Accuracy again??

[jay tomaka]

if the machine moves in .002" steps or even better with the new gears why isnt the accuracy better than the stated .005-.015"? are some cuts right on size and some off? would some one please plot with a fine point pen (this way there is no load) a one foot sqaure and measure it with a good metal ruler to see how close it is perfect? lets put this topic to rest once and for all with a difinitive answer. thanks guys, this forum has really helped me in the leaning towards a shopbot!

[Wdyasq@yahoo.com]

Jay,

I have an old SB. It was originally a cable drive unit. I built my own gear boxes. I still have "Uni-Strut" rails. In other words, in relationship to the new ones mine is 'sloppy'.

When I cut a sliding dovetail, I calculate the needed 'spline'. When I measure the part it is what I calculated. I have never had a part out more than ~.010~0.015".

If I want or need more accuracy, I change to a material that won't move - Like aluminum or steel. And, I use a worn out old Bridgeport Mill and have trouble holding closer than ~.002" there.

There are tools that are more accurate. They cost a LOT more money. I have asked before - what have you got that can measure that accuratly? Have you corrected for tempeature? Is there dust on the gear rail - or a hair? Did your felt tip pen flex??

You need to ask yourself "What do I want to do and build with this tool?" And how much accuracy does it take. I find my sloppy old converted ShopBot my most accurate woodworking tool. I could care less if it is .020 accuracy or .005 accuracy.

Ron Brown

[JimM]

Greetings,
I recently purchased the Anaheim Woodworking show Demo machine, PRT96, off the floor and it was installed a couple weeks ago. This discussion about accuracy had my attention, and I hadn't heard a good definitive answer to backup the factory specs. My Engineering company, among other things, does precision Laser imaging for the electronics and arespace industry. Here we image polyester base photographic film with a CNC laser that resolves 8000 lines per inch, over areas of approx 3 foot square. We use a Swiss 2 axis co-ordinatograph with digital readouts to scope the accuracy of these films in a climate controlled environment, and can measure
+/- .0005", traceable.

After a week of playing with the PRT, I set up a test to see what accuracy it was holding. The quick test involved center drilling or 'spotting' a grid of holes on 1" centers, 24" x 24" area, into a polycarbonate 1/8" thick sheet clamped to the table. Althought this isn't the most dimensionally stable material, I was able to get some feeling for what the machine can do over this 4 square foot area, without tool wander from cutting loads confusing the issue.

On the co-ordinatograph, the drill spots were sampled and the results showed that the centers were never out more than .004 from grid position. This is to say if you drew a .004" radius circle at each inch grid center, no spots fell outside the circle. Many positions were better than .001" to grid. I zigzagged thru the array with out paying attention to backlash (my machine is a new, metal framed, rack and pinion model). There appeared to be no 'accumulative' nature to the error set. Incidentially, this test is how I 'squared' up the x axis to the y axis, using the tilt numbers (initially .010")from the measurement, setup a magnetic dial indicator on the frame touching the carraige, and then moving it the proportional amount and re-tightening things.

I have ordered the 20 tooth pinions just to have fewer 'roundoff' numbers in the display, and will rerun the test to see if there is any change in accuracy. So far, I am very happy and believe the factory specs will hold up under reasonable tool loads with a properly setup PRT96. I have also ordered the Colombo auto tool change spindle and upgrade, which should add another level of precision to the machine (plus the lower sound level)!

[Wdyasq@yahoo.com]

Just curious Jim,

How much is the equiptment to analyze and check the accuracy?

Ron

[larry]

Jim Hi,

I purchased my shopbot with the columbo spindle this spring. One thing to consider is the greater weight of the spindle. This large moment gives the y axis assembly the tendency to swing. Since there is a motor and rack only on one side of the y axis assembly, if you smack that motorless end with the palm of your hand, it will easily move .1" or more. I get some roughness of cuts on curves and attribute it to the lack of a fourth motor to minimize the resonance of the y axis assembly. I have not yet spoken to shopbot about this but I believe it needs to be addressed with these heavier spindles.

Larry

[jay tomaka]

please keep in mind if you are looking for laser accuracy buy a laser!!! if you are looking for the quoted .015" than shopbot is fine. dont expect it to be what it isnt. so many factors go into cutting wood with a moving head. the spindles have a level of play/runout in them that is .005" before you hit the wood!then the rest of the possible inaccuracies start to add up. hey .015-.020" is accurate enough for what this machine is being made for. a lasers head stays still and never makes contact with the material it is cutting. enjoy the machine but be realistic with the accuracy issues. had my bot for 3 years now i know it well and it does a great job for my applications.

[tom allen]

i agree! well stated this board has helped many of us decide to buy a shopbot.

[JimM]

Ron,

The Swiss co-ordinatograph cost 30k in 1981, and another 5k to upgrade it to Sony readouts. It is maintained and calibrated every 6 months, and used daily to QC laser imager output.

Larry,

I only tested the machine in an 'unloaded' condition. I have noticed what I perceive to be a weakness in the design that you mention. I am going to add a spring loaded follower to the opposite side of the Y carriage on a pivot with spring and turnbuckle just like the motor is mounted, to provide the same hold-down leverage on both sides. This should keep all 4 Y axis V-bearings riding down on their tracks, and minimize the obvious. A sharp load might still be a problem, but in general I believe this mod plus the added mass of the Colombo will provide a smoother finish.

Jay,

You are right. I am quite happy with the product as delivered for holding tolerances as advertized when cutting wood. I will have some applications that could benefit from better accuracy and cut finish quality. It's my opinion that the steel table PRT is a transitional machine, bridging the light duty to heavy duty cycle gap. I have been waiting on the sideline for the Shopbot to evolve to it's current state.

Jim

[sheldon@dingwallguitars.com]

What Columbo tool changer?! How does the SB software control the tool changes?

[JimM]

Sheldon,

This is a new spindle option, which will include software as well. The prototype is near completion. The Colombo spindle is the one with air collet change capability.

Jim

[threadfx]

Does anyone know the price of the Columbo with the tool changer?

[Sheldon Dingwall (Unregistered Guest)]

What ever happened to the tool changer prototype?

[Brady Watson]

I *think* a toolchanger Colombo is about $9500 and up. In many cases it's easier to just get a 2nd Z and add another tool with a different bit unless you see yourself using 3 bits or more for one piece.

-Brady

[sheldon@dingwallguitars.com]

We use up to 5 tools per part and not all parts use the same tools, so a 10 or 12 tool changer would be the cats pyjamas.