The "fix your Maho" thread - Page 16
Close
Login to Your Account
Page 16 of 21 FirstFirst ... 61415161718 ... LastLast
Results 301 to 320 of 409
  1. #301
    Join Date
    Jun 2009
    Location
    Norway, Vestfold
    Posts
    183
    Post Thanks / Like
    Likes (Given)
    5
    Likes (Received)
    6

    Default

    So.... it's quite clear we all disagree on what's the best solution to what the best solution is as far as retrofitting a Maho.

    We all agree it's worth the effort though, which means we disagree with the majority of response I got in the start of this thread. Which were: "you will never be able to retrofit a machine like that yourself, so you better have someone do it for you for money if you really want it done".

    So we agree on this at least. Which is a milestone in my opinion. And so we do have a lot in common to discuss.

    I have already written plenty of my thoughts on this previously in this thread, so there is no reason to repeat that. But one thing I'd like to say as a comment to anonimasu is this:

    I fully agree with you Mesa is by no means an amateur in this game, and they really know what they are doing. But more important than this, they have the best support I have ever seen!!!! I mean, the very guy who is developing the cards we use is providing personal support online for everybody to see (so it might help others with similar problems).
    Hey, he even provided schematics details when needed, for free (not all obviously, but what was needed to provide an answer to a question), and volunteered to make custom versions of hardware for advanced users with out of the box requirements. He isn't slow at it either, and he is more or less always able to help too. Obviously some errors are user errors that he can't help figure out, but from what I have seen so far it's better than any other support I have seen. That also includes the paid support from Camsoft!
    The "price" is you have to do it in the LinuxCNC forum, so everybody will see your mistakes..... but at least to me that is absolutely a price I'm willing to pay.

    So this time the "free software will have poor support" rumor doesn't come in to play. In fact it's the opposite way around!

    Then think about the fact that if you decide to go with an expensive dedicated controller you will probably (even though not all controller companies have this policy) have to pay for this same support as well. And the guy who will provide the support will not be the one who designed the thing.

    They do offer to come to your workshop to do the repair work for you, which won't happen with LinuxCNC. That is one thing big difference. But that is something completely outside of the discussion in THIS thread, even though it might very well be what others here preferred.

    And I do understand where you are coming from Ihavenofish. The price you'd have to pay for a modern controller will not be anything close to what you'd have to pay for the machine you will end up with when you are done.
    My answer to that though is, I don't have the money to buy a controller in the league you are considering in the first place. Disregarding the machine I will possibly end up owning. It is purely hobby for now, and I need to be able to say to my wife: "This is a cost I'm able to absorb now, and I won't have to put myself in debt because of it".
    You might be in a different position where you know you will be able to get return for the investment. And that is a totally different thing.
    Come to think of it, this forum is primarily intended for the last group. I get that. But the thing is, that's what makes it have the level of skills and knowledge that makes it worth having discussions on. Again, this forum is the very reason I was able to make my mill run.

    So I do understand where you are coming from, and I will not argue it's not the best solution for you. But for all of us with hobby budgets I would say LinuxCNC looks like the best alternative. And it is in fact also fully capable of serving professionals and their needs. So I'm not worried if I ever take the step to try to make my mill make me some money, even though the heart of my machine is free software running on an of the shelf computer.

    So I conclude we will have to agree to disagree on what is the best alternative for a retrofit. Even though it is useful to us all to hear experiences with other controllers as well. So keep it coming. I find this kind of discussion useful for information, and I think you guys also do so.
    So don't look at this post as an attempt to stop any of it. I'm just stating my take on it, and why I have come to the conclusion I have.

