Especially the ones we’re using they’re very inelastic
3 Likes
@dlang the belts start fully retracted to the sled, you pull them out to the anchor and
then it pulls them back in
There do not appear to be any limit switches, so is ‘pulling in’ terminated by a cessation of belt movement? Does this imply that tests such as belt breakage (which I recall bar mentioning somewhere) are suspended during calibration? (Not a criticism or complaint, just a question.)
Second, there was a mention of ‘zero’ belt length, but am I right in assuming that belt length is actually calculated from the center of the router spindle, to the center of the anchor pivot? If so, is it correct that the ‘homed’ length of a belt is the distance from the center of each arm’s point of rotation to its outer extent, plus the distance from the edge of the belt anchor block to the center of its pivot point?
NB- I’m undoubtedly mangling the names of components, if there’s a glossary somewhere I apologize.
@dlang There is also the Z axis to consider. If the belt anchors are all at the same level…
I hadn’t considered this. It seems this might be an advantage of a horizontal/tabletop configuration- since there is no need to apply a downforce, each of the anchor points can be planar with its spool.
@dlang This is not quite a valid assumption. you can have points inside the anchors
that are too close to the line between the anchors (your first point above)
Point taken. I was actually thinking of ‘safe’ in terms of crashing. (Not that I have ever crashed a CNC… However, at the very least I realize that there is an area around each anchor equal to at least the diameter of the sled that is inaccessible- or ‘unsafe’.
@dlang In theory, 3 anchor points should be enough to work.
One can essentially consider the classic Maslow as having three anchors, the third being the earth’s center of mass…
1 Like
I can think of a few example scenarios;
-
For a given stand, you need to be able to define the workspace. There has been discussions to the effect that anchor points a foot beyond the workspace is adequate, but eighteen inches (or more) is better- so the calibration can’t assume the size of the workspace based on the anchor dimensions.
-
Even after the workspace is defined, there may be a need to constrain the movement to a portion of the area. For your G code, you can do this, but for an autonomous, automatic calibration process this could be a problem.
-
As an edge case, the workspace might not be rectangular. Here is a potential real world case I have been considering- setting up an M4 horizontally over a circular or oval table. Yeah, that’s probably outside any typical use case, but it’s something that can be machined.
There are not limit switches, but there are current sensors on each of the servo motors so we can detect when the belt is fully retracted by watching for the current spike when it is at the limit.
You are exactly correct. I haven’t found a good way to describe that yet so sometimes I say “zero” length, but you are spot on. It is never truly zero.
1 Like
Having re-hung a Maslow sled multiple times- and realizing that being 6’4 makes it perhaps less unwieldy for me than some others- and also realizing that if anything the M4 is going to tend to have at least some of its anchor points detached and reconnected even more than a Maslow, I’m thinking about the process of extending and attaching the four belts.
For my Maslows, I would simply hang the sled above the middle of the workspace to get it out of the way, but with an M4, it would seem that it would be necessary to disconnect at least the bottom two belts.
(I’m going to note that like the overwhelming majority of the human race, I have never yet been in the presence of an M4, much less used one, and I am in no small part addressing this to the honored few who have (its creators), so like a Greek philosopher, I will eschew experimental data for thought experiments.)
For a horizontal configuration it seems dead simple- put it on the table, extend the belts, connect them to the anchors- voila!
For the more traditional easel configuration, are the belts extended one at a time? Do you connect one of the top anchors, extend one or both of the top belts enough to connect the second top anchor, then extend the bottom belts and connect them?
Digressing slightly, there was a statement, or at least what I took to be an implication, that once a stand is built and calibrated, it might be possible to save the calibration data and eliminate or reduce the need to re-calibrate every time the belts are re-attached.
Assuming that this is correct (and based on my track record, that’s not a given…) I would consider adding a cradle at the very bottom middle of the stand, well below the bottom material rest, to support the sled at a specific position. The reason for putting it at the bottom is that it can be low enough that the sled can travel the entire usable workspace without colliding with it. This would probably put it outside the effective operational range, being too low relative to the lower anchor points, but this wouldn’t be an issue for lifting it via the upper anchor points.
The position of this cradle would not be random- it would be very specifically defined after previously calibrating the M4 to the stand. After calibration, the sled would driven to this low position. Although it is outside the optimal envelope, its position near the middle of the long axis should provide good accuracy. This location is then saved as part of the saved calibration as a registration point. (Not an origin.) Then a hole is bored at this point, securing the sled with reasonable precision. With the sled immobilized at a known position, the cradle can be carefully attached to the stand to maintain the position of the sled and the Z axis retracted.
Now, when you want to disconnect the belts, you would send the sled to this rest position. Then each belt in turn would be slightly extended to provide enough slack to disconnect it and then be retracted. Then the sled could be simply picked up and put away. Remounting would essentially be the opposite- the sled is placed in the cradle, each belt would be extended exactly the necessary length to reconnect it, and again- voila!
But wait- there’s more.
This assumes that the calibration data is being stored, but ‘trust but verify’ is (almost) always a good approach. So, since we have placed the cradle in a known location, we have an initial reference against which to compare the belt lengths and thus a single-point validation of the calibration. Furthermore, since the cradle provides a mechanical stop, the lower belts could be tensioned against it and then the uppers. For an added level of verification, a bit (or pin) could be lowered into the hole to ensure registration.
