Adaptive Control for CNC

allloutmx · Nov 18, 2020

Good Day All-
we are currently reviewing options for tool monitoring adaptive machine controls.
Does anyone have any experience with the system from either Caron Eng., Marposs or the Fagor version of DMC and be willing to give some feedback?
We would like to implement one of these systems into our grinding machines and looking to gain as much outside knowledge about each of these products before we set out.
Our end goals would be, but not limited tool- monitoring conditions of the grinding tool(tool loading/drawing more amps) and reduction in cycle time by increasing feed of areas with less material(air cuts)
Any feedback you may have will be greatly appreciated.

:cheers:

DouglasJRizzo · Nov 18, 2020

About the only thing I can add is don't be too trusting with these. I used to work for a MTB, we had our own software and we evaluated other systems. Some were better than others but none were (to me anyway) 100%. Nothing beats the eyes and ears of experience.

PROBE · Nov 19, 2020

DouglasJRizzo said:
. Nothing beats the eyes and ears of experience.

I agree. But those are becoming rare and rare nowadays.

TeachMePlease · Nov 19, 2020

What kind of machine, and what kind of machining?

I've demo'd the Caron Engineering system and the 3PM Machine Boss (Might want to add them to your list)

Both are good. Neither fits my type of machining. In my world, the the 1000PSI high pressure coolant creates more disturbance than the .017" endmill does, so neither can actually catch when my tool breaks.

If you're running 1/2" endmills for roughing, yeah, they'll catch when your rougher breaks, or they'll catch when your drill breaks before you send a $400 threadmill into a 1/2"Ø hole that doesn't exist because your drill broke.

Edit: the 3PM was pretty neat with showing the areas of decreased load, and would definitely be helpful in tweaking your cycle times. I didn't get enough hands on with the Caron system to comment on that part of it.

Mike1974 · Nov 19, 2020

TeachMePlease said:
What kind of machine, and what kind of machining?

I've demo'd the Caron Engineering system and the 3PM Machine Boss (Might want to add them to your list)

Both are good. Neither fits my type of machining. In my world, the the 1000PSI high pressure coolant creates more disturbance than the .017" endmill does, so neither can actually catch when my tool breaks.

If you're running 1/2" endmills for roughing, yeah, they'll catch when your rougher breaks, or they'll catch when your drill breaks before you send a $400 threadmill into a 1/2"Ø hole that doesn't exist because your drill broke.

Edit: the 3PM was pretty neat with showing the areas of decreased load, and would definitely be helpful in tweaking your cycle times. I didn't get enough hands on with the Caron system to comment on that part of it.

No knowledge of the OP question, but Teach makes a good point. If you are using small tools, even normal coolant/machine noises/way cover noise/etc will all be louder than a small tool breaking. We never know until the part is "finished" and notice a feature/hole missing. On more critical parts, or trying to do lights out, we will do a tool break check...

allloutmx · Nov 19, 2020

We build a variety of machines here that mimic traditional cnc mills but are more so cnc grinders than cnc mills. 3-5 axis. Our 5axis machines have a b axis on the head and c on the table
Tool sizes range based of feature sizes- and also surface speed desired. 1.5mm dia tools - 200mm dia

Parts are symmetrical in nature....usually, but they may start out with irregularities such as with a cast blank.

We really aren't looking to catch tool breaks. If the tool broke- there is a very good chance our part is toast and we are too late. We are mainly looking to identify areas of lower material removal and allow the machine to process through those areas fast. Additionally- we would like to be able to identify where our tool begins to load and require additional power to the spindle so we may be able to know in advance when to begin a dressing cycle.

Load sensitivity and reaction time of the software are our biggest concerns.

Teach- thank for the heads up on the 3pm machine boss software. I will look into it.

DDoug · Nov 19, 2020

IIRC Bryant grinder had something similar, I recall seeing adverts from the late 1960's showing
the machines.

TeachMePlease · Nov 19, 2020

Both the 3PM and the Caron will do what you want. I found the logging capabilities of the 3PM to be impressive, with data output in a .CSV file that you can manipulate in excel quite easily. I never tried to do that with the Caron, so I can't tell you if it's better or worse.

Upside to the 3PM unit is they make a portable version, so you can buy one unit and take it from machine to machine easily, versus buying one unit per machine, or buying one permanent unit and having to install/uninstall to check multiple machines.

Both companies were happy to provide us a demo, and our machinery dealer installed them for us.

Not sure if they treat everyone that way, though.

TeachMePlease · Nov 19, 2020

Looks like they've updated their website some since the last I looked at it (It needed it)

Our Methods | 3PM Technologies LLC

motion guru · Nov 20, 2020

allloutmx - what control are you using? We have done a number of CNC grinder controls and have used Schmitt Industries (now SBS - What We Do - Accretech SBS) for monitoring grinding load and touch off for dressing. We used process information from the acoustic sensor with various digital filters to sense coolant noise versus touch off. We also monitored spindle current and then wrote software to dynamically adjust feedrate and offsets on the fly to achieve uniform material removal rates, transition points from roughing to finishing and polishing . . . also monitored wheel surface quality, could detect crashes, etc.

This isn’t a third party software package, it is code we developed to query data from the SBS unit and use inside the CNC controller in real time to control feedrate and sequence of operation of the grinding process.

The system works well grinding interrupted surfaces to Rc62+ holding +/-0.0003”

The system learns the wheel wear rate over the first x cycles (user adjustable - typically set to 25), touching off between cycles and then begins auto compensation and then checking every n cycles set on the HMI (most often set to 10-20 cycles).

