Inconsistant cutting errors.

[SOSx]

We are having issues with getting the mech mate we have bought up and running properly as it is cutting quite inconsitantly. We really need to get this working asap (like 2 weeks ago) as we have work for it backing up now.
So any help and suggestions are greatfully welcomed.

Firstly i will give you a run down on the machine.

Right we bought this 1200x1200 mech mate second hand on ebay from a sign company who bought it from the builder. When we bought it we realised it may take some sorting out being a self build machine, but after studying this forum thought it had the potential to be a good machine. Now that we have it we are fairly certain that it has never actually been got running propperly (and we think it was built 07-08 from some of the recipts that came with it).

When we got the machine back to our workshop we plugged it strait into our other machines computer which already had mach 3 installed and started it up to give it a little shake down and well thats what it did "shake". The base table was obviously very underbuilt with fairly poor welding and when we moved the gantry it wobbled like a jelly in an earthquke. So the first thing we have done is add a load of trianglulation to the base using 18mm birch ply (as we dont have welding facilities). With this we managed to get the machine fairly stable now with only the slightest of movement when under hard turns. We then cleeaned and re-lubricated all the moving parts and they all seem ok and everything rolls nice and free. The rails Have not been ground very neatly or evenly but it doesn't visually seem to effect the way it moves.

So we turned to the electronics next. It was only fitted with a very early model CNC4PC C1 control board. So we have upgraded this to the C32+ smooth stepper boards. We did this so that we could intall a little asus all in one PC into the control box. The machine came with Motion control stepper drivers - MSD556, Motion Control Stepper motors - FL86STH65 and 1 Motion control power supply - PS407 (40v @ 7a). We have left these in for now but realise these being under speced for a machine of this mass could possibly be the cause of our issue. We next fitted the C16 opto isolated limit switches.

Then we thought we would be ready to test the machine and start setting it up. Wrong! The Machine came fitted with a porter cable router fitted and on attempting some cuts at varying feeds, speed, acceleration and plunge rates it was evident that this was not going to be upto the job. So we went and bought a 3kw Chinese air cooled spindle (square type) + VFD from a local cnc supplier, got a new mounting plate machined for us and got this fitted.

So the machine no longer wobbles or shakes, it has a "decent" spindle and vfd and better control system. When moving the gantry round on the joystick (xbox controller) it all seems to run quickly and smoothly (and suprisingly quitely). We still need to fit e-chain and route the cabling properly as well as fit a few e-stops but other than that it seems about ready for testing.

So we started doing some movement calibrations to check the distance accuracy. We have done this using some engineering bocks that were acurate to around 5 thousands of a mm. We then used a dial guage on each axis to check the correcsponding movement. We appear to have got it to within a reapeatable tollerance of aound 0.05mm over a 400mm length on all axis.

Mach settings are 25.064 steps per rev 8000mm/min 700mm/s/s

Test cuts were made with 1/4" compression cutter 6mm depth cut at 2767mm/min in birch ply 24mm we ran a basic program with a square with a diamond inside of it, a spline, a circle, a diagonal line and a square.

This is the result.

http://i1108.photobucket.com/albums/h412/scattyv/CNC%20Cuts/2012-10-29103148.jpg

http://i1108.photobucket.com/albums/h412/scattyv/CNC%20Cuts/2012-10-29103022.jpg

http://i1108.photobucket.com/albums/h412/scattyv/CNC%20Cuts/2012-10-29103007.jpg

Some Photos of the machine.

http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-10-29104145.jpg

http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-10-29104127.jpg

http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-10-29104105.jpg

[smreish]

Mechanical Slop. It truly looks like grub screws loose on the shaft or z plate shifting during transitions.

Did you grind a little flat spot at 0 and 90 degree's on motor shaft and install 2 grub screws on the pinion?

[SOSx]

We did not, the pinion does have 2 grub screws at 90 degrees but the motor shaft does not have any flat spots. After reading several of the trouble shooting threads I thought this may be a possibility. I actually tried unhooking the springs thus releasing the pinion from the rack and gave it a tug with a set of large pliers (As suggested by gerald in another thread) but the motor shaft just turns really quite easily rather than the pinion, this is with the holding torque on.

[smreish]

But your turning the pinion on a motor shaft with no load. I bet if you tried the same thing with the motor active and pliers holding to "stall" the motor you would have movement.

It really looks like mechanical slop.

[SOSx]

Another thing I should add is that the motors installed on the machine actually have 12mm shafts not 1/2inch and the pinions are obviously half inch. Who ever fitted them
has fitted an extremely thin strip of brass wrapped around the shaft to act as a spacer of sorts. This does give them a very slightly un-even rotation along the rack. From our measurement tests this does not seem to have much effect. We were a little bit worried about removing the pinion as we think the brass sheet may disintegrate, leaving an extremely wobbly pinion?

[SOSx]

The motors were turned on at the time. Is there any way to get the motors to put more of a holding force on? As without too much of a force the shaft turned so couldn't really test how tight the pinion was.

[SOSx]

By the way thanks for the quick reply Sean. Much appreciated.

[smreish]

The pinions being the wrong dia could be an issue too.
Did you replace the pinions or are they still shimmed?

Do you have a tig welder? You could always just "tack" a spot on the end of the shaft and pinion as an absolute last resort. A quick grind and your back to new again.

[SOSx]

The slight difference in shaft diameter doesn't seem to be creating to much issue from what we can see. As when we did all of our setup measurements on each axis they were repeatable. The pinions are still shimmed as they were. We've been trying to avoid removing them in case the shimming falls apart. We did actually get a quote for new pinions from a gear company over here and it would cost another £160 (£40 each) which to be honest we don't really have spare. We don't have any form of metal welding but there are a few machine shops and garages around our industrial estate that might.