    Anders

  2. #302
    Join Date
    Jul 2011
    Location
    toronto
    Posts
    320
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    20

    Default

    Quote Originally Posted by akb1212 View Post
    My answer to that though is, I don't have the money to buy a controller in the league you are considering in the first place. Disregarding the machine I will possibly end up owning. It is purely hobby for now, and I need to be able to say to my wife: "This is a cost I'm able to absorb now, and I won't have to put myself in debt because of it".
    do you know the prices though? not to push you in any direction, but i was somewhat surprised with the pricing. for the sake of information, even though you are somewhat along the way with your solution, id give siemens a call and ask the price of the 808d advanced.

    to put it another way, when you look on ebay at used siemens motors, the price you see there is probably more than the price new from siemens :x

  3. #303
    Join Date
    Jun 2009
    Location
    Norway, Vestfold
    Posts
    183
    Post Thanks / Like
    Likes (Given)
    5
    Likes (Received)
    6

    Default

    I wasn't able to find the price, but I highly doubt it will be ANYTHING near what I would be willing to pay. As I understand (according to you actually) I will need to replace the motors and servo amps to use this cheapest solution. You were writing about this earlier, and one concern was whether to go for this solution as it would mean replacing all the servo amps and motors.

    And you might consider this, but you should know very well I'm in NO WAY even considering replacing the servos and amps at this time. And I'm not even considering in my wildest fantasies to go and buy new amps and servos for my mill! At least not as the only option to make it move. I'm not even going to make an effort in trying to find out what Siemens will ask for these. I know it will be totally out of range for me, and it's not anything I would even consider.
    It MIGHT be considered as an improvement in the future.... as I would probably be able to make it move faster with higher powered servos. But I'm not willing to install a controller that won't be able to control the servos that are in the mill now. Not only is it a cost I don't think is justified..... it's lots of extra work fitting the new servos as well. And they will need a power supply.
    And does this controller support dual encoder feedback? As a costly option possibly?

    So, you might have been positively surprised on how low the price Siemens is asking for the controller itself. But I highly doubt it will impress me in the same way.
    And your previous posts is giving more of a complete picture. The options I really want, full 5 axis, mill/turn capabilities and so on will not be anything near cheap. These options are available in LinuxCNC! They will require more work from my side. At the same time I'll be more familiar with the machine itself and what it can do because I'm the one who configured it.

    And most importantly I will be able to get all the functions I want without having to buy expensive options as I progress.

    The thing is, I'm not willing to pay for all options as I need/want them. The only thing I'm willing to pay for as I go is new hardware like additional I/O cards and such. That is if they cone at a reasonable cost. And LinuxCNC with Mesa hardware is close to perfect in that regard. They have all the hardware interfaces I need, and they have reasonable prices. And you can get good support as well, totally free!

    So unless you are willing to actually sell me a controller that will do everything I want at a price I find reasonable (which would be lower than Siemens is asking for it for I'm sure) there will not be anything that will convince me to replace LinuxCNC.

    Anders

  4. #304
    Join Date
    Jul 2011
    Location
    toronto
    Posts
    320
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    20

    Default

    crap. 808d advanced doesnt run scales. i forgot about that. youd need the 828d which does cost a fair chunk more money.

  5. #305
    Join Date
    Apr 2015
    Country
    GERMANY
    Posts
    3
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    0

    Default Great post!

    Hi Anders,
    I have been reading this forum on and off for some while and this is the first time I felt the inclination to write something. Your post is excellent. I have no idea what is wrong with some of the so called experts on this forum. Quite frankly they talk from their behinds. They frighten good guys posting, stop other people learning and do not promote anything of note. I really wonder if they ever get their hands dirty. I have many friends (and myself) who have taken great old iron and 'upgraded' to linux cnc. The machines work and perform really well if not better than originally, 5 axis, 3 axis, lathes, milling machines, robots - the list is endless. You need to join cncecke forum to really appreciate what is going on (german speaking unfortunately) - that is where the action is. Don't let a small minority of 'experts' who seem to spend much time typing and not much else put you off.

    Yours is a great project and many fully completeted ones like it are to be seen on cncecke....

    Cheers, john

  6. #306
    Join Date
    Aug 2010
    Country
    AUSTRALIA
    Posts
    82
    Post Thanks / Like
    Likes (Given)
    1
    Likes (Received)
    3

    Default

    Hi All
    Still plugging away at my MH800C, and now trying to sort out the gear change.
    There are 3 positions, 1 at the head, 1 at the back and 1 at the centre.
    Question is dose anyone know which is low gear, high gear and neutral?
    My assumption is at the head is high gear, at the back is low, and at the centre is neutral??

