Spider Maslow, Spider Maslow, does whatever a Spider Maslow does

[RandomDave]

I’m probably going to have to write this up in chunks rather than my usual monolithic style due to…life - so don’t be surprised when I miss big chunks in this first writeup.

Ok, so I’ve had this going on in the background for a number of months (to give you some idea, the green bits are from when I started on it, and they pre-date me buying the printers I’ve used for everything else you’ve seen from me…). Given the awesome discussion in the Next version of the maslow idea? an m8 I thought I should actually stick something up about it even though it’s barely at proof of concept stage, just to avoid anyone wasting duplicate effort.

What is it?

Basically, taking the arms off the Maslow and moving them to the corners. Replacing them at the centre with a set of interleaved rings to attach to:

Core
I guess the core of it is is the equal-height attachment points at the spindle:

This was the seed of it all for me really - I was mulling what you actually would have to do to get equal height for the attachment points, and moving the arms off the Maslow made the most sense. Then you have to work out how to get those equal heights. I figured the best was was a set of interleaving you can see there, with a bunch of washers between (it was what spawned the [PTFE washers for between arms](PTFE washers for between arms thread - ptfe was easier than getting custom washers made).

It’s not the easiest to explain in words so i’ll take more photos later, but in essence:

• 2 plates for opposite side with one washer between and a thicker attachment-point-area fors a 2mm-thick set for 2 arms that are opposite.
• We can offset 2 x 2 plates with spacers the same height as the belt end - then we have 2 opposite side assemblies with equal-height attachements.
• But if we then use an additional 2.5-3mm spacer alternated top/bottom of the attachment points, we can interleave a 2nd set of 2 x 2, to give us 4 bottom plates & 4 top plates, with 4 equal height attachment points, that also have the benefit of only taking up about 50mm height (which i’ll talk about in a sec.

The corners and wiring are their own whole things but I’ll add details later - they’re largely obvious.

Why? Advantages?
I’ll be honest, I did it because I was curious if I could. But it also offered some potential benefits I’m going to start exploring now I’ve had a working POC:

• As above you can get the attachment points all the same height - with all the benefits that brings.
• You can get a sub-5kg assembly - or given I’m working on a chunky-spindle build, I can offset some of the weight gain.
• I think you could run a higher speed lower torque set of motors and push feedrates higher: (hence this thread: Feedrate limits of the Maslow (theoretical).
• The less deep core means you can fit a Makita router easily, or have more control of the height of the bottom of a DeWalt router collet while still beefing up the clamps. It’s a bit blurry, but:

Drawbacks
Cost and complexity are obvious, but it also feels a bit…against the philospohy of the Maslow - it’s no longer a single unit you can (theoretically) plonk down on a sheet of material, hook up the corners to solid attachment points, calibrate and cut. In time, perhaps it can get to be a bit more like that but we’ll see.

The big one yet is i’ve not explored how long the wires can be before you hit issues. I deliberately am only doing the arm-motor and sensor wires rather than the Z-motor ones too because I think they are more resilient to longer wires (as the Z-stepper drivers are on the control board). But I have also only tested 2.5m wires, which give me 1mx1m, possibly 4ft*4ft with careful routing.

What have I managed so far?

A few simple logo test cuts - 3.175mm (1/8th) bit, 4 passes, 3.25 DOC each pass. Speed is…complicated. I did 1000, 1500, 2000, 2500 mm/m, but I believe I was acceleration limited at or above 2000mm/m - I am not claiming that as an actual speed. I saw a perceivable difference between 1500 and 2000 but not above. At that point I was also starting to see a little stuttering.

Next?
With those cuts, I am using a long O-flute single flute bit. I was getting ok-ish chips at speed 4 but it felt like hitting issues I can resolve. The bit is quite long, and the collect of the router is a little way off the top of the cut, so I think I might be seeing bit flex as I push it. I believe with some minor changes (to get the bottom of the collet down to just above the top of the cut) I can get a bit more stiffness in the bit and I can try pushing it a bit more. I want better chip ejection if I can, I think the O-flute isn’t doing me any favours, they are just the most easily available for me. I’m going to get hold of more spiral / open sided bits too.

