Motor distance question

Hi, I have 2 questions about motor positioning.

1- When I use the chain calibration to measure motor distance (motor to motor) it come out as 2986mm but when I measure with a measuring tape its 3000mm. What measurement should I use in GC?

2- I used the alternate frame design found in the wiki as I preferred to have a solid board between the motors to prevent flexing. In the original frame design there is a small piece of plywood the motor attaches to which hangs the motor out over the work surface a few inches (I think so the chains are parallel with the work surface as the sled connection points are a few inches out). But with my motors directly in line with the work surface and the sled connectors a few inches out, the chains arenā€™t parallel with the work surface, they start close at the motors and end up farther out at the sled. should I be fixing the motors farther out?

1 Like

If your tape measurement was made while the chain was taught, there are important improvements in the dev version of firmware and GC that address this and other related issues. If youā€™re comfortable using that version, youā€™ll get an auto-measurement much closer to your tape check. If not, it will be in the released version next Wednesday. Using that new release will probably mean going through the improved calibration routine to gain the accuracy benefits it will bring.

The chain really needs to be close to parallel with the surface, or the chains will jump and lose position on the sprocket. The top-beam descriptions donā€™t emphasize this, I guess. Either move the beam out or use something to extend the motors out to put the sprocket in the right place.

2 Likes

If your tape measurement was made while the chain was taught, there are important improvements in the dev version of firmware and GC that address this and other related issues. If youā€™re comfortable using that version, youā€™ll get an auto-measurement much closer to your tape check. If not, it will be in the released version next Wednesday. Using that new release will probably mean going through the improved calibration routine to gain the accuracy benefits it will bring.

the last released version includes a significant error, so your tape measure is
probably closer to correct, but the same error will cause the chains to be
longer than the machine thinks they are, so itā€™s not as simple as just entering
the correct value.

The chain really needs to be close to parallel with the surface, or the chains will jump and lose position on the sprocket. The top-beam descriptions donā€™t emphasize this, I guess. Either move the beam out or use something to extend the motors out to put the sprocket in the right place.

In the top beam design, the beam is supposed to be able to move out to keep the
chains parallel. Thatā€™s why it used a 2x10 instead of a 2x4. The 2x4 is strong
enough, but would need to be moved out. I think I need to tweak the design to
make this clear.

1 Like

@HydronCollider

Welcome to our group. What version. Of Ground Control are you using? What firmware are you on?

@ - answering folks

Please be specific about referring to versions - ā€œLatest Versionā€ is confusing when searching the group for an answer 3 months from now.

I suggest something like this - in our current version .99 we know there are errors addressed in 1.00 releasing on 12/20/17. If you want to test the beta build of 1.00 it is available at ā€¦I

This way the information will be friendly to searches in the future.

Thank you

Thanks Bee, Im using .99 and .99. I got things working much better today using tape measure and manually entering measurements. I just have to fix up the top beam now to hang out over the work surface as per dlang.

2 Likes

ith the 0.99 firmware, the chains will always be a little longer than the maslow
thinks they are, this is going to affect the accuracy of all your cuts (possibly
not enough for you to notice), but when you upgrade to the 1.0 firmware, you
will want to re-do the calibrtion and your results should be a bit more
accurate.

2 Likes

Good morning!
Top bar design, running v1.02 of GC.
The top bar allows me to attach a tape measure with gaffer tape and measure a pretty accurate 3077mm between motor spindles.
Iā€™ve then run the calibration procedure and it gives me 3073.43mm between motors. So, not a million miles away, but as I look at the chain, even whence itā€™s been pulled tight there is a perceivable sag in the chain, obviously.
The obsessive in me says I need to go with the accurate 3077mm, but DLang, you mention above that it isnā€™t as simple as that. Iā€™m guessing I must go with what Maslow thinks it has so that this doesnā€™t develop inaccuracies in cuts. Is this ā€˜working errorā€™ how we calculate for chain sag? Is chain sag accounted for in the maths? Is there a catenary equation that works out chain sag based on length of chain in play, and the angle it subtends from the motor spindle, based on the chain sag achieved over the distance between motors? Am I talking rubbish? ;Ā¬)

Just looked up catenary maths for a revision session, think Iā€™m gonna need some stronger coffee!

1 Like

First off, it may be worth waiting for 1.03 to get released later today. I know that the UK is ahead of us, so it may not be until the evening that you see it. The new update will include a new calibration pattern that should be easier and more accurate than the current one. Iā€™ve been trying to follow itā€™s progress but Iā€™m sure it will all make a lot more sense to me when Iā€™m actually running the procedure myself :wink:

That being said, I can try and help you through calibrating the machine. I use a manual process myself because I havenā€™t had much success with the automatic routine.

Did you notice any sag in your tape measure? If not, then I would start with that measurement. Maybe itā€™s the steel fabricator in me, but I feel like I trust my tape measure more right now. Thatā€™s subject to change with the new update. With that measurement plugging into GC, cut (or draw) a 4in (~100mm) square into a scrap piece. Check the X and Y legs of it to see how far off the measurements are from one another. Ground Control has a simulator which will help you figure out what variables need to change in order to dial the machine in. In a lot of cases, in my experience, tweaking the distance between motors is how I get the machine within tolerance. So you ultimately may find that your measurement will end up closer to the machineā€™s measured distance.

Do you have a linkage kit, the ring, or the stock brackets? That also has a lot of an effect on calibration. Quadrilateral (stock) kinematics are a bit more tricky, and you have to tweak with the sled mounting points a little more to get the machine calibrated. Triangular (Linkages or ring) kinematics are quite a bit simpler, as you only have to tweak the rotation radius.