[bradm]

The shim gives extra opportunities for slippage, and you say "motor shaft does not have any flat spots" - I think you have to take that apart and file two flats on the shaft. Shims are easy enough to replace - and you get to drink the beer out of the can first ;)

[SOSx]

I have been thinking this. Right I'll be back shortly with an answer on if it is another case of slipping pinions.

[SOSx]

Well there are definite groove marks where it looks like the pinion may have been slipping on the motor shafts both on x and Y (still to get to the Z). I have just ground some flat spots on the shafts and my business partner Matt is making some new shims (using your method Brad nice one). So hopefully we will see a change in the cuts shortly,
Fingers Crossed eh!

[SOSx]

Well still not sorted. We have removed the pinions added flat spots and re-shimmed the motor shaft. But we are still seeing the same kind of issues. Another thing we noticed was the motors were moving about a little on the motor plates when the gantry was changing direction. So we tried tightening these up to pretty much fully locked and this helped a little but not much.

[timberlinemd]

Looking at the pic's. Doesn't look like your machine was built per plan. If the rest of the machine is as bad as the table looks, then I would be examming everything vs. the plans.

[bradm]

Check for slop in the Z-Axis v-rollers in both possible dimensions. (As per Sean's original list of likely mechanical issues). Given what you've found so far, it's likely there are a few other issues to sort. I assume that both the car and the gantry roll smoothly along their entire paths with all four wheels down?

Oh, and I can't take credit for the beer can shims - that appeared years ago here on the forum.

[SOSx]

Just home now. But just before leaving work we started having a good look a the z slide and it does have play in one direction where one of the V-rollers doesn't actually contact it propperly and looking at the cuts it does seem that it may correlate to the change of direction. How do you adjust the v rollers, sorry not looked at that part of the drawings yet? Matt did purchase a set so we had a reference source (as well as you lovely lot of course).

Steve. The base was definately not built very well but the gantry all looks pretty much to the drawings from what I can tell so far. Also the gantry moves nice and smoothly to full extent in both X and Y with no binding of any kind. We have already been looking at getting a new 8 x 4 base made with some nice new v rails and a pully gear reduction system. Infact it was one of the first things we did, but it will simply have to wait until funds allow (and that requires the machine working pretty much with the spec it has now).

[bradm]

Ideally, there are are offset bushings on those V-rollers that can be used to adjust them - a fraction of a turn should bring the V-roller closer or farther. But you'll first need to confirm that they are actually there on your machine - if not, then figure out how to add 'em.

[SOSx]

Well fingers crossed the builder realised there would be a need for adjustment post build. Will check first thing in the morning. Cheers for the help thus far. I know there is a good machine hiding in there somewhere.

[javeria]

Put the machine one and then try shaking up the Z assy like Brad has already mentioned, if not you still have one final mod left to the mechmate - a Z axis with linear elements.

My best to you- been there - done that :)

http://www.mechmate.com/forums/showthread.php?t=573&page=8

[SOSx]

Well! There were the right off set bushings on the Z slide rollers and that is now rigid in all axis. This has made another improvement to the cut quality but it has allowed us to see more issues this time with the machines rails "Hooray". The rails on the x axis are nice and strait but due to the poorly done griding of the rails we can see varying differences in height along its length. Both rails have there variations in different places and it means that at times the gantry is really only running on 2-3 rollers. This allows a little bit of slop in certain places. We have the same issue with the y car rails but it is actually more evedent here as in certain places along its length you can actually rock the Y car a little from corner to corner. Also one of the y rails was very slightly curved/bowed out at one end (although it didn't effect the car running in any noticable way). Matt managed to straiten this out quite easily.

Does any body know somewhere in the uk or eu to get Alu rails of the right type at a decent price as i feel this may be the only option?

[bradm]

I would be tempted to try to knock the high spots off the rails, if you concur that you can manage to not make it any worse. Can you give us a close up photo of what a variation in the rails looks like? It might trigger some group ideas as to how to jig up to knock it down. Do you have a grinder, or a hand held sander?

[domino11]

If the rail situation is not all that bad, it might just be a shimming issue perhaps?

[bradm]

I like Heath's idea (you put shims under the bolts at the low spots). Also, note that there are slots to adjust the V-roller heights at at least one corner each of the gantry and car (can't remember exactly right now), so it may be that you can find a compromise setting that doesn't actually rock at any locations. Some combination of the roller height adjustment and shimming might get you home.

[SOSx]

Will try and take some photos of them in them morning. We have both grinders and sanders but I dont think I would want to try it free hand (as it has obviously been done). At a minimum I would want one of the rail grinding jigs to do it (we could potentially cut one using our other small axyz cnc machine). Personally i like the bolt on rail style more than ground (due to less possibility for error).

Heath: i will have a look to see if any of the low spot correlate with the bolt holes. my gut feeling is the variation is a bit to randomly spaced for it to work effectively. But all options are open and will be looked at!

Brad: We'll ceratainly be giving it ago but again I feel that if we adjust it in one place to get it all level from one of the dips it will just rais another wheel at a high spot.

I'm not sure if

[vickers2012]

Hi all

Thanks for your help so far.
Here are some photos of the rails.


http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-10-31101053.jpg
X-Rail

http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-10-31101104.jpg
X-Rail

http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-10-31101123.jpg
Y-Rail

http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-10-31101157.jpg
Y-Rail

[Surfcnc]

Nooooo !!

Such a shame, I can see nearly every error one could make in those shots.
The beauty of the Mechmate Design is that the rails can be remade fairly cheaply as angle is inexpensive.
Contact your seller and get the grinding skate they used or have a good look at the site and make your own version of it.

