Add not cutting square to the list of things my machine won't do. It's a deepgroove, from eBay, they come with no documentation whatsoever.
I'm just getting the hang of the design side, drawing a 4-inch square in a cad prgoram and converting it to gcode and watchign it go on the sim and then... big moment, run it on teh machine and... out pops something that's not a square. A 4-inch square is about a 16th out of square.
So I think, 'Hmmm, what's causing that?' And I start tweaking the carriage frames a bit. Got them shimmed up for some compensation, let's see.... nope, still out of square.
I'm cutting half-inch plywood with quarter-inch passes, it ought to cut through like butter, but it's not square. SO I redce both speed and depth and make more passes. Better but still nto square. I can get about a 64th of an inch out of square for a 4-inch square.
And then just for grins I decide instead of cutting conventional, I'll cut climb.
And it's square.
So wassupwidat? Frame torque? Huh? What? Bit deflaection? Huh? How the heck do I handle THAT? THe spindle's a Bosch Colt, it's extremely securely mounted, so of course, it's probably moving.


Anyone got any Hints? Tips? Donations? ;)

I doubt it's bit deflection, you're not pushing the bit very hard. Are all the sides perfectly straight? You were on the right track adjusting the machine to get it square. The small deviation from conventional vs climb cutting might be due to backlash.

I know what climb cutting is, but it was my understanding that routing direction only had relevance in manual routing for operations like patterns, templates and edge forming using bearing bits.

Is there a "conventional" direction for cnc cutting and if so what is it?

Is there a "conventional" direction for cnc cutting and if so what is it?
Cutter outside of frame travel clockwise or cutter inside of frame travel counter-clockwise

Is there a "conventional" direction for cnc cutting and if so what is it?
Cutter outside of frame travel clockwise or cutter inside of frame travel counter-clockwise

Why? This opposite the way you route with a handheld.

if you cut the other way its called climb milling G41 and G42

Climb Milling
Cycle: Pocketing

In the Pocketing cycle, the tool path moves always in clockwise direction. If this option is turned on, tool path movement is counter-clockwise.

So, is this just convention, or is there a reason to cut one direction verses the other?

when you climb mill there is little tool defection but if your cutting wood that not an issue chip of edge may occur in a climb mill more than a convention path

So, is this just convention, or is there a reason to cut one direction verses the other?

Whether hand routing or CNC routing, Conventional cutting would be going around the outside of your part CCW. Climb cutting is going around the outside of the part CW. You should never climb cut with a handheld router, for safety reasons. The bit will grab and want to pull the router out of your hands.

What I've found when CNC routing wood, you usually get a much better cut when conventional cutting. But there are a few circumstances when climb cutting is needed. When cutting hardwoods, and cutting along the grain going CCW, if the bit is cutting along the edge of the wood, it's very likely that you'll experience tearout. Some woods are worse than others, with oak being about the worst. Climb cutting will allow you to make the same cut without getting tearout, almost all of the time. One exception is cutting arcs. Its possible to get tearout along an arc with both conventional and climb cutting, due to the grain direction changing during the cut.

The best way to avoid tearout all together, is to always leave a little material on the waste side of the bit when cutting. This will always keep you from getting tearout, but depending on bit size, can waste a lot of material.

if you cut the other way its called climb milling G41 and G42

Climb Milling
Cycle: Pocketing

In the Pocketing cycle, the tool path moves always in clockwise direction. If this option is turned on, tool path movement is counter-clockwise.

Apparently your talking about a specific software here. G41 and G42 are tool radius compensation. With a standard spindle rotation, G42 (offset to the right) is always conventional cutting and G41(offset left) is climb cutting. It doesn't matter if your cutting pockets, or the outside profiles.

when you climb mill there is little tool defection but if your cutting wood that not an issue

Tool deflection CAN be an issue when routing wood. But, it's not very likely on a homebuilt machine unless your using bits smaller than 1/8".

