I have been making boards commercially since 1968, and technologies have moved a long way since then. Now I rarely make boards, but plan to re-activate that at some near future time. Over the years I have worked quite a bit in improving the mechanical processes (I'm not a chemist so I had no hand in the though-hole-plating process that we used), and can tell you that precision is not as easy to come by as you may think. Your response to Phil about squaring and trimming to a couple of thou seem quite naive to me. Even the commercial operators would not expect to achieve that. Now, those accuracies are probably not required, but DO expect to have lots of headaches getting RELIABLE and REPEATABLE results. .... This is why a camera is required, so you can calibrate things.
Now, I'm not looking to put a negative spin on your enthusiasm, and wish you every success, but, unless your requirements are very nominal, be prepared for some set-backs. For example, are you aware that even commercial precision lead screws may have an accuracy error of a few thou per foot? And even then not consistently over the length of the rod? So precision measurements are made and loaded into tables to "calibrate" the positions of the motors to remove those inaccuracies from the final position result of less than a thou.... Again, a camera here would be very useful for you.
A "few thou" sounds simple enough, but is VERY, VERY tough to achieve, and ought not be glossed over if you really mean to get those kinds of precisions.
It is not my goal to create commercial boards. If inaccuracies become an issue, I will just increase pad sizes. I certainly have to give it a go, because I have come a long way not to even try.
Your response to Phil about squaring and trimming to a couple of thou seem quite naive to me.
You would be amazed at the accuracy that can be achieved by combining the gearing of an old Zebco 202, an old 14.4 DeWalt drill motor, some recycled linear shafting, and a new Dremel 545 diamond wheel.
Yes, I get surprised every day... lots of clever thinking is possible.
But if inaccuracies that crop up simply change the specifications, or pad sizes as you say, then success is almost guaranteed because the goal line is a moving target. And there is nothing wrong with that. I simply took issue with you glossing over making something within a few thou as seeming trivial, and in my experience it is FAR from that.
Most of the devices i have seen use a light barrier because mechanical switches are to sloppy to use for repeatability over time for other than limit detection. You may also want to consider either using the limit switch or a barrier at the opposite end from zero so that should you decide to drive away from zero then to zero (to avoid under run limit) in case the unit starts to seek zero at the far end of travel. Also, do you plan to use an encoder (absolute or simple quadrature) to determine / measure position from zero? Randomquick thoughts
It is rarely fun, but quite often, it is very rewarding and satisfying. Over the last several days, I have been drilling and tapping many holes, as well as making many other needed parts. As you might have guessed, since it is an XYZ machine, I needed three carriage assemblies. This afternoon I finished the remaining parts for the carriage assemblies, and I must say that all three look identical, very nice, and very unique.
To my knowledge, there will be no other assemblies like them in the world, but I could be wrong. The fun part is creating new ideas and concepts.
The way to deal with backlash when drilling is always to make your final approach to the hole position in the same direction (e.g. +X and +Y). If you're coming from another direction, simply overshoot the position and return along the proper vector.
-Phil
Funny you should mention this little trick, I can recall calibrating certain dot matrix printers (alps and epson?) this way for plotting and certain printing functions. and these prints would always print from the same direction across the page. (vertical was not an issue on the printers I guess because the paper always fed up and out). Same thing for galvo scanning in a laser based film printer system, the laser theoretically could scan in both directions, but that would complicate the he77 out of the calibrations and ultimate accuracy of the print.
There is always change. Everything has a radius, not an edge. Surfaces are never flat, or pure, or fair. Nothing is truly square, etc... No two physical things are exactly the same. As far as I am aware, the only real constants, that always perform to spec are molecules, and even those things probably have some variances in them the theoretical physics guys can tell us about. Since I can't manipulate them in an atomic way, the whole affair is "above my pay grade", essentially!
It is all about management of acceptable tolerances, and those need to be matched to the manufacturing process variances to understand what repeatable results are possible, and verify they are in-fact happening, and or practical.
Machine wear makes no difference regarding accuracy and repeatability. Also, a failing feedback device can be identified before the machine operator is knee-deep in scrap.
If you're looking for really good repeatability, many people who've made the CNC machine I have use hall-effect switches and small rare-earth magnets. This may be affected by ferrous metal debris, but if you don't have that issue it's fine, and not subject to mechanical wear or debris getting in the way of optics.
I have not had the opportunity to toy around with hall effects yet, but I did put an AB proximity sensor or the wire bender, and it worked great. I embedded it in an aluminum table and it still picked up a 0.032" piece of music wire passing by.
