To Scott_bob
Your input of: "By the way, the feed rate is constantly changing depending on the upcoming geometry, so there would be no point in determining a single feedrate that may represent an ideal "low chatter" frequency."
Dear Scott _bob. There is no “low chatter frequency” as you speak of, that is called process dampening. Your about 90 years behind times.
In the real world there is a natural frequency and flexibility - period. These readings can be taken to determine a constant feedrate to maximize machining operatations.
Last edited by hardmill; 05-10-2004 at 03:23 PM.
camminc,
Well I suppose you could tell me all about it...
It's always interesting to me when someone goes on the attack when their ideas are challenged.
No harm, no foul...
What do you say to the question:
If a cutting tools feed rate is constantly changing, then what is the optimum feed rate for that tool?
Is this what you mean by "constant feedrate"?
Scott_bob
Scott-bob:
I don't mean to attach you, I just get tired of saying the same thing over and over. You know what I have written in this forum and it is the truth. I don't play games. I have fought antiquated ways of everyone for too long.
Bottom line: A cutting tools feed rate is constantly changing due to programming. Chip load is chip load and that is how they program it. Programming changes feedrates (Chip load) because of depth of cut, or it will chatter. (Because the cutter exceeds stability and they don't know this with out data given to them) If you know the maximum depth of cut (Stability) for a cutter then programming will not have to change feedrate. In another words, if you know a 1" cutter in an assembly with a specific gage length, in this holder, in this machine, at a certain rpm, can go 1" deep with a chip load of .010, you can program it to do that all day long, full depth of cut of 1", even in corners. But if you exceed the depth of cut or change the rpm it will fail. Example of this cutter being in the 1x multiplier of NF and Flexibility.
Problem being: Programming has no idea of the optimal depth of cut for a cutter, (Stability) if they did then chatter would not occur. They simply program a tool for a chip load, that is it. But when that cutter reaches to high a depth of cut it begins to chatter or forced vibration takes over. Such as in a corner, etc. The chip load has nothing to do with it, the cutter simply overloads the machine tool dynamics causing chatter or forced vibration.
To maximize one must know the frequency of the cutter at the tool tip, period. It will tell you exactly what rpm to run to maximize. Flexibility will tell you depth of cut. If you can't reach the optimum rpm, then you go to the next best thing, the 2x, 3x, etc. The 1x is the optimum rpm to maximize that cutter on that machine due to measurements taken. It will give you the highest depth of cut, feedrate will stay upon the chipload required of that cutter at that rpm. It might be 40,000 rpm, but you can't reach that rpm, you divide that by 2, to get the 2x, or 3 to get the 3x. Each time you go to a higher multiplier you give up depth of cut.
Natural Frequency X 60 / number of teeth. This is the equation to know, to maximized rpm.
Example:
1000 hz natural frequency x 60 x 2 teeth = 30,000 rpm = 1x multiplier, maximized condition.
/ x 2 = 15,000 rpm the 2x multiplier (Less depth of cut then the 1x)
/ x 3 = 10,000 rpm the 3 x multiplier (Less depth of cut then the 2x)
The trick is: Dynamics of that cutter of the machine tool, do not exceed those parameter. Rpm will allow you to reach depths of cut that you could not imagine before.
Simply put: If you know the deepest depth of cut of a cutter in full diameter and run it at the desired chip load it will cut without chatter or forced vibration.
The way to know this is with knowing the Natural frequency and flexibility of that cutter, then program it. Limits are not exceeded, chatter, forced vibration do not occur. Maximizing. Controllers play a part but if you go an impact test it will give you parameters. If those parameters do not transpire then it is the controller, cpu, etc.
Dynamic response at the tool tip reveals machining capability of that tool, period.
I have seen things happen. Take a Marwin 40,000 rpm machine. Tolo use to have them then sold out to BF Goodrich. They still have the same problems because they won't listen.
