| | |
Hello all, I just got back from a two month working streak in nowhereville and have the spare monies to put together my pcb mill. Below are my ideas and I ask anyone to drop me their input!
Ok so I was going through website looking at linear slides and pillow slide and thought they where great untill I saw the prices, with the best deals I was looking at $400 just for slides. I then looked at how they where built in the first place. Alot have crazy bearing races but the cheapest used a 3/4 bearing roller system on a T shaped rail. I figured I would put together a drawing and run up a price list so here is what I found out:
From the picture you get the idea of using an L shaped rail on either side of the axis and you would screw the tooling plate to the top of the slides (2 per rail) each slide has 5 bearings to keep it well in place.
I used common sizes of 6061AL and got my prices from onlinemetals.com
Rail is a .75 square with a 1.25 x 0.25 flat bar on top 12in in length ($5.75 total) 2per each axis, 1 for z as its 6in, $29 for rails.
Slide is 2 x 0.25 flat bar cut to 1.75, 4 for x-axis, 2 for y-axis, 2 for z-axis ($6 total)
Its a gantry design so I threw in 4 pieces of 2 x 0.25 for side posts and bearing blocks ($10 total)
AL sheet for table top 6 x 6 x 0.25 ($14)
$60 for all the 6061AL I need to build this design, I figured I was on a good roll but bearings are gonna kill me...
I will need 40 bearings (1/8 id, 3/4-7/8 od, 5/16-3/8 wide) are $14/10pack so another $60 for bearings. not bad.
All the hex screws Ill need I guessed around 150 and a 2in hex is $0.25 so at most $40 for fittings
So for less than $200 ill have myself a 6" x 6" x 2" milling table solely designed to mill PCB's with 0.01 isolation & traces. My question to everyone is will it work as I imagine or will their be binding issues? also are bearings good enough to give me the 0.005in resolution im looking for?
P.S - I figured I would put an eccentric shaft on the bottom bearing so I can tighten each slide to the rail nicely.
Last edited by 2_many_hobbies; 01-23-2009 at 03:05 PM.
You don't say anything about how you plan to treat the extrusion used for the rails. Without accurate machining on the surfaces the bearings ride against you will have bearings tight in one spot and loose in others. You've also got hardened steel bearing races rolling on 6061. With no moment load on the axis, bearings roll fine, but put any sort of momentary load on them (bump the spindle into something) and your bearings will dent the rail they ride on.
Thanks for your input, I went back to automation overstock after seeing zealcnc's mill and what it could actually do on youtube. Figure for the higher cost I could mill aluminum (quite light passes of course) and that would suit more than just my pcb needs.
Pretty simple design:
Hiwin LG55 rail with one truck on the Y-axis ($160)
Hiwin LG35 rail with two trucks on the X-axis ($205)
Hiwin LG35 rail with one truck on the Z-axis ($117)
Misc 6061AL for mates, tooling plate, blocks & mounts $150
12-5mm or 10x2mm ballscrews (three at 300mm each) $200
for ~$800 would yeild a 12" x 6" x 6" mill that from the videos is VERY smooth and accurate. Do you think even though its substantially more money it would still be able to hit my 0.005-0.007" accuracy mark?
I have looked at the x2 & taigs but their about $1500 in the end so with my budget of $1000 I am really interested in this single rail design.
I would expect that to create a wobbly table. Linear rails are almost always used at least in pairs, and the typical application would have four trucks per axis. The trucks have very little slop, but there is undoubtedly some, and multiple trucks can be adjusted to remove this little bit.
Regards,
Ray L.
Yes, it would be quite doable to be able to hit 0.0005" when necessary, if you build it well.Originally Posted by 2_many_hobbies
You should see good repeatability too, which is arguably more important than outright accuracy for milling PCB's, I'd suggest.
Regards,
Jason
See my setup at millpcbs.com pretty simple fixed gantry. Spindle is ultra important if you want to do small traces. Also you need mass to reduce vibration.
Phil, Still too many interests, too many projects, and not enough time!!!!!!!!
Vist my websites - http://pminmo.com & http://millpcbs.com
I have read the Hiwin spec sheets many times and believe a one rail setup would be perfectly fine. My reasons are as follows:
The Y-axis rail is LGW55 which with truck is 140mm wide static load of 3700ftlbs with rail support of ~90ftlbs in all directions. Only one Y rail is needed because the spindle is always placed over the center of it.