    Not sure if it is neutral at all?

    Also the colured dots have gone off the mechanism/disc and cannot tell what position it is has anyone a photo of a good one?

    Any thoughts greatly appreciated.

    Burgs

  7. #307
    Join Date
    Dec 2011
    Location
    Ohio-USA
    Posts
    20
    Post Thanks / Like
    Likes (Given)
    8
    Likes (Received)
    0

    Default

    Does this help--https://www.youtube.com/watch?v=E0rVbVixMw8

    Wish Mr. N. Mueller was still around.

  8. #308
    Join Date
    Aug 2010
    Country
    AUSTRALIA
    Posts
    82
    Post Thanks / Like
    Likes (Given)
    1
    Likes (Received)
    3

    Default

    Quote Originally Posted by HBOB View Post
    Does this help--https://www.youtube.com/watch?v=E0rVbVixMw8

    Wish Mr. N. Mueller was still around.
    Hi HBOB
    Thanks but already checked it out, not the same as the HM800C.
    gear-changer.jpg
    The gearbox stays in low gear until the spindle is commanded to exceed 143 Rev/min, then it changes to HI gear with a maximum spindle speed of 6,300 Rev/min.
    The coloured lines(Blau, blue?, Siber, Silver?, and gran, green?)you can just make out in the photo I think.

    Regards
    Burgs

  9. #309
    Join Date
    Jun 2009
    Location
    Norway, Vestfold
    Posts
    183
    Post Thanks / Like
    Likes (Given)
    5
    Likes (Received)
    6

    Default

    Hi Burgs,

    I think your gearbox is similar to mine. And if it is it has two gears and neutral, as you expected.

    I don't remember what position is high and low, but that's quite easy to check. It's just a matter of trying to rotate the spindle when the machine is off. If it's relatively easy and you feel like the motor is doing about the same rotations as the spindle you are in high gear. If you are in low gear it will be relatively hard to turn the spindle, and you will feel the motor is making many more revolutions than the spindle.

    And it's possible to change gear manually. If you look in the center of the aluminum wheel in the lower center in your photo you will see a hexagonal hole. That hole is made to insert a 12 mm (I think it was) Allen key to turn the wheel manually to change gear.

    This is all explained in the manual.

    The manual also warns about changing gear manually without disconnecting the electrical wire to the changer motor. The reason for this is that this motor will generate a voltage when turned (and it's geared quite hard to be able to rotate the gear changer wheel). I haven't had any problems even though I didn't disconnect this cable. But I wanted to mention it so you are aware. What I did was to turn it very slowly to prevent the motor from turning so fast it will generate a dangerous voltage.

    If you look in the electrical schematic it will tell you what outputs you need to use when changing gear. It also tell you what sensor will be high for each gear position. So I suggest you look in to that to figure it out. If you have the normal documentation you should be able to get all the information you need.

    Anders

  10. #310
    Join Date
    Aug 2010
    Country
    AUSTRALIA
    Posts
    82
    Post Thanks / Like
    Likes (Given)
    1
    Likes (Received)
    3

    Default

    Quote Originally Posted by akb1212 View Post
    Hi Burgs,

    I think your gearbox is similar to mine. And if it is it has two gears and neutral, as you expected.

    I don't remember what position is high and low, but that's quite easy to check. It's just a matter of trying to rotate the spindle when the machine is off. If it's relatively easy and you feel like the motor is doing about the same rotations as the spindle you are in high gear. If you are in low gear it will be relatively hard to turn the spindle, and you will feel the motor is making many more revolutions than the spindle.

    And it's possible to change gear manually. If you look in the center of the aluminum wheel in the lower center in your photo you will see a hexagonal hole. That hole is made to insert a 12 mm (I think it was) Allen key to turn the wheel manually to change gear.

