584,846 active members*
4,532 visitors online*
Register for free
Login
IndustryArena Forum > MetalWorking Machines > Benchtop Machines > Pico systems closed loop servo boards.....
Page 1 of 2 12
Results 1 to 20 of 26
  1. #1
    Join Date
    Apr 2007
    Posts
    2580

    Pico systems closed loop servo boards.....

    Just got done reading an article in the Digital machinist magazine I have been getting and noticed a small ad at the bottom of the page touting a true closed loop board that uses Gecko drives somehow. Checked their website and it has a gecko board that uses some other board and allows closed loop. IS this true and is anyone here using this system on their machine?

    It says in the ad that the system will maintain dro control even when the geckos are powered down. Sounds real cool and the board is only about $100.00 . Do you need to use the other board or can you just use this with an existing breakout board like my C11g? Any and all information on this would be most appreciated...peace

  2. #2
    Join Date
    Mar 2004
    Posts
    1804
    Pete,
    John makes interfaces for EMC!

    From his page listing the board your referring to:
    "Pico Systems has developed several products to interface the EMC motion-control program to motor drivers"

    Sorry bout that!
    Art
    AKA Country Bubba (Older Than Dirt)

  3. #3
    Join Date
    Apr 2007
    Posts
    2580

    WOw man bummer dude....

    Did not see that it was only for emc.... Sounds like Jon has to get his act together and work on a system that will run with MAch3.... Would not know the first thing about Linux or emc.... '


    HEY BUBBA!!! hows it going man..... I have been out in the shop working on the enclosure and beleive it or not it is starting to look like a control unit. I got the Vector drive mounted, the fan mounted and ported and I am now working on all the little fuse mounts and switch mounts.

    I ordered a female 110vac socket for the control unit and a fuse holder and bought a power switch at the shack yesterday. I also ordered a bunch of shrink wrap, some buss blocks, a cool new wire stripper that I used at Bills that I just had to have and a few other odds and ends. Should be here mid week I suppose. I am cutting all the holes for everything I have here and what I can from the specs on the net for each particular connection point and I am hoping to get this bad boy painted tonight. Then I maybe can insert the control board inside and start to permanently wire everything tomorrow and over the next week. I still gotta debug my darn computer and I am thinking I will wipe out all the harddrives and reload windoze and start with a fresh load without all the crap from the internet and my kids programs and all... I was also thinking about your saying that you mounted the control unit on the back of the machine and I had an idea that I might mount it to the wall behind the machine since my mill is on a large welded rolling stand I made when I bought the machine. That way I do not have to make such a large enclosure which will maybe save me a few pennies... What say ye? peace

  4. #4
    Join Date
    Mar 2004
    Posts
    1804
    Pete,
    Been hanging in there trying to stay cool. Darn HI is above 100 so am staying in the AC!!!

    As far as: "Sounds like Jon has to get his act together and work on a system that will run with MAch3.... Would not know the first thing about Linux or emc.... '", He MIGHT say you should get your act together and jump on the EMC bandwagon

    There is a lot to be said about both systems and I am NOT trying to start a flame war here. To each his own.
    Art
    AKA Country Bubba (Older Than Dirt)

  5. #5
    Join Date
    Apr 2007
    Posts
    2580

    yup......

    figured you would say that....haha...


    I would think that if it can be done in one program it could be done in another.... no idea about how tho.... is Jon a member here? peace

  6. #6
    Join Date
    Mar 2004
    Posts
    1804
    Yeh, I have seen several posts from Jon, but looks like he is busy on the EMC list this afternoon.

    By the way, there is a board that I think Rogers Machine makes to do what you want.
    Art
    AKA Country Bubba (Older Than Dirt)

  7. #7
    Join Date
    Apr 2005
    Posts
    1778
    Quote Originally Posted by pete from TN View Post
    I would think that if it can be done in one program it could be done in another.... no idea about how tho.... is Jon a member here? peace
    Well, that would be a definite maybe. Mach was not designed as a real time controller. It is designed to work around some of the deficiencies of Windows (I am only talking about deficiencies related to real time control). So, if someone had access to the source code of both Mach and Windows, it might be possible to solve the real-time access problems.