The X & Z use LGW35 which is 1700ftlbs static load with ~25ftlbs rail support. Two trucks are used on the X axis to fully support the work table so their is no give under spindle plunges.
So in the worst case scenario I would rate the machine at having a static load of 800ftlbs with pressure in any direction upto 15ftlbs before their is bindling or chatter of any sorts which is plenty enough even for machining aluminum.
The picture is my solidworks rendering with the correct block and rail specs, 0.5in aluminum bar is used for the base, x/y mate, and z pole (I would use a hollow or channel for rigidity though)
Yeilds a 300 x 150 x 150mm working envelope in a 600 x 390 x 450mm table space.
NOTE: The Hiwin's are quite a bit taller then the THK's my work table is 200mm up which is a pretty good distance, with THK's if I manage to find some on ebay would be ~110mm up.
Comments or suggestions?
Load is not the issue. They clearly have far more than adequate load-carrying capability. The issue is backlash. These are fundamentally no different from ballnuts. Achieving low, or preferably zero, backlash requires significant preload. Preload can be provided in two ways: over-sized balls, or multiple trucks loaded against each other, as described in the HiWin data sheet. Over-sized balls can reduce, but will almost never (other than by luck) completely eliminate backlash. Multiple trucks, with one set being adjustable, is the only reliable way to achieve zero backlash. Due to the geometry of your design, any backlash in the rails will be multiplied several times by the time you get to the cutting edges of your tool, so 0.001" backlash at the truck (a reasonable, if not very good value for use of over-sized balls) could well end up with 0.005" of positional uncertaintly at your cutting tool. And that's not counting spindle runout, flex in the frame, positional accuracy of the axis drive, etc. Plus, the uncertainty of all three axes will add to each other, making the overall error potentially considerably worse. I would be amazed if you could achieve any better than +/-0.005" overall accuracy with the setup you propose, and I could easily see it being considerably worse. Which means you can forget about doing 0.010" traces and spaces. 0.015-0.020" would be more like it.
Regards,
Ray L.
Hi Ray,
You're obviously skeptical about the backlash potential in this design - I would be too, however I do have a mill with a similar layout to that proposed - I am thinking maybe I could offer some empirical evidence to confirm or deny the design should be sound for the intended purpose.
All I have on hand to show you is a pic of a coaster (I had some acrylic left over after a job, so I made some for the customer as a thank you for the work). Not exactly the best example for the question, I'm sorry, but the cut was very clean and the edges are sharp with no chipping, and no evidence of any backlash as far as I can tell. A bit of dirt, but I've been using it daily for 6 months
If you'd like to propose a simple test, I'd be happy to give it a try on the similar machine and post the results, pics and video if you like? Perhaps engrave some very small text, or cut a simple PCB design? Text would be easier, but a PCB might be more helpful to the OP.
Best regards,
Jason
Jason,Hi Ray,
You're obviously skeptical about the backlash potential in this design - I would be too, however I do have a mill with a similar layout to that proposed - I am thinking maybe I could offer some empirical evidence to confirm or deny the design should be sound for the intended purpose.
All I have on hand to show you is a pic of a coaster (I had some acrylic left over after a job, so I made some for the customer as a thank you for the work). Not exactly the best example for the question, I'm sorry, but the cut was very clean and the edges are sharp with no chipping, and no evidence of any backlash as far as I can tell. A bit of dirt, but I've been using it daily for 6 months
If you'd like to propose a simple test, I'd be happy to give it a try on the similar machine and post the results, pics and video if you like? Perhaps engrave some very small text, or cut a simple PCB design? Text would be easier, but a PCB might be more helpful to the OP.
Best regards,
Jason
What is the size of the coaster? It certainly looks very nice. That is with a single rail per axis??
Regards,
Ray L.
Hi Ray,
The coaster is about 3" square.
Yes, there's one rail per axis. The Y has a single big wide bearing block, the X has 2 smaller wide blocks, and the Z has one of the smaller wide blocks. The rails are wider and lower than the proposed Hiwins so the comparison is not exact, but the moment load ratings are very similar. The basic load ratings for my blocks are about half those for the Hiwins. I couldn't say how the accuracy or potential flex due to elasticity compares, but my blocks are the standard non-preload ones (THK)
Best regards,
Jason
Jason,
Well, that's impressive. Better than I would have expected.
Regards,
Ray L.
Posting Permissions
| | |
LinkBack URL
About LinkBacks