    This is all explained in the manual.

    The manual also warns about changing gear manually without disconnecting the electrical wire to the changer motor. The reason for this is that this motor will generate a voltage when turned (and it's geared quite hard to be able to rotate the gear changer wheel). I haven't had any problems even though I didn't disconnect this cable. But I wanted to mention it so you are aware. What I did was to turn it very slowly to prevent the motor from turning so fast it will generate a dangerous voltage.

    If you look in the electrical schematic it will tell you what outputs you need to use when changing gear. It also tell you what sensor will be high for each gear position. So I suggest you look in to that to figure it out. If you have the normal documentation you should be able to get all the information you need.

    Anders
    Hi Anders
    My manual has nothing other than the power/speed and torque graph unfortunately, the schematic although dose as you say and that is well documented.
    I believe the spindle is currently in 1st gear as you correctly say it is hard to turn, due to the gearing.
    Dose your mill still have the colured dots or lines on it for the positions?
    Next trick is to work out whether the gear change is controlled by the controller or the motor drive parameters.
    Unless someone can tell me which way it is I will have to wait until power is back on and check the driver (Indramat KDA) parameters?
    Regards
    Mark

  11. #311
    Join Date
    Jun 2009
    Location
    Norway, Vestfold
    Posts
    183
    Post Thanks / Like
    Likes (Given)
    5
    Likes (Received)
    6

    Default

    Quote Originally Posted by John von insel View Post
    Hi Anders,
    I have been reading this forum on and off for some while and this is the first time I felt the inclination to write something. Your post is excellent. I have no idea what is wrong with some of the so called experts on this forum. Quite frankly they talk from their behinds. They frighten good guys posting, stop other people learning and do not promote anything of note. I really wonder if they ever get their hands dirty. I have many friends (and myself) who have taken great old iron and 'upgraded' to linux cnc. The machines work and perform really well if not better than originally, 5 axis, 3 axis, lathes, milling machines, robots - the list is endless. You need to join cncecke forum to really appreciate what is going on (german speaking unfortunately) - that is where the action is. Don't let a small minority of 'experts' who seem to spend much time typing and not much else put you off.

    Yours is a great project and many fully completeted ones like it are to be seen on cncecke....

    Cheers, john
    Hello John, and thanks for those kind words.

    I have been over at cncecke, but for me there is a major reason I'm not participating..... My skill in the German language is so limited it's no use for me trying to read it.

    So... I tried to use google translate to make some sense of it. But unfortunately that doesn't work at all! All it accepts to translate is the login page. As soon as I have logged in there is only German words on the screen, even though it think it's translating. If I switch back to original page and back to translated it jumps back to the login page.

    Which is a bad thing, because I would very much like to have a closer look as it seems to be a forum with lots of great discussions. I know it should be possible as I have looked at other forums with google translate. DaBit on the LinuxCNC forum has a great (!!!!!) and long thread on how he built his mill on a Dutch forum, and I was able to make it try to translate it. Not a good translation though, so it was hard to understand what was being said. But a lot less hard than trying to understand the original text.

    So until that bug is fixed you won't see me over there as I won't be able to make any sense of what you are saying.

    Anders

  12. #312
    Join Date
    Nov 2009
    Location
    Sweden(Luleć)
    Posts
    113
    Post Thanks / Like
    Likes (Given)
    1
    Likes (Received)
    0

    Default

    Anyone here got a clue if i could safely plug the through spindle coolant i am having a minor leak still even after making new innards for the svivels. (Is it used for anything but cooling the svivel itself)?

  13. #313
    Join Date
    Apr 2011
    Country
    UNITED STATES
    State/Province
    Colorado
    Posts
    17
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    2

    Default Siemens 828D price

    Does anyone have a quote for the 828D control hardware? And the cost to have it installed? Or even a guess? I'm seriously considering the 828D for my 2000 Lagun VMC 3516. I contacted Siemens, but the response was that the engineer is on vacation for the next week. I'm just looking for a ballpark figure. thanks!