These machines have high rpm but they also have very small holders / cutter assemblies. (Smaller assembly higher frequency) Programmers program them at 40,000 rpm and it chatters, when they maybe should be programming them at 36,000 or 28,000 rpm. They have no idea. To run a machine tool at maximum rpm does not give you maximization, due to dynamics. You have to know the natural frequency at the tool tip to know the dynamics of that machine tool to cut, period.
It is simple, you take readings with an impact test, you put these reading in a database, it gives you the maximum doc for that cutter on that machine, send it to the programmer, he sends it to presetting to set up and away it goes. The database makes tracking simple for tool crib, presetting and programmers. Reports, etc.
You do this for about 2 years and you have 1,000 tools ready to go, maximizing operations of all machines. Once a reading is taken of a machine tool it will continue later, if not then the machine tool dynamics have change alerting you of maintance of that machine tool. (Another quaility of impact testing called "Modal Analysis" very usefull) All it takes is someone to do it and someone to make it happen. Otherwise, it is a guessing game, production is sacrificed, profits and redundant procedures occur making for inefficiency.
Thanks, Randy
CAMMINC,
I respect your commitment.
But, IMO your insight is a bit myopic (tunnel vision). I don't want to be too direct but the process variables from your posts reflect old school methods.
High Speed Machining (HSM) would not use the "maximum depth of cut" that you mention.
This is one of the fundamental realities of HSM. I agree that there is significant benefit to finding the sweet spot for a cutter assy but, if the program is not going to be cutting at the maximum depth of cut, then it may not chatter anyway right?
The point of HSM is to maximize metal removal rate by very high feeds with lighter cuts. Obviously, not too light of a cut else the process will not yield faster results.
Another observation I have is that you refer to the feed rate changes as if that is a result of the programmer making those changes:
"Bottom line: A cutting tools feed rate is constantly changing due to programming. Chip load is chip load and that is how they program it. Programming changes feedrates (Chip load) because of depth of cut, or it will chatter."
A good HSM control will compensate the feed rate depending on the approaching sharp corner, or small radii, or density of the point data. A smart control looks ahead and buffers many steps ahead, thus it is able to decelerate before sharp corners, this is called geometric look ahead. This type of control adjusts the feed rate depending on the algorithms that are used to handle high speed motion. The CNC program needs just one feed rate say a maximum (perhaps determined by your impact testing) and surface finish requirements, then the "control does the rest".
There are two other problems that are often ignored, namely acceleration and servo lag, which affects high-speed contouring more than most people think.
1) Acceleration and deceleration are features of the servomotors. If a program tries to accel/decel faster than the servos can handle gouging or overshooting occurs. This is not chatter; it is the result of inadequate control...
2) Servo lag, 2 factors control the smoothness of the motion (and its inaccuracy as well): the servo lag and the jerk factor. Both are doing similar things but in a completely different way.
Servo lag is an inherent feature of the servo loop and jerk factor is part of the motion generation by the control. The servo lag is exponential; when step acceleration is sent to the servo the result is a little rounded. The jerk factor converts the rectangular acceleration to trapezoidal or triangular.
Not all controls have jerk factor, and some have exponential one, which is easier to calculate but introduces greater errors. Without jerk factor the machine jerks upon accel/decel. This is noticeable with short jogs.
Both servo-lag and jerk-factor are time delays and they make the servomotors move a little behind the program. For a linear move that's not a problem, because all axes are behind equally (if the servo gain is equal), but for an arc there is a decrease in the radius:
It can be seen that the contribution of the jerk-factor is much less than that of the servo-lag.
Conclusion.
Not all HS are created equal...
Scott_bob
Coolant issue:
http://www.cnczone.com/forums/showth...1013#post41013
Scott_bob
Camminc and Scott_Bob,
Both of the issues you have brought up are important aspects in high speed machining. The idea behind high speed machining is to optimize the cutting condition to the machine, tooling, and control. To do this you must know the machine and spindle power rating, the vibration frequencies and limitations of your tools, and the limitations of the control. If you know these you can optimize the machining operation. This means you will achieve less power usage, maximum metal removal at lower cycle times, optimal tool life, and lower maintenance costs. Without all of these you can get pretty good using trial and error. Only problem is there isn't much error in high speed machining. If you have one it's usually costly.
