The "fix your Maho" thread - Page 5
Close
Login to Your Account
Page 5 of 20 FirstFirst ... 3456715 ... LastLast
Results 81 to 100 of 398
  1. #81
    Join Date
    Jun 2009
    Location
    Norway, Vestfold
    Posts
    181
    Post Thanks / Like
    Likes (Given)
    5
    Likes (Received)
    5

    Default EXE cards working

    I received 2 off EXE cards I ordered a few days ago.

    Last night I was finally able to do some initial testing of it, and its working!

    It was a straight swap with the original "pulse shaper" card originally in my controller. Same connectors and everything. Even seem to be consistent on wire colors. But I'm still not 100% sure on this.

    If it is then the original wiring in the Maho is a bit off. I have used what information I have from the schematics to try to get the correct outputs connected to the new controller. And it seemed a bit odd to me that they didn't wire it so that each output was grouped in pairs. The connectors are rows of 2, so it would seem logic to have each output paired up so they are lightning up as pairs in a row.
    Well, I looked up the documentation of the EXE cards to verify if the signals are compatible to do a straight swap. This is when I found out that Heidenhain actually do pair up the outputs as I expected.

    With this information I found that I have switched -B and -I, which won't stop it from working. But it might mess up the indexing..... I will have to look in to that.

    Then there is the resolution.
    I wasn't sure on what multiplication factor was on the original pulse shaper card so I tried to calculate it. The scales say there is a period of 20um. And from testing with CamSoft I found that I had to set it to 1000 steps/mm to get correct readings with the original converters.
    I wasnít sure if any of these included the factor of 4 for quadrature or not, so I figured it should be a multiplication factor of 20.
    I did some research on the EXE boxes and found they come in lots of variations. Most of the cheap ones had only 5 factor, and I imagined that would not be enough since I would possibly loose resolution. And since I have complete control over the configuration now and can set whatever resolution I want (not sure if that is possible in the 432 controller). Increased resolution is also no problem with the Galil card as it has very high frequency limitation (12 MHz or something).

    I also have a spare linear encoder I got from work, 1.2 meter or so long. It was thrown away because the protection on the cable is breaking up. Except from that it's working. I have been using that as test for this process and found it has a lower resolution (possibly depending on higher multiplication factors in modern versions?). And I plan on using one of the two I bought for this, and didn't want to low resolution on that.

    So I found that the EXE boxes with designation 650B 25/8 was the one I wanted to try. As far as I know it has a multiplication factor of 25, and a frequency limitation on the output of 8 MHz. And it turned out there were some of these that weren't so much more expensive than the more common 602 type. And I found 2 with cables that were in perfect condition and ordered them.

    When I swapped one of them in for a test I found that it now has 5 times the resolution. So the conclusion is that the original ones actually do have a factor of 5 only. So it must mean that the factor of 4 from quadrature comes in to play, and the scale has a resolution of 5 um. When dividing with 5 you get 1um resolution which is 1000/mm.

    So for you guys that are going to use this you only need to get the 602E ones. These are cheaper and much more abundant. I saw one on ebay now for $30 + shipping, so they are really cheap now!

    Alternatively you could get the 610C (2MHz) or 612 (8MHz) ones. The key is that it's a ~11uApp type. That seems to be all the ones named EXE.

    The documentation also mention that the 602E are "Without and 5-fold" interpolation. I'm not sure if that means they come in 2 versions, or if it is settable inside. It would be bad if one got one and ended up with one without 5-fold interpolation and ended up losing resolution. At some point I might end up getting some of these too, being as cheap as they are. If they are one or the other only be sure to read the description and get one that has a 5-fold interpolation factor!

    In my card there are the possibility to set something with hardware switches, not sure what. It might actually be the multiplication factor, but I'm not sure. Initially I thought the designation told what the factor was, and I see they come with the 50/8 designation as well. The documentation I can find also indicates the 650B cones with either 25 or 50 factor. So I'm not sure what this is. Possibly frequency limitation? That would actually be kind of bad, so I hope it's not.

    Anyone here that knows this? It would be nice to know what these settings do set when they are there.... There are some mentioning on "Clock frequency" in the documentation, but I'm not sure what that means and how it's clocked. The 602E type is "Non-clocked" whatever that means. The 650 B has 8 MHz frequency anyway.