I would recommend remaking the rails for several reasons.
1 - the cut off height is inconsistent
2 - the rails are not ground evenly on each side
3 - the ends of the rails are ground differently to the middle of the rails.

The ends get like that because the maker did not allow and extra margin on the length when grinding the rails.
If a bit extra has been ground then you can cut any wobbly bits off the ends.
it is not my idea but I always push this one as it is just so good to "sand" the rails not "grind" the rails.
Use a grinding disk as a backer for a course sanding disk and the removal rates will be so much faster as the sanding disk will not glaze.
You can also use 120 grit as a final polishing pass, they will come up like a mirror.

You are correct in saying that there is a good machine hiding in there somewhere and with some effort you will be rewarded.
I note that the motors are smaller than spec so once your going you will need to slow the feed rates down a little to compensate.
Check the wiring quality as well as the specification is generally for a braided shield on the motor wires.
It may not be necessary to use shielded wire but if signal noise becomes an issue down the track it is always a good thing to know about.

Good luck.

Regards
ross

[SOSx]

Well we are have now removed the y and z cars so that we could check the rails and bottom of tha y car spacers more carefull. The Y rails are not strait or perpendicular to x. So we have unbolted them now to get them strait and well go from there.

[SOSx]

HRoss,

I think your post came in just as I was writing that one. Its such a shame as we bought this one thinking it would be a quicker route than either getting a chinese machine or building our own. But I am fairly certain we could have had one built in a similar amount of time as we have now spent messing around with this one. But hay ho we live and learn.

[SOSx]

Oh yeah the motor wiring does all have braided shield which we have grounded.

We had this problem when we retro fitted our old AXYZ machine.

[bradm]

Yeah, Ross covered it. Those rails are probably not worth trying to recover :(

Great photos, though, they make the diagnosis easy. You guys should be in the home stretch now, though.

[SOSx]

Right, Well. We have straitened the y rails and have the y car sitting much more firmly on them. This has improved the cuts some what but we are still seeing an issue as the cutter enters and exits the timber. We are also having calibration issues.

http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-11-01214654.jpg

http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-11-01214700.jpg

http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-11-01214713.jpg

http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-11-01214719.jpg

http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-11-01214723.jpg

We used the rack and pinion calculation spread sheet to work out the steps per mm and the machine would cut short of the correct distance.
We have 20dp Pinions and rack with a 20T Pinion gear and are using Motion control stepper drives MSD556 set to 10 micro steps giving a steps per mm of 25.064.

We have then tried to use the mach 3 axis calibration function for steps per unit, as for working to a measured distance. We did this using a dial gauge and the setup as shown below.

http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-11-01221209.jpg

http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-11-01221217.jpg

http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-11-01221236.jpg


Even after doing this (and both being spot on to the dial gauge) we are still finding the cut length to be short by 0.3-0.4mm in both x and y over a 150mm cut. The amount of error does reduce/increase as the distance gets shorter/longer. Also something that seems odd is that to get this the steps per mm setting are different from the x to y different.

We have actually pretty much finished designing a new 9x4 base table and have found a supplier for the bolt on rails (actually only 30-40mins away).

http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/3D-Model.jpg

But the simple fact is we need to get the best out of the machine at hand now (well 3 weeks ago really!!). We simply cant afford to have this made or even get new rails until we can free up some money from selling some old assets.

I think for now if we can at least get the steps per mm right for the distance on both x and y we should just about have an "acceptable" cut quality with "reasonable" accuracy.

[bradm]

You're making great progress! The replacement table looks good as well.

For the calibration question, check your acceleration / deceleration values. If they are too aggressive, you might be losing steps, which would explain your symptoms. Use conservative settings to get consistent calibration, then ease them up as you find the limits of your machine.

Great photos, again: clearly, you're experiencing a little wiggle across the Y / in the X dimension, which shows up on the lead-in and lead-outs. That could be in your spindle mount, in the Z plate, in the Y car perpendicular to the rollers, or in the X pinions. Check for where the flex is. One thought is that the hold down idler opposite the motor on the Y car isn't doing it's job; it could be any of the springs not holding their motor or idler firmly.

[Surfcnc]

Hi Brad

Take a look at the springs on the y car motor and the Z axis in the earlier pics.
They are long and not very taunt, you probably picked one of the issues there with the Y car moving around. SOSX remember their is another spring on the backside of the car where the roller bearing is too.

SOSX you are persistant, so you will get there. If you did a good job on the flats and shims for the pinions that is less likely.

Cinch the car down and see what gives. That also makes the x axis pinions probably less than firmly engaged if the same springs have been used.
From memory you have to pull around 12 kg (but don't quote me on that)

Regards
Ross

[SOSx]

Thanks Brad, it feels like were going round in circles.

Were about to run a load more test cuts in a minute after lunch.

[Surfcnc]

Also check the motor swing plate on the side opposite the pivot point that it has not run out of travel.
In some circumstances the bolt can bottom out on the end of the slot in the plate and hold the pinion slightly away from fully engaging in the rack.
There are two countersunk holes where the bolt can be mounted so it might need to be shifted upwards.
The symptom of the rack not fully engaging is a very small clunk when the car is pushed along the Y axis. Same for the x axis for that matter.

While we are at it, there are also supposed to be PTFE or teflon washers behind the pivot points and the bolt in the elongated slot in the motor plates to ensure the plates move freely.
Any plastic that has been greased up will do as a substitute. Before I found the right stuff I used a thin plastic cutting board.

Regards
Ross

[SOSx]

Hi Ross,

We had already tightened the springs (Drilled new bolt holes for the bolts they attach to) and the pinions do seem well engaged with the rack and the Roller bearing seems quite firmly against the steel. But we will look at it again. We have considered they may need to be heavier duty, but it does take a fair bit of effort to get them engaged. We had also noticed that we were missing the teflon washers and the lever action is a bit sticky with just steel washers. We have had a look but cant seem to find them any where. We can get hold of nylon washers as a possible substitute if they will be better than the steel ones?

