The "fix your Maho" thread - Page 3
Close
Login to Your Account
Page 3 of 20 FirstFirst 1234513 ... LastLast
Results 41 to 60 of 398
  1. #41
    Join Date
    Nov 2012
    Country
    UNITED STATES
    State/Province
    Wisconsin
    Posts
    2,142
    Post Thanks / Like
    Likes (Given)
    3390
    Likes (Received)
    2607

    Default

    Peace brother......my Maho experience was so bad that when I see the name ......... I remember wwhen we would get a bad storm the things would crap out the parameters (all binary)...it took 2 guys about 3 hours to read them in manually...one guy saying1 1 1 10 0 1 1, the guy typing saying1 1 1 0 0 1 1 1 ??? no 1 1 1 1 0 0 1 1. OMG so w eventually downloaded them to a dnc system at a baud rate of about 3 lol.....then when the day came that we needed to read them back in (thunderstorm) it turned out that they actually read in WRONG for some reason and this machine acted like it thought it was a gd dishwasher or something.....awwww man.....you enjoy yours.....I am gonna have a beer just remembering that thing lol

  2. #42
    Join Date
    Jul 2011
    Location
    toronto
    Posts
    281
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    11

    Default

    Quote Originally Posted by toolsteel View Post
    Peace brother......my Maho experience was so bad that when I see the name ......... I remember wwhen we would get a bad storm the things would crap out the parameters (all binary)...it took 2 guys about 3 hours to read them in manually...one guy saying1 1 1 10 0 1 1, the guy typing saying1 1 1 0 0 1 1 1 ??? no 1 1 1 1 0 0 1 1. OMG so w eventually downloaded them to a dnc system at a baud rate of about 3 lol.....then when the day came that we needed to read them back in (thunderstorm) it turned out that they actually read in WRONG for some reason and this machine acted like it thought it was a gd dishwasher or something.....awwww man.....you enjoy yours.....I am gonna have a beer just remembering that thing lol
    the parameters arent binary. numbers yes, but but not binary. my battery is on the way out, so i tend to lose the params if the machine is unplugged for more then a day or 2. my machine has more params than most, and it takes maybe 15-20 mins to reload them by hand. i need to get me a new battery, ha!

  3. #43
    Join Date
    Jun 2006
    Location
    Munich / Germany
    Posts
    2,338
    Post Thanks / Like
    Likes (Given)
    342
    Likes (Received)
    1068

    Default

    in EMC on my other machines, my acceleration was set to 0.2G or 2m/s2. the machine is capable of this easily. in mach3 the same program runs fairly constant at 450ipm (machines top speed). under emc it would have a highly variable running speed, capped at 250ipm, but as low as 40 or 50 ipm when the precision of following was set to .001". if you increased this to .01, it ran faster, but created a contour not even vaguely respembling what was programmed.
    The truth behind constant velocity (CV) and G64:

    Acceleration is change of speed over time:

    a = v / t #in m / s^2

    Centripetal acceleration is:

    ac = v^2/r
    r => v^2/ac
    v => sqrt(ac)*sqrt(r)

    how small a radius can your path be, if in cv-mode?
    Speed is 5000 mm / minute.
    Acceleration is 2 m/s^2.
    I convert m to mm and minutes to seconds on the fly:

    r(5 / 60, 2) => 0.0035

    Radius in mm:

    0.0035 * 1000 => 3.5

    So the smallest possible radius is 3.5 mm. G64 would have to be 3.5 mm. If you set it to 0.25 mm, what speed could you get in corners?
    That's in m per second:

    v(2, 0.25 / 1000) => 0.0224

    Or in m/minute:

    0.0224 * 60 => 1.344


    So it is quite obvious that you tricked yourself with the cv-mode of Mach. It physically CAN'T do it that fast. No matter how clever the software is.

    Nick

  4. #44
    Join Date
    Dec 2000
    Location
    Coastal Dogpatch, SC, USA
    Posts
    51,193
    Post Thanks / Like
    Likes (Given)
    2714
    Likes (Received)
    5517

    Default

    Quote Originally Posted by toolsteel View Post
    Peace brother......my Maho experience was so bad that when I see the name ......... I remember wwhen we would get a bad storm the things would crap out the parameters (all binary)...it took 2 guys about 3 hours to read them in manually...one guy saying1 1 1 10 0 1 1, the guy typing saying1 1 1 0 0 1 1 1 ??? no 1 1 1 1 0 0 1 1. OMG so w eventually downloaded them to a dnc system at a baud rate of about 3 lol.....then when the day came that we needed to read them back in (thunderstorm) it turned out that they actually read in WRONG for some reason and this machine acted like it thought it was a gd dishwasher or something.....awwww man.....you enjoy yours.....I am gonna have a beer just remembering that thing lol
    Uh...ok...still shy about saying what they actually were or just too slack to know ? An early 80's MH series can be a whole different animal from a 1986 and newer MH for example. Re parameter (constants) loss, since this forum started over a decade ago you are the first person I recall to mention loosing constants due to power surges or brownouts. But I could see it if your backup battery was old and needed replacing.

