The Z-bot power drawbar works exactly like this See http://littlemachineshop.com/products/product_view.php?ProductID=3034&category=209730719. Brian has one running at the CNC workshop last summer, and we had it running at Cabin Fever in January..

I talked to the guy that developed it and he told me about 600-800 pounds was enough for the Tormach tools.

I built a similar, but simpler one on my Jet knee mill. I have my table move for the z axis, so my quill is free to operate by hand. I put the bellville washers under the drawbar, then I added a simple metal bridge above the top of the drawbar. To change a tool I simply drive the quill up with the hand feed until the top of the drawbar hits the bridge, then push a bit further to compress the springs and pop out the collet.  It works fine, but I am a little light on the bellvilles, so I did have a collet slip once.

Fred Smith of IMService has shown one  of them running at the last Cabin Fever and NAMES shows. He was running a program that just kept changing tools, but it seemed to work OK. I think it is his unit shown on the videos links of link posted.

The holder is a simple plastic finger, you program the tool to slide into it, then lift the z. The finger grabs the bottom of the holder.

Ive had a few discussions with Brian about changes for the next version. At this time I dont think we will try to make any functional changes, just make sure everything works with the new comp and all the other changes. When we ar esure its all stable under the new release Ill look at making other changes.

You are right that the cut circle could be more efficient, but the basic idea of the wizards was to be simple and reliable.

The 2.83 version was done to fix problems created by the new comp code. When I did it I thought Brian was putting the new comp into the released version so I did 2.83. I later found the new comp is only in the one test version, but will be in the new Mach4

The change has to do with the rule that a comp move must be a linear move in the new comp. The old comp allowed an arc move. The new wizard should be OK doing a straight comp move with the old comp code.

Is that confused enough ?

The M9 is there because no coolant is asked for, so it issues an m9. The code is in there to make that an M7 or M8 if you choose mist or flood. Remember, the wizards are written so you can chain them together to generate one big program. It is possible (although unlikely) that one could do an operation with either flood or mist, then do a cut circle and want no coolant, so the m9 is issued.

The ramp code was written by Brian a long time ago, and I admit I have never exactly understood it. The angle is the important value, zero means straight plunge into the work. A value up to 90 is allowed, but it seems to me at 90 it would never get down. Although the ramp dist is an input the code calculates a new ramp distance value based on the tool dia and the ramp angle, so I dont think the value in the DRO really matters.

Sorry I have not got a better answer.

The reset is happening because the screen update time is long enough to trip t0eh watchdog timer in Mach. It happens on slower machines and large screen sets like the NFS wizards. You can turn off the 'use watchdog' in the config page, or get a faster machine.

Thats how the wizards are programmed, always going through the material select and tool select. No real way to change that. For a single screen wizard the "run last wizard" button is more useful.

The Zstart or Ztop issue is one tha tis oftem discussed, but since the current way it works is burried in many screens its just to big a change to make. It would impact every screen of the set.

There is something very strange here- line 10 in your example is a peck drilling cycle. There is nothing in the cut circle wizard that does peck drilling. Have you edited that code? Since there is no G80 to cancel the peck cycle the rest of the code is very strange.

It also appears you ran the wizard 3 times in the one program.

How about a simple example, just one run of the wizard.

A shellac or other sticky way to hold parts is fine for one off stuff, but if you really want to make these in large numbers make a holding fixture. Id buy my stock in a width just enough to make one part. Start it in a vise, with just  a bit more than one half sticking up out of the vise, Cut your part half way down.

Next make a soft vise jaw and cut into it the image of the part- some call this a 'nest'. You flip the half cut part into the nest, tighten the vise, and do the other half.

Id also program an arc in and out on the cut to minimize the cutter marks, and Id program the part over size by about .010. Then Id make one final pass around the part at exact size as a finish cut.

I agree with the Bridgeport size machine. You ought to be able to run much faster than on the Taig. If you really want to buy a ready to run machine also look at the Tormach.

Can you post a sample code that does this? It sounds like some init condition, but you say a reset clears it.

Ron, I think your motors where just rated too high a voltage. I think if you had 2V motors, you would have gotten acceptable performance.

I bought the xylotex kit, with their motors, board and power supply. When I found them to slow I replace the board with geckos and raised the power supply voltage to about 48v. That worked well.

The simple fact is that voltage equates to speed with steppers, and the Xylotex board with its 24v useful limit is just slow. With the Gecko 251s available (they were not when I started the project) I see no reason to ever buy a Xylotex board.