Just about to run some more cuts and have another look at it.

[Surfcnc]

They need to glide smoothly and at the same time be free of sideways slop.
This keeps the pinion running true in the rack and also properly engaged.
Ross

[SOSx]

The rails and rack do slide smoothly and the pinion seems well seated.
We have just added some more tension to all of the springs for good measure.

Does any body know where we can purchase teflon washers in the uk or europe? I'm not sure we are getting the right ballance between being nice and rigid yet still being able to slide smoothly. As it is, they are either nice and rigid yet don't slide smoothly or they slide smoothly yet have a bit of flex sidways.

I think the biggest issue we are having now is getting the steps per mm to correlate with the cut lengths. As it just isn't right to the calculated settings and for some reason our measurments over a 400mm measured length using the dial guage are not working out either.

[danilom]

If the machine travels lets say 1000mm and returns to the same point then its ok in mach, but you may have bad quality rack with inconsistent tooth pitch maybe

[Surfcnc]

Washers can be cut from 3mm Teflon sheet with a hole saw.
Your machine will do a good enough job cutting them as well at this point.

Are you checking for slop with the motors turned on?
The images still indicate mechanical slop as the primary problem.

Drill a hole at a known coordinate, run a cutting sequence, then jog the bit into the drilled hole and check the Mach3 DRO to confirm you have not lost steps
Reduce your motor microstepping to half step to increase the motor torque and run the cut test again if you have lost steps.

As Danilo suggests, measure the accuracy of your racks.

Regards
Ross

[cdev]

I have some 2mm PTFE and can punch out the washers for you - (5 off 40 x 10, and 8 off 20 x 8). PM me your address and I'll send them out first class on monday if you still need them.. I used a slightly different combination but the ones listed above are what the drawings show. 1 off 40 x 10 on all motors and the idler and 2 off 20 x 8 on all but the z motor
Chris

[SOSx]

Hi Chris,

That would be great. Sending Pm now.

[zumergido]

i use washers made of plastic from a pet bottle. works like charm.
letme see if iam getting this right.
you are getting errors of 0.4mm on a 150mm travel? you have belt reducion? if not how many teeth you pinion have and module rack?
i have 24teeth on direct motor module 1 rack i think i have 0.3 mm error but i found that normal. do i have to spect more accuaricy?
maybe its a good idea to do a .tap normalised test file on mm and inches to compare machine errors. using most standard bits. like 6mm
btw i have a 6mm bit that is 5.96 mm

[rischoof]

When I see your cutting pattern i think you have some clearance somewhere where you should not have. Especially the begin and ends from your lines are moving away.
Check your spring load. should be 12kg I checked my springload whith a bucket filled with 12 liter water. and measured the lenght of the spring.

when your machine is running (cutting) you can place your hand on the y cart and feel it. movements should be smooth.

power you motors on, and try to push the x, y and z slide by hand. if you can move them, it's not OK. should be solid as a rock.
1) check if the gears are fixed well
2) check spring load
3) check if your nuts for your motor mountings are not to tight. check this by removing the spring, you should be able to move the motor up and down
4) check the bolt which are used to "hang" there should not much clearance between hole and nut
5) check if you x slide and y cart are proper aligned in the guiding rail
6) check if you gear is nice in your rack, it's possible that the flange of the gear is touching something, so your gear is not properly in the rack
7) are your V bearings mount properly, or do you have clearance at the V bearing, check if the v-groove piece is fixed on the bearing.
8) is the same type rack and pinion used or did they mixed up metric / imperial

Teflon sheet can be easily cut with a heavy scissor. They don't need to be perfect round. drill a hole in the middle and ready (place the Teflon between two pieced of drilling to avoid that the teflon "run's" in the drill

[SOSx]

Well an update. We have managed to get things a fair bit better.
We now do have it repeating cut lengths in both x and y to within +-0.1mm. Phew just about acceptable!

Fernando: I would certainly want and expect a better tollerance than 0.4mm from any CNC machine. We achieve pretty much +- 0.05 on our other machine. Before doing the test cuts we measured our tool 6.2mm rather than 6.35. This wa taken into account previously. What we ended up doing to get this right was we wrote a programm with a load of equal length cuts in each axis and took the average length of them all with a set of vernier calipers. Then knowing the steps per unit, pulse rate, pinion siz and pitch (20dp, 20T) etc we calculated back to what it needed to be. Just out of note the X was close to what the theory says but the Y was a bit off.
http://i245.photobucket.com/albums/gg62/sostmalone/20121104_154529.jpg
http://i245.photobucket.com/albums/gg62/sostmalone/20121104_154511.jpg
Still not quite got the z as accurate but I think with the rails as they are we are never going to get the same tollerance.