    Quote Originally Posted by ihavenofish View Post
    insane is a relative term. ive decided to go with a propper control. the cost is less than you may think, and the result is worth it i think. total investment is still much less than any comparble machine i could buy, and the new control (and motors if i go digital) have a 2 year on site warranty.

    these are what im looking at:

    Siemens 808D Advanced - pros: very cheap with new motors, all digital. cons: does not supports scales, so you need zero backlash in your machine, new motors are mandatory, 4 axis only. NOT CSA APPROVED. this means you cant buy one in canada right now without alot of hassle (basically paying to get it approved which is thousands of dollars).

    Siemens 828D Basic - pros: medium cost with new motors, more flexible options and faster block processing than 808d, all digital, supports scales. cons: new motors are mandatory, 4 axis only (5 axis on the higher end 828d models but they cost considerably more).

    Siemens 840D SL Basic - pros: highest performance contouring, simultaneous 5 axis, can use existing drives, supports scales, more optional features like digitising. cons: costs the same with no motors as the 828d does with motors.

    i also looked into heidenhain. they are much more expensive siemens and much less willing to support the individual user.

    im leaning to the 828d. the price of the new system plus motors, minus the selling price of the old motors works out a fair bit less money than the 840d, and will have warranty on the enw motors and drives, and similar performance.

  14. #314
    Join Date
    Jul 2011
    Location
    toronto
    Posts
    320
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    20

    Default

    Quote Originally Posted by JHarrelson View Post
    Does anyone have a quote for the 828D control hardware? And the cost to have it installed? Or even a guess? I'm seriously considering the 828D for my 2000 Lagun VMC 3516. I contacted Siemens, but the response was that the engineer is on vacation for the next week. I'm just looking for a ballpark figure. thanks!
    i have a quote specific to my maho including 4 axis motors drives and spindle motor and drive and glass scale interface.

    i have another quote for a 4 axis small router with no spindle and no scales.

    both are canadian dollars with duties tax etc

    theres no breakdown so its hard to guestimate any individual parts cost.

    on installation, i have no idea there. i intend to instal mine on my own with basic "support" from siemens.

  15. #315
    Join Date
    Jul 2011
    Location
    toronto
    Posts
    320
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    20

    Default

    question about velocity control on maho servos.

    im building a test system that will just run motors and magnetic scales and other systems one at a time just for getting everything in order while we wait for a new control (or rather money for said control)

    one thing has me puzzled and the linux guys might know the answer.

    on a 12 bit DAC, like the philips 432 has, we get 4096 speed levels. on my machine this equals 3.6mm/m increments. this seems awefully coarse, but of course we also know that it works as the maho was super precise "stock".

    but what actually happens when the machine needs to go 1.2mm/s, for example in a cirle motion. does the control just create a dithering pattern every 1ms to approximate that speed, with motor's comparatively slow response effectively filtering the signal?

    thats what im guessing it does, but maybe im way out.

  16. #316
    Join Date
    Nov 2009
    Location
    Sweden(Luleć)
    Posts
    113
    Post Thanks / Like
    Likes (Given)
    1
    Likes (Received)
    0

    Default

    Quote Originally Posted by ihavenofish View Post
    question about velocity control on maho servos.

    im building a test system that will just run motors and magnetic scales and other systems one at a time just for getting everything in order while we wait for a new control (or rather money for said control)

    one thing has me puzzled and the linux guys might know the answer.

    on a 12 bit DAC, like the philips 432 has, we get 4096 speed levels. on my machine this equals 3.6mm/m increments. this seems awefully coarse, but of course we also know that it works as the maho was super precise "stock".

    but what actually happens when the machine needs to go 1.2mm/s, for example in a cirle motion. does the control just create a dithering pattern every 1ms to approximate that speed, with motor's comparatively slow response effectively filtering the signal?

    thats what im guessing it does, but maybe im way out.
    That would be incorrect the scale signals are sinusoidal which gives you 4096 increments per scale period which gives with the numbers used in your calc ends up at somewhere around .001mm

  17. #317
    Join Date
    Nov 2009
    Location
    Sweden(Luleć)
    Posts
    113
    Post Thanks / Like
    Likes (Given)
    1
    Likes (Received)
    0

    Default

    Quote Originally Posted by anonimasu View Post
    That would be incorrect the scale signals are sinusoidal which gives you 4096 increments per scale period which gives with the numbers used in your calc ends up at somewhere around .001mm
    Sorry ignore that whole post I read adc. (it's before coffee this morning).

