585,991 active members*
4,604 visitors online*
Register for free
Login
IndustryArena Forum > Machine Controllers Software and Solutions > Centroid CNC Control Products > Suitability of Centroid Acorn controller for gear hobbing application
Page 3 of 3 123
Results 41 to 57 of 57
  1. #41
    Join Date
    Jun 2010
    Posts
    4256

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Hi Craig

    Don't get me wrong. I am sure ethercat works fine to control those motor drivers. And I am sure that those motor drivers can do what is claimed, both for motors and for sync and for accuracy. The big-name machine companies would not use them if they didn't.

    It is just that those drivers are quite expensive compared to an ESS and a handful of Gecko drivers. And I am quite sure that when Mach(n) tells the ESS to tell the several Geckos to go to point x,y,a, that the machine will go to that point in full sync, to within one or two steps - and those steps can be sub-micron (0.8 um on my CNC). Whatever smarts the ethercat drivers have can be and has been reproduced in the ESS/Gecko combination. Both of these have powerful FPGAs in them.

    The only limit I see at present for the ESS/Gecko combination (or UCC or others) is a power limit.
    The stepper Geckos (G203V) I use are rated to 80 VDC/7A (1/2 kW).
    The servo Geckos I use are rated to 80 VDC/20 A (1.6 kW).
    If you need more than that you definitely out of my humble class!

    Biased opinion follows. I rarely see any of my servos pulling more than 50 W. I rarely see the spindle pulling more than 300 W, and that is machining metal. The combination of a 3:1 gearing on the motor and a 5 mm ball screw gives huge forces. Granted, a commercial user may want to go much faster than I do. For that you need a LOT of cast iron.

    Cheers
    Roger

  2. #42
    Join Date
    Nov 2013
    Posts
    4372

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Hi,

    It is just that those drivers are quite expensive compared to an ESS and a handful of Gecko drivers
    Perhaps, but the price difference is steadily reducing.

    Mach is not realtime and cannot in itself do realitme comms, eg Ethercat. There are controllers (Hicon for example) that extend Mach4 to do realtime comms and therefore be Ethercat capable.
    These controllers are by in large more than an ESS and not sufficiently differentiated from regular Mach4 to be considered compelling.

    In more recent times a company called Kingstar have collaborated with Interval Zero. Interval Zero make software that means that a multicore Intel processor has one core (at least) as genuine
    realtime while interfacing to the remaining Windows cores. Kingstar provides the software that enacts Ethercat while Mach4 is the trajectory planner and GUI.

    Such a system is on sale, it comes with a repurposed (ex lease?) PC, with the appropriate high quality Ethernet card, Mach4Hobby license,the Kingstar Ethercat runtime license and the Interval Zero runtime
    license for $1600USD. The competing combination, ie re-purposed PC, Mach4Hobby license, an ESS and MB3 breakout board would cost about $1100USD. Thus the Ethercat solution is
    rather more but not over the top, and remember this gives you up to 100 Ethercat nodes, be they servos and data IO nodes, or some combination thereof.

    Ethercat servos, Ethercat closed loop steppers, Ethercat data nodes are available and all attract a premium, say a Ethercat servo might cost $50 more than a Step/Dir servo. So there is a price to
    pay if you were to go Ethercat but its not unreasonable. Siemens were, maybe still are the owners of Profibus, and like anything Siemens they want a bloody forutune for it. I won't even entertain
    Profibus, but Ethercat is vastly more affordable.

    Ethercat is, or was started by Beckhoff, another German company, but they made the licensing that much better that Ethercat is coming to dominate the market. There is NO license fee to make an Ethercat master,
    but you'll use an IC to make an Ethercat slave, say a servo or data node, and that IC has a license fee attached. The IC's are about $10.00 each so the license fee can't be much! This licensing arrangement has
    encouraged many manufacturers to participate. Yaskawa is a big and long term supporter of Ethercat for example.

    Craig

  3. #43
    Join Date
    Jun 2010
    Posts
    4256

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Hi Craig

    I continue to learn new things. Thank you.
    I do have some questions.

    Mach is not realtime and cannot in itself do realitme comms,
    Um. That sort of depends on what is 'real time' and why do you need it. I don't know whether one would count Mach(n) as 'real time', but I do know they have to output signals in real time if you don't want the CNC machine crashing. It's crucial. The difference may be that Mach is now mostly a trajectory planner, so to speak, and that the real-time stuff has been devolved to an ESS. The combination works very well on thousands of machines (?tens of k?)

    Ethercat may give you up to 100 nodes, but that is of zero value to me with one CNC. Are they even the same market? A lot of us here are hobbyists.

