Minimum Computer Functionality Needed for CNC

[Too_Many_Tools]

I am trying to determine what the MINIMUM requirements are for a computer to run CNC.

What are the MINIMUM requirements for the popular CNC applications frequently used today?

Does it change with the OS that you use...and if so, by how much?


Thanks

TMT

[Controller]

Hi
I've just been looking on ebay UK and there is an ORAC CNC lathe item 140010207061 , this has 12 G codes printed on the headstock, I presume they think this is enough for a training model which the ORAC is.

[NC Cams]

Depends on the O/S and software - Period. Extremes presented for consideration as they are NOT out of the ordinary:

The DX32 system used by Bridgeport subsequent to BOSS ran 4 axis CNC VMC and 2, 2.5 and 3 axis Eztrak and 2 axis Lathe (plus spindle speeds) plus 3 axis Ez surf ALL ON DOS 6.22.

Mach needs XP

DX32 runs on dog dirt simple 386SX, 386DX, 486 and Pentiums up to 133 mhz - all depends on which version of machine you have as to the level of PC in vogue at the time.

XP's run the gammut on speed and memory required.

With respect to your question: "Does it change with the OS that you use...and if so, by how much?"

YES and LOADS, respectively.

Your question needs to be rephrased to obtain an answer that does not go into a dissertation on every OS, computer and or software permutation that could be in use. What are you SPECIFICALLY trying to do?

[Too_Many_Tools]

Originally Posted by NC Cams Depends on the O/S and software - Period. Extremes presented for consideration as they are NOT out of the ordinary: The DX32 system used by Bridgeport subsequent to BOSS ran 4 axis CNC VMC and 2, 2.5 and 3 axis Eztrak and 2 axis Lathe (plus spindle speeds) plus 3 axis Ez surf ALL ON DOS 6.22. Mach needs XP DX32 runs on dog dirt simple 386SX, 386DX, 486 and Pentiums up to 133 mhz - all depends on which version of machine you have as to the level of PC in vogue at the time. XP's run the gammut on speed and memory required. With respect to your question: "Does it change with the OS that you use...and if so, by how much?" YES and LOADS, respectively. Your question needs to be rephrased to obtain an answer that does not go into a dissertation on every OS, computer and or software permutation that could be in use. What are you SPECIFICALLY trying to do?

Thanks for the response...it is appreciated.

The reason why the question is general is that I have a number of homemade CNC systems in the works. I also have a generous supply of older QUALITY computers that I would like to use for these CNC systems. When one looks at what different CNC applications need for their operations, you seldom see an accurate response...their response is usually "faster is better"...which does not tell you anything. So I am asking the user base for their experiences.

Has anyone done a review of the major CNC applications and have they done a CPU loading test? If so, I would be interested and I would guess others would be too.

TMT

[NC Cams]

If you open up the boxes, you'll probably see that now 2 are alike. Besides layout, you'll specifically note the the support chipset will differ from board to board.

Consequently, each and every machine will do things just a little bit differently.

Then look at processor. An AMD clone may not do things exactly the sme or as well or even bettre as a true Intel Pentium, Celeron or whatever whazoo name they come up with next.

Some guys plug stuff together and it works. Some do it and it has to be trouble shot. In the plug and play life of today, the order in which you insert support cards into a PnP computer will determine interrupt settings the computer self generates which can affect operation.

With the myriad of computers, chip sets, ancillary cards, processors and software out there, the likelihood of finding out exactly what you're looking for are slim and none.

The computer mags have done comparison ratings via "computer shootouts" for years. They clearly show the each one does things differently and some do it better than others. Sadly for the CNC enthusiast, CNC testing is NOT included in any of the computer rags I've ever seen.

Computer software developers will tend to specifically develop their product with certain PC's in mind. You're better off using what they recommend and THAT's where its been tested. There is no "standardized PC hardware" that CNC software developers use. Reason: too many computers out there for there ever to be a consensus developed on/for.

