[resistor]

Pardon if i'm intruding, have followed this thread with great interest.
Having cut tons of plastic for large and small parts, many requiring high precision.
I would like to mention one characteristic of most plastics that may
have a detrimental effect on it's use as a rack to drive a resolver or encoder.
“Thermal Coefficient of Expansion”.
That is to say that plastic changes radically dimensionally with temperature change.
Nylon for example has a typical linear coefficient of expansion of 4.5 in/in/°F x 10^-5.
Solving this equation tells up that nylon typically expands at a rate of
0.000045” per degree °F per inch.
So if you think of a rack that is 50” long,
in a loose temperature controlled environment,
with a 50°F total swing,
a difference in gross length of 0.1125”.

As politely as I can ask, perhaps the CncBrain could comp. the expansion as well?
It is logical that if one needed to scale the cutting path for a cold or hot
plastic part to comp for thermal characteristics, the control would be the place.
It seems logical that if the CncBrain is reliant on the feedback from a plastic rack,
that it could track the required thermal expansion.
Can the control pass a variable string to the CncBrain like we do with an aux axis?
Perhaps you would have to manually change the encoder ratio depending on temp?

Simplistically, just a modal call, which scales the commanded position
based on input from several thermocouples and a variable that indicates the material.
From my point of view these modal commands where created on the eighth day,
just after womens spandex ski attire.

res out

[rokag3]

I'm thinking about using two pulleys located on the center of the moving part of the axis and then have the rotary encoder connected to one of the pulleys

the big problem with this approach is vibrations if you find an efficient way to prevent it no problems
I sugest you make some trial with your encoder at different position and read the value of your encoder while using a small excentric mass attached to a small electric motor !!!
if you find a good solution tell us i was thinking about using small hooks in a tube like a curtain and foam rubber but i am affraid of the over length needed

“Thermal Coefficient of Expansion”.
That is to say that plastic changes radically dimensionally with temperature change.
Nylon for example has a typical linear coefficient of expansion of 4.5 in/in/°F x 10^-5.
Solving this equation tells up that nylon typically expands at a rate of
0.000045” per degree °F per inch.
So if you think of a rack that is 50” long,
in a loose temperature controlled environment,
with a 50°F total swing,
a difference in gross length of 0.1125”.

Thank you for the information my willing to use plastic was to reduce the momentum so I think i can use plastic pinion and stainless racks but in order to be very precise it will be nice to have many teeth on the rack backslash on rack and pinions is very easy to solve

[cncor]

I installed Visual Basic Express 2008. And the CNCBRain GUI application compiles and runs perfectly.

When is V1.0.6.4 expected to be released? I would like to add something, but I heard that in this next version a lot will be changed. So I think I better wait for it.

[oodanner]

the big problem with this approach is vibrations if you find an efficient way to prevent it no problems
I sugest you make some trial with your encoder at different position and read the value of your encoder while using a small excentric mass attached to a small electric motor !!!
if you find a good solution tell us i was thinking about using small hooks in a tube like a curtain and foam rubber but i am affraid of the over length needed

Can you elaborate on why vibration would be an issue? I don't understand how that would come in to play or why it would cause an issue. Any responses are appreciated.

[rokag3]

Can you elaborate on why vibration would be an issue? I don't understand how that would come in to play or why it would cause an issue. Any responses are appreciated.

The minimum you want from your encoder and specially the load encoder is an absolute stability the vibration of the rope are going to create false transitions but if you are using a laser or plasma might work if it's for a lathe or a mill i have a big concern

[snoopy27]

Hi all

I`ve been lurking in the shadow`s for a while.
Sins i`m quite new to cnc in general, and don`t want to ask any
darn right annoying questions that the veterans here are more than tired of answering. (thers supposed to be a smily here but din`t get that function to work)

Anyhu, i Must credit you bruce for creating the cnc brain.
Even though there are hundreds of controllers and breakout boards.
It doesn`t seem to be any product coming even close to yours.
All honors to you.

Even though i have two suggestions for improvements.

The first one demands little explanation. It might even have ben mentioned previously in this thread. (feel free to verbally abuse me if that is the case.)
Right, the cnc brain as it is at present supports 6 axis. And this is fine for most applications, but if one where to build slightly more advansed systems.
Like if one wanted to add a automatic tool changer etc you migt want atleast one or two more axes. So a possibility to supplement with one or more cnc brains to achieve 12, 18 or 600 axis if one felt like it. Only limit would be the number of usb ports on the computer.