To take this a step further, additional registration holes could be drilled and saved. An obvious place for these would be at each end of the bottom material rest, as this is both available and very close to the workspace. The sled could be sent to each registration point and verified against the router spindle to determine if recalibration is needed.
1 Like
Oof- you have mentioned more than once the stresses placed on parts. One of the systems I work with moves television cameras around- a lot of mass and momentum especially once you add teleprompters- and the thought of running anything into a hard end stop is terrifying.
I would put the rest position up top. With the frame leaning back the footprint
is smaller, and the sled up top is at eye level where you will see it and avoid
it while down low you will not see it and trip over it far more frequently.
David Lang
one of the things someone did with the original maslow was they got slices of
tree trunks and engraved on the face of them and sold the result.
David Lang
@dlang the belts start fully retracted to the sled, you pull them out to the anchor and
then it pulls them back inThere do not appear to be any limit switches, so is ‘pulling in’ terminated by a cessation of belt movement? Does this imply that tests such as belt breakage (which I recall bar mentioning somewhere) are suspended during calibration? (Not a criticism or complaint, just a question.)
Second, there was a mention of ‘zero’ belt length, but am I right in assuming that belt length is actually calculated from the center of the router spindle, to the center of the anchor pivot? If so, is it correct that the ‘homed’ length of a belt is the distance from the center of each arm’s point of rotation to its outer extent, plus the distance from the edge of the belt anchor block to the center of its pivot point.
without looking at the code, we can’t say if it homes to a fixed belt length, or
if it homes to a zero and adds an offset, either approach is valid
NB- I’m undoubtedly manging the names of components, if there’s a glossary somewhere I apologize.
@dlang There is also the Z axis to consider. If the belt anchors are all at the same level…
I hadn’t considered this. It seems this might be an advantage of a a
horizontal/tabletop configuration- since there is no neet to apply a
downforce, each of the anchor points can be planar with its spool.
with the old maslow keeping the chains parallel to the workpiece didn’t have
problems getting the bit to drive into the workpiece unless you had a dull bit
(well, in part because ‘driving into the workpiece’ was such an incredibly slow
process)
@dlang This is not quite a valid assumption. you can have points inside the anchors
that are too close to the line between the anchors (your first point above)Point taken. I was actually thinking of ‘safe’ in terms of crashing. (Not that
I have ever crashed a CNC… However, at the very least I realize that there
is an area around each anchor equal to at least the diameter of the sled that
is inaccessible- or ‘unsafe’.
not to mention places inside the anchors where you would be hanging in the air.
@dlang In theory, 3 anchor points should be enough to work.
One can essentially consider the classic Maslow as having three anchors, the third being the earth’s center of mass…
you could, but that third anchor is rather difficult to adjust the tension on
David Lang
1 Like
it all depends on how hard you are running it when you hit that hard stop, with
no tension on the belts, even spinning at max speed only takes a very low power
setting.
David Lang
From the storage standpoint, I completely agree- that’s where I’ve hung my Maslows’ sleds- out of the way. And as I said, this is all visualized in my head. But in order to machine out to the edge of the workspace, half the sled has to project beyond the workspace, (Slightly more for the bit to cross the edge.) and therefore nothing can project above the working surface for this distance all around the workspace. A single pin from which the Maslow sled hangs is well above this. My concern is that a cradle in which the sled could rest would be too low to provide the needed clearance.
The second issue is that if the cradle is high, it would be difficult or impossible for the belts to lift the sled up out of it. That might require the user to lift the sled away from the work surface to clear the cradle and then extend the cable sufficiently to lower it clear the cradle from below and then proceeding. From a cradle at the bottom, all four belts can be connected and then the sled can be lifted up into the workspace by the upper belts.
((furiously burying massive quantities of depleted uranium under my garage)) I’ll show him…
2 Likes
So it’s not the absolute torque/current, it’s a (sharp?) change in torque/current?
Plus we’re looking at a very different Em VeeSquared regime here.
even if you are running at 10% PWM, there will be a current spike (and movement
will stop) when you hit the limit, but it will be FAR less violent than if you
are at 100% PWM when you hit.
David Lang
1 Like
I did say ‘difficult’ not ‘impossible’
David Lang
1 Like
Any merits in adding some ‘soft’ limits (something that can be sensed), before hitting the ‘hard’ limits?
If so, I’d add ‘soft’ limits at both ends of the belts.
not sure how you would do that.
there is no sensor-switch on the belts, just the encoder telling you how much
the belt moved (you don’t know where you started)
the only feedback you have is the motor current which tells you that you hit the
hard stop.
David Lang
Perhaps something optical e.g. a certain colour painted on the belt at the ‘soft’ limit location. Use 2 different colours (or colors if you prefer) for the soft limits at each end. And of course the required sensor…
You obviously wouldn’t want something that ever wore off.
In the end, all I’m thinking of, is something that ‘identifies’ the ‘soft’ limit location on the belts themselves that can be very simply (and cheaply and reliably) sensed. Whether that’s something doable at the moment I obviously don’t know
I may have started something here- that was not my intent. I’m just a guy with strong opinions, a big mouth, and a little experience, some of it in systems that are vastly different than the M4 in scope, characteristics, and requirements, and I have zero firsthand experience with the M4. Nothing I say should be taken or even implied to be indicative or predictive of the M4.
2 Likes
Opinions are welcome here
David Lang
4 Likes