Backlash would be an issue, wouldn't it? In a climb situations, bi deflection could be countering it. I took all backlash compensation out of the config after I was cutting circles and getting fun offsets;) And removing the backlash comp did make the squares a bit more square too, but after thinking about it, it would still be a factor. DAngit.
Guess I need more brass shim material. ( sigh )
Trust me, this deepgroove machine ain't all that.

Is this machine ever going to be able to cut accurately? It's got backlash, it's out of square... IS there really any hope for this hunka junk or am I screwed?

If you can make it square and get rid of the backlash.

But how to remove the backlash? Tyring to cut circles tells me that backlash compensation is not the answer there. Dunno how to adjust for it. I've emailed deepgroove and asked bu so far no response. I don't see any anti-backlash adjustments to make on teh machine itself.

You might have to change the screw and nut, if you can't get rid of it.

I was afraid of that. Might just try to upgrade to ball screws then. You're the link master, got any links to decent ball screws in the 18 and 36 inch lengths?;)

Well I think you need to reduce the number of variables.....I would start by routing a straight line of a certain length using only one axis and checking the length....then I would perform the same set of tasks using the axis that is perpendicular to the one I just tested.....

ROuting a straight line on one axis of a certain length is sort of redundnat to the carriage movement measurement, but it WILL tell me if there's flex in the router mount somewhere. If I can measure .017 backlash on the carriage then if there's not a .017 discrepency on the line then I've got flex, right? This would be assuming that I ran the machine the opposite direction from the cut just prior to cutting, too.

Is it racking at all?

Even if screw is central, it might still rack at the edge.

Andy

I don't know what 'racking' is, sorry.
But I tore it all apart and started playing with the axes one at a time and fixed most of the issues- but it still won't cut square, and the reason is a design flaw.
The carriage runs on two large metal rods , one on each side, that run through two metal collars on each side of the carriage. These metal collars don't fit the pipe exactly, so you can 'wiggle' the carriage out of square with the metal rods. Since this is on the Y axis and it's carrying the rest of the machine, this is a problem that affects the X Axis- wherever the Y axis carraige ends up, it's got some play, so sometimes the X axis cuts a little out to one side, and sometimes it cuts a little out to the other side, depending on which direction the Y axis last travelled in.

Is that 'racking'? Sounds like it might be.

At least, thi sis the vast majority of the problem. I only discovered this because I couldn't see any backlash on the Y axis at all but was still getting some. I finally realized it after about three hours of staring at the thing, going back an inch, measuring, going up an inch, measuring, and it never varied. Finally I measures the other side of the carriage... Yikes.

So, anyone got any bright ideas for fixing this one? I'm wide open for suggestions;)

Is this the machine, you have?

If so, The "Y-axis" supports, look flimsy.

That's the machine, alright. He's got it wired with the X axis on the gantry and the Y axis under the table. The gantry's not the issue, it's the carriage along the sides under the wood.

I would run two additional rods in parallel to the existing with their own collars and also change the existing collars.

I reckon you would get exponential improvement.

You would have to ensure the existing rods are fully straight first otherwise you'll have new problems.

Perhaps do some z calibrations against a laser level first to prove your existing rods have no curve.

Andy.

I wouldn't know where to begin to install those rods, I dont' have acces to a machine shop. It's probably what it needs though.

Create a sub table and fix it to the existing table. Attach the new rods to the sub table. The rods don't have to be identical to the existing. If the new rods are under the old, that would help remove any existing sag issue (if it exists that is).

The new collar assembly would also have to be attached to the gantry frame.

Complicated but worth it.

Andy

I'm having a hard time picturing this. I attach two new pieces to the ends of teh table frame, and put the new rod mounts into those, then attach two new pieces to the gantry carraige,and feed the new rods through those?

Ed_R,

Post a pic. of the bottom of the machine.
So we can see what you already have to work with.

No camera;)

Quote : I'm having a hard time picturing this. I attach two new pieces to the ends of teh table frame, and put the new rod mounts into those, then attach two new pieces to the gantry carraige,and feed the new rods through those?

