Grinder design questions: tracking wheel & drive wheel position, over-center tension, etc.

[Hubert S.]

Fitzhugh, be careful with a router jig. I think you would be better off drilling a few holes and connecting them with a fret saw and/or file. I have had some pretty scary kickbacks with a router both on wood and aluminum, especially on the router table. I particularly do not like cutting slots. Spiral up-cut bits seem to do better, straight flute bits always pack the slot full of sawdust.

I did the math, at least a quick approximation using a knife edge instead of a plate to push against. I could not wrap my head around how to compute the cam shape having a fixed tangent as a constraint. I will look at it again tomorrow. With the lever going through a 120 degree arc, the numbers I tossed out earlier should work for friction coefficients in the 0.1-0.25 range. So, unless there is grease on the cam, it should lock with a good safety margin. Obviously, the larger the cam the better. There is also the option to make the cam lift more rapidly through the beginning portion of the arc where it does not have to lock. I think this would allow for a smaller cam that still locks where it matters. There are some space constraints, but I think I can get the cam near the tracking arm pivot point by using a bolt through ball bearings mounted in a support on one side of the arm only. The cam could then swing past the tracking arm pivot point on the other side. Sometimes, I really wish I knew CAD, but every time I attempt to learn it I find it so much quicker to just make a sketch by hand.

I think the Northridge and KMG TX both use a ratchet without any springs. The TW90 has a spring, but it appears to me that it must be very stiff and not allow for a lot of displacement. I might be wrong on that. I am also not sure that the system is necessarily short throw/stiff spring. This depends not only on the stiffness of the belt, but also on the change in path length generated by the rotation of the tension arm. It seems to work for many different geometries and I think you would hear a lot of complaints if belts popped off in use because the belts stretched a bit.

If I can figure out the cam geometry, I will find some time this weekend to make a prototype out of plywood.

 

[Ken H>]

Something I've played with on my grinder to test the idea of a fixed tension rather than a spring is to wedge a piece of wood between tension arm and grinder frame to hold pressure on belt. Something like this:

The more the wood is moved rearward the more tension is applied.

 

[Hubert S.]

Ken, has the piece of wood ever come loose while grinding? My cam is the same idea, only using rotation instead of translation for the wedge.

I think I have figured out how to solve the cam shape for a flat face instead of a knife edge using cubic B-splines. At least my code produces a matrix that can be inverted, and that's half the battle. The shape looks plausible and if my analysis is correct, it can be made to work for pretty low friction coefficients. I will make a simple mock-up out of baltic birch scraps as soon as I find some spare time.

 

[Fitzhugh]

How does it feel with the stick compared to the spring?

This shows me I need a grinder so I can wrap my head around why rigid tension setups don't fling belts as soon as they stretch, like my poor brain insists they should. Who knows, I might even grind a knife or two.


Hubert, just out of curiosity, what are you programming in and what's your background, in broad layman's terms?

 

[bjansen]

Catching up on this good thread. I found xometry cuts fairly precise- precise enough for what you need for this build. When i cut 6061 for these grinders i cut it with a portaband saw and that works well and is safer than a chop or tablesaw. I think the bigger questions is how are you going to clean up your saw cuts? I clean the cuts up now with my mill, but before i had the mill i used woodworking sanders and that worked just fine (its just a tedious task).

I have never used a ratchet style grinder (I can see how they would be excellent and superior) but the old KMG style tracking system is super simple to make and works great every time.

 

[Hubert S.]

I wrote the code for computing the knife edge follower cam geometry in python. For the flat faced follower I switched to C++ because I am not very familiar with python's linear algebra and svg libraries. Of course, the code was written using emacs on a linux box.

My background is process and mechanical engineering, but I got interested in signal and image processing in college and have been doing that ever since.

 

[Ken H>]

Hubert, I envy your ability with code. I do some playing around with code, Arduino, etc and many years ago played with Basic some. When I got my first computer (Commodore 64) I wrote a simple program for navigation using a sextant, star and sun reduction. I do a great job of designing programs for easy use, but not so good at the actual coding - I have to work at that!

Never had the wood stick to move while grinding and it works good, but the spring works pretty good also. The biggest problem with the spring is how much pressure it takes to compress the spring by hand when changing belts for heavy grinding with 36 grit ceramic belts. For finer grits (120, etc) a weaker spring works pretty good, and when I'm running in reverse for a couple of operations I use a "not so tight" belt. I "think" the rigid tension system is best, but it does require a tracking system that allows the tracking wheel to stay level, and adjusting the side to side movement of belt either by moving either the drive wheel (TW-90 style) or the tracking wheel side to side KMG-TX, Reeder, etc.

Maybe Hubert will come up with a good design and I'll change from spring to rigid. I've looked at a couple different methods but not felt good about building any of them, but am still looking.

Later

 

[Hubert S.]

Catching up on this good thread. I found xometry cuts fairly precise- precise enough for what you need for this build. When i cut 6061 for these grinders i cut it with a portaband saw and that works well and is safer than a chop or tablesaw. I think the bigger questions is how are you going to clean up your saw cuts? I clean the cuts up now with my mill, but before i had the mill i used woodworking sanders and that worked just fine (its just a tedious task).

I get much cleaner cuts with a chop saw than the band saw. But I agree, the band saw is safer, and it is also a lot less noisy and messy.

Since you have a mill now, do you still use the method outlined in the build-along of transfer punching the holes one bar at a time, or do you just drill all the holes in one go on the mill? I think I am going to try that. My DRO has five decimal places, it's got to be accurate enough, right?!?

I have never used a ratchet style grinder (I can see how they would be excellent and superior) but the old KMG style tracking system is super simple to make and works great every time.

I think a lot of this may have to do with the overall design of the grinder. With a face mount motor, the tracking arm and the tool arm all aligned by a (fairly) flat plate, it is probably a lot easier. I have looked at a lot of grinder builds online and many people have issues with tracking. And even on commercial grinders, there seem to be quite a few tracking issues...

 

[Hubert S.]

Hubert, I envy your ability with code. I do some playing around with code, Arduino, etc and many years ago played with Basic some. When I got my first computer (Commodore 64) I wrote a simple program for navigation using a sextant, star and sun reduction. I do a great job of designing programs for easy use, but not so good at the actual coding - I have to work at that!

The Commodore 64 was my first computer also! That's how I learned to code, first in Basic and then 6510 assembly language. I wrote a little jump-and-run game that was published in a German computer magazine back in the day. After the C64 I went straight to Unix systems. I am afraid I am not that good at designing for ease of use. I like command line programs, editing configuration files with vi, programming in emacs and word processing using latex.

Never had the wood stick to move while grinding and it works good, but the spring works pretty good also. The biggest problem with the spring is how much pressure it takes to compress the spring by hand when changing belts for heavy grinding with 36 grit ceramic belts. For finer grits (120, etc) a weaker spring works pretty good, and when I'm running in reverse for a couple of operations I use a "not so tight" belt. I "think" the rigid tension system is best, but it does require a tracking system that allows the tracking wheel to stay level, and adjusting the side to side movement of belt either by moving either the drive wheel (TW-90 style) or the tracking wheel side to side KMG-TX, Reeder, etc.

Good to hear there are no issues with the rigid setup. Can you fit a longer lever to make it easier to push the stronger spring down? Do you think the side-to-side movement is the critical aspect, or the ability to adjust both axes?

Maybe Hubert will come up with a good design and I'll change from spring to rigid. I've looked at a couple different methods but not felt good about building any of them, but am still looking.

Later

I will let you know what I come up with. I still think there must be something wrong because it seems so simple.

 

[bjansen]

Hubert, I have not used my mill w/ DRO to drill all the holes as of yet (one of the reasons is because I never actually had my plans in CAD/PDF design (until recently, which Fitz helped me with) so I have been doing it the old school way - works well though! I am building a 2x48" right now for leatherwork and I may use the DRO on that one.

 

[Fitzhugh]

Thanks Bjansen - I planned to try files and disk sander for minor cleanup. I can use a mini mill if I find I need to but it's a drive to get there. I've updated my drawing so all the mating surfaces are original bar stock sides, not cut ends, so the cuts don't need to be machinist level precise. I do want it nice though, and will want to round over some parts.

In my case I'm looking at other tension and tracking styles out of curiosity and future possibilities. This one will be traditional, nearly identical tension & tracking to yours from risers on.

Hubert, how would you cut out the cam neatly? Is it a disk with an offset pivot?

That's a nice skill - it's cool how what we have in our quiver influences how we see the world, not just how we solve problems [referring to your ability to whip up a model, not emacs you degenerate].
Also, the matrices mention made me shudder. I needed linear algebra for a program I cobbled together for a scientist friend last year [using vim like any decent person would]. The programming was easier than expected considering it's been years and a new language (it was also happily simple), but I foolishly expected the math to all come right back when I haven't touched it since college. Man, could I feel how my brain had atrophied.

 

[Ken H>]

With a face mount motor, the tracking arm and the tool arm all aligned by a (fairly) flat plate, it is probably a lot easier.

All the functions aligned by a single flat plate is one of the things that appeals to me about Bjensen's design.

Can you fit a longer lever to make it easier to push the stronger spring down? Do you think the side-to-side movement is the critical aspect, or the ability to adjust both axes?

Naw, I've got as long an arm as I wish and I can pull it down just fine, as long as the shoulder isn't acting up. Those are the times when I'd like it easier.

YES!!! The side to side moving is VERY critical. If you move the tension arm sidways just a tad you'll see the belt move a good bit on the pulleys. shucks, a way to adjust the tension arm sideways would make a good tracking adjustment. Actually, the design Reeder is using does just that - well it moves the tracking wheel as if the tension arm was moving.

Later

 

[Ken H>]

One of the issues I see with a cam for tension adjustment would be how to vary the height it held the tension arm. It would be easy to make the cam hold in a single position, but different belts are going to need different heights, varying as much as 1/2" at least, if not a tad more. I guess it would work good to have the cam a smooth curve, and at the point of proper tension have a lock to hold in position. Hmmm, that might work, let me think about that some.

 

[Fitzhugh]

Here's an update if anyone has feedback.

To get the tracking risers to reference off the side plates I extended them down between the side plates, and changed the top spacer to make it fit. Top spacer is now 3" stacked 1/2" x 1" pieces, stack on side, with middle piece extending between the risers. It's not the most elegant, but it locks together and saves all of $5 by not introducing another size of stock. Bolted through from side instead of countersunk screws from underneath & through top spacer, which I did not like.
There's a middle spacer between the risers as well. I have it stick out towards the front as a brace, though not screwed from bottom, but really only because that way it's the same length as a couple other pieces (the spacers between the D Plate and platten mounts). If I cut my own parts I'll just cut it flush with risers - the 4 bolts should hold the assembly just fine, I think.


...sometimes you still have to color parts to see where they are. Here's how f360 does it if you turn on the feature to help you distinguish parts - can probably see why I do it manually:

One potential source tracking nightmares addressed, I think. I still have to deal with the motor being separate, but I should be able to align it - kinda like adjusting bandsaw wheels. A long enough stick against each wheel should make any misalignment apparent.

Bjansen, if it's CNC you can generate the gcode (or other) from fusion 360, or just I can send it - would just need some machine parameters. You can simulate test runs of the code in the program and tell if you're going to crash the bit - though I've still managed it by being stupid in zeroing.



[My first computer was a sad Tandy. I got a free Kapro luggable soon after, though - got to keep it after taking photos of the guts for the fcc report, a job I had in junior high. Got to keep most stuff... made more selling CB radios, cordless phones and clock radios to other kids than from the job itself. Never thought of it this way but the cell phone has replaced all those things.]

 

[Hubert S.]

I built a prototype that illustrates the cam mechanism. It is cobbled together from scraps, but works surprisingly well. Let me explain the components: The 2x8 at the bottom is the grinder frame, the platen/contact wheel or what have you would be on the left, motor on the right. We are looking at the belt side. The horizontal bar with the rubber band hooked around a small bolt is the tracking arm. The wheel would be where the rubber band is attached. The pivot point of the tracking arm is on the right behind the cam. The cam is the white piece that has a handle bolted to it. The 2x4 is what the cam pushes against.

The mechanism locks in any position, you can really push down on the tracking arm and the cam won't let go. It should work for friction coefficients of 0.2 or higher. The mechanism does need a small pre-load (a weak tension spring), otherwise the cam lifts off the support underneath. This does not prevent it from working, but I think it could be annoying. The handle is in the same plane as the tracking arm, and it interferes a little bit earlier than I thought. I think I can only get to about 100 degrees, not the 120 that I had planned. That's easily fixed by extending the tab where the handle bolts on.

I have to think about this for a bit. My initial impression from the prototype is that it will work great.

 

[Natlek]

Something I've played with on my grinder to test the idea of a fixed tension rather than a spring is to wedge a piece of wood between tension arm and grinder frame to hold pressure on belt. Something like this:

The more the wood is moved rearward the more tension is applied.

OK Ken, here is much better solution even from ratchet ..........
First get rid of that tracking arm, you don t need it .Fix tracking wheel in one place /which will track side by side , not up-down / Then find how big of center cam you need to cover all your attachment ...flat platen ,contact wheel ,small will ...etc .
Then attach that off center cam on your tool arm ...Push off center cam as much you want to tension the belt and ...LOCK that tooling arm ........flawless

 

[Fitzhugh]

Ah, nice! I had pictured it with the cam mounted to the chassis and pressing on the arm.
I'm sure it will take a bit of dialing in to get the specifics just right but it's an elegant solution.

You could mount the cam handle to the other side of the cam pivot, so it's on the far side of the tension arm. Just one possibility.

Natlek's idea makes a lot of sense, but I think you'd want the cam on the chassis pressing against a protrusion on the arm... that way you have one cam for all the arms.

 

[Hubert S.]

Natlek, I think that might work. You could have a series of holes in the tooling arm that are spaced just under the maximum displacement of the cam and use those holes to put a shaft attached to the pivot point of the cam in. Or fix the cam to the grinder and just use a bench dog to put in the right hole. Definitely something to explore. I think it would be a bit more inconvenient to change belts, but you gain stiffness by eliminating one rotational degree of freedom on the tracking arm.

 

[Fitzhugh]

I do think you'd find cad worthwhile. You can mock it up quickly and define joints and contact surfaces to easily study range of motion, identify interference, etc.

It won't replace these tests (and you have to build one on the end, of course)... F360 at least does not handle friction of cam on surface type stuff, from what I found searching just now (or it seems it's an iterative hassle - I'm reading into a tangentially related thread I found on their forum, possibly all wrong).

 

[Hubert S.]

Hubert, how would you cut out the cam neatly? Is it a disk with an offset pivot?

Best way would be a laser or a CNC. Short of that, a belt grinder!!! Seriously, though, I think it could be done with a file in a couple of hours, provided you do a good job rough cutting with an angle grinder or a band saw. The plywood prototype was cut on a band saw and then sanded smooth.

The cam is not just a circle with an off-center hole. I am not sure how well that would lock, I have not analyzed it. The cam shape I generated has a lift that is proportional to the rotation angle and pushes against a flat plate. You can get a sense of the shape from the picture of the prototype I posted.