So for the second and most interesting thingy. The tripple closed loop option.
This was an idea that came to me, as i tried to figgure out how to level the rails on "hudge" machinery. You see i own a company that manufacture varius items in fiberglass. Among others hudge (13 by 3,5 m) flat panels with intricate cutouts. And then to achieve high accuracy. With the lengths of rails required to machinery of this sise isnt easy. I'm even fiddling with the thought of milling casting moulds out of aluminium in sisabl dimensions. Meaning even higher accuracy is required. And today it came to me how to possibly not fix the problem but making it smaller. Now the system i am going to revele and explain might possobly help to increase accuracy on smaller machinery.

Note: i haven't tested the theory. Pluss i don't know if something like this already exist. And even though it all seems ingenius in my head. It might aswell be comlete Crapadudle.

Anyhu.
The cnc brain currently has a double closed loop system that littelay means that you know the whereaboutst of the "slead" in the moving direction.
Shown in picture one. With the help of rotary encoders or linear scales. Plus a rotary encoder mounted on the motor shaft.
Attachment 65994

Now what i'm suggesting is to know the whereabouts of the "slead" in two new directions shown in picture 2. A so called tripple closed loop.
And how are we gonna do that. We are gonna achive that with a 20$ laser pointer and something like a 200$ web cam.


The setup is shown in picture 3
Attachment 65997

Now what we do is we measure so that the laser pointer is correct at each end of track. Meaning it is the same distance "away" from the track at each end, in all directions. Now the next step is to mount the web cam on the "slead" on the current axis. The cam should be lined up so that the laser beam shins in the center of the web cam. (this is of course not an absolute)
Now what you get wen you look at your computer is a picture of a red dot.(the laser)

And now here is the part where i can no longer do anything sins i know ninkinpoop about programing. Now the idea is that when the slead moves along the track. If the "slead" is out of alignement the red dot wuld move on the screen. We should then have a script or program, that counts the number of pixcels the dot moves and then compensates. Sins the laser is 100% straight in all directions if it moves on the screen. The track must be out of alignment.

This would also help compensate for bending due to temperature etc. Causing an extreamly accurate machine.
This could be applied to all axis. And one could run the machine trugh the axis recording how the dot moves to make a referense. So the machine know prehand where to compensate.

But i can see two possible problems.

- Resolution on camera.
- Expansion of laser beam.

This is an interesting idea, but I believe you are correct with the issues you see, along with machine vibration during cutting. However for certain robotic applications this idea could be useful for tracking an object and Artificial Intelligence modeling. For instance with a web cam with a specific resolution one might be able to calculate the distance to identifiable objects based on the number of pixels up/down left/right to the center of a target. You could perhaps determine the size shape distance and relative identity of an object. A robot might be able to do in path corrections to a target with this concept as well. Interesting!

[snoopy27]

from Macona


I am very interested by rack and pinions , using 2 small pinions press on the rack in opposite direction it's very easy to eliminate backslash
a small teeth nylon or delrin coupled to a high def. encoder could make the job but
Where can i find small rack and pinions in plastic ?

When considering cost, isn't a precision rack and pinion as costly as comperable scales? As for the rope and pulley, doesn't this setup have a built in error, mainly the diameter of the rope changing as extracting and the distance of it's travel as the diameter of the rope at the pulley keeps changing the number of pulses counted each encoder revolution?

[Al_The_Man]

As for the rope and pulley, doesn't this setup have a built in error, mainly the diameter of the rope changing as extracting and the distance of it's travel as the diameter of the rope at the pulley keeps changing the number of pulses counted each encoder revolution?

I have some compact commercial units, no name or make on them, they have a steel cord wound on a spring loaded drum, similar principal to a steel measuring tape, as the cord leaves the drum, it drives over the encoder wheel. They measure up to around 6ft and a very accurate, because the wire travels over the encoder wheel, there is no discrepancy due to diameter change etc.
Al.

[oodanner]

As for the rope and pulley, doesn't this setup have a built in error, mainly the diameter of the rope changing as extracting and the distance of it's travel as the diameter of the rope at the pulley keeps changing the number of pulses counted each encoder revolution?

Are you saying the wire rope will change diameter from being stretched? I'm thinking if the wire rope is properly tensioned that it would work because you could tension it so very minimal stretching would occur as there wouldn't be a load on the wire itself, just G force as the axis moves back and forth right? Would that be enough to stretch the wire and make the encoder measurements inaccurate? I was thinking this wouldn't be much different in principle than what is listed on celesco's website:

http://www.celesco.com/stringpots/oem.htm

I'm all ears as I don't fully understand how using a rotary encoder in this arrangement might impact the resolution. Sorry, don't mean to high jack the thread, I'm just looking for a cheap way to measure axis movement that can be hooked up to the brain, if there is such a thing. EDIT: I know there is a CNCBrain, I'm talking about cheap axis measurement for it.

