Thanks! Do you happen to remember what your measurements were for the calibration routine?
No, but I can measure again. Itâs past midnight and last working day of the week tomorrow.
2 days of Maslow time ahead.
2 Likes
If you would give the updated calibration process in 1.07 which we just released when you have a chance I would love to hear your feedback. One of the changes it makes might address the problem you are seeing.
(The change I am thinking of is that we got rid of using the chain to measure the y offset and instead you get to enter it manually which I think makes more sense)
Edit: There are also a couple other fixes in 1.07 that might be reliant
1 Like
the calibration should calculate what the yoffset is at the same time that itâs
calculating the rotation radius.
If you are using triangular kinematics, you should never have to enter the
yoffset.
I believe with 7" horizontal arms the rotation radius should be 7".
Iâm adding only 1 link, as the pin for the other is incompatible with my eyesight.
That link is on during calibration, so I think I donât have to add anything.
In the pic the ruler is resting on a 6mm shaft, so the rough eyeballing would suggest a rotation radius of 7" and 3mm.
P.S. The tilting has been tracked down. It is an unbalanced sled. My mount is to high, to there is pressure at the top of the sled, so the top is dragged behind. By mounting the vertical arms on the down side, I could partly make it better, but not good enough. Iâll either add few washers on the outside or take my mount down a bit. Boring videos of dry runs are on youtube.
Sorry no test patterns this weekend.
I want to have confidence in the release and with the outcome of calibration, before doing accuracy tests. 1.02,1.03, and 1.07 did not give me that.
2 Likes
sorry, 7" plus links, you are correct
remember that the 0 point is with a link straddling the 12 oâclock position, so that half link needs to be accounted for somewhere.
thinking about it more, that half link needs to be subtracted from the rest
so 7" + 1/4" (one link) - 1/8" (the half a link past noon), so you should have calibrated the rotation radius to be just shy of 181mm, what did the calibration routine find the rotation radius to be?
179.6 and 180 on 2 runs. Pretty close.
one other quick test you can try, can you put the machine a little closer to vertical (10 degrees instead of 15 degrees for example) and see if it makes the problem better or worse.
I am currently at 8° and still have a unbalanced sled with top-pressure. Going lower is not an option till Iâve lowered the mount.
Lowering weight reduces the error more then lowering the federate does. but do any eliminate the error and make a straight line? Or is it just straight enough.
It is worrying me why the target and position indicator do not part. Only at the last part the y-error is discovered and the sled is pulled up.
I am currently at 8° and still have a unbalanced sled with top-pressure.
Iâm not sure which axis you are saying you are unbalanced in. Z or Y?
Lowering weight reduces the error more then lowering the federate does. but do any eliminate the error and make a straight line? Or is it just straight enough.
once you lower things to the point that there isnât a motor maxing out (which is
what the warning message indicates), you should be good, not just âstraight
enoughâ, without the warning message, you have no way of knowning when you get
to tht point.
It is worrying me why the target and position indicator do not part.
That is a good point, are you sure you were zoomed in enough that you would have
seen them off from each other?
Only at the last part the y-error is discovered and the sled is pulled up.
No, the error is discovered much earlier, but the left motor is at fill power,
so it canât pull any harder. Itâs only when the right motor stops pulling that
itâs able to catch up. the move command keeps going until you are at the final
position, not just until you have moved for the time at the speed you set.
The right motor is feeding out, the left does not pull in enough. Close to the end the right motor is still feeding and the left has suddenly gained the strength to pull up? repeatable image with x in minus and x in plus.
With a mount to high, the sled tilts forward at the top. Specially when the sled is moving up, I can tip at the bottom and feel the gap.
1 Like
no, the left is always pulling in as hard as it can, but at the very end of the
cut the right motor stops and the left keeps going.
the âoverspeed warningâ goes off is a motor has been commanded to run at 100% of
itâs capability for any noticable amount of time (the theory being tht it should
be just below that for any sustained movement, only maxing out at the beginning
of a move)
that seems to be whatâs happening in your case, but we canât be 100% sure
without the warning in the code. If that warning was in the code and you are NOT
seeing it when you are having this hooking problem that indicates that we arenât
telling the motor to go as fast as it needs to.