True high speed machining isn't guess work. Using the the power graphs, a few formulas, and the frequency analysis chart you can determine the maximum (and most optimal) speed and metal removal rate for your operation. You can actually see the effect of depth of cut vs power and stay within your most optimal range. HSM reduces the amount of power needed to remove the same amount of material as a conventional machining operation thus reducing wear and tear on the machine and tooling. The control is important because of the acceleration and deceleration at the higher feeds and speeds. With an old controller you may never hit the the programmed feedrate the the HSM rates. You need fast "G" acc/dec and good look ahead control capabilities to be successful. Now you come into the drive part of the machine. You need the positioning accuracy to be able to handle the information from the control. This gives the advantage to linear motors/drives which can postion faster and are more accurate than the ball screws.
Briefly, here are a few more important issues in HSM.
*Tooling - The tooling must be balanced for high speed machining. An unbalanced tool will decrease accuracy and will increase vibration as RPM increases. It will also cause premature damage to spindle bearings. Max RPM ratings on tools must be followed and put into your HSM equation if alternate faster tooling can not be used. HSK style holders are a must when working with spindle speeds over 24K. They are more stable and rigid at high spindle speeds than the CAT type. The frequency analysis for each tool at the cutting edge would also be added to your HSM equation. Note, if you change a tool you would need to repeat the frequency analysis and make any required adjustments.
*Coolant - high pressure... thru the tool if possible. Dry methods should be explored but if coolant is used misting can be a problem and mist evacuation should be considered.
*Fixturing - conventional techniques and methods are OK for conventional parts. Picture framing is an option for thin wall parts where vibration or distortion are possible.
There is a lot more to this but I'm getting tired of writing. I've only highlighted some of the things I learned at the "Introduction to High Speed Machining" seminar I just attended at Tech Solve in Cincinnati, Ohio. I strongly recommend this to anyone interested in HSM. It lays it all out but it's up to you to take the next step. You both are right, each of your areas are important, but, high speed machining is a system with all areas being of equal importance. If one area is weak, you will not be able to acheive the optimum.
Gunner
Gunner,
On any CNC machine the control is the primary limiting factor for HSM. Most controls cannot deliver the high performance motion that is considered High Speed Machining (HSM). The limitations of our control prevented us from using this methodology for max metal removal rates. When we replaced the control, we were able to use this new process. Before the retrofit, we could not.
All the contributing technologies only contributed small improvements to our processes reduction in cycle time or improvement to quality.
I appreciate your input on the list of the variables...
If it's ok with you, I’d like to disagree on the equality of each variables effect on the optimum...
Sincerely,
Scott_bob
HSM factors - not involving controllers
The machine control is not the primary factor to HSM. You can't run a cutter at 25,000 rpm when it wants to run at 23,000 rpm, etc no matter what kind of controller you use. See these facts: Impact Testing
F18 E/F Fighter: MetalMAX impact testing has been in use on this program for over ten years. MetalMAX was a critical component in the cost reduction program that saved the US Navy F18 E/F fighter program over 1,000,000,000 (one billion dollars).
Aircraft Engine Mount: The American Helicopter Society article, “Productivity
Improvements through Collaboration” (June, 2002), credits MetalMAX technology for driving down the cost of a titanium aircraft part from approximately $43,000 to $28,000 per part, a savings to the program of $2,000,000 over the life of the contract. This is a 35% cost reduction made possible by the MetalMAX system. MLI earned the prestigious Pickney Award form the AHS in 2003 for this accomplishment.
Automobile Head: In rough milling the internal ports of a cast aluminum cylinder head, the MetalMAX system, in less than one hour, created dynamically optimized cutting parameters on three milling cutters. The cycle time was reduced 33%, from 1.5 hrs to 1.0 hrs. This single cost reduction and ROI justified the purchase of the MetalMAX system.