After much fetteling with the z plate, rollers, concentrics etc we do have the run out of cuts looking much better but we still are seeing it particularly on the plung in. On panel cutting it doens't really show up on square profiles but does a small amount at the plung point of circles.
http://i245.photobucket.com/albums/gg62/sostmalone/20121104_154615.jpg
http://i245.photobucket.com/albums/gg62/sostmalone/20121104_154622.jpg

The first couple of jobs are relatively simple cabinet work so I am just about happy enough to run them as it is. Last night I cut and attached some 18mm birch ply which
to day I am in the process of turning into our vacuum table.
http://i245.photobucket.com/albums/gg62/sostmalone/20121103_203943.jpg
This morning i skimmed the panel with a 2mm cut to level it off and at the moment (while i type now) it is cutting out the 6mm deep channels. All a bit labourious as the program for the rebated channels is 4 & 1/2 hours long and the surfacing was 2hours. One thing that was nice to see earlier and confirms the machine is not loosing steps is that even after surfacing the panel with a 1/4" end mill (yes I know, but we dont have any larger tools at hand) the machine went back to the home position I had set perfectly.
http://i245.photobucket.com/albums/gg62/sostmalone/20121104_144930.jpg
http://i245.photobucket.com/albums/gg62/sostmalone/20121104_152511.jpg
http://i245.photobucket.com/albums/gg62/sostmalone/20121104_152452.jpg
http://i245.photobucket.com/albums/gg62/sostmalone/20121104_154457.jpg

Hi Rik: The gears do seem quite well seated but I think I am still going to check the spring strength (i'll try your bucket and water test). We have some teflon washers coming thanks to the generosity of our fellow member cdev (Chris), hopefully this will help with the smoothness and rigidity of the motor mount action. We've been though and re-aligned/ straited our y rail and the Y car is now well seated all through its travel and the x already was, but of course the rails are quite inaccurate in the first place which will through things off a bit as it moves. One thing you mentioned may be worth a re-check one of the v-rollers on the Z slide did seem to have more lateral moevement (in X) than the rest, I may take this off and have another look at it to see if the v roler is propperly seated on the bearings.

Any way will keep hunting to try and track the last issues down.

ON a side note what vacuum pumps / motors are most mech mate users using?
We have a nice rotary vein Becker unit on our other machine but couldn't afford a new one of them. We did miss out on a nice Busch roary vein that went for penuts the other week but havent seen anything like it since for sensible money.

Anybody here using the shopbot style cabinet with multiple Lighthouse vacuum motors? Anybody know of good european equivalents (have looked and cant find lighthouse ove here).

[Quadro]

SOSx,
What feeds and speeds are you using to carry out your test cuts?
Looking at the photos it maybe the bit flexing or the stepper motors not holding on the micro step.
To determin which problem mount the dial indicator on one axis and cut a straight line only with the other.
Anthony

[zumergido]

your v wheels have bearing? there is any loose on the them? i did have problem with them becouse one loose wheel.

[SOSx]

Hi there,

Many thank to Chris (cdev) for the teflon washers. It has definately improved both rigidity and action of the stepper mounting plates.

Antony were fairly certain were not loosing steps, tool bit returns to home bang on every time.

Still haven't had a chance to check the v wheel I was concerned about yet. Next thing on my list is stronger springs. Have looked at the specified spring and cant find anything identical with out having them custom made. Firstly does anybody know of any springs that are a direct fit in the uk or europe. Or the other option is I have found a possible substitute what do you all think of the T42480 on this page
http://www.assocspring.co.uk/p/16/extension_springs-stainless_steel/?PAGE_ID=1&INDEX-order=0&min_1=14.2&max_1=18&min_2=1.5&max_2=3.2&min_3=46.4&max_3=124.3&update=#startcontent

It has a slightly higher maximum force about 16KG but this is when at maximum pull which is 20mm longer than I need.

What do you think?

[cdev]

I used these springs. http://www.strutsdirect.co.uk/components/product_info_springs.php/cPath/55_80/products_id/154
They come in at under 12 kg but by drilling a new hole about 20mm further out in line with the other holes but clear of the v rollers I get a bit more than 11kg force. The only thing to be careful of is to make sure that the stepper motors can take the the radial load. From memory, if the spring tension is around 12kg then the radial load on the motor shaft is around 20kg. I'm sure someone will correct me if I'm wrong.. This only applies to direct drive motors of course..

[Surfcnc]

SOSx

Not in the least bit surprised the telflon washers improved the machine over using a steel washer.
How the original builder could seize up a joint that is supposed to move is beyond me :(.

Did you actually test your existing springs on the machine using the bucket method.
I wrote earlier they seemed a long way off the mark.
If you have a figure then it is always good to feed it back to the forum.
It shows the rest of the builders by how much they were out and gives others a chance to learn.

The table surfacing time you gave was too slow, way too slow.
I suspect even with a 6mm end mill to take 2 hours on a 4x4 you have not used the tool step over correctly.
There is no need to use a 10% stepover to surface a table - use 90%.
The operation removes very little material it is more like jogging than cutting, very little change of lost steps.
Smooth the motor acceleration on the x axis and wind up the speed - time is money you know :).

I also suggested that you lock the motors and wiggle the spindle.
You absolutely NEED to do this as your cut still indicates mechanical slop in the Z axis or a loose Y car pinion.

Regards
Ross

[SOSx]

Well time to back track a bit now.

It is not after the last calibration cutting distances accurately.
We thought we had it sorted with the method I described earlier.
We can get it to cut a specif distance with this method accurately and it will repeat that cut to within +-0.1mm but as soon as we do cuts of different lengths the measurments go out. They will repeat quite happily but they are still out and were talking 1-2mm here over a distance of 600mm.

We thought that it may be that our rack goes out of pitch or is not very accurate to it supposed pitch (or even if it might actually be mod 1.25 not 20dp), but today we have taken off both of the x racks and both meshed them together and measured them/counted the teeth and they are definitely dp20 and they mesh together nicely so it is not the rack. As Ive said before the error seems to be an incremental increase over distance which would suggest the settings for the steps per mm is not correct. But the theoretical calculations are out and every way of calibration that we have tried has not yielded success other than at a specif distance in both x and y axis.