[fer_mayrl]

What is meant of the wire and rotational encoder is the following:

The rotational encoder is fixed to the moving axis carriage, a pulley is fixed to the shaft of the encoder.
A wire is then fixed and stretched on both sides of the machine.
Something like this

______________O___________________

Just think the wire is wrapped around the pulley (O) once, maybe twice...

When the carriage moves the wrapped cable forces the pulley to turn, hence translating linear motion into rotational motion.

simple, cheap, and i guess it should be very accurate.

Regards
Fernando

[NEATman]

I think this has been mentioned before in this thread, but for long travel with a rotary encoder you could just replicate a "trav-a-dial" http://www.southwesternindustries.co...surement.shtml
It is a friction wheel pressed against the side of the machine. Very simple and with felt wipers on each side, it should be trouble free. This is what is used on every Prototrak machine as linear feedback.

Alternately, you could attach an encoder to one of these belt drives:
http://www.cnczone.com/forums/showthread.php?t=59570
just back-drive it and you should get good repeatability.

As far as the webcam actively compensating for abbe offsets due to temperature change, that is a very interesting idea. It is close to the look up tables that are used on Delta-Tau control systems, only an active version. Large high precision machines are usually tested with a high accuracy laser system. For gantries with travels of about 2 x 2 meters, it is possible to get them to a repeatability of about 1 micron. The trick to doing this is with air bearing guideways, linear motors, and non-contact linear encoders in a temperature controlled environment. Granite construction and zerodur low expansion glass encoders also help.

The machines that I designed and tested like this were dual drive. We drove and had encoders on both sides of the Y axis, and were purposefully designed to allow about ±0.5° of rotation of the X beam in relation to the Y. We could purposefully make the machine go out of square by 0.5°. Each axis on the Y (Y and Y') had thier own encoder and thier own reference mark. When the machine did it's homing routine, it was cool to watch it wiggle when driven by only one motor or the other as they each found thier home index mark. Then, each axis was programmed a certain offset from thier index mark (derived from diagonal length mesurements done by a laser interferometer) the servos were activated and the machine would become square. As far as compensating for linear variation of the encoder, a simple linear slope correction could be used to drop the error for each axis.

In addition to this, a table could be programed based on the error measured by the laser to compensate for X, Y and Z errors in straightness or flatness. For instance, if the primary Y beam is a bananna shape in the XY plane, as the machine traverses down the Y axis, the X axis would move opposite the error, and accually move in a straight line. So the look up compensation table would move the X carriage in relation to the position of the Y axis. If the out of straight was in the YZ plane, the Z could move as the Y axis traversed. In theory, it is very much like a hexapod, where for any linear or rotary move, all six axes are moving at the same time.

Taking this to the extreme, you can make machines that have axes in non-traditional layouts such as this:
http://www.dovermotion.com/SpecificI...Page.aspx?ID=8
and get XY and Theta (rotation) from three linear axes arranged in a triangular fashion. The center moving puck is air bearing, so it floats over the granite surface.

NEATman

[dgapilot]

I've read almost this entire thread and couldn't find an answer to my questions, so ... Will scales similar to these work for the feedback loop? http://www.mitutoyo.com/TerminalMerc...spx?group=1298

These type scales are quite reasonable on Ebay, see http://cgi.ebay.com/DRO-0-6-Y-AXIS-V...QQcmdZViewItem

Thanks,
David

[minimum]

I've read almost this entire thread and couldn't find an answer to my questions, so ... Will scales similar to these work for the feedback loop? http://www.mitutoyo.com/TerminalMerc...spx?group=1298

These type scales are quite reasonable on Ebay, see http://cgi.ebay.com/DRO-0-6-Y-AXIS-V...QQcmdZViewItem

Thanks,
David

In short (as I understood), NO!
The data refresh rate from the scales is poor (comes in bursts) so there's no way to use these (or at least with good results).

Besides, Mitutoyo and cheap Chinese scales use different data protocol. Shumatech who sells DRO kits for Chinese scales, has a brief description:
http://www.shumatech.com/support/chinese_scales.htm

[macona]

zLike minimum said, nope, these wont work. You need scales with a quadrature output. Heidenhain, RSF, Anilam, Acurite are big names. You can find scales pretty cheap on ebay.

There are also a couple chinese manufacturers out there as well.

[ger21]

I copied this thread to the controller forum, and am going to close this one in the Product Announcement forum. Please continue further discussion here.
http://www.cnczone.com/forums/showthread.php?t=64794