Inconsitent dados

[andre]

I have a new prs alpha and things are going well.
However, when I am doing area clears for my dado's I am finding that if I do outside to in path,that the dado is around .01 wider than if I do an inside to out path. It is a noticable difference. Is this a common problem or am I the only one. Things are tightened down well so I should not be getting that kind of deflection.

[richards]

Andre,
I run into that problem all the time. Climb cutting gives me a different size on a part than conventional cutting. Gabe Pari showed me how that could work to my advantage when cutting out parts. If I first cut the part almost to depth with a climb cut and then did a finishing pass with a conventional cut, the part had much better edge finish and was sized right.

You might try making a few test dadoes. Do one with a climb cut and do the next with a conventional cut. Carefully measure the dadoes and then make another test with an offset added in to get you to the correct size. The second test should be right on. You may find that different materials need different compensation. You may also find that different cutters 'flex' more and require different compensation.

[ckurak]

Andre,

I agree with Mike's first paragraph: climb cut first, then finish with a conventional cut. I use this strategy always when cutting out cabinet box parts (essentially rectangles with holes and/or dadoes cut first).

For cutting dadoes, I do an "area clear" in PartWizard using the following options: OFFSET, CONVENTIONAL CUT, START FROM INSIDE.

The START FROM INSIDE strategy also prevents having a loose piece of thin wood flopping around the center of the dado. If I try START FROM OUTSIDE this happens quite often as the center breaks loose instead of being routed into chips.

Another reason for START FROM INSIDE: bit deflection is at a maximum during the first pass. I prefer to have that cut in the center where it will be "cut away" anyway. Bit deflection on the final pass will be minimal, so you should get the correct size.

One other note: if your dado is only being cut in two passes, therefore bit deflection is causing problems, you might try lowering the stepover on the tool. This may give you more passes, but the first pass, with the bit deflection, should not be on the finished edge.

[knight_toolworks]

hey thanks for that I was wondering why some of my pieces that are mirrored are larger then the other.

[andre]

Mike
I did what you said a while ago, I did about ten different test with the different sizes and tool paths and I found I was concistantly off. It sounds like this is the nature of the beast. I have swithced to inside out and have tighter fits now. Maybe I will try a conventional cut vs climb and see what happens also. I am using a .25 bit with a .5 shank and really like it.

another issue
When I cut a square, the dimmension along the x comes up .0625 short. Any thoughts on what might be going on there?

[steve4460]

Hi Andre
When I have to make datos I usualy run a couple of test datos in a wast part of the first sheet .
Lets say I have to make a 0.500 dato 0.25 deep . then I draw 1 at 0.500 the next one 0.515 , 0.520 , 0,525 then I cut out a small pice so that I have a pice to test the fit in the datos once I find the one that fits the best then I draw all the datos with that dimention always using that bit starting on the outside and conventional .
Hope this helps .

[richards]

Andre,

You didn't buy one of the 'short' models, did you? Joking aside, all mechanical things have a little variation. The maker of the spur gears and the maker of the rack make their parts to fit within a tolerance window. Mathematically, we would normally just figure the ratios involved to compute the Unit Values based on the number of stepper pulses required to move an axis one unit (inch or millimeter).

It might be better to lay a piece of scrap on the table, drill a small hole near the 0,0 point and then do a MOVE (not a JOG) to a point near the 96,0 point and then drill a second hole. Measure the distance, compensating for the radius of the cutter used to drill the two holes and then tweak the unit values, if necessary, until the move is exactly as desired.

I've done that each time I've replaced the spur gears. So far, no tweaking has been necessary, but having an axis come up 1/16th of an inch short on a cabinet sized move of thirty-inches or so could be easily corrected.

[andre]

Mike
How would I calculate the change in the ratios to make up .0625?

Stephan
thanks, Thats what I have done. ITs funny some materials from the same pallet can very in thickness by .01. I am realizing I have to make my dados a little wider to compensate. some are looser than others. oh well

[richards]

Andre,
I'm not very good at math, but this is how I would figure it out:

(Distance Desired / Distance Traveled) * Unit Value = New Unit Value

Let's plug in some values to see if it works. (By the way, I'm using the Unit Values for a PRS-Alpha that Shopbot furnishes with version 3.5.3 SB3.) We'll use a desired distance of 24-inches for the example:

((24 / (24 - 0.0625)) * 2482.8171 = 2489.2997

Now, to check out the new unit value of 2489.2997 to see if the new unit value will move the X-axis a little further:

1 / 2489.2997 = 0.00040172 inches per stepper pulse. Thus 0.00040172 * 2489.2997 * 24 = 24 inches actually moved.

We're dealing with some very subtle changes in the math. It is very possible that other factors would negate the change made to the Unit Value. Also, keep in mind that I have no idea how SB3 treats the Unit Values when multiple axes are moved. It could be that having the X-axis Unit Value slightly different from the Y-Axis Unit Value might add a small error in the SB3 computations.