I'll try and answer above.

1 I'm assuming your machine has a fixed base with 2 parallel rods either side. Let's say the total footprint of your machine is 500*500mm
2 cut a rigid piece of thick MDF 500*600 and screw/clamp/glue it in place under your machine thereby leaving 50mm sticking out either side of the existing rods.
3 construct rigid and adjustable mounts on the MDF so as to place two rods both next to and parallel to the original rods. This will create an arrangement of I I-I I where the first and last I are the new rods and the I-I represents the original machine
4 put collars on the new rods and attach those collars to the gantry. That will mean you have both the original collars attached to the gantry and the new collars.

If you do those steps 1 - 4, it will be a lot more rigid but a problem exists if the original rods ( that is the I-I rods) are not rigid in the first place.

So, before you start constructing, you have to determine if the original rods are rigid - under load. A laser level is one way of proving this.

I also suggested putting the new rods under the original rods rather next to the original rods but on reflection that could negate some of the benefits of the supplementary rods.

Not elegant I know and a lot of mucking around but worthwhile considering.

Good luck

Andy

Hm. I'm going to try to use a little green loctite under the aluminum on teh bearing housing first.
;)

Here is one possible option.
Create a sub base for the machine to be mounted on.
The base would need to extend past the front of the machine and the back.
Then mount a cable system on this sub base.
An extension of some sort would need to be added to the gantry to connect cables to.
Rollers would be fixed to the sub base and turn buckels used to tention the cables.
Would think you could source everything required from a decent hardware store without it costing too much.
Paul

Yes, that actually makes a little sense. I'm shocked;)
It all depends on where the deflection is , I think. If it's in the bearing on the rod, I'm going to have to do something external. IF it's the carriage where it ataches to the bearing, though, that I can fix.

hmm. these ideas seem sort of difficult and unsatisfying to someone who just purchased a new machine... perhaps get some cheap $20.00 calipers(if you havent got any) and measure the bearing rod. then, unfortuneately, disassemble the thing enough to find out what sort of bearings the gantry is running on. maybe you can simply purchase some better bearings.. or bearings with an adjustable preload.. this could help..

i would be livid.

Resigned, more than livid. I talked ot the guy who built th e machine today and he told me I can't do jack about the backlash beause of the design. I CAN do something about the racking though.

Looks lie this machine will cut plastic and pearl and not much else.

racking meaning binding, i take it.... 1 side of the y axis lags behind?

i dont think its true you cannot do anything... i mean.. it just will cost $ and time.. however,such a machine was never going to cut much more than plastic or modeling materials anyways i would imagine... but it should still tradeoff ruggedness for accuracy..

are those screws actually plain threaded rod? id imagine a modest leadscrew upgrade on the y might fix alot of the issues. those look small.. perhaps 5/8ths 2 or 4 start leadscrew material would help you? its affordable..


also, mcmaster has ballscrew screw at $1.24/inch and the matching ballscrews are around $55. The trick there is bearing blocks and machining the ends... however, id bet even on good rollerblade bearings, and with a fair lathe job to get the screw to fit, it could be a big improvement... or, maybe not...

It'll cut wood, if I am patient. But it'll never cut square.

But it'll never cut square.

Sure, it will cut square (If you really want it to). You'll just have to do some work on it. :cheers:


I'm curious, are you the winning bidder that paid (Us) $1,562.00 for the cnc ?


.

No, I bought this one back in January and haven't gottenj to it till recently.

Ed,

Put the machine back on Ebay, take your loss and let everyone here help you build a machine that will work. You've had a lot of good suggestions but I keep seeing the statement "sure you can, it will only cost $$". I have been down that road with a couple of pick up trucks! The correct term is " sure you can, it will only cost you LOTS OF $$ and when you get through, you still have a junk heap!"

