The 2014 DMU65 Ultrasonic arrived

A little background: I posted my plan to buy a 5-axis Haas for fabricating titanium firearm components and steel robotics stuff to this forum about a year ago, looking for feedback. Did people who know what they're doing think this is a good idea? After some quick discussion, I settled on a DMG Mori DMU65 Ultrasonic as my ultimate dream machine, and I made the down payment. I also bought a 1992 VF-2 with Machmotion control in order to learn and fill orders until the DMU65 arrived.

The DMU65 arrived a couple months ago. It's a beast: a 2014 DMU65 UL in perfect condition with 2000hrs on the spindle. It has turning capability, which I haven't yet made use of, except to spin up the 4th axis to 1100 rpm. It definitely does that. Pretty wild.

The spindle tops out at 18krpm, is HSK63a, and with ultrasonic engaged w/ ultrasonic toolholders and solid carbide end mills, it chews through titanium and hard steels, and the bits last a lot longer than they do on my VF-2. Much, much longer.

I'm ordering diamond bits for rotary ultrasonic grinding in order to do glass, ceramics, and make tungsten carbide pocket knife blades.

I use NX 11 for CAD/CAM (no support contract - I'm planning on buying one once I'm making more money). For post processing, I started off by modifying one of the posts included with NX, but using TCL to process such large volumes of data is just a terrible idea. Frankly, it completely sucks. We're topping out at a few hundred kilobytes a second... and dealing with gigabyte-sized programs. I ended up writing my own post in C++ and using ctrl-alt-c in NX CAM to dump out a CL file that my little C++ post ingests and converts to Siemens 840D code. The machine cycle GUIs in Siemens 840D did a wonderful job of showing me what machine cycles to call and what the parameters mean. I'm going to post it to github as GPL3 with the option to commercially license it - the same strategy Trolltech used for QT3.

Anyhow, this is turning into a wall of text, so I'll leave off here. Questions are welcome, and feel free to msg me if you have work you'd like me to run. I'm not yet using the machine 24/7; rates are negotiable.

Wow. I remember when you first came in asking questions about which machine would fit your need and then later telling us the route you chose. Seemed crazy at the time, but you are obviously taking off well. Congrats on the incredible machine.

Can you show a picture of what you made with a gigabyte sized program?

a gig program? something sounds off.

OP pointed out that the gigabyte program emerged from a sub-optimal post. So I think OP would agree something was off.

I hadn't realized the ultrasonic machines could do "conventional" machining - what does "ultrasonic" mode do when you have a carbide endmill in the machine? (I had always thought the machine would be used that lateral translating tools and some kind of ultrasonic horn - my ignorance shows.)

With .00001" in and out tolerance for curves and without spline output enabled, the program for these titanium grips was over a gigabyte. I pared it down to 120MiB by taking a deeper depth of cut with the 1/8" ball nose used for the eagle. This was actually back on my VF-2, and that's what first prompted me to put together a quick and dirty post in Python that I cleaned up and ported to C++.

erikh said:

A little background: I posted my plan to buy a 5-axis Haas for fabricating titanium firearm components and steel robotics stuff to this forum about a year ago, looking for feedback. Did people who know what they're doing think this is a good idea? After some quick discussion, I settled on a DMG Mori DMU65 Ultrasonic as my ultimate dream machine, and I made the down payment. I also bought a 1992 VF-2 with Machmotion control in order to learn and fill orders until the DMU65 arrived.

The DMU65 arrived a couple months ago. It's a beast: a 2014 DMU65 UL in perfect condition with 2000hrs on the spindle. It has turning capability, which I haven't yet made use of, except to spin up the 4th axis to 1100 rpm. It definitely does that. Pretty wild.

The spindle tops out at 18krpm, is HSK63a, and with ultrasonic engaged w/ ultrasonic toolholders and solid carbide end mills, it chews through titanium and hard steels, and the bits last a lot longer than they do on my VF-2. Much, much longer.

I'm ordering diamond bits for rotary ultrasonic grinding in order to do glass, ceramics, and make tungsten carbide pocket knife blades.

I use NX 11 for CAD/CAM (no support contract - I'm planning on buying one once I'm making more money). For post processing, I started off by modifying one of the posts included with NX, but using TCL to process such large volumes of data is just a terrible idea. Frankly, it completely sucks. We're topping out at a few hundred kilobytes a second... and dealing with gigabyte-sized programs. I ended up writing my own post in C++ and using ctrl-alt-c in NX CAM to dump out a CL file that my little C++ post ingests and converts to Siemens 840D code. The machine cycle GUIs in Siemens 840D did a wonderful job of showing me what machine cycles to call and what the parameters mean. I'm going to post it to github as GPL3 with the option to commercially license it - the same strategy Trolltech used for QT3.

Anyhow, this is turning into a wall of text, so I'll leave off here. Questions are welcome, and feel free to msg me if you have work you'd like me to run. I'm not yet using the machine 24/7; rates are negotiable.

