Triangular Linkage Evaluation Criteria and Measurements

[MeticulousMaynard]

@dlang

having to wait an entire week

Sorry to make you all wait, life got in the way last weekend. The good news is that my family is going in on a property together, and me and my brothers will get our shop! The sale closes at the beginning of April, so the Maslow will soon get a more suitable home!

I was able to do the tests for the top mount yesterday. I spent about the same amount of time calibrating the machine. I got these results from the auto calibration process:

I’m scratching my head a little as to why the chain sag value changed. Maybe it’s because the rotation radius is smaller, so the chain is a little longer than it would be for the 45?

I had a couple of issues with the linkages, but I was still able to complete the tests. The results aren’t as good as the 45. There is quite a bit more distortion that happens with the top mount. The X-Y tolerances aren’t atrocious, but I had much more trouble with square tolerance and positional tolerance. The top mount I used also scored a 2 : 0.75 on the benchmark test, so it’s a little worse than what I got with the 45 degree.

The chain slipped just before the machine cut square 4, so that particular square is certainly an outlier. I was able to reset the chains and get consistent results for the rest of the test.

and for comparison here is the 45 degree’s results (so you can see them side-by-side):

I would be very curious to see the difference in the square and positional tolerances with the new version of the top mount. I think that the tighter tolerances on the holes will mean that the linkages will rack less when they’re at their most tension. @pillageTHENburn and I investigated just how much play there was in the linkage assembly a little while back. I measured between centers for the top bolts and where the chains attach to the linkages.

Well, really, I measured to the outside of the bolts for the top and to the outside of the pins, then subtracted the diameter of the pins/bolts to get the actual between centers distances. I used my 12" calipers to get as accurate a result as possible. You can see my math in the table below:

The difference in the two distances gets very extreme at the extreme ends of the bed. In some places, it’s almost an inch difference between the top distance and the bottom distance.

I also encountered an issue with the linkages when the chains get close to vertical. Once it gets past a certain piont, the linkage bars slip on each other and throw off the cut.

This may simply have been because my top beam is shorter than most, so chain is more vertical than it would be with a “stock” top beam. It started happening on the outer two rows of squares, I would have to work within 600mm (~24") of the centerline of the bed to avoid this issue. However, I was able to cut out to this area with no trouble with the 45 degree linkage kit, so I believe the system works more reliably across the entire bed.

So my conclusions on the comparison between systems is that I think the 45 degree is a better system all around. If the tighter hole clearances mean that there is less error in the linkages than it should have similar performance to the 45. However, design considerations will need to be made to make sure the linkages don’t slip.

As a last little bonus, here are the summaries of the torture tests I’ve done so far:

And here is the current version of my test spreadsheet in case anyone wants to stare at a wall of numbers:
BedAccuracyTest.xlsx (42.3 KB)

[dlang]

I will get you a new version of the top mount kit for comparison

[dlang]

for the linkage to fold like it did, one of three things has to have happened

1. the center holes are just slightly too close to each other (easy to check by
   comparing their distance with the holes in the verticals)

2. the bolts flexed slightly (temporarily making them too close to each other)

3. the slop in the holes give just enough play for them to clear

The first two are taken care of by the new center piece (I just ordered another
run of kits and ordered a bunch of extra center pieces, so if it just turns out
to be the first two, I can start retrofitting the people who have this first
version of the kit)

if it’s the third, the new kit should do much better.

While I’m disappointed at the poor showing, I’m very happy you tested it

when you get the new kit, can you try just using the center support with the old
linkage first? that way we can tell how good we can get if I just send the new
center supports out to everyone, and then test the new kit and see what it’s
accuracy ends up as.

David Lang

[MeticulousMaynard]

@dlang

for the linkage to fold like it did, one of three things has to have happened:

1. the center holes are just slightly too close to each other (easy to check by
   comparing their distance with the holes in the verticals)

2. the bolts flexed slightly (temporarily making them too close to each other)

3. the slop in the holes give just enough play for them to clear

This makes sense to me. I did check the distance between centers with one of the verticals when I was setting the sled up. However, I could see this being a combination of 2 and 3. I could see there being an opportunity for the bolts to have flexed. I didn’t want to torque down the center nylocks to the point where there would be no motion in the assembly because then it could make the linkages “stick”.

@dlang

The first two are taken care of by the new center piece (I just ordered another run of kits and ordered a bunch of extra center pieces, so if it just turns out to be the first two, I can start retrofitting the people who have this first version of the kit)

if it’s the third, the new kit should do much better.

I’m glad to hear you found a new manufacturer for the parts. Were they able to do common line cutting to save on cut time?

Yes, I think that center piece would solve the issue of the bolts flexing asymmetrically. I can add the center link into the existing linkage assembly to see how much that will help with the linkage folding. I was actually more than a little tempted to machine my own center support. I have some 1/4" aluminum sheet that would have worked well enough for testing. If I still am having trouble because there’s enough slop in the holes, the new kit should show significant improvement, both in terms of the folding and in terms of the distortion I was seeing.

@dlang

While I’m disappointed at the poor showing, I’m very happy you tested it

This is part of development. Personally, I still like the top mount as a design. It has a smaller footprint on the sled than the 45 degree one, which means larger routers fit better. However, if the 45 degree still outperforms it, than I think that it stands to reason that it makes a better contender for the stock setup. A larger version, which uses 8" instead of 4" could allow people to use it with larger routers.

The linkage fold was exactly the sort of thing I was worried about when I said:

@MeticulousMaynard

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.

back in the Can we switch to a linkage arrangement with the next shipment of maslows? - #14 by MeticulousMaynard thread. It’s not to say that the top mount won’t be a good option. I’m really curious to see what the results are from the improved version.

[dlang]

a couple more thoughts

when you first got the top mount kit, you didn’t have the jumping behavior, that
may have been that the mounts you had didn’t let it get that close to vertical,
but it could also have been that the mounts you had held the center bolts better

you mentioned that the nylock nuts had to be a little loose to let things move,
which would also let the bolts move with the spacers.

try putting nuts under the linkage (betwen the spacers and the linkage) and
crank them down tight.

also try making a wood spacer that’s 3/4" wide or so instead of the ~2" wide
spacers you tried the first time.

[dlang]

Another test I’d like to see you do is to go to the outer edge and see how far
you can go to (and off) the side before the sled starts tilting.

do we really need a skirt off the side to machine all the way to the edge.

David Lang

[MeticulousMaynard]

@dlang

when you first got the top mount kit, you didn’t have the jumping behavior

I saw it once or twice before, but it was usually under extreme circumstances. I thought it was just me doing something dumb. This is happening regularly now.

One other difference between before and now may also be the HDPE pad on the bottom of the sled. It’s possible that the earlier plywood sled had enough friction that the sled couldn’t rotate to a position where the jumping could occur. Just a thought, I’ve been wrong before.

@dlang

that may have been that the mounts you had didn’t let it get that close to vertical,
but it could also have been that the mounts you had held the center bolts better

That could entirely be the case. When I first set up the top mount, my plywood risers were almost flush to the bottom of the linkages. In the test, I used 1" long derlin spacers to make sure I had enough clearance for the bolts on the backside of the linkages. This does mean that the bolts have more play than they did before.

I will try to use spacers much more similar to what I used with the 45 mount. For that, I cut two 3/4" plywood circles, with a hole drilled on center sized to the bolts, the other with a hole on center sized to the derlin spacer. That way, the derlin spacer has more support and is less likely to flex under operating conditions. Also, that should help make the results a little more consistent.

@dlang

you mentioned that the nylock nuts had to be a little loose to let things move, which would also let the bolts move with the spacers. try putting nuts under the linkage (betwen the spacers and the linkage) and crank them down tight.

To make sure I understand you correctly, this is a drawing (section) of the riser/spacer assembly now:

And here is what you’re proposing (also adding in the plywood spacer like I was talking about above):

The only issue I can see with this is needing a fully threaded bolt (or threaded rod) to get the ~2" of thread on the end. Or I use partially threaded ones and use a die to cut more thread into it.

@dlang

Another test I’d like to see you do is to go to the outer edge and see how far you can go to (and off) the side before the sled starts tilting.

do we really need a skirt off the side to machine all the way to the edge.

We’ll have to see how far over I can get the sled before the chain is too vertical to move it further. I only have about 6" of top beam beyond the 8’ bed. I’ve never tried to move the sled all the way to the edge.

I’ll give it a go while I’m testing everything else.

[dlang]

@MeticulousMaynard

That could entirely be the case. When I first set up the top mount, my plywood risers were almost flush to the bottom of the linkages. In the test, I used 1" long derlin spacers to make sure I had enough clearance for the bolts on the backside of the linkages. This does mean that the bolts have more play than they did before.

I will try to use spacers much more similar to what I used with the 45 mount. For that, I cut two 3/4" plywood circles, with a hole drilled on center sized to the bolts, the other with a hole on center sized to the derlin spacer. That way, the derlin spacer has more support and is less likely to flex under operating conditions. Also, that should help make the results a little more consistent.

@dlang

you mentioned that the nylock nuts had to be a little loose to let things move, which would also let the bolts move with the spacers. try putting nuts under the linkage (betwen the spacers and the linkage) and crank them down tight.

To make sure I understand you correctly, this is a drawing (section) of the riser/spacer assembly now:

I’m suggesting two things

first, add a nut (doesn’t even need to be a nylock nut, it can be a slim ‘jam nut’ which may fit on the thread), where you show it, without the plywood spacer support that you show. See what happens (you don’t need to do calibration, just see how easy it is to get the linkage to ‘snap’

second, go back to the original all-plywood riser/spacer approach, just make the top layer be 3/4" wide (you may still have to notch it slightly to let the washers clear). NoDerlin spacer involved at all

@MeticulousMaynard

@dlang

Another test I’d like to see you do is to go to the outer edge and see how far you can go to (and off) the side before the sled starts tilting.

do we really need a skirt off the side to machine all the way to the edge.

We’ll have to see how far over I can get the sled before the chain is too vertical to move it further. I only have about 6" of top beam beyond the 8’ bed. I’ve never tried to move the sled all the way to the edge.

I’ll give it a go while I’m testing everything else.

for this, make the ‘edge’ of the plywood that you are testing be about 1’ in from the motor to replicate what happens on normal machines.

Thanks for your testing

[MeticulousMaynard]

Sorry for the delay in responding to you. I was reminded of the linkage slip in @Gero’s troubleshooting thread.

@dlang

first, add a nut (doesn’t even need to be a nylock nut, it can be a slim ‘jam nut’ which may fit on the thread), where you show it, without the plywood spacer support that you show. See what happens (you don’t need to do calibration, just see how easy it is to get the linkage to ‘snap’

I tried to do this first, but I didn’t have enough thread on the bolts to make it work. I might be able to give it a try with threaded rod, but…

@dlang

second, go back to the original all-plywood riser/spacer approach, just make the top layer be 3/4" wide (you may still have to notch it slightly to let the washers clear). NoDerlin spacer involved at all

This worked as a solution to the issue. I took one of my old plywood risers and ripped it to 3/4" wide to make sure that there was enough clearance for the bolts. Where the previous setup would jump at about 610mm (~24") from the centerline of the machine, now it slips at 1000mm+ (~40"+). And where it does slip, it no longer seems like the bolts aren’t aligned well as much as I’m reaching the limit of travel of the mechanism:

@dlang

Another test I’d like to see you do is to go to the outer edge and see how far you can go to (and off) the side before the sled starts tilting.

do we really need a skirt off the side to machine all the way to the edge.

I also was able to test this. I took a video, but I don’t have my camera with me right now to post the resulting video. I will try and post that tonight when I have access to the video and my editing software.

I would recommend a skirt around the edge of the sheet though. At about 1’ in from the edge of the machine, I had pretty good stability. I have a 4’x8’x3/4" sheet of MDF as my spoilboard, so I effectively do have a skirt for the edges of my machine. When I jogged the machine over to the edge of the sheet, however, things got harrier. With the bit at the edge of the sheet, the sled was webble-wobbling quite a bit. I think having a skirt that is a similar thickness will improve stability greatly.