    EMC2 was designed from the ground up to be real time control software. It runs on version of Linux that has had the kernel modified to provide real-time access. Even so, some hardware grabs the processor and doesn't play nice, so that EMC2 won't run well on that hardware.

    Alan

  8. #8
    Join Date
    Feb 2007
    Posts
    592

    Just a thought

    A very popular card with the EMC2 crowd is the Mesa systems 7i43 Anything I/O card.

    Well it appears it has caught the attention of the Mach developers... This $99 card can do stepgen as fast as the smoothstepper AND take back encoder position.

    You are still at the mercy of the G320/340 internal fixed +/- 128 count following error - but they might be able to trip an E-Stop before it gets that far.

    Eventually I figure they will write code to use this card for rigid tapping to.

    A :wave: from the EMC2 camp.

  9. #9
    Join Date
    Mar 2004
    Posts
    369
    Quote Originally Posted by skullworks View Post
    A very popular card with the EMC2 crowd is the Mesa systems 7i43 Anything I/O card.

    Well it appears it has caught the attention of the Mach developers... This $99 card can do stepgen as fast as the smoothstepper AND take back encoder position.

    You are still at the mercy of the G320/340 internal fixed +/- 128 count following error - but they might be able to trip an E-Stop before it gets that far.

    Eventually I figure they will write code to use this card for rigid tapping to.
    Of course, EMC2 has had rigid tapping for over 2 years, now. I also have a board that interfaces through the parallel port and support up to 4 axes, and also has digital I/O for limit and home switches, home to index pulse, and can do either step/direction or PWM for a digital servo amp.
    It is not as cheap as the Mesa, but you don't need a bunch of breakout boards, adaptors and connector panels with it.

    Jon

  10. #10
    Join Date
    Mar 2004
    Posts
    369
    Quote Originally Posted by pete from TN View Post
    Just got done reading an article in the Digital machinist magazine I have been getting and noticed a small ad at the bottom of the page touting a true closed loop board that uses Gecko drives somehow. Checked their website and it has a gecko board that uses some other board and allows closed loop. IS this true and is anyone here using this system on their machine?

    It says in the ad that the system will maintain dro control even when the geckos are powered down. Sounds real cool and the board is only about $100.00 . Do you need to use the other board or can you just use this with an existing breakout board like my C11g? Any and all information on this would be most appreciated...peace
    No, you need BOTH the Universal Stepper Controller (USC) and the Gecko interface. If you already have the Gecko servo drives, it is a good solution to a number of problems. If you don't already have the Gecko drives, then it is cheaper to use my own servo drives.

    The Gecko interface is designed to copy the encoder signals, via opto-isolators, to both the Gecko 320 and the USC. if you want to use this with steppers, there is no reason to use the Gecko interface. You put encoders on your motors and connect directly to the USC. You can flip a switch (on a per-axis basis) to convert each axis from open-loop (board counts step pulses) to closed-loop, using the encoder.

    For the most part, none of this is designed to work with Mach or other breakout boards. In theory, if you had Mach set up to read encoders, the Gecko interface could be used to copy the encoder signals to the computer and Gecko 320 drives. I don't know if anyone has done this, but it should work.

    Just to be complete, here's what the Gecko interface does :

    1) provide power to the encoders all the time.

    2) encoder signals come directly out to be sent to USC or computer.

    3) copy encoder signals with opto-isolation for the Gecko 320 servo drive

    4) power up and then reset geckos for one second when enable command is given

    5) If any Gecko faults, power all Geckos off, connect to braking resistor to safely decelerate motors and light an LED to indicate which drive caused the fault. Signal the computer a fault has occurred.

    6) When powering the Geckos on, the voltage is ramped up gradually to avoid current surges.

    Jon

  11. #11
    Join Date
    Feb 2006
    Posts
    7063
    Quote Originally Posted by pete from TN View Post
    Just got done reading an article in the Digital machinist magazine I have been getting and noticed a small ad at the bottom of the page touting a true closed loop board that uses Gecko drives somehow. Checked their website and it has a gecko board that uses some other board and allows closed loop. IS this true and is anyone here using this system on their machine?