View attachment 218177View attachment 218178View attachment 218179View attachment 218180

Click to expand...

Congrats Erikh, That's a hell of a "Package".

When you have time would be cool if you could post pics of how the ultrasonic head attaches and detaches to the spindle, I think there's something special about the receiver/coupling (can't remember).

Is that beasty direct drive motors + twin drive left and right of the trunion?

Interesting migration/last minute hack. Very cool.

120 Still seems pretty big. Couldn't you likely get away with a .0001" tolerance? What's you run time on those?

bryan_machine said:

OP pointed out that the gigabyte program emerged from a sub-optimal post. So I think OP would agree something was off.

I hadn't realized the ultrasonic machines could do "conventional" machining - what does "ultrasonic" mode do when you have a carbide endmill in the machine? (I had always thought the machine would be used that lateral translating tools and some kind of ultrasonic horn - my ignorance shows.)

Click to expand...

The ultrasonic motion is configurable via frequency and amplitude control up to approximately 10-20 microns of vertical travel at between 10khz and 40khz. Using it for conventional machining seems to be a very novel concept; I learned of it because the local DMG Mori sales guy, Chris Klepacki, is very sharp and well informed. He had heard that Luke Ivaska of the DMG Mori advanced technology group was seeing amazing results with ultrasonic for conventional milling of titanium. We did a conference call which convinced me that the benefits for conventional milling and the ability to occasionally do glass and carbide would be worth it. I'm coming from a programming background and have only been milling since fall of 2016, so I don't have many preconceptions or set ways of doing things, but I'd done enough to know that tool wear with titanium and heat treated steel is a real pain. Ultrasonic is not a panacea for bit wear; sensible surface speed and lack of chatter are still essential. However, with good feed and speed, my tools hold up about 10x longer.

There is a PhD to be earned studying precisely *why* ultrasonic provides benefits for conventional milling of certain materials. There's no benefit for 6051 aluminum except for deep side cuts, in my experience, and the only difference there is a somewhat smoother finish. Luke's thinking is that the ultrasonic motion causes microfracturing at a very small scale, similar to how rotary ultrasonic grinding (a well-studied process) functions, and that galling is also reduced by the rapid up and down motion. I understand that DMG Mori is securing more patents covering ultrasonic milling processes and will be introducing interesting features in the next few years.

The most profound results I've seen are with deep drilling of steel. Cheapo cobalt steel half inch diameter bits from Menards that are only good for a few holes with a peck cycle on the Haas take just the slightest wear on the first hole and then show no appreciable change. It's kind of weird. I do perhaps a dozen 3" depth holes a week for making jigs from random scrap obtained from the yard across the street. Sometimes, this stuff is heat-treated nightmare steel of unknown nature.

Random note: stainless 321 and 304 cut very well at 150sfm with ultrasonic, which is absolutely crazy. The metal dealer I buy from was pretty shocked that I was milling the pieces he sold me and strongly suggested that I really don't want to do that. We got to talking about it, and now I'm making him some nice furniture out of 321

huskermcdoogle said:

120 Still seems pretty big. Couldn't you likely get away with a .0001" tolerance? What's you run time on those?

Click to expand...

At .0001", NX had some kind of sampling error that resulted in a huge pit in the toolpath. I didn't notice that until I ran it on six pieces overnight. Live and learn. Run time was six hours per grip before I figured out that I could jack the feed through the roof. These days, about an hour and a half per pair of grips. FYI, they will be available at coldbane-armory.com once I get my web shit sorted out.

cameraman said:

Congrats Erikh, That's a hell of a "Package".

When you have time would be cool if you could post pics of how the ultrasonic head attaches and detaches to the spindle, I think there's something special about the receiver/coupling (can't remember).

Is that beasty direct drive motors + twin drive left and right of the trunion?

Interesting migration/last minute hack. Very cool.

Click to expand...

Yup, direct drive with tandem motors on the trunion.

The blue part that extends below the spindle emits a magnetic field used by coils in the gold part of the ultrasonic toolholder pictured to somehow cause the ultrasonic motion. I'm not sure if Piezoelectric effect is involved. I'll ask Luke next time we chat. He explained it to me, but I forgot the specifics.

erikh said:

Yup, direct drive with tandem motors on the trunion.

The blue part that extends below the spindle emits a magnetic field used by coils in the gold part of the ultrasonic toolholder pictured to somehow cause the ultrasonic motion. I'm not sure if Piezoelectric effect is involved. I'll ask Luke next time we chat. He explained it to me, but I forgot the specifics.

View attachment 218206View attachment 218207

Click to expand...

Thanks for that VERY interesting.

What does it sound like to cut various materials... I know its "ultra" sonic but are there other ear splitting harmonics or excited frequencies? Or is it quiet to run... Trying to imagine "Ultra sonic" chatter?

Does the tool cutter or tip always need water flow/immersion or are certain materials cut dry?