Again, I’ll post the video tonight when I get the chance so you can see better.

---

I have switched the linkages back over to Logan’s. I need to start cutting projects, and currently it has the least distortion so I will be using that for now. This will be really important for cutting my new sled.

In response to this criticism I had for this linkage kit:

@MeticulousMaynard

The other design consideration is that the lower right mounting point really crowds the Z-Axis. It’s not impossible to fit both the standoff and the axis’s brake, but it’s something to be aware of:

I took @madgrizzle’s suggestion on how to mount the bracket, and I think it’s quite a slick solution to the problem:

So far so good!

---

And finally, I’ve been able to put my test sheet to good use! I needed to make a countertop for another project. It’s going to get nosing on the front, and tile on top, so you’ll never see all my test patterns once it’s done!

[dlang]

are the chains above or below the CG (on the Z axis), I think having them
slightly above the CG will make it more stable off the side.

[dlang]

how far apart are your motors? These are designed to be able to go up to an angle of 80 degrees, which is 66" below and 12" over from the motor. IIRC you have a short top beam, so you may be in that range 40" over from the edge.

your new linkage kit got out in the mail this morning (you should have received an e-mail notification of it)

[MeticulousMaynard]

I hate to disappoint, so here’s the video:

Skirt Test

@dlang

are the chains above or below the CG (on the Z axis), I think having them slightly above the CG will make it more stable off the side.

My chains mount about 2 5/8" from the bottom face of the sled, or the part that contacts the work surface. I think that, given the top-heavy nature of the rigid base, the CoG is above the plane the chains are in. I really, really need to make time to add in a unistrut top beam. That would help immensely as far as keeping the chains above the CoG.

@dlang

how far apart are your motors? These are designed to be able to go up to an angle of 80 degrees, which is 66" below and 12" over from the motor. IIRC you have a short top beam, so you may be in that range 40" over from the edge.

The distance between motors in Ground Control is 2761mm (108.701in). That makes sense because my top beam is 112" long. By your math, that means that the red hatched areas below are where that will happen on my bed, given my top beam length:

So on a stock frame, where there is roughly 120" between motors, this won’t happen. If you’re weird like me, don’t expect to get 96" at the bottom of the machine. I always knew I’d be giving up usable bed with my short frame.

[Gero]

@dlang @MeticulousMaynard @all
I’m a bit lost on the status of comparable test-cut patterns.
I have a BedAccuracyTest2_6mm1FL.nc here with date Dec 17 and
from the ->CG folder the Calibration Benchmark Test.nc<- last updated on Git Jan 18.
The fist has a nicly done xls-sheet, that till date 1 person has filled out and the later is missing the ‘SUBMIT’ button completely
Edit: And yes, on what release of FW/GW should I run it or them?

[MeticulousMaynard]

Sorry, this thread has gotten pretty long. I would like to condense a lot of this down to a wiki page at some point.

The two files I have been using are these:
BedAccuracyTest2_6mm1FL.nc (7.6 KB)
Calibration Benchmark Test.nc (640 Bytes)

I had setup the google doc back when we first were discussing recording the results, but I’ve been using my own spreadsheet to track the data. I added visualizations to help interpret the data better.

You could add your data to this sheet, and it will give us comparable results:
BedAccuracyTest.xlsx (42.3 KB)

Looking forwards to having more data than just from my machine. Having results from many different machines will help us determine patterns with the different systems.

@dlang: the new kit arrived in the mail yesterday. I put it together but I’m not going to have time to run the machine this weekend. The clearances on the holes do feel much better than the other kit. I also really like the added half-link, that should make it much easier to remove the sled for calibrations. The new version is certainly an improvement over the first one Now I just need time so I can put it through the paces.

[dlang]

part of what I want to figure out is how much adding the central bar improves
the original set (do those people who got the first kit need to replace
everything, or just add the central bar)