  18. #318
    Join Date
    Jul 2011
    Location
    toronto
    Posts
    320
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    20

    Default

    :P
    yeah, its the dac. the analogue output im concerned with. i just went over this with someone on a stepper controlled variant, which is very easy to see the math, and its obvious that you can hover back and forth above and below the desired speed 1000 times a second and have it track perfectly at low speeds, since one time increment is on the single digit nanometer scale in the lower speeds (1mm/m) where you need the most precision (coming to a stop or corner and "holding").

    this performance at high speeds is basically what will limit contouring (not rapids) - basically once the controllable speed "steps" and compensation loop makes the desired position deviate by more than a handful of microns. the is easy enough math. basically how fast is 1 micron per millisecond in mm per minute. so thats 60mm/m. at 6000mm/m, we devite by a possible 100 microns. in reality it wouldnt deviate near this much but it could in sharp rounded corners where we arent decelerating enough.

    increasing the DAC bit depth, and increasing the servo loop frequancy will both help tracking. so 16 bit and 2500khz loop would lower the deviation at 6000mm/m to 40 microns at high speed. modern controls use a high precision digital command (more than 16 bit) and have loops as high as 10khz from what ive seen. but these machines move at 3000+ipm with 2g acceleration.

    so basically if you want the "same" as the 432 (which isnt bad). 1khz loop and 12 bit DAC will do it. if you want to be faster AND not lose precision at that new speed you want a higher bit depth on the dac, and a faster servo loop.

    i think that makes sense to me, haha, does it makes sense to anyone else? :O

  19. #319
    Join Date
    Jun 2009
    Location
    Norway, Vestfold
    Posts
    183
    Post Thanks / Like
    Likes (Given)
    5
    Likes (Received)
    6

    Default

    I have been looking more in to encoders and resolution lately. More precisely I have looked in to using sin/cos 1Vpp encoders.

    The reason for this it to figure out how to get high enough resolution on my 4th (and 5th) axis. Earlier I was considering using a read head from Renishaw and a magnetic tape. This was a solution that could have given me reasonable resolution for a 5th axis with low rotational speed.

    The thing is that for my 4th axis I want high speed as well. 5000 RPM in fact. So the resolution I want is impossible to reach with normal incremental encoders. A resolution of 1.000.000 CPR (which is what I want as a minimum for low speed accuracy) gives encoder counts in the ~100 MHz range.

    So analogue output encoders it the way to go. iC-Haus make several Sin/Cos encoder interpolators that is able to increase the resolution to way beyond the resolution I want from my rotary axes. I have a lathe head I'd like to use for my 4th axis as a start. This has a 512 CPR Sin/Cos 1Vpp encoder built inn which I wasn't aware of until recently. With that I started to look in to using that form of encoders. And it's quite clear that this is the way I'm going!

    First I was under the impression that all the iC-Haus interpolators would only do 1Vpp encoders, but I found a schematic in one of the data sheets (for the IC-NQC) that showed it can handle 11 uA encoders as well! This means I can use them on my linear scales too! I was under the impression that the 11 uA encoders were inferior to the 1Vpp, and not possible to interpolate to the same degree. And that might still be the case. But when 1Vpp encoders can be interpolated to 16 bit I would assume I can interpolate the 11 uA encoders to at least 9 or 10 bit. In particular if the interpolation chip is positioned inside the read head (or built on to it) for very short leads that will reduce noise. The more expensive EXE boxes from Heidenhain can interpolate both types of scales to at least 400. That means that it should be possible for me to get at least the same interpolation rate if I do it inside the read head. The new chips from iC-Haus also incorporate more powerful calibration and filtering technology, so I'm sure it will work.

    I will then end up being able to increase the resolution on my linear scales to a resolution not possible to achieve mechanically. But it's still a point in doing so. Take a look at this video here: https://www.youtube.com/watch?featur...&v=dbGLybLOX8U

    They use 1 Vpp Sin/Cos encoders to get about 130 million CPR resolution. Nothing is being said about the overall accuracy, and there is no way they have an absolute accuracy in the range they get the repeatability. But with that high reliability it would be possible to get as high accuracy as your calibration equipment would allow for.

    Anyway, you are correct Ihavenofish. Bit depth of your drives (and analogue +/- 10V outputs) do play a role here. The original 432 controller were high end at the time, but it's hardly noting to shout about now. In particular if you get out of the +/- 10V and incremental encoder way of doing things.

    Reading the position at given intervals (that is coordinated with your servo loop periods) and then feeding the drive speed (or torque) back to the drive by a fast digital interface is WAAAAY better! It will give you way higher accuracy, and on top of that the required hardware for interfacing will in fact be easier (and should also be cheaper) to make. Going from digital in the controller through a DAC to transfer a voltage level to the drive, and then in the drive go back to digital in an ADC isn't the most elegant way of doing things.....

    As far as DAC bit depth goes. The Mesa cards use 12 bit DAC's in the 7i77 cards, so there isn't much to gain by getting drives with higher resolution ADC's if you use LinuxCNC with Mesa hardware like I do. That is unless you change to digital communication.
    So my plan is to try to make it work with BiSS C for encoders. Not sure about the drives yet, but I'm wondering if the Mesa 8i20 drives (which have a digital interface only) will work here.

    There will always be AD and DA converters involved though. And the drive will need to read the currents (for feedback), and the output stages will need to be digitally modulated. All of these will impose digital limitation. The the Kolmorgen rotary system in the video surely have high resolution components all the way, and claim to be true sinusoidal output drives. I'm not sure what accuracy will be possible to get with "standard" drives. And I'm not expecting to get the resolution shown in the video. But I am expecting to be able to gain both accuracy and speed by using digital communication between encoders, controller and drives.
    I also expect to get better low speed performance as a result. The challenge now is to replace the TTL level encoders I have on my servos (to be used for the P and D part of the PID loop) with 1Vpp ones. That is to get some reasonably priced ones that is....
    Or hack in to the ones I have (which was free, so I'm not to concerned about destroying them) to see if it's possible to get a 1Vpp signal out of them. That signal should exist somewhere in there. It's just a matter of finding it I would guess.
    A quick search on eBay wasn't too encouraging. Not many reasonably priced 1Vpp sin/cos encoders available......

    Anders

  20. #320
    Join Date
    Jul 2011
    Location
    toronto
    Posts
    320
    Post Thanks / Like
    Likes (Given)
    0
    Likes (Received)
    20

    Default

    for the linear axis loops, i think were ok. ive done all the math and basically, the stock 12 bit DAC with a medium speed loop (4khz) will be more than adequate for anything im doing that requires high precision. the only place it will loose out (a small amount) is in a very high speed 3d contouring job, and those generally are not critically precise tasks. being out 25 microns isnt a showstopper if it allows contouring at 6000mm/m

    on the rotary encoders. my machine has a rod270. 18000 line encoder. this gives 72000 counts per rev. the axis only spins 20.5rpm though and im fairly sure the limiting speed is very low. its also extremely expensive at several thousand dollars new. i dont intend to replace it, but if i did, i would use the same AMS magnetic encoder as i am on the linear axes, with a magnetic multi pole ring that they provide. this ring has 128 poles, for a 128mm outer circumfrence. it has a resolution of 64436 counts per revoluton in quadrature incremental mode. the limiting speed on this encoder is only going to be in the 100's of rpms.

    they make standard rotary encoders that spin fast, however they are limited to 16384 counts per rev.


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
  •