Can we switch to a linkage arrangement with the next shipment of maslows?

“In theory, there is no difference between theory and practice. But, in practice, there is.”

The pendulum effect means ovals instead of circles, tabs that don’t fit slots, corners that aren’t square. No automatic calibration will correct that.

The calibration routine has become good enough to achieve the target accuracy over a reasonably large portion of the build area, and that area increases with refinement. Triangular Kinematics is simpler, but in the current forms physically harder to implement.

The circular linkage brings something that the current linkage kits lack, a complete mounting arrangement. If it is manufactured accurately, measurements made on the plane of the sled will correctly align the whole unit, independent of the tools available or skill of the user. It looks like installation will be a matter of assembly rather than construction. In this, it takes the next step beyond a ‘proof-of-concept’ device. The chain attachments provided with @pillageTHENburn’s kit is another example of that kind of refinement, making it a simple to use as the ‘traditional’ arrangement.
The triangular Kinematics approach is still developing, the calibration is changing and improving. It’s still ‘beta’, though. Maybe an ‘advanced beta’, not yet v1.0

Going to try to stay on-topic here as opposed to my last attempt to respond to a thread…

While I do think that some form of the linkages should come with the machine, but I agreed with @blurfl that it’s still too early to implement it. I completely agree with @dlang when he says that the linkage is at 90+ out of 100, whereas the stock setup is more like a 30. However, the linkages still need some refinements, such as the mounting pattern, before they are good enough to become the “stock” setup.

I think the mounting issue that’s being discussed can be solved if we provide an SVG of the sled with holes for the risers. That way, once the temporary sled is dialed in “well enough”, the operator can cut out a sled with the needed holes cut relatively accurately. I know from my own experience that letting the machine handle locating the linkages helps greatly.

The other issue we have is that currently there are 3 different versions of triangular kinematics and no clear “best” version. Each have the advantages and drawbacks. I think we need to test each of them head-to-head to see which one would be best for the stock version. I have David’s top-mount setup on my machine, and I’m still having issues getting the level of accuracy I’m looking for across the entire bed. I have @pillageTHENburn’s 45 kit on order and I plan to test the two to see what differences there are between the two. I probably should also get @bar’s ring setup and test that against the linkages. I think it’s crucial to test all three versions on the same machine so that we can get controlled results.

Once we have a comparison between each of them setups, then we can decide which one should make it to the stock version.

5 Likes

While I do think that some form of the linkages should come with the machine,
but I agreed with @blurfl that it’s still too early to implement it. I
completely agree with @dlang when he says that the linkage is at 90+ out of
100, whereas the stock setup is more like a 30. However, the linkages still
need some refinements, such as the mounting pattern, before they are good
enough to become the “stock” setup.

then the question is what we would need to do to get to the point of making one
of them the stock setup.

I think the mounting issue that’s being discussed can be solved if we provide
an SVG of the sled with holes for the risers. That way, once the temporary
sled is dialed in “well enough”, the operator can cut out a sled with the
needed holes cut relatively accurately. I know from my own experience that
letting the machine handle locating the linkages helps greatly.

This is one place where I think the 45 kit is behind, it requires more mounts in
precise locations. It’s hard to be precise before you have an accurate machine
to work with (although as I noted earlier, you can drill the holes based on the
linkages)

the top mount kit requires only two holes in the riser, vertically in line with
the bit, so it’s easier to mount

the ring kit can be centered with a printed circle jig (as long as that jig is
not distorted too much when it’s printed)

The other issue we have is that currently there are 3 different versions of
triangular kinematics and no clear “best” version. Each have the advantages
and drawbacks. I think we need to test each of them head-to-head to see which
one would be best for the stock version. I have David’s top-mount setup on my
machine, and I’m still having issues getting the level of accuracy I’m looking
for across the entire bed. I have @pillageTHENburn’s 45 kit on order and I
plan to test the two to see what differences there are between the two. I
probably should also get @bar’s ring setup and test that against the linkages.
I think it’s crucial to test all three versions on the same machine so that we
can get controlled results.

I would LOVE to see direct comparisons between the different options.

Once we have a comparison between each of them setups, then we can decide
which one should make it to the stock version.

What do we need to do to make this happen? can we do it fast enough to not force
the next batch of users to have to buy something extra or suffer with the stock
kinematics?

“In theory, there is no difference between theory and practice. But, in practice, there is.”

I use that quote a lot, but here I will say that things are bad in theory, but
in practice there is not much different between the results of the different
options :slight_smile:

The pendulum effect means ovals instead of circles, tabs that don’t fit slots, corners that aren’t square. No automatic calibration will correct that.

it depends on the size of the pendulum effect. The maslow is moving fairly
slowly (and can be run even slower if you want to have it cut a more accurate
set of risers for example), and we are not talking about several inches between
the pivot point and the bit, but rather a few mm if someone is particularly
sloppy.

