Z axis on Knee mill

[ballbearing00]

I need a third axis for knee mill
Trying to decide if the quill or the knee is a better choice.
Thinking about driving knee screw with an ac servo can accuracy be good with a lead screw?
Any input greatly appreciated!

[xr4x4ti]

I don't have any experience, but you could always put a scale on that axis to get the required accuracy.

I will also add, that I have a bed mill where the quill is not motorized and the head moves up and down. This is a great setup for the home shop.

If you could motorize the knee and leave the quill manual, that would be nice. I had an Analam machine for a while with CNC quill and it had its drawbacks.

Tim

[cnckeith]

https://centroidcncforum.com/viewtopic.php?f=57&t=9816

[cnckeith]

this is a nice setup.
has scale on knee connected to the Centroid (W on DRO)
and power feed on knee.

[CRM]

I might have a unique perspective on your question. I am a journeyman machinist with just shy of 3 decades of experience in the trade. I have direct experience with BOTH configurations, each being controlled with a Centroid CNC control. In addition, and I feel an important detail, is that I also ran 2 axis Bridgeport EZ-Trak knee mills with R8 spindles and glass scale readouts on the quill. I feel there are SERIOUS limitations with a CNC controlled quill.
With only about 5 inches of effective CNC controlled movement it almost becomes a liability to program in 3 axes if your tools are not relatively close in length. Consider a part that requires a 3/4" tapershank drill AND a 3/4" end mill. The drill, after being held in a R8 to Morse #3 adapter is likely longer than a 3/4" end mill being held in a 3/4" R8 collet by more than the available quill travel. Forget even having available spindle travel to satisfy the program Z movements. Where I currently work, we bought a Clausing knee mill originally equipped with a Fagor 3 axis CNC control.
The Fagor control advertises a feature called "axis summing" wherein you can theoretically associate the Z axis position (Quill) with the W axis (Knee). Sounds great, but it doesn't work in practice under program control. Additionally, it had a horrible user interface. Based on previous Centroid experience (detailed further below), I lobbied to spend the money to scrap the nearly new Fagor control and replace it with a Centroid based on promises from Centroid that THEIR axis summing function actually does work. So we cut a P.O. and had Granite Controls (the regional Centroid reseller in our area) remove the Fagor control and install a Centroid system. Guess what? It doesn't work either. After much phone time and many service calls we have given up on using it as envisioned. No amount of parameter changes allows the mill to successfully handle the transition between tool changes involving tool length compensation on tools with significant differences in length. The program invariably errors out with "z axis overtravel" alarms. We were SUPPOSED to be able to track the manual movements of the knee via a glass scale that to make rough compensation for significant changes in tool length. As far as I'm concerned, it was a waste of $25k, based on a promise that was ultimately never delivered. My suspicion is that if the CNC controlled axis is the one with the majority of the machine's travel (the knee), it would be a different story. I can't verify that theory though...YET.
Which brings me to the second example. My personal mill in my home shop. I bought a Southwest Industries TRAK DPM bed mill that was in absolutely mint condition but with an archaic Prototrak MX3 control. I DIY'd a Centroid Allin1 DC conversion utilizing the existing servos and am quite happy with the results. The machine is a bed style machine in that the whole Bridgeport-style head moves on the column with about 19" of travel. This axis is the CNC controlled "Z" axis. The quill is still a standard Bridgeport style unit with 5" of manual control like any manual Bridgeport. (It does have a NMTB 40 spindle) This configuration works flawlessly. In fact, I use the same GibbsCAM post processor for the Haas VF-3 we have at work to generate G-code that runs on my Centroid equipped Trak DPM here at home. Eventually, I want to install a glass scale on it and assign it to the W axis with axis summing to the Z axis. Having a manual quill is nice to be able to mix with the 2 axis positioning for certain hole drilling operations and especially power tapping. (Because I don't have a spindle encoder to facilitate rigid tapping.)
That brings me to my last point: the 2 axis Bridgeport EZ-Traks I ran at a previous employer. They simply felt much "handier" and productive for the onesie-twosie parts were were making than the Centroid 3 axis control on the knee mill at my current employer. The simple 2 axis programs with prompts to manually control the Z axis always ran without alarms for phantom Z axis overtravel moves which meant very little operator frustration, even though it was a less autonomous operation.
If you managed to read this novel all the way to the end, congratulations! I hope it gives you some points to consider and think about.