From what I understand, that isnā€™t currently implemented, but is under development. Discussion thread here. Thereā€™s a link to a Github pull request at the end of the thread which leads you to a more thorough description of the updated calibration routine, as well. Not sure if chain sag calculations are being included in todayā€™s update, but it would be sweet if it was! :smiley:

You and me both. o.O
I dove a bit into the wikipedia rabbit hole and Iā€™m not sure I actually absorbed anything. I now know a little bit more about the existance of hyperbolic functions, though! xD
Makes me really appreciate the amount of work being put into the chain sag calculations!

1 Like

Hi MM,
Thank you for that. Iā€™m using the triangluar kinematics for a ring that Iā€™ve made. Still moving things along in baby steps at the moment. Figure Iā€™ll get the motors and sled working with a Sharpie before I stick my router on it.
I share a workshop sometimes with an ex-sheet metal worker, and this has led to an unhealthy obsession with the vernier! So Iā€™m very particular about taking measurements and getting it right before moving on. In answer to you tape measure question, I duct taped it to the top bar of the frame so as not to have any sag.
Aim to get a few pictures up in the next day or so to better explain what Iā€™m working with (pic/1000 words, etc).
Will have a look at the Linked dev thread, but doubt Iā€™ll have much to offer seeing as my maths is better when there are numbers involved - I thought letters were for literature ;Ā¬)

Excellent, that makes calibration much easier! Because itā€™s a DIY ring, you may have to tweak the rotation radius to make sure itā€™s correct. If you donā€™t you may get the calibration right at the center of the machine, but it will be off as you go to the left and right side of the bed. Again, playing around in Ground Controlā€™s simulator should give you a good feel for what is happening when a variable is off.

Always a safe bet. Not that the router is especially unsafe, but itā€™s safer with a pen than a spinning, cutting bit. :wink:

I wouldnā€™t see that as ā€œunhealthyā€, I always try to get as accurate a measurement as possible. Thatā€™s actually one of the reasons for the 100mm square. I can easily measure it with a normal set of calipers! :smiley:

Sounds like weā€™re both pretty particular about getting it right! I really need to take a picture of some of my calibration blanks. I have entire scrap pieces covered in calibration squares from dialing my machine in. Sometimes Iā€™ll tweak a value just to see how it affects the cuts. At the very least, it gives me a better understanding of my machine. As good as the simulator is, sometimes there is no substitute for the real thing. Not trying to knock the simulator, though. It is a VERY useful tool for calibration.

Perfect! Based on what youā€™ve told me, it seems like the measurement you took with the tape is pretty reliable. I would use that as my starting point and run a couple rounds of calibration to dial it in from there.

1 Like

Thanks for the advice. The decision with the Sharpie was largely based on my anticipation of all sorts of CoG issues with my sled. Will attach the router later, try hanging the whole assembly and get back with some pics. Also need to clear a little garage space to enable a little more reclination of the frame.

Iā€™d be interested to see some of your calibration pics.

The vernier is my go to tool in the box, and youā€™re spot on with the accuracy thing, but I have caught myself worrying about nths of a mm when I should really of been at the finishing stage. However, when it comes to setting up machinery, I try to achieve a state of Zen aforehand. That way I donā€™t get too uptight if it takes a day to get things square! ;Ā¬)

2 Likes

the fact that you are seeing 4mm of error is enough to be noticable in the
results (see the new features in the simulator to model this error and see how
it affects things all over the sheet), itā€™s also enough to cause the calibration
to calculate incorrect chain sag and rotation radius values.

the chain sag compensation does not apply to the motor measurement.

I had used this calculator (imperial)
https://www.spaceagecontrol.com/calccabl.htm
(metric)
https://www.spaceagecontrol.com/calccabm.htm

the motors produce ~30 Kg-cm torque, and the sprocket radius 1.2cm, so
this is 245N of force from each motor.

even without doubling the force (due to the two motors pulling against each
other

245N of force, 3M (real distance), 0.13 Kg/M chain weight results in

cable = 3.00003048 m
cable sag = 0.00586 m
resulting measurement error = Ā±0.001016%

so are you seeing the chain sag ~6mm in the center? more? less?

test everything with the new version that was released today, it includes a lot
of fixes, but it also includes a test pattern that will let you see how far off
the machine is from perfect in practice. Thatā€™s the part that really matters.

1 Like

Actually, yes, this is a very good point. I was struggling trying to get my machine to be less than Ā± 6mm awhile back and the update to 1.00 (specifically the corrected value for the gearbox ratio) gave me much better accuracy. Iā€™m currently running at a little better than Ā± 0.3mm tolerance at the center of the machine. So this could be an indication of a larger problem.

Ask and you shall receive! We had an unseasonably warm day yesterday so I did some more work in the trailer last night. While I was out there, I snapped a picture of the calibration pieces I could find. I thought I had another one board somewhere but I think I already cut it up and threw it out. These panels are pretty much covered front and back:

This may just be a moot point, however, because I get the sense that the new calibration routine in GC is going to make my methods obsolete.

There are a couple of different styles of tests, although my most commonly used one is a 100mm square with a 90mm diameter circle inside it. When I want to speed up the process, I omit the circle. I also have a 300mm square that is good for checking macro tolerances.

If you look closely at the left panel, youā€™ll see some dadoes in it from parts that failed mid-cut. The chain jumped and I lost calibration. I tried to relocate the home position to make it work but I couldnā€™t align it well enough, so the panel became a calibration panel :stuck_out_tongue:

1 Like