    Siemens - have you ever looked at their version of a CNC language? Weird is an under statement. And remember that Siemans is a single-source supplier, with zero competition and consequent high pricing. Not for me.

    Beckhoff seem to have read the tea leaves, seen the future, and understood what makes a successful product. Competition, choice, and public standards. Like the PC.

    So a question is what more does one need, if my CNC runs well? Are we comparing like with like, or is it more like LandCruiser vs Humvee? Different markets.

    Footnote: Mach(n) and ESS are not the only options in this highly competitive fast-moving Open CNC space. There is even, gasp, LinuxCNC! Or GRBL. Or UCC. Or . . .

    Cheers
    Roger

  4. #44
    Join Date
    Nov 2013
    Posts
    4372

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Hi,
    Mach, or rather the Windows PC on which it runs is not realtime, and Mach cannot run a CNC machine directly otherwise there would be pauses in the data stream which would\screw your machine.
    Mach produces trajectory data that is buffered on a realtime controller board, of which the ESS is an example.

    Ethercat is realitme ethernel communication, and a Windows PC cannot run Ethercat because Windows is not realtime capable. That is where the Interval Zero solution comes in, extremely clever by the way.
    By all means google Interval Zero and watch a couple of their videos, extremely clever and remarkably simple at the same time. The Interval Zero solution allows a Windows PC to do realtime ethernet communication
    and therefore be Ethercat capable.

    Back in the 80's DC servos were the go, and many said the expensive new fangled AC servos would never catch on. Twenty years later and AC servo totally dominate and DC servos are pretty much
    consigned to history. My guess is that distributed control, of which Ethercat is just one example, will become the norm and our Step/Dir servos and so on will in years to come be consigned to history.

    So a question is what more does one need, if my CNC runs well? Are we comparing like with like, or is it more like LandCruiser vs Humvee? Different markets.
    This is a much more germane question....'do I need Ethercat'. Your machine and mine and countless thousands of others run perfectly well without Ethercat, so in that sense the answer
    is no. What happens if however Mach includes a reverse kinetics solver for running a robot arm to load your machine?. Far from impossible, and would be seamless if you had Ethercat
    but not so easy otherwise. I rather think that over time there will be increasing applications for which distributed motion control is the right answer and therefore the uptake will grow.

    When I built my current machine I considered Ethercat. Previously I posted that the cost of the PC and licenses was $1600, whereas PC, licenses and an ESS was $1100. But I already had the ESS and Mach4,
    so it was not a case of paying up an extra $500 but having to find $1600 to buy anew. Then I also had to consider that I would have to spend extra on the servos. I buy the 750W Delta B2's at $438USD
    at a permanent discount, whereas the regular price for an Ethercat version of the same servo is $648. Due to the discount I would have had to pay up an extra $600 for the servos.
    All up, despite my interest in Ethercat I decided against the extra expense. I can use that same money for a fourth and eventually a fifth axis.

    I could retrofit an Ethercat control system.....but why?. I've got paying work to do right now with the machine as is....I'm not of the opinion that Ethercat would add anything to that calculation,
    although I'm quite prepared to do so if there is a business case to do so.

    I imagine it will be another five-ten years before I'll have to consider building a new machine...and then I'll ask my self the same question, 'do a need or want Ethercat?'

    Craig

  5. #45
    Join Date
    Aug 2005
    Posts
    1092

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Hi Craig,

    I very confident that Mach3 and its former incarnations (Mach1 and even Master5 )are real-time.

    If they weren’t, it couldn’t generate a step pulse train with 2us step pulses at a rate of 35k+ pulses/s

    Art performed a number of software gymnastics to achieve this. He basically inserted an interrupt service routine at a very low level so Mach3 got access to the processor before most of the windows processes. Once he was done he gave it back to windows. This way Mach had accurate and deterministic timing, such is basically the definition of a real time system.

    Code:
     What is a real-time application?
    A real-time application, or RTA, is an application that functions within a time frame that the user senses as immediate or current. The latency must be less than a defined value, usually measured in seconds. The use of real-time applications is part of real-time computing.
    The technique he used was basically one of the techniques hackers use to hide viruses on early PCs.
    As Windows OS moved from Win95 to 98, etc, measures were put in place to combat playing around with the os. Eventually the OS (win 7 64-bit) was encrypted which ended using this method. That’s why you need to use a motion controller with an OS later that Win7 32-bit.

    Anyone remember using Master5? Maximum pulse rate of 8k steps/s.

    Cheers
    Peter.