    It says in the ad that the system will maintain dro control even when the geckos are powered down. Sounds real cool and the board is only about $100.00 . Do you need to use the other board or can you just use this with an existing breakout board like my C11g? Any and all information on this would be most appreciated...peace
    Pete,

    That board, by itself, does nothing like what you describe. It does little more than buffer the encoder signals, keep the encoders powered even when the Geckos are turned off, and causes a fault on any one Gecko to shut them all down and E-Stop the controller software.

    I truly do not understand the fascination people have with the whole concept of "closed loop" on a stepper controller. If you design a stepper based system correctly, and operate it within it's capabilities, closed loop is totally unnecessary. Lost steps are a result of poor design and/or improper operation. Period. No controller is going to be able correct a badly designed or badly operated stepper system on-the-fly without resorting to servo-like control algorithms. And if you're going to do that, simply use servos in the first place!

    Regards,
    Ray L.

  12. #12
    Join Date
    Apr 2007
    Posts
    2580

    Actually Himy....

    I am using servos.... I have read a few things from mariss on the net concerning the gecko drives and in a nutshell you are right about the idea that if you run a system the way it was designed you should not need a closed loop and the G320's have a counter that internally monitors position and faults if there is a problem. There is a certain amount of allowance for that and for most things that would be just fine. However if you are really trying to get every ounce accuracy out of your machine would it not make sense to try to close the loop and get positional information fed back to the thing that is running the whole show in the first place? I have had a few conversations with a fellow that uses one of these closed loop setups that was custom made for a small high precision cnc router and he said he is very happy with it. He also said that he nearly never has to reset his position since the machine even remembers it's position when not in use. The idea that the encoders track things even when the geckos are off is pretty cool I think and should make for a more accurate machine I would guess...

    Rigid tapping is also a very cool feature and I would love to see a homebuilt machine do it. I intended to buy a tapping head or possibly the tormach compression system. I know that Hoss was able to do it with his freak somehow, I want to be able to do that at some point as well.. What say ye? peace

  13. #13
    Join Date
    Jul 2003
    Posts
    1753
    Home built machines can rigid tap. Emc2 can do it even through the printer port.

    http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?Pjm

    that is a stepper machine reading the spindle encoder back thru the printer port for rigid tapping (spindle synced motion).

    video here
    [ame="http://www.youtube.com/watch?v=C740zS9R9kk&feature=channel_page"]YouTube - Rigid Tapping M3X 5 750rpm[/ame]

    sam

  14. #14
    Join Date
    Feb 2006
    Posts
    7063
    Quote Originally Posted by pete from TN View Post
    I am using servos.... I have read a few things from mariss on the net concerning the gecko drives and in a nutshell you are right about the idea that if you run a system the way it was designed you should not need a closed loop and the G320's have a counter that internally monitors position and faults if there is a problem. There is a certain amount of allowance for that and for most things that would be just fine. However if you are really trying to get every ounce accuracy out of your machine would it not make sense to try to close the loop and get positional information fed back to the thing that is running the whole show in the first place? I have had a few conversations with a fellow that uses one of these closed loop setups that was custom made for a small high precision cnc router and he said he is very happy with it. He also said that he nearly never has to reset his position since the machine even remembers it's position when not in use. The idea that the encoders track things even when the geckos are off is pretty cool I think and should make for a more accurate machine I would guess...

    Rigid tapping is also a very cool feature and I would love to see a homebuilt machine do it. I intended to buy a tapping head or possibly the tormach compression system. I know that Hoss was able to do it with his freak somehow, I want to be able to do that at some point as well.. What say ye? peace
    Pete,

    The Geckos will keep you where you command as long as they are properly tuned, and you operate them within their capabilities. The "closed loop" boards are of value only when you push *too* hard, and exceed the capabilities of the drives and motors. The *only* times I've ever lost position on my machine was when I did something stupid, and ran the tool into an immovable object, like a vise or clamp. Now that I know what I'm doing, that's rare, and even when it does occur, it only takes about two minutes to re-reference the machine, and get back to work. These things will not help with "accuracy", unless you're willing to accept a HUGE loss in performance. If you believe you can "close the loop" and start operating with zero following loss, get ready to run VERY slow accelerations, and start running your jobs at a snails pace. Following error is a fact of life, and under any remotely realistic operating conditions, it is not a problem at all. PID control loops were developed to give the best performance over a wide range of operating conditions, and you cannot have a PID loop without following error *and* get reasonable performance. The I term will always ensure you end up where you should, which is all that matters. Under light loads, the following error will be very small. In fact, Geckos are often criticized for faulting with such a small following error. Larger, more powerful servos controllers often allow following errors in the thousands, or tens of thousands, of counts, yet they work just fine on large VMCs delivering the kind of performance and precision we can only dream of. Keep in mind too that as long as you're running CV mode, you're not necessarily *exactly* following the programmed path anyway, because Mach will modify the path as necessary to maintain CV. This probably has a far greater effect on "accuracy" than any following error.