    Anyway, I have now confirmed that these cheap and easy to get boxes actually do work as we expected. They do make it easy and quick to reuse the original encoders fitted to our Mahos and keep the accuracy they have.

    We can even increase resolution cheap and easy if we want to!

    Anders

  2. Likes Burgs liked this post
  3. #82
    Join Date
    Aug 2010
    Country
    AUSTRALIA
    Posts
    82
    Post Thanks / Like
    Likes (Given)
    1
    Likes (Received)
    3

    Default

    Hi Anders
    Your progress sounds good and en-lighting.
    I have had a look on EBay for the converter and found this one, HEIDENHAIN SCALE CONVERTER EXE 602 E 246 842-02 is this the ones you have?
    If so looks like it could be a good solution to my problem with the linear Philips scales at least.
    What cable/plug connections do they use?
    Also do you know if the original Philips scales have home and soft stop reference marks on the scales as I cannot find any switches on the 3 axis?
    Are the Philips scales absolute or incremental?
    I see in previous posts there was some confusion over built/manufacturing dates for your MAHO?, when I removed all the covers, I found on my MAHO casting dates of 1991 and stamped dates on the various castings and all the manuals/schematics have 1991 dates so I guess it must be 1991.
    See a couple of photos of my 1991 MAHO MH800C if I have got the process correct? img_1018.jpg


    Best Regards
    Mark
    Attached Thumbnails Attached Thumbnails img_1026.jpg   img_1017-copy.jpg  

  4. #83
    Join Date
    Jun 2009
    Location
    Norway, Vestfold
    Posts
    181
    Post Thanks / Like
    Likes (Given)
    5
    Likes (Received)
    5

    Default A few steps backwards....

    Ok, so I have started the PID tuning to get my mill moving....

    Which proved not to be as easy as I expected it to be.

    First off I found out that the increased resolution on one axis by having a different interpolator on one axis wasn't such a good idea. For some reason the Galil controller doesn't work well with different resolution on the different axis. In addition to that the flex in the belts were too high compared to the resolution I ended up with.

    I'm using Galil's tuning software to do the tuning. It's way better than the tuning built in to the Camsoft software. I tried using that at one point, but that wasn't made to do advanced tuning. They also state that when you try to use it for tuning, so I quickly returned to the Galilsuite software for tuning.

    At first I also had another problem. I wasn't able to get anything useful out of the system, and all axes were resonating wildly almost no matter what I did. I found the reason for this. At first I powered the computer from a different power connector, and due to earth faults in my neighborhood there was 50 Hz on the earth of the computer. At first I tried to look at what the drives got on their input, and was shocked to find +-10V 50Hz on all of them! That was when I started expecting something was wrong with the earth connection. I was even more shocked to find that when I disconnected the earth between the computer (with the Galil card in it) and the mill I found the AC voltage between the two earth potentials to be 170V AC!
    So I decided to use the 220V from a transformer in the mill instead. That did the trick. After all now the power had a galvanic insulation, and it worked as expected.

    So now I was ready to start on the proper PID tuning. But I found the tuning software unable to tune properly. For some reason the automatic tuning methods doesn't allow me to adjust the resolution, and ends up not moving the axis enough to get a good reading. On the two with original resolution (1000 steps/mm) I was able to get half decent results. But on the axis with the new interpolator there is no chance of getting anything useful.

    So I have concluded I need to get new interpolators with the original interpolation ratio. So I decided to buy the EXE 612, which is the one with highest frequency capabilities and 5 fold interpolation. Found them for $44 a piece, so I decided to get some of those.

    So a word of advice to you guys who need to get EXE boxes and have original encoders. Go for the ones with the same resolution. In particular if you are using Galil hardware. This controller isn't capable of dealing with different resolution on the different axis.

    Here is the best guide on the different EXE boxes I have found so far:http://www.lna.br/~det/Projetos/TCSP...heinEXE602.pdf

    Until I get the new EXE boxes I will figure out the homing and indexing issue. I will be back with more on that when I get the time to do it. Right now my 12 weeks old baby girl needs most of my attention.

    Anyone knows approximately what the values Maho use? After all the movement the mill had with the original controller was way better than what I get now.

    Anders

  5. #84
    Join Date
    Aug 2010
    Country
    AUSTRALIA
    Posts
    82
    Post Thanks / Like
    Likes (Given)
    1
    Likes (Received)
    3

    Default

    hi Anders
    One question I have is on my Philips scales they are 12VDC but I see the EXE boxes are 5VDC will they still work or have I missed something?
    It would be so nice if someone Knew some thing about technical side of the Philips scales.