    Sent from my iPhone using Tapatalk Pro
    -------------------------------------------------
    Homann Designs - http://www.homanndesigns.com/store

  6. #46
    Join Date
    Jun 2010
    Posts
    4256

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Hi Craig

    Mach, or rather the Windows PC on which it runs is not realtime, and Mach cannot run a CNC machine directly otherwise there would be pauses in the data stream which would\screw your machine.
    Here I will contradict you. Windows OS is not Real-Time, as you say: there I agree. But Mach3 on a PC running Windows XP WAS Real-Time. How was this done? By extreme clever.

    Art Fenerty wrote Mach3 ( and predecessor) to answer his own needs. He found a way to kick Windows XP down one ring layer on an X86 processor, and to put Mach3 in at the highest priority (ring 0), where it did indeed run in genuine Real-Time. In particular, it took over the system clock, the system interrupts, and the LPT printer port. The end result was that a PC nominally running Windows XP could (and did) run a CNC machine through the LPT port in Real-Time. Fairly soon there were thousands of hobby machines running like this, using LPT and XP.

    However, all good things come to an end. MS released W7 with a trap which blocked what Art had done. There were screams of course, but MS took zero notice. So we could not upgrade from XP to W7 (let alone higher). Warp9 saw an opportunity here, and came up with a USB Smooth Stepper controller which took the Real-Time stuff away, so Mach3 could run under W7, with a special custom USB driver. Unfortunately the USB driver (3rd party) was not really reliable, and soon there were complaints of the CNC crashing.. The root cause was that using this USB interface meant the PC and the CNC shared a common earth - and not very well. You can imagine the rest.

    So Warp9 came up with an Ethernet version, now known as the ESS. The USB version was allowed to die. (I still have one on the shelf.) While the USB was 0/5 V with a common earth, the Ethernet is +/- 15 V and transformer isolated. All problems vanished.

    So now Mach3/4 does the high-level trajectory planning etc, and sends the moderately high-level commands to the stock Ethernet port. The ESS does the rest, in hard Real-Time. In a compliment to it all, you can now buy a cheap Chinese 'copy' of the ESS with a 'free' pirate copy of Mach3 from ebay. However, the Chinese copy cannot handle all the Mach3/NIST g-code commands, as many people have found to their cost.

    I am still running Mach3, rather than going to Mach4, because it works just fine for all my current needs.

    Cheers
    Roger
    PS, written before I saw Peter's posting. I agree./

  7. #47
    Join Date
    Nov 2013
    Posts
    4372

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Hi,
    Machs parallel port driver behaved in a manner like a realtime motion control, but it is not Windows software. Mach, the application, is Windows software whereas the parallel port driver
    was in effect a software hack at low level in the kernel. The quirk that Art used to do it eventually closed.

    The parallel port driver was a very clever piece of work....but it does not mean that Windows is a realtime OS. How would you feel about a Windows PC flying the aircraft you're on?
    No bloody way!

    All the currently supported Microsoft OS's cannot run Mach's parallel port, ergo such a Windows platform MUST have an external motion controller, or a realtime core that can do realtime communication
    for Ethercat such as the Interval Zero solution.

    Craig

  8. #48
    Join Date
    Jun 2010
    Posts
    4256

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Hi Craig
    it does not mean that Windows is a realtime OS
    Good god no! Not a chance!
    Which is why systems with mach on them are not normally connected to the Internet: Windows has a bad habit of going off looking for updates and blocking everything else.

    How would you feel about a Windows PC flying the aircraft you're on?
    I would be on the ground.
    Didn't the US navy have a ship controlled by Windows at one stage?

    Cheers
    Roger

  9. #49
    Join Date
    Aug 2004
    Posts
    780

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Fwiw..
    mach3 has been a realtime controller with closed loop IF and ONLY IF you have the right hw that closes the feedback loop at the ac servo level and or controller level.

    So it can fault within 1 ms, or do 1 ms probing, where the hw controller actually watches for the signal and the sw only acts later, when it gets to it.
    Linuxcnc with mesa cards does exactly the same.

    The cslabs csmip-ips, 1 have 2 of, for 5-6 years, do this with 12 khz update rates and 4 MHz hw rates - with ac servos at 500 khz, 10.000 counts/turn, 3000 rpm, 750W each axis, 2.5 kW servo spindle on lathe.
    The lathe has 90 Nm of peak torque (3 secs), 1:3 via HTD timing belt, 8 mm x 30 mm wide and 30 kg very heavy and very rigid mounts.

    The servo reacts in sub ms rates, and the controller and servos both fault before a hard crash breaks expensive mechanical stuff.
    Happened about 3-5 times.