    Regardless, what is the point of blurting out "I hate these things" (complete with no clarification of exactly which "things" you mean) in an otherwise intelligent discussion between folks that "like these things" ?

  5. #45
    Join Date
    Nov 2012
    Country
    UNITED STATES
    State/Province
    Wisconsin
    Posts
    2,142
    Post Thanks / Like
    Likes (Given)
    3390
    Likes (Received)
    2607

    Default

    Quote Originally Posted by Milacron View Post
    Uh...ok...still shy about saying what they actually were or just too slack to know ? An early 80's MH series can be a whole different animal from a 1986 and newer MH for example. Re parameter (constants) loss, since this forum started over a decade ago you are the first person I recall to mention loosing constants due to power surges or brownouts. But I could see it if your backup battery was old and needed replacing.

    Regardless, what is the point of blurting out "I hate these things" (complete with no clarification of exactly which "things" you mean) in an otherwise intelligent discussion between folks that "like these things" ?
    Quite obviously when Ihavenofish said "it sounds like it may have been a MH700s" and I replied"maybe so"....we resolved the issue.......I am not shy to say....quite honestly it is a piece of my machining history that is something less than pleasant due to not only the unacceptably poor reliability of the machine....but also the poor customer service response and the piss poor service we received. The point of blurting out I hat these things would be to let someone who does not have ome unexplainable hard on about "these things" that they are best left to those who do......they are similar to owning a classic British motorcycle....they look cool but if you own one you better know how, and be prepared to work on one.....because you will be......on a more personal note MILICRAP......it seems you have a little bit of fenderhead in your blood.....good luck with that.

  6. #46
    Join Date
    Dec 2000
    Location
    Coastal Dogpatch, SC, USA
    Posts
    51,193
    Post Thanks / Like
    Likes (Given)
    2714
    Likes (Received)
    5517

    Default

    Quote Originally Posted by toolsteel View Post
    Quite obviously when Ihavenofish said "it sounds like it may have been a MH700s" and I replied"maybe so".
    That still doesn't tell me anything without knowing the year. A 1983 MH700s is a whole 'nuther animal from a 1989 MH700s. That you don't say tells me you don't know that, which tells me you are somewhat ignorant of Maho machines in general, which tells me your opinion of them is highly suspect anyway.

    .on a more personal note MILICRAP......it seems you have a little bit of fenderhead in your blood.....good luck with that.
    I have no idea what that means, but FWIW, since being born again with a Makino KE-55 manual/CNC with Fanuc 20, I'm not that fond of Maho's anymore either or CNC Deckels for that matter (still love the manual Deckels) But I don't see much point in butting in with zero helpful info in a discussion with those who do like the machines, Tool "crap"

  7. #47
    Join Date
    Jul 2011
    Location
    toronto
    Posts
    281
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    11

    Default

    Quote Originally Posted by Nick Mueller View Post
    The truth behind constant velocity (CV) and G64:

    Acceleration is change of speed over time:

    a = v / t #in m / s^2

    Centripetal acceleration is:

    ac = v^2/r
    r => v^2/ac
    v => sqrt(ac)*sqrt(r)

    how small a radius can your path be, if in cv-mode?
    Speed is 5000 mm / minute.
    Acceleration is 2 m/s^2.
    I convert m to mm and minutes to seconds on the fly:

    r(5 / 60, 2) => 0.0035

    Radius in mm:

    0.0035 * 1000 => 3.5

    So the smallest possible radius is 3.5 mm. G64 would have to be 3.5 mm. If you set it to 0.25 mm, what speed could you get in corners?
    That's in m per second:

    v(2, 0.25 / 1000) => 0.0224

    Or in m/minute:

    0.0224 * 60 => 1.344


    So it is quite obvious that you tricked yourself with the cv-mode of Mach. It physically CAN'T do it that fast. No matter how clever the software is.

    Nick
    so your answer to how bad emc's contouring is, is that mach3 must not be as good as i think it is... riiiiiiiiiiiight.

    the fact remains that emc2 to me is an unacceptable choice for a maho retrofit due to its abysmal contouring ability.

  8. #48
    Join Date
    Jun 2006
    Location
    Munich / Germany
    Posts
    2,338
    Post Thanks / Like
    Likes (Given)
    342
    Likes (Received)
    1068

    Default

    the fact remains that emc2 to me is an unacceptable choice for a maho retrofit due to its abysmal contouring ability.
    If you are not able to draw your own conclusions from the physics facts I have given you, there is no need to further discuss how bad a 3 year old version of LinuxCNC was.


    Nick

  9. Likes jCandlish liked this post
  10. #49
    Join Date
    Jul 2011
    Location
    toronto
    Posts
    281
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    11

    Default

    Quote Originally Posted by Nick Mueller View Post
    If you are not able to draw your own conclusions from the physics facts I have given you, there is no need to further discuss how bad a 3 year old version of LinuxCNC was.