The calibration routine has become good enough to achieve the target accuracy
over a reasonably large portion of the build area, and that area increases
with refinement. Triangular Kinematics is simpler, but in the current forms
physically harder to implement.

I disagree that the calibration routine works well enough with the stock
kinematics

The circular linkage brings something that the current linkage kits lack, a
complete mounting arrangement. If it is manufactured accurately, measurements
made on the plane of the sled will correctly align the whole unit, independent
of the tools available or skill of the user. It looks like installation will
be a matter of assembly rather than construction. In this, it takes the next
step beyond a ‘proof-of-concept’ device.

so if the top kit had one more pre-made piece that the links connected to at the
center, and attached to a metal bracket, that would satisfy you? It would be
pretty easy to add one more piece that also had a couple holes in it to attach
to brackets like the Z axis uses. I didn’t bother to do that because a 2x4 on
edge seemed ‘good enough’ and saved a few bucks. but it could be done pretty
easily if needed.

The chain attachments provided with
@pillageTHENburn’s kit is another example of that kind of refinement, making
it a simple to use as the ‘traditional’ arrangement.

The triangular Kinematics approach is still developing, the calibration is
changing and improving. It’s still ‘beta’, though. Maybe an ‘advanced beta’,
not yet v1.0

so what do we need to do to get there?

Yes!

Not to muddy the waters further by throwing a third option into the fray but the goal is to ship the next batch with the ring as the attachment system. I fully agree that the rotation mechanism is a huge step forwards and I want to get it out there.

I like the ring because it is dead simple to mount. All you need is the included allen wrench and a philips screwdriver to attach it to the sled. A moderate amount of care is needed to keep it centered around the router but a paper template makes that pretty simple.

With the ring included in the kit with an exactly known radius with no room for user installation error the ONLY unknown variable will be the motor spacing which means there will be only ONE variable to adjust in the calibration process which sounds like a total dream :sunny:

The ring is also simple to adjust vertically. It is also very easy to have a few rings made of a larger size to accommodate the bigger routers out there.

I haven’t announced it yet because I’m still sourcing all the parts so something could go wrong still. We have enough of the old brackets left that if anyone wanted the old system we could easily give them the brackets, but I don’t know why anyone would. The goal is that the next batch will ship with the ring in the box.

13 Likes

Holy damn that’s awesome! I hope that in my previous post it did not sound like I was ruling out the ring approach. It has some possibilities to actually outdo the linkages as far as accuracy. I know that the early versions it skipped as it moved and Logan had mentioned in the triangular kinematics development thread that the linkages work better with the forces available. However, it doesn’t have the same opportunities for compounding errors that the linkages have if there is any play in their connections. I know that is an issue for the top mount, not sure about the 45 degree version.

Which brings me to:

While what we have right now for linkages might actually be perfectly fine, I err on the side of caution. We really need to do more tests and trials to see how well each system works and where they are going to encounter issues. What if we sent out one of the linkage systems with the stock machines and they have a critical flaw with compounding errors or wear? Even if an issue doesn’t come up during regular use, every machine is going to be pushed to the extreme at some point, and that could trigger a bigger problem. I deal with these kind of issues on the shop floor every week when we make a system change and it has unintended consequences. Are we expecting that all these new Maslow operators are now going to have to deal with these problems?

That being said, in erring in favor of caution, this does slow down the development process. The last thing we need is to become a bureaucratic mess and get in our own way. So maybe the correct way to move forward is to ship the ring as is Bar’s goal, AND send out the original brackets so that the operators have the option to to back to quadrilateral if issues do come up. I know you say that using the original kinematics is “suffering”, but I ran my machine for a few months using that setup and it really is good enough to get acceptable results.

I think there is a bit of misconception with the stock kinematics being “inaccurate”. Yes, quadrilateral will begin to distort at the left, right, and top dead center of the machine. But the stock sled is only 18" in diameter. I was able to cut out a sled for the top mount linkages WITH the mounting holes using a stock sled, and the hole placement is exactly where it needed to be. When I checked the hole placement with the linkage arms, in the same manner that you had suggested to locate the holes on an existing sled, it was dead nuts. I think that any operator going through the build process should be able to get the same results.

In my previous post, I had said that the temporary sled could be used to cut the linkage sled. I misspoke there. I do not like the temporary sled. In the brief amount of time I used it, it was unstable, unpredictable, and a pain to use. I would agree with you there in saying that using the temporary sled for any length of time is suffering. I would like to correct my previous statement and instead recommend that the 18" round sled with the stock mounting solution could be used to cut that SVG file. I would edit my previous post, but I think that anyone reading through this thread might loose a bit of context in our discussion.