Plus longer wires and testing it a bit more generally!

[RandomDave]

@dlang points from the Next version of the maslow idea? an m8 thread:

Ahem, soooooooooo I have been quietly playing around with this idea in the background of the other stuff I’ve been doing, for like…months, I was going to test it a bit more before I talked about it, buuuuut:

The main advantages:

• With some clever geometry you can get the attachment points all the same height.
• You can get a sub-5kg assembly.
• You can fit a Makita router easily.

go to a spindle instead of the router and you can save more wight

Yeah, I considered jumping to that point - I’m not sure how well you can see, but at the moment i’m using a cordless/brushless dewalt running off a PSU like I talked about in Interesting find - brushless Franken-router setup (at least I think I talked about getting the dewalt one, I should check). Run like that it actually doesn’t weigh too much for the amount of power you get. I wouldn’t want to go down to a 500W spindle, but the 800W ones (I think that’s what you use?) are potentially interesting yeah.

• I think all these mean you could run a higher speed lower torque set of motors and push feedrates higher: (hence this thread: Feedrate limits of the Maslow (theoretical).

you will be pushing the rigidity of the system, but you are working on that as
well.

Yep! The other direction is using it to offset the weight in a really powerful spindle, but things would reward continued work on stiffening.

However.

(And this is why I hadn’t posted it yet)
I’ve only tried it with 2.5m leads, and I am not sure how 5+ metre leads would work, which is what you’d need for a full sheet. Voltage drop and interference could well scupper the whole thing.

the electronics are light, so I don’t see a lot of advantage in moving the
controller board off of the sled, but I look forward to see your work.

David Lang

Ah, it’s still on the sled, just on the side because of the PSU adapter - I don’t want to extend the Z-stepper leads by any more than I need to given the drivers are on the board.

[dlang]

the firmware has the code in it to support a fixed Z, so instead of having to go around the router and move up and down, you can mount a ring to the sled and have trollies for the belts to attach to (similar to what the earlier maslow did). that would let you set the height of the belts on the sled to match the center of gravity in the Z direction (it will only be perfect for one height of the spindle, but there is enough other weight that I think it will be more balanced than moving the sled attachment up and down with the Z

It also makes it MUCH easier to handle different spindles as all you have to deal with is the Z movement and dust collection (in my onshape ‘lang maslow’ tab has a toggle button to show that approach)

I also think that a fixed Z, pulling on the sled instead of the spindle will be more rigid for fine detail as well, as the belt forces will transfer directly to the sled (with it’s high friction contact with the workpiece) rather than pulling on the spindle and going through the Z movement axis. I could be wrong and the forces on the cutting bit are higher than the sled friction, but I don’t think that would be the case.

when doing a ring, you can easily have the ring cover 180 degrees top and bottom, pretty much eliminating the belt angle problems in the corners

[dlang]

As far as long cables go:
use a shielded wire for the encoders (and ground it only on one end, probably the electronics side)
the motor power wires may generate interference (it’s high frequency PWM signals being sent over them) but if there is power loss, it will just mean that the motors are less powerful than they should be (more likely to hit the 4000 current limit) but since the movement is detected from the encoders, if the motors are more sluggish, the machine will ‘just work’

[wouldchuck]

Well it’s beautiful. It’s still very much in the spirit of the maslow making a large format cnc affordable accessible and depending on how you set up the corners it still could be portable. I had not thought about hotrodding a maslow by dropping weight for speed. The problem with the threed printer and cnc world is speed of production so if you could raise that even a little bit for projects, it would be pretty cool. My last maslow job took 5 hours. That is very very cool. I think we need to add a library to the wiki for fun builds.

[RandomDave]

@dlang

the firmware has the code in it to support a fixed Z, so instead of having to go around the router and move up and down, you can mount a ring to the sled and have trollies for the belts to attach to (similar to what the earlier maslow did). that would let you set the height of the belts on the sled to match the center of gravity in the Z direction (it will only be perfect for one height of the spindle, but there is enough other weight that I think it will be more balanced than moving the sled attachment up and down with the Z

It also makes it MUCH easier to handle different spindles as all you have to deal with is the Z movement and dust collection (in my onshape ‘lang maslow’ tab has a toggle button to show that approach)

Yeah, I’d certainly be interested to see a ring build! Given time it’s something I might try, but not super high on my priority list.