    Nick
    emc does not follow the path in cv mode, mach does except in the corners. mach3's method is acceptable, emc's is not.
    would you like to post information on making EMC perform more adequately? it would be useful to the OP, who seems to think EMC is a way to go on his machine.

  11. #50
    Join Date
    Jul 2011
    Location
    toronto
    Posts
    281
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    11

    Default

    anyhow, back to the original topic

    Quote Originally Posted by akb1212 View Post
    1. Constants. This could possibly even be presented in a thread of its own. A list of different machines constants and such could be given in that thread. That way we would have a library of constants to look at. But a quick intro on the basics of them should be in this thread as well.
    i have a "new" manual from dmg, with lots of constants with some explanations, however i imagine the values for said constants only apply to my machine. they are also older than yours, the lists are for v300, and v500 (what i have). ill have to see if theres an easy way to scan them up.


    4. E-stop chain. Hints on how to get a machine out of e-stop. Including how to bypass the ****ing door safety system.
    door interlocks should be on the control panel. there IS an estop override switch in the back though. one person needs to hold it on, while another does whatever needs to be done at the control.


    6. How to get the table up if it is at the bottom. There is an easy way to do this by changing constants only. Doing it mechanically is a nightmare on the y-axis because of access. So it's much more efficient to do it by the constants. And it's not straight forward because of the software limits. A detailed description on how to do this will be presented.
    cosntant 80/81 (if i remember) overrides the soft limits. you need to turn the switch at the back to override the hard limits if youve hit them. careful to go the right direction, its not pleasant if you hit the hard stops. my machine has a safety clutch, which slips if its overloaded in a crash. i dont know if other mahos do.

    8. Electrical schematics. I have made an overview of what component does what in an excel sheet. This has proven to be very helpful during fault searching. This is very machine specific. But there is a system to it, and it might be useful for other machines as well. And it could easily be updated for your own machine if you have the electrical schematics.
    i have the full scematic for my machine, but its hard to know how much is applicable to others. id assume some of the basics like transformer wiring and the first few stages of power are pretty universal. BTW, you can BUY new manuals from DMG, the previous owner got one for my machine, i think it was $500. it included everything from wiring, to parts, to maintenence, inspection and operation. good resource.

    i also have some other basic stuff from a maho 700 and 1000. operator maintenence guides i think. i meant to copy them and upload long ago, but i completely forgot. they are sitting in a box.

    also have the original purchase order, and maintenance log, so i know what in the past had gone wrong, which helps in guessing what might go wrong in the future. most of it revlolves around the scales getting dirty.


  12. #51
    Join Date
    Jun 2006
    Location
    Munich / Germany
    Posts
    2,338
    Post Thanks / Like
    Likes (Given)
    342
    Likes (Received)
    1068

    Default

    Here's a fix or enhancement of my MH 700 C:

    A adjustable spindle brake. With a "slightly" worn out bevel gear in the vertical head, making shallow cuts at low RPM with a disk cutter makes a hell of a clappering. Neither good for your ears, the tool and the finish. And the bevel gear won't get better too.

    Pictures:
    Before
    sb4.jpg

    In between:
    sb2.jpgsb3.jpg
    As there was no cylindrical surface to clamp the brake drum to, I had to use the pulley-grove that is used for the quill-feed mechanism (that I don't have).
    Only two of the 6 claws used shown here.

    And after
    sb5.jpgsb6.jpg


    Has more than 10 hours of braking on its belt and still doing perfect. Doesn't get too hot, at least there's no discoloration from the heat.

    Nick

  13. #52
    Join Date
    Jul 2011
    Location
    toronto
    Posts
    281
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    11

    Default

    interesting. although im not 100% sure what it does, haha. is it just reducing "backlash" in the spindle for less vibration?

  14. #53
    Join Date
    Jun 2009
    Location
    Norway, Vestfold
    Posts
    181
    Post Thanks / Like
    Likes (Given)
    5
    Likes (Received)
    5

    Default

    Not much happening around here for a long time….

    I have been tied up in work myself so I haven’t had much time to discuss here.

    Anyway I have taken some drastic steps now lately, and it's all thanks to this thread!

    For many years now I have wanted to use EMC2, or LinuxCNC as it is called now. But thanks to this thread I was made aware it doesn’t do contouring as other systems can. And to me that is a problem. As already mentioned I wanted to replace the controller with one that would have at least the same capabilities, but have a modern interface and be able to expand the capabilities of my mill by adding features.

    So the short story is that I made a major decision not to put loads of effort in to learning all I would have to in order to get a good system working with EMC. In particular when it is lacking major capabilities as outlined by Ihavenofish. The fact that you had done direct comparisons on equal hardware was the eye opener for me.