Scott_Bob
I'm interested on how your HSM projects are going. Are you producing different parts on the newly retrofited machine or was this dedicated to running one particular part? If your running different parts, have you achieved the same cycletime improvements or pcs/hr rates on the them? Are there any new problems you've encountered because of the HSM techniques (fixturing, tooling, coolant, etc)? Have you seen any tool life increases? How about power consumption? I've heard some shops have actually reduced their electric bill noticeably after adopting HSM techniques. I thought you might have some more info now that you've got some time under your belt.
Gunner
Gunner
Hi Scott_Bob.
1: Are you producing different parts on the newly retrofitted machine or was this dedicated to running one particular part?
We work with single parts, multiple parts, new machines, retrofitted machines, different materials, it does not matter. We do Impact Tests which simply take readings of the cutter at the tool tip, which in a nut shell tells us the dynamics of that machine tool, giving suggested rpm and depth of cut to run that cutter on that machine. Material, max rpm, HP, number of teeth, cutter and holder nomenclature are then input to give information, by way of a custom database. This results in maximum cutting of the cutting tool in that spindle on the machine, dynamically.
2: If your running different parts, have you achieved the same cycle time improvements or pcs/hr rates on the them?
Operations can increase by 0-100%. It all depends on if the programmer uses this data each time he programs.
3: Are there any new problems you've encountered because of the HSM techniques (fixturing, tooling, coolant, etc)?
There are always problems, but in general, the same applies to all machining. Of course exceptions such as wall thickness etc are noted. Dynamically tuned cutters will out perform all other cutters in all instances.
4: Have you seen any tool life increases?
Tool life, of course increases. By using dynamics, chatter is minimized. Chatter is the most determining factor of tool life, spindle life and surface finish. Although, excessive chip loads (Forced vibration) which are mistaken for chatter in many instances. Programming makes a boo-boo and operators think it is chatter. Chatter occurs at a frequency of natural frequency of the cutter, forced vibration occurs at spindle speed frequency. Easy to detect by audio.
5: How about power consumption?
I would imagine that the better the cutting operation the less HP used.
HPM progress
Current average cycle time improvement is 50%Originally Posted by Gunner
Smallish lot sizes <100 pcs (JIT scheduling), we chose jobs that are common runners for us, to maximize our savings with this 1st CNC retrofit. We have a total of 40 jobs programmed for this machine now and every one of those has run more than once.
Cycle time improvements range from 10 to 300%, that's why I mention the average of 50%. On jobs with a lot of drilling, we are not going to see a lot of improvement, jobs with a lot of milling we do...
Our fixturing, tooling, holders and most of the other main variables in machining processes have already been improved. Replacing the Fadal CNC Control was really forced on us. When we could not even come close to feeds/cycle times that the industry is showing off, we had to find out why. Anyone who makes parts for a living, I mean the guys who actually use the CNC, then inspect their work and they discover they cannot hold tolerances unless they slow the feed way down.
By the way, let me just say right here, that it is just down right goofy to think that the right RPM is all you need, to do HPM (High Performance Machining). Come on, a lot of us have many years of experience running CNC machines and I gotta say spindle speed is important don't get me wrong but, the effect on the features holding tolerance is not that great. Controlling chatter, oh yah, it's the biggest contributor. Our objective in using CNC machines is not to use them chatter free, but to make parts within specification, as fast as possible. Some day, I too will have to look at eliminating chatter; right now I don't have that problem. I guess that's because my spreadsheet does work, and I use the right RPM when I write the CNC programs.
On short run jobs, tool life is hard to measure. What we don't have is problems with tool breakage at the significantly higher feeds we are running now because we are still within the higher end of the range of each cutting tools chip load. It does help to have 6 high-pressure nozzles and good filtered 100-psi coolant flow. This was part of the upgraded Rubicon / Numeryx CNC control retrofit solution, as there was no way we were going to feed 3x faster without improving the coolant system. Power consumption is another difficult variable to quantify in a shop like ours. I could not guess what kind of savings we get here but I know it is significant.
Simply put, if we can get a given amount of work done in 1/2 the time average with a given amount of resources; power, man hours... Then we could say we saved 50% of the cost of those resources for that period of cycle up time, right?