(I would like someone else to check my math. I have been known to get my formulas messed up.)

[richards]

I just noticed that my 'checking' method is just using reciprocals which really prove nothing, but my timing was off and I missed the time allowed for editing a post. A better check would be to use this forumla:

(Old Unit Value / New Unit Value) * Distance Actually Traveled. The equation should give a number that equals the Distance Desired.

Using actual values would give:

(2482.8171 / 2489.2997) * (24 - 0.0625) = 24.00000047

[andre]

Weird thing
The two cab sides to the left and above were right at 23.25 but the ones to the right came up short by.0625

6501

[andre]

Just talked to scott at SB
I am installing version 3.5.5. I had 3.5.0 and he said they were having problems, so hopefully that clears it up.
I will let you know.
Thanks

[bleeth]

Andre: Did they check out properly for square everywhere?

[richards]

Andre,
We might need some more information to diagnose the 'short' problem.

- What type of material?
- What depth of cut per pass?
- What feed speed?
- What router/spindle RPM?
- Which axis made the cut?
(I know you said the x-axis, but are you measuring two points on the x-axis of a cut that was made with the y-axis, or, are you measuring the endpoints of a cut that was made on the x-axis)?

I've found that when I use a 1/4-inch cutter, I NEVER have the depth of cut greater than 1/4" per pass if I want accuracy. In addition to that, I use Gabe Pari's method of cutting multiple passes in the CLIMB cut direction to a depth of at least 0.50" (on 0.75" material), and then make a finish pass in the CONVENTIONAL direction to final depth (which is MATERIAL THICKNESS - 0.030" with my vacuum setup). My feed speed/RPM are selected to give a 0.020 chipload, depending on the material that I'm cutting. With that method of cutting, I get very accurate results with an excellent edge finish.

I can cut to full depth in one pass with a 1/4-inch cutter, but the parts are not always consistent in size and the edge finish is rather poor. However, there are lots of times when that type of cut is all that is needed. It is faster and I don't like to do it because I would rather be known for quality than for speed, but when the customer just wants parts that he will slap together, I do what he wants.

[andre]

Dave, yes, I squared up sometime ago and things are cutting nice and square. Scott at sb said they had some inconsitancies with the beta version of the control software so I just updated to the 3.5.5 version. hopefully that was the issue. I will cut again tomorrow and will find out.
Mike do you have an older prt?
I cut at around .3 per pass at 4.5 per sec at 19000 on my PC, leaving a .01 skin. I use .25 bits with .5 shanks. Every thing cuts square.
Interesting that out of eight cab sides, 4 were .0625 narrow, I had 4 on the same sheet and two were short. strange.

[richards]

Andre,
My machine is the PRT-Alpha. Usually it is in 'squared-up ready-to-run' shape, but sometimes, I get lazy and let things slide a little. A few ruined parts reminds me that if I don't maintain things, nobody else will.

Your figures look just fine to me. Before I installed a spindle, I used settings very similar to yours. You might want to give Gabe Pari's method a try. I just use PartWizard to select the direction of cut and then cut-and-paste the tool paths depending on whether I want an entire part to be cut at one time or whether I want all of the climb cuts to finish before running the conventional cuts.

One trick that I always use is to have any part that overhangs another part cut 1mm large. For instance, the backs on my cabinets overhang the sides, so they're cut 1mm larger. That lets me run a router with a bearing bit around the back after the cabinet is assembled to trim everything to exact size. The alternative is to have the back occasionally a little too narrow. When that happened, I had to scrap the entire cabinet. In other words, I've tried to find areas where my methods have led to failure and then modified my methods to ensure success.

[fleinbach]

Andre,

Read the link below it may explain what’s happening. I did extensive testing when I first got my PRTAlpha after an error while cutting a sheet of MDF into equal strips. The error primarily occurs with the X axis. If you want to see my actual test made with everything adjusted and as tight as possible I made videos of my tests. They are 1 to 3 MB and I can email them to you or anyone else interested in seeing them. Many methods have been detailed here to alleviate these discrepancies but here is a summery of them.

1. Make sure everything is adjusted so there is as little movement as possible.

2. Cut slowly

3. Make shallow passes

4. Make sure the bit is not dull.

5. Use cleaning passes

All these things should assure accurate results

http://www.talkshopbot.com/forum/show.cgi?tpc=29&post=16165#POST16165

[andre]

I did some test and here are the results if anyone is interested.

I milled four dados at .72 width
4 different tool paths
1.offset climb inside.
result- right on
2.offset conventional inside
result loose by maybe .005-.008
3. offset climb outside
result right on
4. offset conventional outside
result loose by maybe .005-.008

I find with longer dados I will choose the loose fit for easier assembly and shorter ones I will choode the right on tool path.
Hope this is useful someone

Andre

[andre]

Mike
I resolved the problem with my cuts being .0625 short.
You were right the y unit values needed to be adjusted to 2486.0541. Scott at SB helped me with the math and analasys.
Now everything is dead on!

Thanks for the insight!