If we do see it, than it means the motor is going as fast as it can and canât
keep up.
The first would indicate a significant problem in the PID tuning, the second
means weâre hitting the limits of the machine.
Itâs possible that something is preventing full power from getting to the motors
(weak power supply, too small wires, etc), so it would be worth putting a meter
on the left motor and make sure itâs seeing full voltage during this top cut.
Hey guys, @Gero pointed me to this thread based on a comment in another. These subjects tend to get intermingled.
My question is: what additional constraints limit the sled weight and face angle, and are we executing the appropriate tests to understand the constraints? I ask this because, even though the chain-sag correction frees up some capability, there are additional constraints outside of the ability to calibrate the machine. Maybe this has already been covered, and is really the outcome of this work.
Nonetheless, I would like to propose some constraints/considerations, and ask about what tests have been conducted to understand. Maybe there is just a lack of consolidation of the conclusions related to each of these.
- The ability of the sled to counter the force of the bit plunging into the surface. I mention this because I see this as the only reason to increase the surface angle (steeper); all other constraints would suggest steeper is better.
- The maximum chain angle relative to vertical (ability to cut the bottom corners), limited by friction between the work surface and the sled.
- The maximum weight that the motors can move at the top-middle position. The reason I ask mention this because, as a constraint, it is mostly not calibration related.
1 Like
I donât think any organized testing has been done on this. Some thoughts:
Friction isnât constant, it varies with the material under the sled, the material of the sled face and the amount of sawdust or chips between the sled and the material, as well as weight and angle. Measuring this value isnât rocket science, but the chains would probably need to be slackened to get a reading. It would be nice if the measuring tool were inexpensive, verifiable and readily available worldwide (not asking too much here 
) so that users everywhere could contribute - a pull-type spring scale?
The force pushing the bit into the material needs to counteract the force of the the bit trying to climb out not just during the plunge but while cutting as well. That force is related to feed rate, cut depth, bit size and sharpness, and the material being cut. Maybe we could avoid these bit-related variables by testing without cutting.
The âtop middle positionâ is relative to the geometry of the frame, so is pretty variable. The effect of weight might be best measured at âstandard positionsâ below the level of the motor shafts as that will be replicable from frame to frame. Note that Iâm avoiding talking about âthe centerâ, which can be ambiguously defined. The distance between the motors is a factor as well, so perhaps those âstandard positionsâ should be âstandard chain anglesâ instead. That would mean that a standard .nc file probably wouldnât do without recalculating for the testerâs own frame geometry. Feedrate, another factor. Cut path characteristics (angle/direction/curve) too.
Of all these, coming up with values that relate frame angle, sled weight, feedrate, and âchain angleâ seem doable. Friction and plunge force are harder to measure and are much more variable, so harder to include. Removing those there are still so many variables left, though, that it seems more like fine tuning than calibrating.
1 Like
I donât think any organized testing has been done on this.
it has not, itâs needed.
Some thoughts:
Friction isnât constant, it varies with the material under the sled, the
material of the sled face and the amount of sawdust or chips between the sled
and the material, as well as weight and angle. Measuring this value isnât
rocket science, but the chains would probably need to be slackened to get a
reading. It would be nice if the measuring tool were inexpensive, verifiable
and readily available worldwide (not asking too much here) so that users
everywhere could contribute - a pull-type spring scale?
I donât think itâs worth trying to measure friction because it is so variable.
The force pushing the bit into the material needs to counteract the force of
the the bit trying to climb out not just during the plunge but while cutting
as well. That force is related to feed rate, cut depth, bit size and
sharpness, and the material being cut. Maybe we could avoid these bit-related
variables by testing without cutting.
but at that point we arenât testing what the machine can do.
The âtop middle positionâ is relative to the geometry of the frame, so is
pretty variable.
not really, the question is how high in the center can you go.