That could add up to a lot of $'s in a hurry... ROI (Return On Investment) for us on this retrofit, was less than 3 months without considering the operational costs...
I'll try to keep you posted,
Scott_bob
I guess your right, rpm is a waste of time in machining. These articles are really not there, in your world. Welcome to audio chatter recognition - The Harmonizer. See: http://www.mmsonline.com/articles/080302.html
http://www.mmsonline.com/articles/0300rt2.html
Again, drilling: See:
http://www.mmsonline.com/articles/060101.html
Again, See:
http://www.mmsonline.com/articles/010301.html
A database is required for information. Spreadsheets are confusing and cumbersome when it comes to large amounts of information. The database I use is custom made to provide simple screens of information of each tool, thru the convenience of a user-friendly main screen. There are screens for cutter, balance, presetting, tool crib, programming, engineering, inventory, purchasing, contacts, etc. Tracking of stability lobes, frequency response functions created from impact testing, spindle, torque, cutting forces, harmonizer screen shots, pictures of the tool assembled, etc. The database can be queried from any field of the tool, called a stackup id. Show all cutters in a specific diameter, number of flutes, machine tool used on, material of cutter, material of part, specific power used by material being cut, etc. Reports far superior to any spreadsheet. Multiple choice, user input, etc. Form views of the database may also be shown in spreadsheet form, but as I said, it is to cumbersome to view all this information in spreadsheet form. The database can be used on a network so programmers, tool crib, presetting can see just the information they need to use. Parameters can be set by the administrator so certain people in the company can edit, add, subtract information or lock out others.
There you go again!
I did not say selecting the right RPM is a waste of time.
I just don't have to use your brand of dynamics to know what RPM to use. I use experience, and of course I refer to cutter mfgr's specs for the material I'm programming for, and I use Excel (spreadsheet) and it works great. I don't need a database to tell me what RPM to use.
You have said:
http://www.cnczone.com/forums/showpo...1&postcount=16
"Hardmill needs to get over it"?
I know it's not just me who is irritated by your lack of consideration for other people’s points of view. When you make statements excluding all other points of view except your own, you are being arrogant.
http://www.cnczone.com/forums/showpo...7&postcount=17
A lot of us here just believe your input has its place, just not the most important factor. Why don't you take a little of your own medicine. Get over it...
Example:
Here is a guy that is well respected in the mold machining industry.
Todd Schuett, wrote this article in 1997.
No mention of RPM as a significant variable, but a lot about the control and feed rate...
http://www.ctemag.com/viewArticle.asp?ARTICLE_ID=117
BTW, Mr Schuett is a competiter to the control that I chose, but I acknowledge his expertise in the field. You've got to be this way too if you are going to be an effective communicator.
Sincerely,
Scott_bob
Spreadsheet versus Database: The right tool
To effectively manage tooling and to maximize cutting operations in a shop, especially using higher rpms, large amounts of data are required. The reason being that with higher rpm's there may be different rpm's that can maximize cutting operations, called multipliers. You want to try and reach the number 1 multiplier. A cutter in a machine tool might want to run at 30,000 rpm x 1 multiplier (Deepest cut), 15,000 rpm x 2 multiplier (Less depth of cut) or 10,000 rpm x 3 multiplier (Less depth of cut). If for instance a machine tool can then only go 25,000 rpm and it has a cutter that dynamically wants to run at 30,000 rpm x 1 multiplier, then to run it at 25,000 rpm would be disastrous. Big problems would occur. It would be better to run it at 15,000 rpm producing higher depth of cut than at 25,000 rpm, better finish, longer tool life, higher mrr. There are many more scenarios, but this should give you an idea. This information needs to be given to the right people in the right format in the process of the operation in order to work correctly. Tool crib, presetting, programming, operator. Therefore - A spreadsheet does not achieve this by lack of design. Here is an experts words: Database vs Spreadsheet.