We have still not been able to isolate the mechanical slop in the z slide although upon giving it a good tug you can feel it is there. one thing we thought might be cause of error is that the rack is not running perfectly strait along the bottom of the angle section it tapers in toward one end. In particular you can see this in the x rails as it is about 3-4mm difference from end to end of a 1700mm length. In the Y axis it is only maybe 1mm difference from end to end.

http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-10-31101053.jpg

Could this through out our steps per mm calculation by this much? Another option we have thought about is that as I've described our motor shafts are 12mm yet the pinions have a bore of 1/2inch. So we have them shimmed with drinks can sheet wrapped around. This seems to work quite well and definitely centers the pinion much better than the copper sheet shims that were in there before, but they do still leave the pinion with a slight wobbly rotation.

Surf cnc we did not actually measure the springs in the end with the water bucket method, but upon measuring there diameter 12mm and wire size of 1.4mm it is clear to see from looking at LOTS of spring specifications they are not able to produce 12kg of pull, maybe 6-7kg at best. But that has also raised another question our motors are only rated for 60N lateral load which is only 6.11Kg:( Are we going to kill them putting on the correct 12kg force springs (or 11Kg ones cdev recomended)? Maybe a gear reduction system for them is our only route there or larger geared steppers or a belt reduction. But to be honest at the moment I really wouldn't want to through any more money at it for things like that with out knowing they were going to sort our root problems out (which I don't think they will).

As for the surfacing of the vac bed I did a 2mm pass at 18000rpm, 3780mm/m and a step over over 75%.

To be honest this machine is really bugging us now, in many ways the little movement/slop on the z axis on entry and exit seems like the least of the machines issues at the moment seeing as we cant seem to find a way of getting to cut more than one specified distance correctly. Please dont take me as slating the Mech Mate design in any way, I actually think it is a very good design its just we've been unfortunate to buy one which unknown to us at the time implemented the design very poorly.

I really dont know what to try next?

[bradm]

What are your acceleration settings? Set them so slow that you can see and hear the accel and decel on every move, and try the calibrations again. You are still describing something that matches up with missed steps, which will vary with the length of the operation, but be somewhat consistent for a given length.

[danilom]

Try changing the computer, it may be not giving nice pulse train, if all mechanical issues have been checked turn to the electronics.
You can also check is there a nice 5V on the dir pin at the drives (as its harder to measure it on the step input)
Also you can try changing outputs for STEP/DIR , as I found on one computer that it would not output nice signal on only one output pin, PCI LPT port is another idea to try as its cheap.

[SOSx]

I Dont know if Matt has seen the replies but if he does and can remember hopefully he'll put them up.
Brad surely if it was loosing steps it would'nt return to home correctly?

Hi Danilo-we have a smooth stepper board conected to a fairly decent spec all in one micro itx pc.
We have had that issue before with different computers on our other machine, but had pressumed the smooth stepper board would remove the possibily of this.

[bradm]

You are correct that losing steps would destroy repeatability on the return to a known location. So, if you have repeatability after cutting the inconsistent lengths, it's not lost steps. Of course, you could have repeatability of jogging, but loose steps under the cutting forces. If you have the repeatability under cutting conditions, but still have cuts of variable length, then we're either looking for a software issue, or a mechanical issue - we've exonerated the drivers and the motors.

This has got to be driving you guys bonkers. Wish we'd been able to nail it for you earlier.

[Surfcnc]

Ok it's got to be annoying by now, however

1. Ignore any "motor spec" and just pull up the pinions to the "machine spec". Replace the motors with the right motors if or when they fail.

2. Calibrate over a long distance and accept the Mach 3 calibration then do a cut to ensure it agrees with the long calibration when cutting.
When in doubt calibrating over a long distance reasonably accurately beats calibrating over a short distance very accurately.

I would just like to confirm that you are using both the "Gear Steps Calculator" and then secondly the Mach3 axis calibration settings using the "Settings Tab" also labeled ALT F6.
The individual axis calibration routine is located directly above the Reset button.
If X axis needs different steps to the Y axis to be calibrated then so be it.

3. Your electronics laundry list seems pretty good quality.
Not much to be gained here but just kill the network when cutting, don't scroll the tool / table window around when cutting, run an AV scan, update windows... so the normal cautions should be fine.

4. The elephant in the room is the pinions and the racks.. Over the longer term consider some new pinions and have them correctly machined for your motor shafts.
The wobble has the effect of moving the gears slightly out of mesh twice on each rev.
The alignment of the rack to the pinion is also not perfect so as the pinion traverses along the rack it may not be fully engaged along the full 15mm width the rack.
So rather than a nice 100 percent seat you have both the pinion crawling out of square with the rack and the contact area changing as the pinion traverses along the rack.

Spraying the rack with cheap mat black paint will show you the contact area of the pinion on the rack after a few machine cycles to wear the paint off.

It is possible to pack the rack square to the pinion along its length using shims (the same drink cans as the pinion shims). If the rack is consistently out of square Gerald has even recommended bending the motor plate in the past.

Regards
Ross

[KenC]

Ever check for movement in the router, its mounting bracket & worn bearing ??

[SOSx]

We'll back to it this morning. Once again thank you all for your input and support in trying to get his sorted. It is most definitely driving us up the wall a bit now as its just putting us further and further behind our work schedule and the idea was to free it up.

Brad- we do seem to have repeatability under cutting conditions. If we repeat a cut or program it will be the same roughly +-0.1mm, it will just be wrong?

Ross - As for motors/gearboxes I am just waiting to hear back from motion control now regarding prices for there planetary gear boxes and rotary encoders, as I think that more resolution and torque can only be a good thing and with rotary encoders we would know for certain if we loose any steps.

Also I am hoping that the planetary gear boxes can take a higher lateral load so be more capable of surviving the forces applied by correctly rated springs. Also we could use (for now) the 20T 1/2 inch pinions we have currently with out shims as we can get an output shaft of 1/2 inch not 12mm like we have now. I know it would be a bit under geared but I would sacrifice a bit of speed for resolution right now.

I can confirm we were using the gear steps calculator (and confirmed it in excel manually). We have also used the mach 3 axis calibration function. I think we have some black spray paint so will put some on tonight and have a look at the wear tomorrow.

You say kill the network when running? The wifi is disabled but the wired is left active, but we have never had any issues with this and our other machine?

As for calibration we started at 150mm using a small set of verniers then as we could get it accurately this way we then bought a dial gauge, borrowed some 0.0005mm accurate engineering slips and re worked using this method but i think that the tiny differences in the dial gauges angle in comparison the engineering slips face threw those test off. Again we could get it to repeat the one measurement accurately but as soon as we tried other measurements / cuts it would be off again? Then the last 4-5 days we borrowed a 1m long set of vernier calipers and re-did the calibrations using averaged cut lengths and this again would get it to repeat a 980mm cut or movement time and time again but as soon we cut a shorter length it was well out again?

Ken- We had a brand new 10mm thick alu mounting bracket machined and it seems pretty solid and the motor is firmly attached to it. I Think the slop is more in the Z car/slide i think?

While I've been typing this Ive just heard back from motion control and spoken to their head engineer. In case anybody was wondering on lateral forces on the shaft he has just done the calculations and for a pinion at the end of our motor shaft and with a 12kg spring load we are applying just under 20Kg force on the bearings and that they are quite comfortable much higher than this (he said more like 200kg before they break). For the gearboxes hey have given us a price of £170 (about $270) each for the encoders they are £70 each (about $110). Are these prices reasonable are htere cheaper/better options?

[KenC]

Why not re-build the Z-Spider? that should cost less than 170quid...

[KenC]

Some close-up pic of your Z-axis will help a lot. Namely The slide plate's V, the bearing assembly.

[SOSx]

Hi Ken,

I'll take some pics of it in a bit.

[Surfcnc]

Back to basics, are my assumptions correct here (from previous posts)

1. 1.8 degree 200 step per revolution stepper motors
2. 20 tooth module 1 pinions
3. 10 microstep setting on you MSD556 driver

My answer is 38.831 steps per mm ??

All that stuff about tuning things off was just to reduce other possible sources of error.
I'm a massive fan boy of reductionist problem solving.
I run my machine with the wireless on all the time.

So what gives with the steps, have I got something wrong.

Regards
Ross

[Surfcnc]

8 microstep = 25.465 ?

Ross

[viz]

Back to basics, are my assumptions correct here (from previous posts)

1. 1.8 degree 200 step per revolution stepper motors
2. 20 tooth module 1 pinions
3. 10 microstep setting on you MSD556 driver

My answer is 38.831 steps per mm ??

All that stuff about tuning things off was just to reduce other possible sources of error.
I'm a massive fan boy of reductionist problem solving.
I run my machine with the wireless on all the time.

So what gives with the steps, have I got something wrong.

Regards
Ross

I got 31.831 not 38.831

Another one with 8 microstep is ok.

[SOSx]

Hi Ross - 1.8 degree 200 step per revolution indeed. But running on dp20 rack and pinions (20T indeed).

[SOSx]

Oh yeah sorry, yes we are set to 10 micro steps on the MSD556 drive.

[SOSx]

we made that by the maths to be 25.06337

[Surfcnc]

Also thinking out loud...

The mechanical slop in the z axis may be from errors in the Z slide grinding.
It may be something as simple as being tight at one part of the z axis stroke and loose at another part of the stroke.

The check is to run the verniers over the width of the slide in several places to see if it has any significant variation in width.
Maybe tighten the V roller eccentrics while the Z slide is at the bottom of the stroke (where it does the majority of the cutting).

Ross

[Surfcnc]

Typo, thanks viz

Ross

[SOSx]

That does make sense. The rail grinding on the z slide is a little better than on on the x and y but still by no means exact. I had thought they were certainly contributing once again as we had tightened the concentric bushes as much as we could with the slide up.

[rischoof]

When I see this cutting pattern, it's clearance some where. If you take your router in your hands, and you put some force on it, you must be able to move it.
put a magnet holder and a dial indicator to the different axis and figure out where you clearance is.
springs should be any problem to get, if you buy a longer spring, make it shorter. if you can not find the wright strength, mount 2, check my building log, I explain with pictures how to create the right spring. Yes I am spoiled, there's one street in the middle of Taipei with 4, yes 4 specialized sping shops, and a shop where I can buy teflon next door, but even across the supper marked in my naberhood is a shop run by 3 ladys were I can by the proper springs it's open to 22.00 in the evening.

that rail, I can see from 16.000 km away that's not straight at all
I was able to make a rail within 0.03 mm (not inch!) flatness. check my building log to see the measurement device I used.
When I should have this kind of cutting quality, I should not be able to sleep.:eek:

[SOSx]

At the moment were just working out how to re-fix / re-drill the x racks and rail so that it sits both square to the rail from end to end and not tilted down its length. We actually just
tried shimming with a few washers on the outer edge to see if it corrected the slight tilt but it didn't seem to have much effect. The problem there is the bolts that hold it on actually go through the angle section right on the curved part of the angle and due to this have not been drilled strait (it looks like they were drilled from the bottom though).

[SOSx]

Rik - in fairness Ive not been sleeping properly due to this b****y machine. And yes there is definatly some play in the z axis but we have been having a hard time tracking it down. As for the rails we know they are not good but at the same time they should not be causing errors in our calculation of steps per mm and this is my biggest concern right now. We can work on the slop but unless we can get it running different cut/travel distances correctly its absolutely no good to us replacing the rails.

We have ordered springs this morning so they should be here Monday or Tuesday. You are lucky to have supply shops like that locally. There are only a few specialist spring suppliers nationally over here. When I get a chance this evening i'll have perusal of your thread.

[Surfcnc]

Keep up the good work.

OK try 31.831 steps per mm. Sorry about the typo adding to the chaos.
We use Gerald's gear steps calculator, it is an excel spreadsheet to calculate the steps per mm.
It is so powerful it can make dumb guys like me look smart :)

Ross

[bradm]

I think you've proven that you aren't loosing steps, so I wouldn't spend money on encoders - you can read more opinions on encoders in general here in the forum.

Try Ross' long calibration method, and see how that works out. He seems to be working on a good theory, which is that pinion wobble is making it impossible to get accurate measurements for calibration at short distances. A 3mm error over a 1m calibration set is a more accurate result that a .5mm error over 150mm calibration set.

[SOSx]

Hi brad the last calculations / test measurements were over 980mm in both x and y as
our shorter measurements were not giving us good results on larger measurements even though they were less than +-0.1 repeatedly. So we moved to long measuments after borrowing some 1m long vernier calipers. We were able to get the machine repeatedly to cut the 980mm it to within +-0.1mm. so the results matched the program. So after running these test and setting it accordingly so that x and y correlate over this distance we decided to run a couple large panel test cuts of 600x600mm squares.
what we got was basically 601 in the x and 598 in y? It just doesn't make sense.

[Surfcnc]

Apologies the numbers are wrong that I gave you for your DP20 rack and pinions.Went down the wrong path there not paying proper attention and specified the steps for module one rack and 20 tooth pinions.. but you already know this :o
Ross

[SOSx]

No problem Ross. Yeah had twigged onto the mistake.

[smreish]

Mach 3 system has a variable that might affect that.
In the General Config page you will see:
Motion mode
(constant or exact)
Distance Mode
(absolute or incremental)

Try toggling into exact and absolute mode and try your cutting profile. Did you see a change from the original results?

[SOSx]

Will try that tomorrow once we've got the racks back on properly. Once again its not proving to be an easy task. Thank sean

[KenC]

If the exact stop mode test fail, What I'll do is dismantle the Z-axis -> inspect & measure the parts -> re-assemble it the correct way.

I suspect the plate V isn't parallel enough. tight at top loose at bottom or vice verse. This can't be fix by turning the eccentric bush...

[Gerald D]

On the Z-axis, all the rollers must be on the same flat plane, and the slide plate itself must be flat.

Then the eccentric bushes must set the roller pressure quite tight (a lot of "springy" preload). If the holes for the bushes were drilled incorrectly (too far apart) in the spider, then the holes on one side need to be welded up amd re-drilled about 2mm closer to the other side.

[SOSx]

Hi Gerald,

I think we'll be looking/dismantling the z Slide tomorrow now as we have only just finished getting the racks fixed back on strait this afternoon. They are certainly much better now anyway. Hopefully that is another issue with the machine ticked off now.
http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-11-10170128.jpg
http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-11-10170120.jpg
http://i1108.photobucket.com/albums/h412/scattyv/Mechmate/2012-11-10170040.jpg
After a couple of attempts at re drilling through the rack and rail/angle section we decided that we would just try the tape on its own as this way we would not have bolts not drilled and tapped perfectly square twisting the rack. As you have suggested previously Gerald this is actually quite adequate on its own with out the bolts. In fact we have actually just used a good quality automotive tape that we use for quite a few things and to be honest unless you cut it off it is not moving on its own. We did make sure we prepped and cleaned both the rack and bottom of the angle section very thoroughly with both acetone and IPA before applying it then clamped down its length to give a decent amount of pressure to the join.

Sean (or anybody else) - Any chance you can explain the difference between the setting you have noted? They are not ones we have used or altered previously. And I really have feeling that the distance traveled / steps per mm issues we are having could be a software or calibration issue.

[Red_boards]

Forgive me stating what you may have already done, but have you tried the automatic steps calculation from Mach 3? Put a pen in the spindle and move the gantry 2m a few times, measure the actual distance moved and let Mach calculate steps (or you could fit your fancy meter at the end of the 2m stroke to get an even more accurate measure). I did a few rectangles 2mx1.5m for precision (repeatability) purpose.

Remember, you may be able to calculate to 3 or 5 decimal places, but the decimals are only accurate to the number of places you did your measures to. Wikipedia significant figures entry quote:
For multiplication and division, the result should have as many significant figures as the measured number with the smallest number of significant figures.

For addition and subtraction, the result should have as many decimal places as the measured number with the smallest number of decimal places (for example, 100.0 + 1.111 = 101.1).

When performing a calculation, do not follow these guidelines for intermediate results; keep as many digits as is practical until the end of calculation to avoid rounding errors

Ithink this means that, if you can measure 2000.0 (150.0)mm, then you can have confidence in xx.xxx (xx.xx) steps.

Only tape holding my racks on for a year now. No signs of weakness, even though the motors only run over the racks once a week.

[SOSx]

Hi Red - Thanks for replying but we have been using the calibration setting and taking measurements over 980mm with a good quality set of engineering vernier calipers.
I am hoping that having the racks straitened/squared properly will help with this incremental error but haven't had a chance to really check it properly as we are working on the z slide at the moment.

Gerald - The holes for the V rollers (W3) are drilled at a spacing of 138mm and the Z slide measures roughly 99.3mm across from point to point. There are captive bolts welded on the back of the slide. On the Drawings it says the spacing should be approximately 132mm depending on v rollers and z slide.

Sean - we had a look at the Mach 3 manual regarding those settings and they appear to only to refer to the post processing reference/origin points. Ie one references the origin for all cordinates from 0,0 the other takes the origin as the previous point. Our post processing is set up for all points taken from origin.

[SOSx]

Gerald dont worry just found your post regarding this subject http://mechmate.com/forums/showthread.php?t=878:o