    I tried about 7 different solutions for 10 years plus before getting this.
    The cslabs stuff is expensive.
    I believe the centroid stuff is also very good, but have not had engineering samples so don´t have direct experience.

    For your use, I would recommend centroid, or mesa+linux, or cslabs, and not much else.
    There are endless good technical reasons.
    There is another us based hw controller for machx that seems good, but no personal experience.

    It is much much harder than you probably think to make good gears or good threads (better than class 3).

  10. #50
    Join Date
    Nov 2013
    Posts
    4372

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Hi,
    CSLabs is good hardware and works very well with Mach3. Their Mach4 plugin is buggy and CSLabs, whom used to have industry leading technical support, have
    very much dropped the ball with Mach4. I believe they have their own CNC software now, and they concentrate on that, at the expense of the Mach4 plugin. The
    last update was two years in coming. Unacceptable for high cost hardware.

    Craig

  11. #51
    Join Date
    Nov 2013
    Posts
    4372

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Hi,
    It is much much harder than you probably think to make good gears or good threads (better than class 3).
    I think you are right.

    That is why I think I'd rather rely on two modern servos enacting commanded moves with a high degree of authority RATHER than have a central controller like LinuxCNC handle
    the coordination. OP seems inclined to disagree but I think modern AC servos have a much higher bandwidth (tighter more accurate control) than LinuxCNC say.

    Craig

  12. #52
    Join Date
    Jul 2003
    Posts
    1754

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Wow. Some really good conversion here...

    Couple things..
    Linuxcnc has already been used in hobbing applications.
    Linuxcnc works with ethercat.
    You don't have to 'compile' linuxcnc unless you are really getting into the weeds. (But because the motion control is in the computer - the sky is the limit). You don't have to make sure the feature you want is in the external motion controller.

    Again.. Great conversation.

    Oh - and closing the servo loop within linuxcnc - you have access to realtime following error - so you know how 'good' your system is behaving.

    This is what happens with linuxcnc when you get in the weeds and have some covid down time...
    (More spindle synced motion)
    https://youtube.com/shorts/23bEsKMNJH0?feature=share

  13. #53
    Join Date
    Jul 2003
    Posts
    1754

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Quote Originally Posted by joeavaerage View Post
    Hi,


    I think you are right.

    That is why I think I'd rather rely on two modern servos enacting commanded moves with a high degree of authority RATHER than have a central controller like LinuxCNC handle
    the coordination. OP seems inclined to disagree but I think modern AC servos have a much higher bandwidth (tighter more accurate control) than LinuxCNC say.

    Craig
    Over on home shop machinist forum Sir John (RIP) was trying to do hobbing by syncing 2 servo drives. He finally gave up and ended up using linuxcnc - which he hated.. But it worked and it was inexpensive.

  14. #54
    Join Date
    Nov 2013
    Posts
    4372

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Hi,
    kool video, and a very good idea. It is a perfect example of coordination between axes, its limited by your imagination only.

    Craig

  15. #55
    Join Date
    Nov 2013
    Posts
    4372

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Hi,
    I have not made gears by this method but I do make threads by using axis coordination, and that's with Mach4. Naturally you rely on the servos having the torque authority to
    follow the commanded toolpath....but then don't all servos have to have that authority?

    Craig

  16. #56
    Join Date
    Jul 2003
    Posts
    1754

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Quote Originally Posted by joeavaerage View Post
    Hi,
    kool video, and a very good idea. It is a perfect example of coordination between axes, its limited by your imagination only.

    Craig
    That cutter (spindle) isn't an axis. The x-y axis are slaved to the spindle encoder. (Kinda like conventional threading)

  17. #57

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Well Acorn works real time and very fast, motion control is on the board with a linux system, and Acorn6 is even faster with two fast FPGA chips, best Mesa cards for linux CNC have only one slower one...

    I make often complex parts with 4 axis mill or 4 axis lathe, never had a problem, even with very short vectors from a CAM system.

    Gear hobbing will be no problem for Centroid CNC if the CNC is set up right, you have lots of options how to synchronize the axis in use.

    Uwe

  18. #58
    Join Date
    Jun 2023
    Posts
    136

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    Thank you for sharing, I learned a lot, but in China, I can't view the videos you posted(youtube) .

    I just produce mineral castings.Others need to continue to learn from you.

  19. #59
    Join Date
    Aug 2006
    Posts
    6

    Re: Suitability of Centroid Acorn controller for gear hobbing application

    What about this:-

    Dividing Head Controller BOX for Stepper Motor 4th Axis CNC Machine Wood Router Part Tools 220V

    https://www.aliexpress.com/item/1005001839964335.html

Page 3 of 3 123

Posting Permissions

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