I have a piece of camshaft test equipment that was developed or use on a HP computer. The developer said it wouldn't run on anything but. We refused to buy what we felt were overpriced PC's and sent him a generic one (486 at the time, DOS based software). It ran perfectly. WHen the 486 died, we replace it with a Pentium whatever. STill running fine.

Yes, faster is better, especially if you're using Windoze. More memory is better and too much isn't enough, especially if you're using Windoze. However, if you come across a legacy system that uses a BMDC card from a Bridgeport mill, lather or VMC, 133mhz and an ISA bus are the max you can use.

The last sentence in reply #3 above pertains here as much as ever.....

[Too_Many_Tools]

Originally Posted by NC Cams If you open up the boxes, you'll probably see that now 2 are alike. Besides layout, you'll specifically note the the support chipset will differ from board to board. Consequently, each and every machine will do things just a little bit differently. Then look at processor. An AMD clone may not do things exactly the sme or as well or even bettre as a true Intel Pentium, Celeron or whatever whazoo name they come up with next. Some guys plug stuff together and it works. Some do it and it has to be trouble shot. In the plug and play life of today, the order in which you insert support cards into a PnP computer will determine interrupt settings the computer self generates which can affect operation. With the myriad of computers, chip sets, ancillary cards, processors and software out there, the likelihood of finding out exactly what you're looking for are slim and none. The computer mags have done comparison ratings via "computer shootouts" for years. They clearly show the each one does things differently and some do it better than others. Sadly for the CNC enthusiast, CNC testing is NOT included in any of the computer rags I've ever seen. Computer software developers will tend to specifically develop their product with certain PC's in mind. You're better off using what they recommend and THAT's where its been tested. There is no "standardized PC hardware" that CNC software developers use. Reason: too many computers out there for there ever to be a consensus developed on/for. I have a piece of camshaft test equipment that was developed or use on a HP computer. The developer said it wouldn't run on anything but. We refused to buy what we felt were overpriced PC's and sent him a generic one (486 at the time, DOS based software). It ran perfectly. WHen the 486 died, we replace it with a Pentium whatever. STill running fine. Yes, faster is better, especially if you're using Windoze. More memory is better and too much isn't enough, especially if you're using Windoze. However, if you come across a legacy system that uses a BMDC card from a Bridgeport mill, lather or VMC, 133mhz and an ISA bus are the max you can use. The last sentence in reply #3 above pertains here as much as ever.....

Again thanks for the response....


Well, I think I have answered that "last sentence"....I have a number of homemade CNC systems in the works (gantry routers, converted knee mills) in the works AND have a number of quality (rack mount industrial) computers that I would like to use in these systems. That is what I have in terms of hardware to work with to define the software requirements of what CNC application to choose from.

What I am trying to determine is what the actual computing requirements are for the popular CNC applications by asking the user base what they are using and are they satisfied with their selection.

The apparent fact that there appears to be no thoughput testing results available for the different CNC applications complicates the process. Hasn't anyone run two different CNC applications from competitors on the same CNC system, compared the results and then posted the results?

Is randomly selecting a CNC application and computer system the manner in which homemade CNC systems are built today? If so, I am surprised...I would have expected a more standardized process by now as the market place has matured.

TMT

[ger21]

It's very simple. The PC requirements depend on the software. Most homebuilt machines use Mach3, which needs at least a 1 Ghz, up to 2Ghz+ if you need 45,000 steps per secon, or TurboCNC, which will run on a 200Mhz DOS machine.

Pick the control software you want to use, and you'll have the answer.

[NC Cams]

Machinists are not typically computer geeks and few computer geeks become machinists. This CNC thing is either make do with what you have or buy what you're told to get for the software or machine interface you plan to use.

It is hard enough to get ONE system to work than to try to get TWO to work and then pick the better - especially if you're dealing with Windoze.

Computers are a means to an end - as this is not a computer science class experiment, what you seek is simply not done. And I don't think that PC Magazine is not going to do CNC retrofit speed through put testing any time soon.

Post #7 by GER says volumes....