    Regards,
    Ray L.

  15. #15
    Join Date
    Jul 2003
    Posts
    1753
    boy - here I go again. (emc person here.. can you tell?)

    With closed loop - You can set the following error so that if the path deviates from the commanded positon - you will get an error.

    Emc also has a tolerance mode. G64Px.xxxx were you tell emc how close you want it to follow the commanded path. it will adjust the feed accordingly to make the tool path stay within that tolerance - slowing down enough around corners. So you will know that the machine is doing exactly what you want it to do.

    http://wiki.linuxcnc.org/cgi-bin/emc...jectoryControl

    sam

    Quote Originally Posted by HimyKabibble View Post
    Pete,

    The Geckos will keep you where you command as long as they are properly tuned, and you operate them within their capabilities. The "closed loop" boards are of value only when you push *too* hard, and exceed the capabilities of the drives and motors. The *only* times I've ever lost position on my machine was when I did something stupid, and ran the tool into an immovable object, like a vise or clamp. Now that I know what I'm doing, that's rare, and even when it does occur, it only takes about two minutes to re-reference the machine, and get back to work. These things will not help with "accuracy", unless you're willing to accept a HUGE loss in performance. If you believe you can "close the loop" and start operating with zero following loss, get ready to run VERY slow accelerations, and start running your jobs at a snails pace. Following error is a fact of life, and under any remotely realistic operating conditions, it is not a problem at all. PID control loops were developed to give the best performance over a wide range of operating conditions, and you cannot have a PID loop without following error *and* get reasonable performance. The I term will always ensure you end up where you should, which is all that matters. Under light loads, the following error will be very small. In fact, Geckos are often criticized for faulting with such a small following error. Larger, more powerful servos controllers often allow following errors in the thousands, or tens of thousands, of counts, yet they work just fine on large VMCs delivering the kind of performance and precision we can only dream of. Keep in mind too that as long as you're running CV mode, you're not necessarily *exactly* following the programmed path anyway, because Mach will modify the path as necessary to maintain CV. This probably has a far greater effect on "accuracy" than any following error.

    Regards,
    Ray L.

  16. #16
    Join Date
    Feb 2006
    Posts
    7063
    Quote Originally Posted by samco View Post
    boy - here I go again. (emc person here.. can you tell?)

    With closed loop - You can set the following error so that if the path deviates from the commanded positon - you will get an error.

    Emc also has a tolerance mode. G64Px.xxxx were you tell emc how close you want it to follow the commanded path. it will adjust the feed accordingly to make the tool path stay within that tolerance - slowing down enough around corners. So you will know that the machine is doing exactly what you want it to do.

    http://wiki.linuxcnc.org/cgi-bin/emc...jectoryControl

    sam
    Which all has terrific "gee whiz" value, but essentially no practical value in the real world. Real VMCs don't operate that way, yet the can crank out work all day long to very tight tolerances. Is anyone here machining parts for heart valves, or spacecraft? Do we really *need* to care if our parts are off-dimension by a few tenths?

    Believe me, on a benchtop machine, with rolled screws, sloppy ways, low-grade bearings held in nice, flexy bearing mounts, flexible columns and everything else, even discussing these kinds of errors is truly like picking fly poop out of the pepper. People get all carried away worrying about minute errors, and completely ignore far greater ones in pursuit of a level of *perceived* accuracy that is simply unnecessary. For 99.99% of the work any of us will ever do, a tolerance of +/-0.005" is more than plenty, and easily achieved without a lot of unnecessary added electronics. Trying to do better may make for a nice hobby, which is fine, but it's just not necessary.