After getting it dialed in, watching the grinder auto dress, touch off, rapid down to the part, automatically adjust feed to 75% spindle load, rough grind, retract, touch off, finish grind, spark out, and then rapid off the part and nail it to within tolerances and then after learning, skipping touch off and running cycle after cycle all while reducing cycle time by about 30 % and automatically compensating for variable wheel performance, detecting wheel load condition and auto dressing when appropriate, etc. . . . Super gratifying to watch.

EmGo · Nov 20, 2020

There's got to be a ton of literature out there, the technology is old. K&T had adaptive control on the C control, PDP-8, using feedback from the spindle and axis drives. It wasn't that great but what do you expect from a 12 bit processor running at 4 mhz ?

I could swear there were creep-feed grinders using these ideas in the eighties, too.

Anywho, this is definitely not a new idea, there should be a lot of available info.

LockNut · Nov 20, 2020

allloutmx said:
We build a variety of machines here that mimic traditional cnc mills but are more so cnc grinders than cnc mills. 3-5 axis. Our 5axis machines have a b axis on the head and c on the table
Tool sizes range based of feature sizes- and also surface speed desired. 1.5mm dia tools - 200mm dia

Parts are symmetrical in nature....usually, but they may start out with irregularities such as with a cast blank.

We really aren't looking to catch tool breaks. If the tool broke- there is a very good chance our part is toast and we are too late. We are mainly looking to identify areas of lower material removal and allow the machine to process through those areas fast. Additionally- we would like to be able to identify where our tool begins to load and require additional power to the spindle so we may be able to know in advance when to begin a dressing cycle.

Load sensitivity and reaction time of the software are our biggest concerns.

Teach- thank for the heads up on the 3pm machine boss software. I will look into it.

What you might want to look at is the FORCE module in Vericut. That is exactly what that software does. They claim 30% to 50% better efficiency in roughing tool paths. Come to think of it, this may be one reason Sandvik bought Vericut.

PROBE · Nov 20, 2020

I would suggest to rely on REAL specialists in this field : MARPOSS adaptive control for monitoring systems

motion guru · Nov 20, 2020

PROBE said:
I would suggest to rely on REAL specialists in this field : MARPOSS adaptive control for monitoring systems

We certainly wouldn't want any FAKE specialists who have deployed Marposs equipment . . . :rolleyes5:

PROBE · Nov 20, 2020

motion guru said:
We certainly wouldn't want any FAKE specialists who have deployed Marposs equipment . . .

I personally installed and implemented the Marposs system on surface grinder, which task was to control the production of MASSIVE amount of parts to accuracy of 0.01 mm. It worked like HELL. Prior to execution of this project I visited their plant in Bontivoglio in Italy to assure the feasibility of the project, and found the most professional stuff for support. Strongly recommended in any aspect of process control.

allloutmx · Nov 22, 2020

motion guru said:
allloutmx - what control are you using? We have done a number of CNC grinder controls and have used Schmitt Industries (now SBS - What We Do - Accretech SBS) for monitoring grinding load and touch off for dressing. We used process information from the acoustic sensor with various digital filters to sense coolant noise versus touch off. We also monitored spindle current and then wrote software to dynamically adjust feedrate and offsets on the fly to achieve uniform material removal rates, transition points from roughing to finishing and polishing . . . also monitored wheel surface quality, could detect crashes, etc.

This isn’t a third party software package, it is code we developed to query data from the SBS unit and use inside the CNC controller in real time to control feedrate and sequence of operation of the grinding process.

The system works well grinding interrupted surfaces to Rc62+ holding +/-0.0003”

The system learns the wheel wear rate over the first x cycles (user adjustable - typically set to 25), touching off between cycles and then begins auto compensation and then checking every n cycles set on the HMI (most often set to 10-20 cycles).

After getting it dialed in, watching the grinder auto dress, touch off, rapid down to the part, automatically adjust feed to 75% spindle load, rough grind, retract, touch off, finish grind, spark out, and then rapid off the part and nail it to within tolerances and then after learning, skipping touch off and running cycle after cycle all while reducing cycle time by about 30 % and automatically compensating for variable wheel performance, detecting wheel load condition and auto dressing when appropriate, etc. . . . Super gratifying to watch.

All of our machines are currently running Fagor controls
Interesting-you say you built your own software using the data feedback. Had you tried working with any third party companies but things didnt work? We have a small software department and I know the idea has been floated around about trying to have them do something respectively....

allloutmx · Nov 22, 2020

LockNut said:
What you might want to look at is the FORCE module in Vericut. That is exactly what that software does. They claim 30% to 50% better efficiency in roughing tool paths. Come to think of it, this may be one reason Sandvik bought Vericut.

So....Howd you know we had Vericut? lol
Ill look into it for sure!

Search

Largest Manufacturing Technology Community on the Web

Search

Adaptive Control for CNC

allloutmx

Titanium

DouglasJRizzo

Titanium

PROBE

Hot Rolled

TeachMePlease

Diamond

Mike1974

Diamond

allloutmx

Titanium

DDoug

Diamond

TeachMePlease

Diamond

TeachMePlease

Diamond

motion guru

Diamond

EmGo

Diamond

LockNut

Stainless

PROBE

Hot Rolled

motion guru

Diamond

PROBE

Hot Rolled

allloutmx

Titanium

allloutmx

Titanium

Similar threads