled that!
I did not tackle that, I just demonstrated in detail why it doesn't work properly:
th-cam.com/video/CIRKRzw54Zs/w-d-xo.html
Good morning! How can I do a Stepper follow the Rotary encoder of a printer, with acurracy? I need it so much, in slow and fast speed.
Thank you
Marcelo Santos
What exactly do you mean with "a Stepper follow the Rotary encoder of a printer, with acurracy"?
What is the point of it if there's already fast and accurate stepper motor available ?
how do you find time to build all these stuff? what is your qualification and study? u brilliant! is it reading only, please advise! :)
It's lifetime studying and learning by making new stuff, trying to push the limits each new day...
Where can I apply for 25h days? ;-)
HomoFaciens i was thinking more where do you store all this stuff after its built?
...I am running out of space. Some of the projects get disassembled to harvest parts for new machines.
Nice detail.
I guess ill just have to get used to your accent because I subbed.
How do you connect the sensor and what is its output? I got one and dont know how to connect the 6 pins
How to figure out the pin layout of unknown sensors is written on my pages:
homofaciens.de/technics-base-circuits-transmissive-optical-sensors_ge.htm
I love your videos but shouldn't you be preparing for Sarah Conner and Kyle Reese to be arriving from the past? Don't get so obsessed with these machines that you forget you still have to deal with the other machines! The future of humanity depends on you! (Terminator Genisys reference for those wondering)
On a more serious note, what tools do you recommend for building machines like this? I currently have a screwdriver and that's about it. Also, what materials would you recommend starting out with? Wood/metal/plastic/dragon bone?
Thanks for your videos.
A hacksaw, a vise, a drillpress, a soldering iron and wrenches should be in all workshops.
amazing! I have te same sensor, can you give the code for the arduino? Thanks!
There are similar codes available on my page about rotary encoders:
homofaciens.de/technics-base-circuits-encoder-disc_en.htm
Genius!!!
Are you to make a cartesian 3d printer soon?
At least not too soon. CNC v3.x will become first and foremost a CNC router, but 3D printing isn't out of range, so let's see what the future brings...
looks like greate option - why you didnt use it in 3.2.2 machine?
Because the combination of sensor disc and DC motor or bipolar stepper motor and threaded rods results in a clearly higher accuracy than the sensor stripe used in this video.
@@HomoFaciens Yes but...
with the printer sensor you receive more info on backlashes,
while the rotory encoder gives more accuracy on missed steppes.
combining the two may bring very interesting result
simply eliminate the backlash in your mechanics as good as you can. I guess trying to work around with software and another sensor is definitely no good idea. However, try it out and prove me wrong...
@@HomoFaciens so basically you suggest not using linear strips sensor as encoder because less accuracy than rotary encoder for machine that run in single axis? Note: i try to build like linear actuator, and still thinking using linear or rotary encoder.
That depends on the accuracy your application requires. Those linear sensors work fine in printers, but won't perform well in a CNC router.
So what is so bad in an inaccuracy of 50uM or 0.05mm? What is the standard accuracy for industrial cnc machines?
There is nothing bad with an accuracy of 50µm as long as you can't do it better for the same price and that's what I achieved in my previous video:
th-cam.com/video/ULJE4MsJpR4/w-d-xo.html
There is no standard for the accuracy of industrial CNCs. The higher the accuracy the higher the price...
+HomoFaciens what would be the approx accuracy of a 100,000$ CNC machine for example?
Google shopping will give you the answer ;-)
The really good ones start at 1µm and less.
1 micron is usually only valid in high-precision, miniature applications, such as swiss-type lathes.
Most expensive("proper") machines have a rated, tested, and proven accuracy of 5um or more, but that is stil enough for 99% of the applications, and for the other 1% you must already have the money to buy better gear.
Of course size matters. It is hard to command a CNC with a work area of 1x1m with a precision of 1µm.
Another thing to be considered is the difference between accuracy (relative positioning) and precision (absolute positioning). 50µm won't be too bad for a DIY CNC.
When the chassis of my CNC v3.x series is up and running I will do some more measurements and try various sensors and motors to see what gives best results. Stay tuned!
it is possible to use it on touchdro? and how?
Yes. By reading through all the docs of touchDro and finding a scale adapter and reading those docs as well.
You might be interested in Single Track Grey Codes. You can get both absolute and relative position data using multiple sensors.
techref.massmind.org/techref/io/sensor/pos/enc/greycodes.htm
With my latest video I am showing why optical mice aren't that good for linear motion sensing:
th-cam.com/video/CIRKRzw54Zs/w-d-xo.html
Next I will do some more machines, but I'll keep the grey scale sensors in mind...
Grey /code/ not grey scale. They can provide localized absolute position in addition to relative position. E.g. within any block of 30 positions, it can tell you exactly what position it is at. The cost is 5 photosensors instead of 1. More photosensors allows a longer block. E.g. with 9 you can do 256. So that issue of losing it's place when struck could be compensated for... assuming it doesn't move more than the block size.
Sorry, I did not read carefully...
Those sensors combine grey scale with absolute encoders as you said. It's indeed an option for future experiments around linear motion sensing.
Thanks!
Germans always brain of world
7:53 a hard-coded interval is a bad idea, because you loose the precision. PID is much better to suppress the oscillations.
...yes, as I told in my chapter about PID and servos:
homofaciens.de/technics-computer-arduino-uno_ge.htm
@@HomoFaciens oh, good material. pretty useful. 👍
galinha pintadinha mini
Hä?