    Another important factor for me is the work I know I would have had to spend to realize my plans of turning my mill in to a 4 and then 5 axis system a ’la Simpson36 over at the Mach3 support forum. Only I’d like to add another axis eventually as well.
    And with EMC I can’t imagine how much work that would be (if possible at all), but I found out I didn’t want to find out either. Reverse kinematics isn’t my strong side……

    And I happened to know that the former owner of my mill was selling all his hobby CNC project stuff. His company have recently gone bankrupt. So I had the opportunity to get good stuff for a good price.

    So I ended up buying his CamSoft Professional license (with 2 months left of support) and Galil 4183 card. Because of the situation it was a bit “now or never” kind of deal, and the price was good. So I thought I’ll just go for it. I hadn’t heard of it before (mostly because I haven’t been on the market for it), but I found out it was supposed to be able to do “everything” I wanted to and then some. That was what I was able to find in the short time I had to make up my mind.

    I later found out several not so nice things about it, which is one of the bad things about buying something without having time to research properly before you buy it. But I think I should be able to get something out of it anyway.

    The first thing I noticed is the overall look of the software….. it reminded me of my friends Amiga 500 he had in the early 90’s! Or win 3.1. I asked them about it in an email, and all they even bothered answering was the comment I had about their “upgrade” folder on the installation CD. It has files from 1996 and onwards in there, most of them from 1999 and older. These are all shared libraries the software is using. And seeing this tells me why it looks like it comes from before 1999. It is using most of it stuff from what they wrote back then! At least when it comes to the user interface. Think of when 16 colours were enough…. Well, for them it still is! When you configure dials and bars that is the amount of colours you have at your disposal.
    Their excuse is that you can load your own background picture and animate things on top of that to get exactly the “look and feel” you want with that. But the tools you have to do that with are from last milenia, so I’m wondering how they expect users to be content with that……
    Oh, and they use floppy disks to send out update files! I had to call my brother in law that is an IT-admin at a large company to find a way to read them. For I needed to read the updated licence files that came with it, and they were on floppy disks….. can you believe that!

    I didn’t even imagine I would have issues like this when I first looked at the software. Now I understand why they don’t show pictures of screenshots on their webpage (or small ones only anyway).

    I was actually hoping to jump a little more than a few years forward in time when I decided to upgrade my control system. But going from 1990 to 1997 or so is also a progress, although smaller than I was hoping for.

    But apparently this software is supposed to be the best alternative there is for retrofitting. Except buying a full blown modern hardware controller that is. And that has never even been considered.

    But I did come to the conclusion that I should be able to sell most of my 432 controller and be able to cover most of the expenses I had when buying this software. So I’m not too worried. I know how hard it is to get a 423 main board with the software version I have, and I think I’ll be able to get my money back because of this. After all it’s one of the last versions of 432/10 with full drip feed and everything.
    I’m considering doing this with the drives and servos as well, but I’m not doing that for the moment. I want to keep this retrofit as simple as possible. And then possibly do that at a later time when I know everything else is working.

    And the retrofit is still on, let there be no doubt. Now more than ever I’m committed to finishing it. After all I have spent even more money on it now, and need to get it done in order to get my money back from selling my old controller. But it will be more concentrating on getting the functionality I really want in there. So all my effort so far is still just as valid for the retrofit now, only now I have better tools available for the job. And I do intend to extend the service agreement (even if it’s $500 a year), at least until I’m done with the retrofit. I think having professional support available will definitely be a benefit. At least until I’m done with the retrofit and have the machine working like I would like it to.

    One problem I have though is the lack of digital I/O on the Galil card compared to the Mesa setup I have (basic 5i25 and 7i77). First off all there are not nearly enough I/O as it is. It has loads of inputs for limit switches and all that I won’t be using. But apparently there are no way I can use those for other functions like I could on the Mesa hardware. And to top that off they aren’t even 24V relay compatible like the Mesa cards….. so I need to add drivers circuits for them to be able to control the relays in my mill…..
    On all other aspects Galil seems to know what they are talking about. But in regards to putting enough (and the correct type) I/O on there they are way off. 8 axes of motion and only 16 inn and 16 outputs….. For a small 3 axis card yeah, but I would think anyone who want something like 5-8 axis also would like way more I/O than that. Even the simplest use of 5 or more axis needs more I/O than that….. doesn’t it? Their answer is that I needed to buy one of their PLC and use that for the additional I/O. Another unit on the Ethernet to slow down things…… great….. what update rate can you get from that reliably then? For me probably high enough, but I find this solution to be very inelegant and unnecessary expensive. The cost of this PLC alone is more than all my Mesa hardware….. And most surely not as fast, not to mention I’d have to order extra options to make it able to drive the relays on my mill….

    Not sure what to do with the Mesa hardware I have now, but I might be able to use it for something else. And it wasn’t that expensive either, so I’m not too worried about that.
    On the other hand, would it be possible to use that for I/O on the Maho with CamSoft? I mentioned it to them on the phone, but he basically laughed at it when I showed him their webpage. Han his answer was: “this must be only unusable hobby stuff, and we only use professional hardware here….. “
    But he hadn’t even heard of them before this.

    So there are still some problems to solve. Only now I have a whole new set of problems to solve. But hopefully this software will be able to get me to where I want with less work than EMC. And hopefully also be able to do everything I want to do, not just some of it like EMC seems to be capable of right now. And I also expect it to be accurate, unlike I hear EMC isn’t capable of being yet (without full stops that is) (and with no knowing when it will be fixed either).

    So this thread took a different direction than I anticipated. But I hope it was a good direction. A good direction for me anyhow. And It wouldn’t have happened like this if not for this thread.
    Not sure how many of you are interested in this retrofit now as it now involves costly software that most of you probably doesn’t consider using.

    But I would like to go back to what this thread started about. Which was giving tips on how to get an old Maho working. And I hope to have some more time to do that in the coming days.

    This post was to inform of the progress and direction taken in the matter. If there is an interest I could keep giving updates on the progress in the future, just not sure if it is of interest or not given the new approach.
    As I have most of the hardware already (a computer and the Galil card). I only need some connectors (which are on their way by now I hope, ordered them on eBay while I’m out). I hope to be able to start the work when I come home from here in a little less than 3 weeks. Then I have about 4 weeks off, and hope to be able to spend much of that time doing this retrofit. I also have some mechanical stuff to finish in that time as well, and attend to my wife and stuff like that. But having time of after being away for so long is always a good way to get some progress a project like this. So it’s definitely worth being down here in Ghana for 4 weeks. It means I get the same amount of time off when I come home. And I earn some extra cash as well, which is also helping a lot. I hope to get the opportunity to do more trips like this soon. But I don’t know when that will happen since there isn’t more work planned until next year for this project (oil field development in Ghana). But I do look forward to that as it gives me time of for long periods of time. And that makes it so much easier to get things done.

    Ok, enough rambling. Anyone know of someone with a Maho retrofitting it with CamSoft?

    Anders

  15. #54
    Join Date
    May 2012
    Location
    A Barn In Oklahoma USA
    Posts
    7
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    0

    Default

    I got my MH700C going this week. I've had it for a year and a half and it sat 2 years more, was running when disconnected. I managed to put a battery in and load constants from 3 lists not knowing which was last. From the constants description list it looked close enough to run but I couldn't keep the hydraulics latched. Holding the hyd button down I got X and Z to move but was getting errors that made it appear Y servo drive was bad, maybe bad connection from control to machine or board damage somewhere. I saved up to hire the smartest guy in town to help. He spent a few hours getting familar with manuals and machine before calling Karl the Mahodoctor. He was giving me the look of Are you sure you really want to go through with this?? I was too busy to watch the whole process but at the end of 3 days of going through bad Input Outputs, finding a constant change fixed it, oh wait it changed other constants, recorrect those but they don't stick so move to another problem then find when we get this Output working several other things start working. It would have saved time if we had known what order order to go through or where to start but every situation will be different. I'm happy because we have every function I could think of working. I forgot to see if I have axis clamping or if it works. The last things to fix were tool clamping and spindle gearbox. Karl wasn't near a phone so we tried all spindle constant options the last one being 0 that seemed least likely to work. My machine has less than 3000 hours, looks like new other than the 12mm socket wrench holes to manually change gears are wallowed out as if it has never worked. It now has the slickest working gearbox i've seen.

    To own one of these machines you need to learn the electrical diagrams, read Input Outputs, how to follow volts and google translate German. My service guy was blown away with the help Karl gave us. Not a single part replaced, it was all in the constants.

    My machine has version 10 upgrade and i'm looking for manuals.

    Luke

  16. #55
    Join Date
    Oct 2004
    Country
    UNITED STATES
    State/Province
    Iowa
    Posts
    3,063
    Post Thanks / Like
    Likes (Given)
    54
    Likes (Received)
    338

    Default

    Anders- keep up the progress here. I found, amidst my single-phase conversion threads, that there will be those who have NOTHING worthwhile to contribute to the thread.

    My MH600E came to be with a very limited history, and has been sitting for quite some time, so I have no idea what it's physical operating condition is. I have managed to power up the 432 and it's display, and other than the CRT having an intermittant offset to one side, it LOOKS like it's alive and capable of being sane.

    I have not fired up anything else yet, and in order to do so, I'll be powering up it's primary segments one-at-a-time from single-phase power sources. When done, it'll ALL run on 240v single-phase.

    My intention for this unit, is to have the capacity to take a drawn object, and have the machine cut it, basically to make patterns, special tools, and objects that I couldn't rapidly do with my manual machines. I'm not anticipating large-quantity production, but rather, prototyping and odd shapes.

    My philosophy of MY MH600E project, is that I have a solid foundation, fitted with ballscrews. The Indramat servos may be good, might be junk. The Indramat servo drive system doesn't show any sign of major smoke, so it may still be useable. The glass scales may still be fine too... but if not, it may become a subject of retrofit. From an NC control standpoint, I'd like to plug this thing into my LAN, draw up a part concept on my computer, send it to the machine, load in a tool and a block of material, turn it on, and stand back. Speed, particularly the contour issues noted before, probably won't be a very high concern for me, as the machine will obviously be able to do what it does, much faster than if I were doing it by hand.

    From a use standpoint, my MH600E will be used more as a development tool, than as a production platform. I do varying aspects of engineering for my company, which means I come up with different solutions, make some trial parts for several concepts, put them all through testing, and when we come upon solutions that make most sense, I pass the parts over to production houses to do large-quantity manufacturing... then move on to the next challenge.

    If I can make the MH600E function in that realm well, I'll have met my goals. Replacing the 432 NC system with some other platform is certainly within my options, so I'm watching your progress closely.

  17. #56
    Join Date
    Dec 2013
    Location
    Cotati, CA
    Posts
    30
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    2

    Default

    Hello All, I am very interested in this thread. I have an MH400E I am going to convert to linux CNC using mesa boards so this is just about the perfect thread on earth.

    My background is more in computer graphics programming professionally, so perhaps I can help with some of the linuxcnc contouring code issues when I get that far down the line. We shall see.

    Firstly though I want to say that the mesa cards are great and having the wonderful luck to live within driving distance of their shop I have had the great pleasure of meeting Peter. He has helped me immensely as we reverse engineered the fanuc serial encoder together for linuxcnc. He is a wonderful guy, really knowledgable, and very willing to help if you ask. They offer just about any board you could want, including fully programmable FPGAs in a variety of flavors.

    Anyway, so I have some of his hardware for a different fanuc project I am working on, but my 432 controls went sour a few weeks ago and so am highly motivated to make the swap. I would love to collaborate with those who are willing to give positive encouragement to those of us willing to put energy into our projects. Naturally healthy criticism is always appreciated. Anders' work here is greatly appreciated in doing all the wire tracing!

    I have a similar interest as DaveKemp here. This machine has good bones, and with a modern control could do a huge amount of precision work. No reason to give Haas all our money when we can use all this great old iron.

  18. #57
    Join Date
    Aug 2007
    Location
    Holmen, Wi
    Posts
    194
    Post Thanks / Like
    Likes (Given)
    1
    Likes (Received)
    14

    Default

    I too have used mesa hardware with linuxcnc. Very happy with both.

    As far as the trajectory planner in linuxcnc.. Currently it blends segments together for smooth transistions from one to the next. The current limitation is that it must be able to stop at the end of the next segment. This means, depending on the acceleration of the machine, the the velocity of the tool path may be slower than it could be. Most people doing normal machining (long cuts - low speed) would not notice it. Where you start to see the issue is a program that has lots of short segments. (because linuxcnc has to be able to stop by the end of the next segment). your mileage may vary. - it is really dependent on machine acceleration, segment length and tollerance (you can do a G64Px.xxx where x.xxx is how far you allow the path to deviate).

    That being said - there is work right now to increase the read ahead depth. The little testing I have been doing on it shows quite an improvement. A 533228 line engraving program with the old TP took about 2.5 hours the new TP took about 1.5 hours. (and this is pretty early alpha testing)

    sam


    Quote Originally Posted by asah View Post
    Hello All, I am very interested in this thread. I have an MH400E I am going to convert to linux CNC using mesa boards so this is just about the perfect thread on earth.

    My background is more in computer graphics programming professionally, so perhaps I can help with some of the linuxcnc contouring code issues when I get that far down the line. We shall see.

    Firstly though I want to say that the mesa cards are great and having the wonderful luck to live within driving distance of their shop I have had the great pleasure of meeting Peter. He has helped me immensely as we reverse engineered the fanuc serial encoder together for linuxcnc. He is a wonderful guy, really knowledgable, and very willing to help if you ask. They offer just about any board you could want, including fully programmable FPGAs in a variety of flavors.

    Anyway, so I have some of his hardware for a different fanuc project I am working on, but my 432 controls went sour a few weeks ago and so am highly motivated to make the swap. I would love to collaborate with those who are willing to give positive encouragement to those of us willing to put energy into our projects. Naturally healthy criticism is always appreciated. Anders' work here is greatly appreciated in doing all the wire tracing!

    I have a similar interest as DaveKemp here. This machine has good bones, and with a modern control could do a huge amount of precision work. No reason to give Haas all our money when we can use all this great old iron.

  19. #58
    Join Date
    Jun 2009
    Location
    Norway, Vestfold
    Posts
    181
    Post Thanks / Like
    Likes (Given)
    5
    Likes (Received)
    5

    Default

    Quote Originally Posted by skunkworks View Post
    I too have used mesa hardware with linuxcnc. Very happy with both.

    As far as the trajectory planner in linuxcnc.. Currently it blends segments together for smooth transistions from one to the next. The current limitation is that it must be able to stop at the end of the next segment. This means, depending on the acceleration of the machine, the the velocity of the tool path may be slower than it could be. Most people doing normal machining (long cuts - low speed) would not notice it. Where you start to see the issue is a program that has lots of short segments. (because linuxcnc has to be able to stop by the end of the next segment). your mileage may vary. - it is really dependent on machine acceleration, segment length and tollerance (you can do a G64Px.xxx where x.xxx is how far you allow the path to deviate).

    That being said - there is work right now to increase the read ahead depth. The little testing I have been doing on it shows quite an improvement. A 533228 line engraving program with the old TP took about 2.5 hours the new TP took about 1.5 hours. (and this is pretty early alpha testing)

    sam
    I already have the Mesa hardware, and I'd have to say I'm kind of sad to replace it with the Galil card that came with the Camsoft package I bought. At least from a physical implementation point of view.

    Just now I'm doing the physical assembly of the control unit. I have a nice industrial 19" rack computer case that will have everything in it (including the control computer).
    And as far as connecting things up I'd have to say Mesa is light-years ahead of Galil, which is kind of funny given the Galil is in the order of 10 fold the price! But then again I expect most of that price difference to come from the built in motion controller, so it's not a direct comparison as such.

    The Galil only comes with those insanely small and flimsy D-sub connectors with 3 rows of connectors, just like the 15 way connector used on VGA. Only here they have one 26 way connector for each axis and 44 way ones for common signals.
    First of all this means its a royal pain to get wires soldered on those connectors. These are definitely neither DIY friendly nor industrial roughed in any way. It's probably meant you should buy their break out cable units. But I find the Mesas solution so much flexible for a DIY'er. And because of these connectors it's only possible to use small wires as well.

    To top that up the largest (8 axis) version of the Galil card I have only have 16 digital ins and 16 digital outputs that could be used for general use. All the others are hardwired up to other functions and can't be reconfigured. So with my need for 32 in and 32 out I need to add additional hardware. And also in this regard Camsoft comes short of Mesa. The only way to solve this is to buy additional hardware that is almost as expensive....
    And funny enough Camsoft doesn't even support the hardware Galil provide for additional I/O. Only Opto22 compatible PCI cards (and a cheap and useless RS232 unit).
    At least if you buy them from Camsoft they are expensive. The 96 port type is costing $1100 from them. They are standard 7296 cards. But I found them on eBay for $50 including shipping. So I'm not that bad off. But they are TTL level with no opto insulation, so I need to add this as well. That is even more added hardware.
    For this I ended up adapting the original I/O card from my 432 controller. This card had already been worked on so much (including replacing logic chips) that it would have been impossible to sell, so I figured it might as well be a good idea to adapt that to take the I/O from the 7296 card.
    This is what I'm working on right now. It shure proved to be far more work than I'd expected though. And it bothers me to think of how much easier this would have been with the Mesa hardware...... just strip of wires and attach them to the screw terminals....

    So boy do I wish I could have used the Mesa hardware for this software! They have built inn enough I/O with optical insulation, or if I need more it's not difficult to add more as I need it.
    And to top it up it's also possible to use those outputs directly since they are able to control relays directly, unlike the I/O on the Galil cards. Although from what I understand the Galil also have 500mA I/O options available, which is what they sell absolutely most of. It's only Camsoft who specify the 20mA type I/O when they buy them from Galil, so that happen to be the one I have.....

    One major problem with the present solution on Galil hardware is the fact that all the I/O for each axis is on one connector. That means that the cables going to the encoder doesn't have its own connector. So there is no way to swap around on these or do fault finding. The only way to get that is if you buy or build a separate break out box.
    Also in that regard the Mesa system has a clear advantage. The cards are built up more or less as break out boards with separate connectors (capable of screwing in cables, no need for a solder iron). And each function has its own connectors.

    So as you might have guessed I find the Camsoft/Galil solution to be vastly inferior to LinuxCNC/Mesa combo in most ways with regards to the physical implementation. The thing making me still hold on to the Camsoft solution is the fact that it can (from what I have heard) control the machine better than LinuxCNC. And it has 5 axis out of the box.

    So this is what it boils down to. Will LinuxCNC fix the trajectory planner to make it in to something similar to other controllers? From my latest experience I really hope so!

    Funny enough Camsoft hadn't heard of Mesa when I spoke to them at first to try to figure out if it was the correct software for me. I wanted to know if I could use the hardware I already had. He basically laughed at them and called Mesa "amateur stuff" when I had the service tech check it out online while I was talking with him on the phone.
    Apparently the stuff they sell is professional stuff.....