This would certainly bring it more to the same level of “completion” that Bar’s ring solution has. I still think that an SVG file would be the best way to move forward, but giving the user the ability to easily modify an existing sled. As @Blurfl said, then it becomes more assembly and less construction.

Me too! :smiley: It’s on my to-do list, along with feedrate tests, a proper z-axis, implementing M codes, and far too many other items. And this is exactly your argument. We are relying on a group of people to do these developments at a hobby level, which means that work, family, health, etc are always going to get in the way and slow the process. Any chance @bar will hire us to do these developments as our 9-5? :stuck_out_tongue:

Ultimately, however, I do agree with you David. The linkages/ring should be the stock option for the machine. The issue that I, and I believe @blurfl are arguing is how do we go about effectively implementing it in a way that doesn’t negatively affect people coming into this?

3 Likes

what size holes are needed to use the brackets that the Z axis uses? I’m adding
a bar that would use them to the top mount kit

As a placeholder, I’m putting two 1/8" holes 1" apart, both in the vertical bars
and in the new bar (so that you can locate those holes with one of the vertical
bars)

The next issue would be getting that bar straight. I think doing it without the
router in place and just folding the linkage to it’s narrowest point would work
pretty well.

I think the Z axis type holders are going to have enough slop in them that I
would still recommend just using a 2x4, but if that’s deemed too complicated or
something, this will work.

David Lang

1 Like

While what we have right now for linkages might actually be perfectly fine, I
err on the side of caution. We really need to do more tests and trials to see
how well each system works and where they are going to encounter issues. What
if we sent out one of the linkage systems with the stock machines and they
have a critical flaw with compounding errors or wear? Even if an issue doesn’t
come up during regular use, every machine is going to be pushed to the extreme
at some point, and that could trigger a bigger problem. I deal with these kind
of issues on the shop floor every week when we make a system change and it has
unintended consequences. Are we expecting that all these new Maslow operators
are now going to have to deal with these problems?

That being said, in erring in favor of caution, this does slow down the
development process. The last thing we need is to become a bureaucratic mess
and get in our own way. So maybe the correct way to move forward is to ship
the ring as is Bar’s goal, AND send out the original brackets so that the
operators have the option to to back to quadrilateral if issues do come up. I
know you say that using the original kinematics is “suffering”, but I ran my
machine for a few months using that setup and it really is good enough to get
acceptable results.

It all depends on what you call acceptable results? IMHO, fussing about the
possible compounding errors in the linkage kits while ignoring the far more
serious errors that we have no way of accounting for in the original kinematics
is massivly misunderstanding the problem.

I think there is a bit of misconception with the stock kinematics being
"inaccurate". Yes, quadrilateral will begin to distort at the left, right, and
top dead center of the machine. But the stock sled is only 18" in diameter. I
was able to cut out a sled for the top mount linkages WITH the mounting holes
using a stock sled, and the hole placement is exactly where it needed to be.
When I checked the hole placement with the linkage arms, in the same manner
that you had suggested to locate the holes on an existing sled, it was dead
nuts. I think that any operator going through the build process should be able
to get the same results.

moving vertically in the center works well with the quadrilateral kinematics,
but with so many variables that distort things in different ways (and
calibration doesn’t even try to address most of these, it just changes one of
them until it produces ‘good enough’ results in the one location)

This would certainly bring it more to the same level of “completion” that
Bar’s ring solution has. I still think that an SVG file would be the best way
to move forward, but giving the user the ability to easily modify an existing
sled. As @Blurfl said, then it becomes more assembly and less construction.

I can produce a pdf file that (if printed without distortion) gives you the
right placement of the linkages on a 2x4 pretty easily, I’ll do one tonight.

svg files have the problem of scaling, but it’s pretty trivial to convert from
one vector format to another.

Ultimately, however, I do agree with you David. The linkages/ring should be
the stock option for the machine. The issue that I, and I believe @blurfl are
arguing is how do we go about effectively implementing it in a way that
doesn’t negatively affect people coming into this?

Ok, it was sounding as if you were arguing that because the linkage kits are
’only’ 90+ on a scale from 1-100 that we are better off shipping a few hundred
more kits that limit people at 30 on that same scale.

If it’s just a matter of trying to improve the existing kits, fine, let’s work
on improving them.

David Lang

Here is a vector sled drilling template (which also has a scale to check for distortion/accuracy) and here is an SVG of the same file.

I feel like adding much of my opinion to this thread could end up being a “too many cooks in the kitchen” scenario, so I’ll mostly hold off for now.