    Regards,
    Ray L.

  17. #17
    Join Date
    Apr 2007
    Posts
    2580

    Ray as I said for the most part I agree with you.....

    These machines, yours included are nowhere near the precision most here kid about and while a closed loop system sounds nice, for most of us here the parts we are making are more than adequate. I know for myself the vast majority of the parts I make are purely aestethic in nature and except for a dimension or two would be fine within five thou or so. I do occasionally make a high pressure part for my precision airguns and they gotta be pretty good to be able to hold back 3000psi but those are mostly lathe turned parts and I make them by hand or manually. The mill parts thus far have not needed to be perfect. However I do intend to try to make some pretty elaborate breech blocks similar to my old Air Arms Pro-target at some point and there are a lot of points in there that must be pretty tight.

    I also realize that I have much to learn and am really just getting started with this cnc adventure so I will leave the opinions to the pros on here. The pico system sounded pretty good on the net and in my magazine but I am already a fan of Mach3 and since it is so widely regarded by so many HSm's I intend to keep that course. Hell if it will do all the things hoss has done with his machine I imagine it will be just fine for me..... peace

  18. #18
    Join Date
    Feb 2007
    Posts
    592

    G64 Pxxxx In the real World

    Ray;

    Are you saying that FADAL and Mori Seiki are not real VMC's - because they have variations of that mode to do just that.

    A 20 year old FADAL with the original CNC88 had M codes to adjust the servo gain in the program to adjust following error in the program.

    Fanuc has a number of options to do the same thing, such as G08 P1 (Fanuc 16 and up)

    And the real world Gee Wiz Value is reduced cycle times while keeping the parts in print.

    I have to work with some machine that don't have this feature - and your right sometimes you have to crank things way down to get what you need in spec. - Its a pain but thats what your stuck with.

    EMC2 is slowly (and at times not so slowly) evolving to meet real world needs - G64 Pxxx was a great step forward.

    As to accuracy of benchtop machines - those owners that take time to to tune there machines can get great results. There are people running on acme screws holding tenths because they took the time to pitch error map the screws. Its not a perfect world, and these people have to remap acme screws every so many hours to check for possible wear.

    GIGO - Garbage In Garbage Out

    The reverse is also true, those who do there due diligence will get the best results from there equipment.

    This "Hobby"'s biggest problem is you have people jumping in with both feet that have never seen real machines running and have no idea what there getting into. I'm approaching 30yrs in trade and I am still learning all the time.

  19. #19
    The new G320X can be dip-switch set to +/-256, 512, 1,024 and 2,048 count following error limits.

    Mariss

  20. #20
    Join Date
    Apr 2007
    Posts
    2580

    Mariss.....

    NIce to meet ya man, I have read and heard an awful lot about you on the net. I am really glad you posted here about this closed loop idea. What is your take on it? Is a closed loop system worth talking about? Can it be done accurately in Mach3 and would it even be feasable to try? I have the Gecko G320 drivers going into my first conversion now and I am very happy with them thus far as it seems most are. Have you experimented with this type of system and would it result in an accurate yet slow machine as some have said? Your input would be much appreciated here.... peace....

Page 1 of 2 12

Similar Threads

  1. Open Loop to Closed Loop Stepper Conversion
    By beamhome in forum Stepper Motors / Drives
    Replies: 11
    Last Post: 03-09-2014, 11:27 PM
  2. Syil x4+ Closed Loop Servo retrofit ?
    By Chipboy in forum Syil Products
    Replies: 3
    Last Post: 09-12-2009, 09:01 PM
  3. Mach EMC Maximum Encoder Count Closed Loop Servo EMC/MACH3
    By bad bearings in forum Open Source Controller Boards
    Replies: 11
    Last Post: 05-28-2009, 05:44 PM
  4. question on closed loop vs open loop (servo systems)
    By boonie in forum Servo Motors / Drives
    Replies: 20
    Last Post: 11-09-2007, 07:30 PM
  5. Closed Loop Driver vs. Closed Loop Computer
    By ojibberish in forum Gecko Drives
    Replies: 3
    Last Post: 06-08-2004, 05:30 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •