Taking some slop out of the Bot

[chunkstyle]

Thanks for the input Brady,

IF the machine's specked at +- 1/64" over length then have I reached it's point of diminishing returns?
By spinning one of the boards around and placing it next to the other one, isnt any resulting gap between the boards a doubling of any deviation from a straight line? If so, then it would seem that I have gotten it as far as factory tolerance? The pic of the MDF cut results show a gap in the middle of the two boards (one of which was spun 180) that is apx -1/32". Divide in half and were at the 1/64" deviation from straight. Would this mean then that I'm out of the rain and on the porch as far as the house of shopbot PRS is concerned? Do you think it could get better than this?
Also, with respect to the rail dip, I had to take out shims after it was found that the middle sagged and the middle feet had to be lowered. One side needed none whereas the other side needed a couple of shims still between the middle and end legs. I think the MDF test cut board comparison picture show what had to be done to that one rail, shim wise. Does that look like typical shimming?
As far as the X table rails being bowed, even if that were the case (and I will check on it now that the thought is in my head) wouldn't the amount of 'slop' in the mounting of the X rails to the top of the table rail allow for any deviation here. If the X rail is straight to the fish line then would that take any 'bow' of the x rail out of the equation?


Machine does run improved now and you can hear it in long X jogs. Noticeably less rumbling and more evenness. Still, if it can get better than this and stay put after getting it there then I'm game to keep going.

Thanks for any insights into this dialing effort from this board. It's greatly appreciated.

Tim

[bleeth]

IMHO the SB tolerance has to do with distance and not trueness of cut.
A straight line should be just that and "micro tweaking" should solve it. Since you are looking at a bare 64th of deviance (I agree with your math) you will likely have some difficulty seeing where the bow is on the rail itself but may have an easier time by using the cut boards as a guide. Since your rollers are now tracking so much better it would seem that you may have a consistent bow in both sides. Easily explained by one rail having a slight bow and the other made so well parallel to it.

The fly in the ointment is that a slight dip or rise in one of the rails can have the same affect, as raising or lowering of one side slightly will cause the gantry to tilt a bit and lead to the bit moving towards a high spot or away from a low. You might run your mdf test near y,o; y,24, and y,48 to see if the gap remains consistant. If it does than it is more likely a slight bow and if it is different than more likely a slight hump or sag.


FYI: Although the all extruded side rails are inherently more true than the old steel C-Channel the sheer strength of the C-channel helps keep things aligned once adjusted IMHO.

[daski]

If the work you do primarily involves melamine and particle board then stop fussing because the tension in the board is more of a factor than the straightness of your rails on the final straightness of the cut. Your bander is designed to band board that has been cut using a saw that will always have a far less straight cut that what you have already achieved on the Bot. If you aren't getting good banding results with the cut like that of the Bot... fix the bander it's the one with the issue! Remember you're building cabinets not doing eye surgery with the Shopbot.

[chunkstyle]

Hmmmm......
A bander that has been designed to take table saw cuts.... I'm gunna have my doubts on that. Even my friends 14" lagazzi wasn't tight enough for his holtz-her. I'm running an scm 201. What brands of edge banders are you referring to here daski? It's always been my understanding that edge banders require a straight cut. Less than that and the edge of the panel will wander away from the glue applicator. The only remedy will be to open the glue metering gate to provide more glue to the edge of the application roller in order for the glue to get picked up by the wandering panel edge, correct? More glue than what is necessary makes for a lot of glue slinging under the hood, not to mention the extra work the end, top and bottom trimmers have to do. Then all that clean up after a run of parts.
I believe the sweet spot of the amount of glue application is that you apply no more than is needed to get full coverage and a good bond. Having the pressure feed rollers adjusted correctly is also important. To little and you will get dry spots, to much and you get what I have.
Here's an example of what I'm talking about for anyone interested
My bander, a k201 that has far more machining with tighter tolerances than my bot:


Parts run yesterday and banded at the end of day. Longest parts are saved for last as those are the ones that have the worst deviation from the Shopbot. Try running at the glue gates current setting for the shorter cab parts (all of which banded fine):


Well that's not so good! Time to 'open er up' and start the glue slinging! As you can see in following photo, there is proper glue application on the leading and trailing ends of a panel. Notice how the banding has chunks of the core material stuck to those areas whereas the middle of the banding has very little.


The front and rear of the banding are getting proper glue amount without any unnecessary glue.
That bow in the panel is similar to the bow In my test cutting while I struggle to bring my off the shelf Bot parts, packed in a box to be later assembled, into tolerance of my bander.
Lastly I will say that for the first 5 yrs of my cabinet existence, I was having my cab parts milled on a Thermwood. I did not see ANY deviation from straight with two long parts butted up to one another. I'm not kidding myself into thinking that a bunch of off the shelf parts bolted together in my shop is going to even come close to the accuracy of a Thermwood. But I will say that any melamine cut on my Bot that has a bow in it is being introduced by the Bot and not from any stress introduced from the panel cutting.
I'm just trying to get my machine to run as 'true' as a Shopbot can be reasonably made to run in an effort to get consistent results at the bander without having to dive under the hood or pour on the glue.

[bobmoore]

You may want to have a machine shop dowel pin the v rail to a more substantial plate of aluminum. Machinists use dowels for accurate placement of things. I wonder if you aren't getting flex from the difference of expansion and contraction between the steel bed and the aluminum support. You can't tighten bolts enough to stop that movement, something will always give.

[bleeth]

Although I sold it when I moved to a smaller shop and miss it dearly I ran a Holzher Sprint for years.
Parts were cut on everything from a Unisaw to a 10' slider to the bot. As long as bander was adjusted correctly we never had an adhesion problem, including on long panels.

I totally am with you in wanting to tweak the "curve" out of it but you may want to play a bit more with adjusting the tape feed pressure roller as well.
I don't know that the "little" SCMI is more or less challenging to dial in but I do know a lot of folks using it and swearing by it, particularly the plus and k-203 versions.

[chunkstyle]

Perhaps my problem is running a glue pot instead of a cartridge glue system? Was the Holzher Sprint a cartridge bleeth?
My lil K201 does all right and will run all day without much fuss. Cleaning it after every run seems to do the trick. There's a sweet spot for the amount of glue applied to the applicator roller that produces minimal squeeze out after the pressure rollers and cleans up nicely from the trimmers. Unfortunately it's to light a coating on the roller for any pieces longer than, say, 42". That's why I roll the longer pieces at the end of a job with the glue gate opened up.
It's annoying to me to see the cab part getting pulled thru the bander with the leading edge and trailing edge making contact with the bander's guide fence but the middle bellies away. Opposite contact is made if the cab part's edge is on the convex side of a long part.
So getting realistic, confirmed expectation of 'trueness' from the PRS is a work in progress and will ultimately rely on the owner's resourcefulness to come up with whatever level of accuracy is needed. I wanted to share what I had done and will do for getting that last 64th dialed in.
Also, I wanted to get an idea of how long it could be expected to hold that level of accuracy from others that have gotten their machines as tight as what I'm trying to get to. I'll get into it some more, when things calm down a bit in the shop.
Bleeth, I think your suggestions will be helpful and I'm planning on bringing some more MDF in for test cutting in different spots in order to examine how the 'trueness' changes across the table, as you suggested.
Again, I don't know if this is pushing past the point of diminishing returns of what can realistically be achieved and maintained.
Gotta ruminate on it a bit, come up with a plan of attack....
If anyone has any good tips or techniques I'd be very interested and appreciative to learn about them.

Thanks again for all the help,
Tim

[Ajcoholic]

I dont know how it would be any different with the CNC, but when I cut 3/4" melamine on my panel saw, at least half of the sheets move (usually pinch after the riving knife) due to case hardening. I know if I again take a cut to straighten the edge, it stays put and doesnt move again. To my understanding this is pretty usual when processing melamine panels. If I try and edge band a concave edge, my bander too will not do a great job in the low point.

With the saw, I rip the sheets slightly oversize and recut. Since I dont cut parts bigger than 48" on my CNC I dont have any issues. But if you are cutting large gables on the CNC, do the parts stay straight or do they also move due to the case hardening?

[chunkstyle]

Hi Andrew,
I get the same curve in the same attitude of the parts wether they get cut from melamine, Mdf or ply.
curves always in the same orientation. It has always been present to a greater and lesser degree. Figuring it may be my approach to aligning the x rails is why I decided to start this thread. It has very helpful to have the advice of others more familiar with this kind if wrenching giving advice for where to look and do...
Its better than it was when I first started this down this road. I'd be very satisfied if it holds it where it is at present. Where it's at is probably good enough for most. Again, I may be going past the capabilities and expectations of what's realistic here.

[bleeth]

Yes-It was a cartridge system.