    Oh well.... it comes down to this trajectory planner for me then. I guess it's all about speed. And again this is all about how it's implemented. I read a very interesting article on cnccookbook on the subject. But I'm not able to find it again now....figures....
    Anyhway, it was on the subject of the capability of a CNC machine to move fast while making small moves. And he used the example of making one move of 100" in 1000 000 lines of 0.0001" steps. That means it will be the controllers capabilities to plow through as many lines of G-code as possible in the shortest time. And even if this is a totally useless program it will tell a lot of the capabilities of the controller. A controller capable of running this kind of g-code in speeds of hundreds of IPM was also mentioned, but I'm assuming that is high end (and high cost) stuff that I can only dream of.
    If LinuxCNC will be demanding that the machine should be able to stop during the next segment, and the next segment is only a few lines of code ahead it wouldn't be able to move very fast with this program. But from what I understand this is where things are going. Modern CAM software will assume your machine can plow through thousands of lines of g-code each second, and there is no need to make an effort to keep the code short and efficient. But I guess this is in line with development. Take normal computer software. Most modern software require more of the computer in the newer implementations. This is considered a normal sacrifice if you are able to get benefits from it. In software that might be better and more advanced features. On a mill that would be better trajectories from a machining point of view, giving better removal rates, less tool wear, reduced machining time and so on.

    But there is also the issue mentioned that LinuxCNC WILL be wrong by the allowed inaccuracy even if it doesn't need to (like it wouldn't need to in the example above). At least that is what I understand is happening from ihavenofish's posts. And it really sounded like he had tested it out. Is this something that is looked in to while you are looking at improving the trajectory planner? If not it should!

    My reason for wanting this is to be able to do HSM, as in constant engagement and material removal to get better finish and less tool wear. I know my machine isn't really capable of HSM machining since the spindle is only 5 krpm. But that doesn't mean it can't have great benefits from using more optimized tool paths.
    Although I have 18 k rpm motor-spindles from a different machine (a wood working spindle with MK taper) I was considering putting on there as a test (but I'd lose ATC by doing that).
    So I know most people are still doing "normal" machining. But I don't see a reason to do that if optimized tool paths can lower tool wear and have many other benefits as well?

    Another reason for me to keep Camsoft is the fact that it has 5 axis machining right out of the box. And I guess it will be a long time until that is "plug_and_play" in LinuxCNC? And I'm not even considering learning reverse kinematics to be able to do that in LinuxCNC...... So I guess that mean's I'll keep on working out ways to get around the practical implementation problems I have with the Galil hardware. I'm expecting to get a fairly decent controller when I get past those issues.

    I hope to be able to finish putting the controller together and start testing soon now, hopefully before Christmas (or preferably a lot sooner than that as well). I will keep you posted, as I have understood that there are several of you who have an interest in it. I'm glad several of you are showing interest and are following this thread. It have indeed been a game changer for me and the progress of my mill, so I hope you guys are still interested and will keep on giving input.

    And I do hope Im able to prove that it IS possible for an advanced DIYer to retrofit a machine like this. For the moment I feel confident that it will be possible. For now its more a question of how much Im able to do before my life will change considerably. My wife is due with our first child (a girl) on the 2nd of January 2014, so I expect to have a pause in the retrofitting for some time because of that. But until she arrives I have the green light from my wife to spend most of my time on my machine. After all its not that much I can do for her that makes that much difference, except from helping more with household stuff and such. But as long as she is ok with that Im happy!
    And since I have time off out the year from working offshore I hope to be able to get some tests done in a week or two, but that isn't a promise..... yet...

    I did have an oil leak from the vertical head. But I have been able to take it all apart now, and I think I know where the leak is. I just need to get a new oil sight window (anyone knows where to get the correct one?) and a new o-ring (that one will not be a problem).
    Hope to be back in a few days with initial tests of the machine with new controller hooked up. After that the REAL fun begins. Then I have to make the ATC and stuff like that work.

    Until then...

    Anders

  20. #59
    Join Date
    Aug 2007
    Location
    Holmen, Wi
    Posts
    194
    Post Thanks / Like
    Likes (Given)
    1
    Likes (Received)
    14

    Default

    Ihavenofish:

    Would you be willing to share some gcode that runs faster on mach vs linuxcnc? We are in the process of testing some improvements on linuxcnc.

    (like what you showed in the videos)

    Thanks
    sam

  21. #60
    Join Date
    Dec 2013
    Location
    Cotati, CA
    Posts
    30
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    2

    Default

    Anders, glad to hear you are going to have a solid chunk of time to work on this. After the kid comes you are going to have three of the hardest months in my experience!

    I am in the process of planning out my retrofit and working to move as much to single phase power as possible. I am seriously considering replacing the motors, but I can't find specs on what the physical interfaces are of the motors on my 400E. Anyone know this information?

    Also Anders, I tried to download your excel spreadsheet and the link was broken. Is there any way to get that posted again or messaged to me?

    A big box of mesa hardware arrived and I am eager to put it to use! I am having to move the machine from a location where it has 3 phase power to my home shop with only single phase and am weighing my options in rotary phase conversion, going with a VFD to power the transformer, or changing the spindle motor / servos out. If the axis motor physical interface is easily adaptable to say fanuc servo motors, then I would rather go single phase, use mesa amps and swap the motors out to fanucs, than to stay at 380V for the indramat and the motor spindle. I have some budget to do this part of the conversion, and experience with these amps and servos.

    Any thoughts from the group?

    thanks.

    Asa


Bookmarks

Posting Permissions

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