Mike

If he can't fix this one, hows he going to build one ? :rolleyes:

You already own a cnc, just make it so you can adjust the table! (and start cutting)

i would look into the easiest and most economical way to replace the y axis guides and\or leadscrew. both seem like they would help, but the leadscrew would probably be easier to replace.. let me suggest some ebay auctions..


http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&rd=1&item=7601077638&ssPageName=STRK:MEWA:IT
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&rd=1&item=7601101413&ssPageName=STRK:MEWA:IT
http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=7555991409&rd=1&sspagename=STRK%3AMEWA%3AIT&rd=1

Good luck!

Also, if you end up getting new lienar bearings and need some new parts machines to fit them to your gantry, id imagine someone on the site here can probably give you avery good rate to make the simple 2.5D aluminum parts.. If i had a machine that would work, I would do it for you.

This deepgroove person seems inept. It isnt fair that he thinks he can sell machines he has designed, then uses his design-flaw as an excuse! (but you know this... not to, you know, rub it in)

Robert.

Sometimes it's easier to start from scratch, than to fix "someone else's mess."

-- Chuck Knight

I currently don't have the resources to put much more into it other than time. If I had a choice I'd have a MaxNc MaxPath 42 or a moving gantry of similar size, but I don't have 5 grand. HEck, I don't have 5 bucks any more!
CLosed loop moving gantry, 3 feet in x and y, 8 inches in Z, would suit me fine. As long as it cuts square and within 5 thou or so.

moving existing screw to the side and slave to another on the other side through a timing belt would resolve racking and should be pretty cheap.

Doesn't interfere with electronics or require replacement parts only additional.

All it really takes then is time.

Andy

moving existing screw to the side and slave to another on the other side through a timing belt would resolve racking and should be pretty cheap.


That is not necessarily correct. We have no idea of the hardware in use and Ed_R would have to replicate what ever is being used. He has already stated he is on a tight budget.

Ed_R is there any possibility of getting some pictures of the bearings and screws. Its proberbly the only way we can provide more ideas.

Paul

Nto withouta digital camera, and I dropped mine a few weeks back.

What should have been asked first is can you send the thing back and get a refund?

If not.
Can't borrow a camera from someone?
Can you describe the screws. Are they ball screws, acme thread or just threaded rod?
What sort of nuts?
Is the thread securely mounted so it can't move back and forth?
Can you sketch the bearing assembly to give us some idea of how it fits together?

Paul

Nto withouta digital camera, and I dropped mine a few weeks back.

CVS has a cheap one use digital for $10 or $20 with an lcd screen. it is a 1.1 megapixel camera and with a couple of downloads and an easy to make cable you can use it over and over.

http://myweb.nmu.edu/~tstrothe/PV2mods/PV2mods.htm

this is the link I used to convert my camera

Ed_R

The consensus seems to be that more info is required.

Couldn't you borrow a camera or perhaps a mobile with camera?

Help can then be much more targetted and relevant.

Andy

I'll find a camera to use soon.

I've ran mills before - metal types (not router mills - except handhelds).
Being familiar with climb/conventional cutting and G41 and G42 ... ...
Have you tried using an dial indicator to rule out that the table is in-line with the spindle (x,y axis)? 1-2-3 blocks should be square enough to check this.

I don't think you have a gear/backlash issue if a CNC type dealing with the spindle:
Conventional is normally used on Bridgeports because of the gears causing backlash - thus added deflection.
CNC motors if (for wood) are similar for cutting metal do not use gears thus can climb cut resulting in better finish, tool life and less deflection. I would think at the high rpm's for woods there would not be much difference except hardwoods and the tearing that was mentioned earlier /as it apears climb cut is the way to go CW.


Check that the table is on first - a cheap indicator should read accurately enough.
If not available chuck a cutter shank down (or some solid ground dia)
Moving the table L to R secure a square block - tapping it know you have one side (one axis) in-line with your spindle.
Then go to the other edge (90 degrees) using a piece of paper as a feeler gage move in that axis direction check the catch points - if that block is square the table is off.

Good luck.