Pointing down, to lessen the chance of debris settling on optical part.
And have a matt black surface at a distance, and when a shiny part gets under it to set it off.
Or go with a slot sensor.
But personally I would go with a cheap USB microscope camera mounted next to the grinder, and "look" for a start point.
Comments
I have been making boards commercially since 1968, and technologies have moved a long way since then. Now I rarely make boards, but plan to re-activate that at some near future time. Over the years I have worked quite a bit in improving the mechanical processes (I'm not a chemist so I had no hand in the though-hole-plating process that we used), and can tell you that precision is not as easy to come by as you may think. Your response to Phil about squaring and trimming to a couple of thou seem quite naive to me. Even the commercial operators would not expect to achieve that. Now, those accuracies are probably not required, but DO expect to have lots of headaches getting RELIABLE and REPEATABLE results. .... This is why a camera is required, so you can calibrate things.
Now, I'm not looking to put a negative spin on your enthusiasm, and wish you every success, but, unless your requirements are very nominal, be prepared for some set-backs. For example, are you aware that even commercial precision lead screws may have an accuracy error of a few thou per foot? And even then not consistently over the length of the rod? So precision measurements are made and loaded into tables to "calibrate" the positions of the motors to remove those inaccuracies from the final position result of less than a thou.... Again, a camera here would be very useful for you.
A "few thou" sounds simple enough, but is VERY, VERY tough to achieve, and ought not be glossed over if you really mean to get those kinds of precisions.
But I know you are having fun, so have at it!
Cheers,
Peter (pjv)
It is not my goal to create commercial boards. If inaccuracies become an issue, I will just increase pad sizes. I certainly have to give it a go, because I have come a long way not to even try.
Yes, I get surprised every day... lots of clever thinking is possible.
But if inaccuracies that crop up simply change the specifications, or pad sizes as you say, then success is almost guaranteed because the goal line is a moving target. And there is nothing wrong with that. I simply took issue with you glossing over making something within a few thou as seeming trivial, and in my experience it is FAR from that.
Again, have fun!
Cheers,
Peter (pjv)
F
It is rarely fun, but quite often, it is very rewarding and satisfying. Over the last several days, I have been drilling and tapping many holes, as well as making many other needed parts. As you might have guessed, since it is an XYZ machine, I needed three carriage assemblies. This afternoon I finished the remaining parts for the carriage assemblies, and I must say that all three look identical, very nice, and very unique.
To my knowledge, there will be no other assemblies like them in the world, but I could be wrong. The fun part is creating new ideas and concepts.
Bruce
Funny you should mention this little trick, I can recall calibrating certain dot matrix printers (alps and epson?) this way for plotting and certain printing functions. and these prints would always print from the same direction across the page. (vertical was not an issue on the printers I guess because the paper always fed up and out). Same thing for galvo scanning in a laser based film printer system, the laser theoretically could scan in both directions, but that would complicate the he77 out of the calibrations and ultimate accuracy of the print.
FF
There is always change. Everything has a radius, not an edge. Surfaces are never flat, or pure, or fair. Nothing is truly square, etc... No two physical things are exactly the same. As far as I am aware, the only real constants, that always perform to spec are molecules, and even those things probably have some variances in them the theoretical physics guys can tell us about. Since I can't manipulate them in an atomic way, the whole affair is "above my pay grade", essentially!
It is all about management of acceptable tolerances, and those need to be matched to the manufacturing process variances to understand what repeatable results are possible, and verify they are in-fact happening, and or practical.
Machine wear makes no difference regarding accuracy and repeatability. Also, a failing feedback device can be identified before the machine operator is knee-deep in scrap.
Mickster
I have not had the opportunity to toy around with hall effects yet, but I did put an AB proximity sensor or the wire bender, and it worked great. I embedded it in an aluminum table and it still picked up a 0.032" piece of music wire passing by.
Bruce
Photomicrosensors Reflective Optical Sensor
http://www.mouser.com/ProductDetail/Vishay-Semiconductors/CNY70/?qs=%2fjqivxn91ccGZsGvLZBEvNzIDLcVhZAa
Pointing down, to lessen the chance of debris settling on optical part.
And have a matt black surface at a distance, and when a shiny part gets under it to set it off.
Or go with a slot sensor.
But personally I would go with a cheap USB microscope camera mounted next to the grinder, and "look" for a start point.