So Who Is A Computer Programmer? I Got A Thought.

[Keith Foor]

I had a thought for those that are trying to put a CNC project together and having problems with the idea of buying ball screws for their mill / lathe and fitting them to the machine. If a CNC program was created that took the output of some of the less expensive DRO scales that are being used with a tablet or other device and actually looked at the table position for making a cut, the backlash of the standard screws wouldn't be relevant. It would simply go to a position on the DRO scale and be in the correct position as long as the scale was accurate. By this I mean if the X axis was to make a cut of 1 inch from left to right then back up .5 inches and then the Y axis was to cut in some distance. the backlash on the machine lets say is .1 (alot but easy math). The machine will return to .6 and not .5 due to the take up of the back lash. If the DRO was referenced then the table would return to .5 because your not counting turns on a stepper motor, you are actually referenceing table location.

I think this would get alot of people past the hump of converting machines due to the cost and complexity of ball screws and a machine tear down and retrofit of the ball screws.
I realize that there are specific kits for Bridgeport mills that the screws are a straight forward direct drop in, but something like an Excello 604 (what I have) is not gonna have a direct fit, no modification needed ball screw kit to convert it.
I know there is Moglice (or something like that) to use to drill out the threads on the nuts and rebuild the nuts on the machine but that frankly is scarier than the ball screw retrofit as drilling and ruining nuts that can't be simply replaced off eBay for a few bucks.

Don't know if there is a programmer out there that would want to take this on, but if so, there's your concept for it. If you choose to take it on I claim no patent on the idea, but would ask for a copy for personal use.

 

[JimDawson]

You are correct Keith. The backlash is self correcting to a point when the scales are on the load (table), it just depends on how fast the servo loop updates and how sensitive the amps are as to what the accuracy is if there is backlash in the system. That is exactly the way I set up my mill, my router, and Alloys mill. In those cases using 1 micron magnetic scales rather than glass scales, but standard DRO 1 or 5 micron glass scales would work fine and do exactly the same job as the magnetic scales.

Most glass DRO scales output a standard 2 or 4 channel quadrature signal that a motion controller can read.
.
.

 

[Steve Shannon]

In answer to to question posed in the subject line, I used to be, using primarily C to program realtime systems and editors for realtime systems. Then I programmed mapping systems (GIS). Then I went back to programming a realtime special protection system for a 500kv electric transmission network.
Using DROs for feedback in positioning an end mill makes perfect sense when you can wait until you're at the correct location before engaging the cutter, such as positioning holes, but if you need to precisely follow a path while the cutter is cutting, such as cutting a circle, having backlash introduces mechanical hysteresis which makes it impossible to follow the path with precision.


Steve Shannon

 

[JPigg55]

I'm no programmer, but have had a little introduction lately while attempting to hack my iGaging Absolute DRO scales for use with the Touch DRO App.
I'd asked a similar question to a friend who's helping me do this and is a programmer.
My idea was to use a record function to record stepper/servo info like revolutions/steps, direction along with position feedback and use that info to reproduce manually inputted commands.
I won't say it isn't possible, but there are many issues to overcome.
First, electronic components like this use clock cycles and refresh rates. My scales, for instance, has a 2.5 khz clock cycle which is fast enough, but only a 10 hz refresh rate meaning the position only update 10 times per second. May seem fast to you and me, but would be slow to electronics if your trying to hold to a couple thou.
Second, unless you're willing/able to write a program to your specific set-up, trying to write a program that would work with all the different scales, drives, and motors would be a daunting task to say the least.

I'd still like to try my idea someday for simple projects not requiring close accuracy. It would be a lot cheaper and easier than going full on CNC for reproducing simple items.
This will only work for simple, single axis commands unlike CNC which provides multi-axis control for complex designs.
In my concept, scale position input would only be used for error detection and could probably be ignored all together.

 

[JimDawson]

Second, unless you're willing/able to write a program to your specific set-up, trying to write a program that would work with all the different scales, drives, and motors would be a daunting task to say the least.

I would say this this is more a function of the hardware rather than the software. If you have the hardware (motion controller) to connect to the field devices (motors, drives, encoders) and the field devices are up to the task, the software is pretty easy. This is off-the-shelf equipment.

Edit: In this case encoders include both rotary, linear, and any other types that could be hung on a system. From the controller perspective it makes no difference.

 

[Steve Shannon]

In answer to to question posed in the subject line, I used to be, using primarily C to program realtime systems and editors for realtime systems. Then I programmed mapping systems (GIS). Then I went back to programming a realtime special protection system for a 500kv electric transmission network.
Using DROs for feedback in positioning an end mill makes perfect sense when you can wait until you're at the correct location before engaging the cutter, such as positioning holes, but if you need to precisely follow a path while the cutter is cutting, such as cutting a circle, having backlash introduces mechanical hysteresis which makes it impossible to follow the path with precision.


Steve Shannon

After thinking about this I'm pretty sure you didn't mean the kind of programming experience that I have, but rather about CNC programming, of which I have virtually none. I shouldn't have said it would be impossible to follow a path; I'm probably wrong. I can see how it might be possible to have CNC controller firmware wait for backlash to be taken up in order to compensate, then, when the sensors detect movement to resume movement along that axis. I really look forward to learning about such things from people like Jim Dawson.

 

[wrmiller]

I've designed similar systems and one (of many) of the critical requirements is sampling rate from your position monitoring device. A cheap dro or digital caliper scale won't have the necessary frequency of updates (sampling rate) necessary to monitor real-time positioning with any useful accuracy. The laser interferometers I used updated thousands of times a second, which was necessary to calculate ramp up/down profiles and at-speed positional monitoring in real-time.

 

[JimDawson]

Steve, there are many books written on this subject. It's really a matter of understanding the general theory of closed loop systems of which CNC machines are a subset. I'm sure you have a lot of experience with this in your power grid programming.

 

[Keith Foor]

In answer to to question posed in the subject line, I used to be, using primarily C to program realtime systems and editors for realtime systems. Then I programmed mapping systems (GIS). Then I went back to programming a realtime special protection system for a 500kv electric transmission network.
Using DROs for feedback in positioning an end mill makes perfect sense when you can wait until you're at the correct location before engaging the cutter, such as positioning holes, but if you need to precisely follow a path while the cutter is cutting, such as cutting a circle, having backlash introduces mechanical hysteresis which makes it impossible to follow the path with precision.


Steve Shannon

You bring up a good point that I had not considered. I was looking at it in a 2 dimensional way and not taking into consideration that a circle or other curve would require the screws to reverse at some point. Well that shoots that squarely in the foot. But I at least have an answer.

Looks like I may be looking for a bridgeport at some point to convert, or just bite the bullet and find a used CNC mill of some sort.
I can't bring myself to CNC my Excello 604. Odly because it's such a good manual mill and is in terffic shape for the fact I only paid $1200 for it.
I actually believe the size of it scared folks away from it. I see Bridgeports selling in this area for twice that, and actually selling.

 

[Steve Shannon]

Steve, there are many books written on this subject. It's really a matter of understanding the general theory of closed loop systems of which CNC machines are a subset. I'm sure you have a lot of experience with this in your power grid programming.

Yeah, everything was closed feedback with lots of PIDs. My lack of experience is with the CNC controllers, their features, and their algorithms. I intend to fix that.
Thanks!