Cheers,

Eric

______________________________________________________________________________________________________________

This machine has the capability to do grinding also (doesn't it?) i.e. you can have a grinding head in the spindle (+dressing cycles) and the trunion's rotary table has a high enough /decent RPM you can carry out genuine grinding ops without knakering the machine?

erikh said:

At .0001", NX had some kind of sampling error that resulted in a huge pit in the toolpath. I didn't notice that until I ran it on six pieces overnight. Live and learn. Run time was six hours per grip before I figured out that I could jack the feed through the roof. These days, about an hour and a half per pair of grips. FYI, they will be available at coldbane-armory.com once I get my web shit sorted out.

Click to expand...

When I get jagged paths like that I filter them with a VBA program I wrote. Slow, but it can chug through a few megs of code in a few seconds filtering out moves that don't match the general motion pattern. One of these days I'll get visual studio loaded on the machine and make it into a proper application. I rarely run a tolerance lass than .025mm these days, but I have found that surface quality for that issue is 90% dePendent on incoming geometry. Add surface normals controlling tool vectors and then is the 99% driver on an efficient path. Next filter I need to write will be to smooth vectors for rotary surface paths. Too cheap to buy high speed smooth tcp....

I have done some testing with intol/outtol set on NX at .0001 to .0005 and could not see enough better machined surface to justify the additional program size and cycletime with the small values.

Based on the image of those grips, I doubt using .000001 intol/outtol made them look any better than if you had used .0002.

cameraman said:

Thanks for that VERY interesting.

What does it sound like to cut various materials... I know its "ultra" sonic but are there other ear splitting harmonics or excited frequencies? Or is it quiet to run... Trying to imagine "Ultra sonic" chatter?

Does the tool cutter or tip always need water flow/immersion or are certain materials cut dry?

Cheers,

Eric

______________________________________________________________________________________________________________

This machine has the capability to do grinding also (doesn't it?) i.e. you can have a grinding head in the spindle (+dressing cycles) and the trunion's rotary table has a high enough /decent RPM you can carry out genuine grinding ops without knakering the machine?

Click to expand...

Interesting question; I was also very curious how ultrasonic cutting would sound. The ultrasonic motion is maximized by setting its frequency to match the first resonance of the toolholder + bit, whatever that happens to be, such that the entire toolholder + bit are moving up and down together. So far, that hasn't been below about 27khz. I incorrectly wrote earlier that the max frequency is 40khz, it seems to go up to 60khz. The power the ultrasonic generator is willing to supply starts to drop off around 30khz, scaling down toward 10khz. Feeding the max power it will generate into the smallest toolholder at 13khz, not much happens.

At the first resonance and max power, touching the toolholder or tool itself is a weird experience. It feels slippery. If you grab it with enough force, you can get purchase, but it still feels odd. The recommended method for manually tuning the frequency (which is recommended for second generation but not first generation tool holders, for some reason - there is frequency scanning support and it works for both) is to lightly touch a long, thin piece of metal to the tip of the tool while adjusting the frequency so as to maximize screeching sound. It has to be a thin piece of metal like a ruler that is deflected away by the ultrasonic motion so that only every 2nd, 3rd, 4th, etc oscillation of the tool hits the metal.

Ultrasonic won't remain engaged for more than about 30 seconds if the spindle isn't rotating, so it can't be used for turning. For that reason, ultrasonic + turning is just about never ordered. I may well have the only example. I wanted it for rapidly roughing out rocket stages, wheel rims, and other large, radially symmetric parts. Turning mode only goes up to 1100rpm for a little while before the controller wants to drop down to 800, which it can maintain indefinitely - so turning mode is probably not useful for grinding, I would guess, but I'm pretty new to this.

FYI, the tools for Rotary Ultrasonic Grinding (RUG) are specialized: Supplier of ultrasonic diamond abrasive tools. | Jandrik-Lukas These always need to be used with fluid, I think.

Rotary ultrasonic cutting is a different story. The considerations are the same as for traditional milling. Anything that needs a ceramic cutter, such Inconel 718, and even some that can be cut with carbide such as crazy hard treated steel, simply cut too hot, even with ultrasonic. Coolant flashes into vapor before it gets close enough to the cutter to do anything useful. If the piece you are making would be deformed by high temps, you have to switch over to RUG with a diamond bit or conventional grinding with an abrasive disc like you'd find in a CNC grinding machine.

Incidentally, the coolant plant supplied with ultrasonic DMG Mori machines includes a centrifuge. According to the manual, if a lot of grinding and RUG is done, that centrifuge must be cleaned out regularly. I'm not there yet - I need to finish out a big order before I start messing with RUG - but I expect that little ceramic chips, diamond dust, and other junk that would trash the 1000psi through-spindle pump collects in there.

Edit: If it sounds like I'm a nutcase who bought a super-expensive machine mostly because I felt like it, you're not alone. As soon as she finished her residency, my girlfriend dumped me, left the country, and married some other dude - rather than deal with one more instant of all this