Except for this:

I agree completely and I REALLY want to see that happen! (btw, @MeticulousMaynard your kit will be in the mail shortly!) I feel like Bryan is the right guy for the job too; I’ve witnessed fantastic data collection on his part. I know it appears that the ring will be the best choice to ship (and I’m not arguing there) but I’d still love to see a comparison just for curiosity’s sake! Data is one thing but ease of use is also important.

I even have hypotheses about the results! But I can save that for another time :wink:
-Logan

6 Likes

That’s good news.

The riser arrangement needs to provide stability in the X and Y directions. Also give some thought to a quick-release for the chains. The other two linkages provide this, and it’s necessary, not just desirable.

We all will. We should start a list of measurements and observations for the testing. Volunteer?

2 Likes

The riser arrangement needs to provide stability in the X and Y directions.

It’s really hard to beat the stability of a 2x4 :slight_smile:

especially with something that’s supposed to be adjustable height.

how critical is it that this be a metal adjustable part vs a printed template
that you could put on the end of the wood to mark the holes to drill?

Also give some thought to a quick-release for the chains. The other two
linkages provide this, and it’s necessary, not just desirable.

what about a link like this:

https://www.mscdirect.com/product/details/58645474

possibly with a pin similar to what ships with the stock machine instead of the
small pin?

1 Like

Adjustability means the system is going to have to be more complicated to have the same stability as a static one. I think that the template is at least the easier route to go for now.

I like that a lot. I have had to take my sled on and off the chains a lot as I’ve been getting things dialed in, so having that link would help speed up the process significantly. I’ve had to spend quite a bit of time looking for the master link parts when I either dropped them or they flew off when I released them. :stuck_out_tongue: Entirely my own fault for not being careful enough.

I think I inadvertently did when I mentioned I was already planning on testing the two linkage systems together. I’m fine with that, but it will take time for me to do it right. The other thing that that my machine is smaller than the stock version, so results will be more extreme at the far sides of the bed. It may be a good idea to have a couple of people try this, especially with all the frame variants that are being made. I would be curious to see how much that also plays a role in the accuracy we see across the bed.

What do we want to see in the tests? Would something like I covered in this thread be good? I checked to see how well accuracy in a small shape in regular intervals across the bed. If I’m missing something crucial, though, we can work out a good test for it. I know I would like to have some way to measure compounding errors (aka slop) for each version. Should I also test the stock kinematics so that we have a control?

@Bar, I should probably go ahead an order one of your ring setups as well :wink:

Admittedly, part of my hesitation about adding this as part of the stock package is that I am still working out the issues with my own setup. My observations so far could be based in my own poor hardware choices. Or I haven’t dialed in the calibration as much as I think I have. I would absolutely LOVE to be wrong here, because then I would have more time for making cool stuff :wink:

1 Like

by the way, why are you having to take the chains off so much?

I’ve been very impressed by everyone on this forum so far. You have all done and are doing amazing things to make the Maslow an even better machine. I perhaps can’t speak for everyone but I’d love to get the best, easiest and most accurate set up possible when the next batch ships and if that means waiting an extra month to get the last final details figured out, I’d be happy to wait. The comment about a possible ring shipping is exciting (since it has such potential to improve things).

There have been some huge strides made by the community and I’m sure this will continue to get even better. I bought in knowing it was in beta phase and am excited to see what happens and to hopefully be able to join in the improvements when my Maslow arrives.

4 Likes

@Jayster

You too are part of what makes this an amazing community. Your desire to dig in and be part of what improves the Maslow.

Thank you for your participation.

3 Likes

For awhile I was having trouble with the chains jumping so I would need to re-calibrate the chains quite frequently. I also had to change the hardware configuration a few times, but that’s more my own fault than it is a problem with the system overall. The quick-release is something that is offered by the other kits, so I believe that is an expectation from our operators. I can imagine that taking the sled on and off could also wear out the master link components faster.

We can’t really say this enough. The involvement of the community is what makes this such a great project to be involved in! :smiley: We’re lucky to have you, @Jayster!

3 Likes

I take the chains off of the sled (but not off of the sprockets) every time I load a new sheet of plywood. I don’t need to recalibrate, but I do detach them from the sled and re-attach them with the cotter pins. Does everyone else load plywood without taking the sled off?

3 Likes

I tried once, ended up scrapping the finished side of my plywood. Now I de-chain every time.

2 Likes

I also think we should move to a top beam approach for the next shipment, the
improvement is ridgidity is significant.

If you have a top beam, then you can have a hook there and just hook the sled to
the top beam when loading the plywood, faster and easier for all sled types. :slight_smile:

David Lang

5 Likes

I have a rope with a hook to suspend my sled above the height of the sheet, but with testing various things often go through chain calibration. Getting the sled out of the way, either by hanging or disconnecting happens several times a day.