[cnckeith]

Dean,
hello.
not sure of the details of your failed Z summing experience on the knee mill, wish i did though. how long ago was that?
i can confirm that Centroid Z axis summing (combining two axis , a quill and a Colum or a quill and a knee) does work and is active on the machine in the photo i posted. travel errors are common with this setup since you have to keep the knee within the quill travel range for a multi tool job setup if you plan on NOT moving the knee. or if you do move the knee you have to plan ahead and know exactly where to move it so the tool being used will be able to do the work that it is commanded to do within the quill travel.

glad your experience with the prototrak bed mill was good.

[CRM]

Hi Keith,
Let me make sure that it is clear to everyone that my PERSONAL mill (the Prototrak) was an Allin1 DC retrofit I did myself, and the one we are having trouble with is at my place of employment; a Clausing branded knee mill that originally came with a Fagor 8055 "conversational" control that was billed to be able to run G-Code. If memory serves, we contracted with Granite Controls about 4 years ago to swap out that barely used Fagor control with a Centroid. I can say confidently that the Centroid control is superior to the Fagor in ease-of-use (being much more intuitive). But, as you eluded to, we struggle mightily with getting a G-Code program to actually run because of overtravel alarms. Honestly, I am not able to immerse myself in it enough to really get a handle on sorting out the issues because in the end, I need to focus my available time on MAKING parts, not troubleshooting and experimenting with machine controls. It is important to note that we don't really use the built in conversational programming; we use GibbsCAM to program the Haas machining center and Haas lathe that are the other two machines in our small prototyping department. We actually are fighting two different problems at once; we do not have an "edit free" post developed yet for the Centroid control on our knee mill; we post our output using the Haas post. But the reason that is the case is that we can't really get a program to run decent enough to go thru the iterative process to tweak the post due to the overtravel alarms. Admittedly, I believe some of the problem is due to the differences in when and where work offsets are turned on and off (G54,G55, etc) and when TLO (G43) is activated. Also, I have never been able to figure out (or been instructed) on the exact, correct procedure to set tool length offsets when it is necessary to move the knee to accommodate those tools.
A further complication is that I believe the physical hardware configuration is not entirely correct. At one point, we had 4 axis readouts (X, Y, Z and W). The W axis is the knee, using a glass scale. The Z axis WAS read from the servo rotary encoder when the retrofitter (Granite Controls) shipped the mill back to us, but that was not satisfactory to me because there is a glass scale that reads quill (Z) position in the original Fagor control. This is important not only because it reduces position error in the quill drive but because when the the control is set to 2 axis mode, we have no digital quill position readout if the quill drive (the Elrod quill drive kit) pinch bolt is loosened. Granite Control's solution was to configure the display for yet another axis readout for manual quill using an encoder input for the quill glass scale (so, Z, ZW and manual Z in addition to X & Y for those keeping track). It is my understanding (and intuition) that it should be possible to use the quill glass scale exclusively for both manual Z operation AND servo Z operation (which would significantly reduce backlash/position error in the Z axis.
I realize there's a lot to keep track of and maybe it isn't even realistic to have a control function under so many different modes of operation. If it wasn't for our occasional need to swing the turret to the side to do end work on long workpieces hung off the back side of the table, I would have definitely gotten a bed style mill instead. I just want to reiterate to all who read this thread that you should adjust your expectations if you plan on summing a pair of co-linear axes if the one under CNC control is shorter than the manual one. If someone DOES get it sorted out relatively well, they should write a supplemental operator's manual detailing how to properly set the TLO table, and specific procedures for tool changes while running a program. It would be a Godsend. Don't read this the wrong way; I'm very pro-Centroid--my mill at home runs that Haas post G-code flawlessly.

[cnckeith]

thanks for the input, and no worries, we always want to make things better and easier.