The effect of weight might be best measured at âstandard
positionsâ below the level of the motor shafts as that will be replicable from
frame to frame. Note that Iâm avoiding talking about âthe centerâ, which can
be ambiguously defined. The distance between the motors is a factor as well,
so perhaps those âstandard positionsâ should be âstandard chain anglesâ
instead. That would mean that a standard .nc file probably wouldnât do without
recalculating for the testerâs own frame geometry. Feedrate, another factor.
Cut path characteristics (angle/direction/curve) too.
testing lower on the frame doesnât find the limit. âcenterâ is very easy to find
(both chains the same length)
we donât really care about some arbitrary distance down from the motors or
arbitrary chain angle, we care âwith this machine, can you cut at the top of the
workpieceâ
Of all these, coming up with values that relate frame angle, sled weight,
feedrate, and âchain angleâ seem doable. Friction and plunge force are harder
to measure and are much more variable, so harder to include. Removing those
there are still so many variables left, though, that it seems more like fine
tuning than calibrating.
we donât care about most of these values. We care about how well the machine
works at different angles and sled weights. And that doesnât require all these
separate things, just testing some fairly standard patterns at different angles
and weights.
David Lang
I thought the idea was to characterize the relationships in general. âWhat is the feedrate limit at various given chain anglesâ might be a relationship that is nearly universal across various frame geometries. Where do these cross boundaries? As the limits seem most restricted centered between the motors, that would help reduce the number of tests needed. It would help one decide how to lay out parts on the stock, or whether to choose different feed rates at different areas. Who knows, the firmware might implement a zone approach to feedrate limits. The relationships between frame angle, sled weight, feedrate and cut location might be useful to investigate. They might suggest changes in frame geometry that would improve performance.
Characterizing an indivual machine for âcan you cut at the top âŚâ is as simple as moving the sled and trying a cut at various speeds until the result is satisfactory. Knitters knit a âTest swatchâ, woodworkers make âTest cutsâ. Finding what works is part of learning the tools. Having an idea of the General relationship between these limits could help to find that.
I thought the idea was to characterize the relationships in general. âWhat is
the feedrate limit at various given chain anglesâ might be a relationship that
is nearly universal across various frame geometries. Where do these cross
boundaries? As the limits seem most restricted centered between the motors,
that would help reduce the number of tests needed. It would help one decide
how to lay out parts on the stock, or whether to choose different feed rates
at different areas. Who knows, the firmware might implement a zone approach to
feedrate limits. The relationships between frame angle, sled weight, feedrate
and cut location might be useful to investigate. They might suggest changes in
frame geometry that would improve performance.
there are different limits to feed rates depending on your direction and
location.
in the top center you have the highest chain tension, and the highest load on
the motors.
in the bottom corners you have the lowest chain tension, so if you are moving
away from the motor you are limited to the speed that gravity moves the sled,
but if you are moving towards the motor, you can move at the max speed of the
motor.
The request was to find out how the machine performs when set closer to vertical
and with less weight on the sled (we know that we are close to the max weight
that the motors can handle for the cuts along the top, because people have been
hitting the limit when they have heavier than normal sleds)
Characterizing an indivual machine for âcan you cut at the top âŚâ is as
simple as moving the sled and trying a cut at various speeds until the result
is satisfactory. Knitters knit a âTest swatchâ, woodworkers make âTest cutsâ.
Finding what works is part of learning the tools. Having an idea of the
General relationship between these limits could help to find that.
right now we need to find out some limits of specific machines before we can
start generalizing.
David Lang
Did you get a chance to test at different weights yet? My sled is by default 15lbâs so would love to be able to not use any weights if possible.
I have not added weight to my sled, nor cut anything yet, so take this with a grain of salt, butâŚ
my understanding is that the sled needs to have a weight distribution such that the CG is below the bit to ensure that everything tracks well as the sled moves. I remember that there was a posting by @blurfl which indicated that a symmetrical weight distribution caused problems.
4 Likes