http://www.smartlink.net/~stevet/signal/sprd-db.htm
Note: (This makes a good comparison of controllers to dynamic's also. I have experienced both at the same time and know that dynamics - could do what controllers could not. They do work together, but controllers can not shine without dynamics being involved. Controllers (a spreadsheet) vs Dynamics (a database) I don't care what controller you use on a machine tool, sure there are good, better and etc of controller, if the controller does not include dynamics at the tool tip or audio monitoring which is at the tool itp during the cut - then it does not have all the information it needs to maximize, it is only guessing of a stable rpm by input of the programmer. (The unit that does take in dynamics in audio form is called CRAC - Chatter Recognition and Control) See:
http://www.rapid-response-consulting.com/knohow1.html
Article: Database vs Spreadsheet
I have spent several days in Beverly Hills this month training programmers on how to develop database applications in Microsoft Access. MS Access is a relational database management system for the Windows environment. This impressed on me again what a versatile programming tool MS Access is.
Yesterday, a client of mine called once again with the same problem. He has repeatedly messed up a formula in a spreadsheet he uses to keep track of jobs and man-days. So I bailed him out again.
When considered together, I thought I should address a common practice of PC users issue in this article. They use a spreadsheet to try to do database management.
This is an easy trap to fall into. The developers of major spreadsheet programs, such as MS Excel and Quattro Pro, make the bait even harder to resist. They keep adding more database-type functionality into their spreadsheet products. A spreadsheet lends itself to just throwing some data in there and massaging it later. It does not take thinking the project through and setting things up pretty much the way you want them from the outset. It requires less knowledge and experience to work with a spreadsheet as opposed to a database.
On the other hand, spreadsheets can so easily be altered that they are prone to the types of problems my client had. Not so with databases. With a database you pretty much have to deliberately change the structure of the database, or the way it displays information, etc. An inherent problem with trying to use a spreadsheet as a database is that often one day you may want to view or combine the data in a way that the spreadsheet cannot do (whereas it would be child's play for the database).
Spreadsheets are real good at calculations and crunching numbers. That is what they are for. They are not databases, and do not lend themselves to the tasks that databases do so well.
*Some shy away from databases because they do not know how to develop database applications, do not want to learn how, and do not want to spend the money to have a consultant do it. In the long run, often the productivity and wages expended on trying to get a spreadsheet to suffice outweighs what it costs to have a database developed.
Take a look and see if you know anyone who could benefit from this.
Another good article for HSM:
http://www.ctemag.com/pdf/0203-lookahead.pdf
I wonder who taught Todd everything, well maybe not everything, he knows...
Does anyone know?
Sincerely,
Scott_bob
I know that answer. Gil for Numeryx.
Donovan
Scott_Bob,
Are these algorithms based on keeping a constant surface footage or are they just for controlling overshooting / Jerk? As the control alters feedrate will/can it also alter spindle speed?
Gunner
Gunner
Automatic feed compensation
The ansewer to the question on feed compensation is better described in this article writen by Todd Schuitt:
Feed compensation, not speed compensation...
For that you'd need the dynamic "system" like what Camminc talks about...
Donovan is right, Todd worked for Gil in his early years at Sharnoa where they developed the Tiger control...
Now that team have all speciallized individually and developed their own products. Excelent products, that out perform even the GE Fanuc giant in our industry IMO.
I'm sure the Lathe guys out there could not imagine not having the constant surface speed control capabilities of modern Lathes. This motion control option automatically compensates the RPM and Feedrate depending on the Tool tip relative to the diameter of the rotating part. In other words, RPM and Feedrate are increased up to the Max RPM, Feedrate is matched to spindle speed, maintaining a constant chip load. This is extremely desirable to maximize feed rate, tool life and is key to better surface finishes, maintaining a constant chip load affects all these...
In Milling, it is nearly imposible to do the same thing. But Rubicon/Numeryx control does this better than anything else I've seen. Just yesterday, a job running on our newest Fadal with their Mp32 control, we had to slow the programmed feed rate down to F30. where this same job running on the Rubicon / Numeryx was programmed at F230. or 7x faster, Gotta make good parts...
Sincerely,
Last edited by Scott_bob; 01-06-2005 at 04:54 PM.
Scott_bob
Posting Permissions
Reply with Quote