It’s an interesting question which is stiffer, and one I have no idea and think some testing would be needed. I think of it as ‘moving the spindle directly and having the sled just be a way of setting Z and keeping the router perpendicular to the XY plane’ vs ‘moving the sled directly and the rest of the sled be about maintaining the stiffness through to the spindle in X,Y and Z’.

And I am really not sure which can be made better for the same weight etc - a ring needs to transfer the force from the belts through it’s structure to the tip of the spindle so sled-spindle stiffness remains the primary problem, but you’re right that you have more freedom geometrically on how you do that

I haven’t really delved into the old Maslows - how were the trolleys and the path of the belt ends kept rigid - was it largely making them metal components and reasonably beefy?

@dlang

I also think that a fixed Z, pulling on the sled instead of the spindle will be more rigid for fine detail as well, as the belt forces will transfer directly to the sled (with it’s high friction contact with the workpiece) rather than pulling on the spindle and going through the Z movement axis. I could be wrong and the forces on the cutting bit are higher than the sled friction, but I don’t think that would be the case.

when doing a ring, you can easily have the ring cover 180 degrees top and bottom, pretty much eliminating the belt angle problems in the corners

Sled friction forces vs spindle cutting forces, again, another area we need to investigate!

I did some verrrrrry rough measuring a while ago and found (horizontal) breaking static friction took 2-3kg (30-30N) of force and dynamic friction took 1-2kg (10-20N), with a couple of versions of my builds on plywood (so difficult to extrapolate). I think that would be less than the force used when cutting if you were going for a decent level of material removal from doing some manual routering over the years, but there’s so many factors for that too

[RandomDave]

@dlang

As far as long cables go:
use a shielded wire for the encoders (and ground it only on one end, probably the electronics side)

shield just the encoders, not the power wires? That does make sense. I’m not sure the easiest place to ground it to on the Maslow? I probably should look at that for something else anyway!

@dlang

the motor power wires may generate interference (it’s high frequency PWM signals being sent over them) but if there is power loss, it will just mean that the motors are less powerful than they should be (more likely to hit the 4000 current limit) but since the movement is detected from the encoders, if the motors are more sluggish, the machine will ‘just work’

Ah, I did wonder if that was the case, good to know. Time to solder up some 5m wires and see what’s what!

[RandomDave]

@wouldchuck

Well it’s beautiful. It’s still very much in the spirit of the maslow making a large format cnc affordable accessible and depending on how you set up the corners it still could be portable. I had not thought about hotrodding a maslow by dropping weight for speed. The problem with the threed printer and cnc world is speed of production so if you could raise that even a little bit for projects, it would be pretty cool. My last maslow job took 5 hours. That is very very cool. I think we need to add a library to the wiki for fun builds.

Thanks, encouragement is good for motivation

Yeah, more sped is always good! I see it as trying to get into the sweetspot of the routers RPMs for chip removal while maintaining DOC - tricky at the best of times, but doable I think.

[RandomDave]

@RandomDave

Next?
With those cuts, I am using a long O-flute single flute bit. I was getting ok-ish chips at speed 4 but it felt like hitting issues I can resolve. The bit is quite long, and the collect of the router is a little way off the top of the cut, so I think I might be seeing bit flex as I push it. I believe with some minor changes (to get the bottom of the collet down to just above the top of the cut) I can get a bit more stiffness in the bit and I can try pushing it a bit more. I want better chip ejection if I can, I think the O-flute isn’t doing me any favours, they are just the most easily available for me. I’m going to get hold of more spiral / open sided bits too.

On this next step, did a bit of checking, the collect currently sits about 8mm off the plane of the sled, which isn’t a huge surprise, as the 20mm Turtle clamps take away roughly that much of the Z travel (which I should have remembered ).

With the unused ~40mm the stack gives me, I think for now i’ll just chop 10mm off the columns and bump the bottom clamp clamping depth to full 20mm double clamps - that gives me a little wiggle room for where the collet bottom sits.

Given the columns house the linear bearings, that might have a minor effect on the X-stifffness

I’ve experimented with moving the linear bearings into the clamp on the big spindle build, I’m not sure that makes sense here though. Possibly I could do linear bearings in the top clamp but not the bottom one sort of similar to the linear-rail top clamps, especially as an easy extra step

[dlang]

Dave wrote:

I haven’t really delved into the old Maslows - how were the trolleys and the path of the belt ends kept rigid - was it largely making them metal components and reasonably beefy?

I’m not understanding which parts you are questioning?

@dlang

I also think that a fixed Z, pulling on the sled instead of the spindle will be more rigid for fine detail as well, as the belt forces will transfer directly to the sled (with it’s high friction contact with the workpiece) rather than pulling on the spindle and going through the Z movement axis. I could be wrong and the forces on the cutting bit are higher than the sled friction, but I don’t think that would be the case.

when doing a ring, you can easily have the ring cover 180 degrees top and bottom, pretty much eliminating the belt angle problems in the corners

Sled friction forces vs spindle cutting forces, again, another area we need to investigate!

I did some verrrrrry rough measuring a while ago and found (horizontal)
breaking static friction took 2-3kg (30-30N) of force and dynamic friction
took 1-2kg (10-20N), with a couple of versions of my builds on plywood (so
difficult to extrapolate). I think that would be less than the force used
when cutting if you were going for a decent level of material removal from
doing some manual routering over the years, but there’s so many factors for
that too

I’m thinking that (with a sharp bit), the cutting forces are low.

but we would need to test pulling just the sled over a material vs pulling the
sled while cutting over the same material.

David Lang

[dlang]

Dave wrote:

@dlang

As far as long cables go:
use a shielded wire for the encoders (and ground it only on one end, probably the electronics side)

shield just the encoders, not the power wires? That does make sense. I’m not sure the easiest place to ground it to on the Maslow? I probably should look at that for something else anyway!

both would be better, but since the motor wires are output-only, noise there
won’t affect the maslow (may affect other things, but not the maslow)

but noise on the encoder wires can cause problems reading

there is a ground pin on the connector, just connecting a shield to that will
help.

David Lang

[wouldchuck]

Friction is very messy and will change drastically with dust and humidity and the exact surface but just now I was getting pretty repeatable results for kinetic friction for the sled (couldn’t see a big jump for static friction but i can get out the electronic balance and try that later)

For a 7.75 Kg Maslow

Standard rough plywood 33-35N to keep it moving with Kinetic friction

Smooth plywood was around 28-33N

Plastic film surfaced Ikea board 17N

When the snow clears I could try some while the machine is cutting. It will depend a lot on the speed of the bit, type depth and material.

I mean even with this simple experiment it would suggest that skiwaxing the maslow sled or a quick sand of a surface could change the forces a lot.

[wouldchuck]

OK did it with the router running. I pulled it through Plastic coated particle board and a pine 2 x 8 both had very similar results. 17 Newtons to pull the 7.75kg sled sideways at a constant rate with no blade. I had a 1/4 inch 2 flute up cut standard router bit. The dewalt router was set at top speed. I dropped it in by increments and pulled. In both the pine and particle board it seems to add a little more than a newton per mm as the blade drops in. Perhaps that is helpful Again friction is weird but maybe this would give someone a sense of the range of forces acting on the machines. Routers run really differently when they are doing pockets, when running against and with the grain and with different materials so any data is going to be rough but knowing the scale of the forces might be useful. I was just trying to drag it at a constant speed but with the blade in that would definitely change with more speed. I would guess I was dragging at about 400mm per second.

[RandomDave]

Oh wow, that’s awesome work!

What kind of cut finish were you seeing - router on full with a 2 flute 1/4" I’d expect to be getting quite hot / making dust, but it’s also probably a normal usage case for Maslow as I think slow and deep is often what people do?

It seems like cut and friction for es are in the same order of magnitude then, but either could be dominant.

I think I saw that people used to put UHMW tape on the old Maslows where you made your own base - that’s something I’ve meant to try, similarly actual sheets of PTFE, the kicker being it would make it difficult to adjust the z-motor screws (and oth stuff) so you’d need to be pretty happy with where you were build wise!