    Best Regards
    Mark

  6. #85
    Join Date
    Jun 2009
    Location
    Norway, Vestfold
    Posts
    181
    Post Thanks / Like
    Likes (Given)
    5
    Likes (Received)
    5

    Default

    Hi again,

    Just thought I'd clarify some of my findings since my last post was a bit unclear.

    First of the earth problems. This has been a reoccurring thing for me, and I wasn't surprised to come across it again. I just forgot about it when I did my first tests, and found out how important it is to watch out for it. I hope this can make others aware if it as well.

    One contributor to this though is the simple fact that the Galil controller only has single ended motor drive output. If it had been differential I wouldn't have to bring the earth of the computer in to the driver as a reference. Iím not sure what the common mode rejection (the capability of the drive to reject the noise found on both signal wires) of the drives are, but Iím sure it would have made a huge difference.

    And the fact that after removing the earth fault the resonance problems disappeared only proves that the noise caused by the ground fault finding its way in to the driver and exciting the servos with 50 Hz constantly. And this happened because I had to use bring the ground from the computer in to the drive as the reference for the driver.

    When this was going on I could feel the 50 Hz vibration in the whole machine. And when I looked at the servo belts they were vibrating very visibly. The servos themselves got somewhat warm as well, indicating they were constantly working to follow the noise found on the inputs.

    Normal driver design is always using differential signaling for sensitive signals like this, and I was surprised to find Galil didn't.

    To my luck I was able to remove the ground fault completely, but if I wasn't I would end up with more or less a useless system because of this.


    I'm not an expert in this. I have only learned about this from practical experimenting. I'm an electronics engineer, so I have electronics background. But as far as CNC systems I have only practical experience from my own machines. So please beware of this when reading this. I just thought I'd clarify.

    Anyway, I took a look at the Mesa hardware used for LinuxCNC to see how they do it. And to my disappointment they also use single end signaling only for the servo signals. In the manual they state that normal analog servo drives use differential inputs to avoid ground loops. And then go on to say you should use the ground pin adjacent to the output as the other input. Not Shure how that will avoid problems like this (I expect it won't..).

    So I have found the only way to get by problems like this is to use a transformer to make a galvanic insulation between the computer and the shop earth. So I took 220V power from inside of the electrical cabinet what I knew were insulated from show ground, and only have the same earth potential as the rest of the mill.
    And this worked. I'm just not sure every DIYer is aware of this. So I wanted to share this with you guys to make you aware of it as well.

    This is also possibly mostly due to the bad way the electrical distribution in Norway is sorted. Itís with two live wires and no earth connection from the supplier. Each house needs to make its own earth connection by sticking a copper wire in the ground. All kind of bad things comes from this, the problems I experienced being one of them.

    In regards to tuning software:

    I understand certain assumptions have to be made on software like this to make it easy to use. But for me I get in to problems because my system is having resolution that is higher than normal. And there is no way to tell the software this, and I end up with software trying to excite the axis, but don't move the axis far enough each way.
    What this results in is itís trying to measure the system but is only working in the nonlinear first part of the move.

    In practice it is exciting the servos so short (in so short time) itís only tensioning the belts before reversing, and then the axis might or might not overcome the initial friction and jump forward. I could see clear evidence of this from the response on the traces of the scope in the software.
    I did the test over and over, and the response was more or less random. Some of the time it didn't move at all, giving large errors. Other times it did move, but not very accurately. And this only happened to the axis with increased resolution.

    That being said, the increased resolution is possibly the biggest problem here, and as mentioned I will change this. I expect this will make this axis behave as the others.

    The software have manual tuning available, and I was able to get better results from that. But there are still limits to the allowed length of move here as well (5000 steps, 1 mm movement on my machine). This was considerably more, and I was able to make the axis move more reliably. But I still have the problem of the slack of the belts are considerable in comparison to the resolution.


    I wonder if the controller is capable of decrease the resolution back down by a factor of 5 and that this would also cure my problems? This is something I havenít looked in to yet, but I just now came to think this might possibly be my best way to solve this. I just need to find out how.

    Maybe this is what the ďgearingĒ factor in the setup is for? I will take a look in to this tonight to try to find out. I hope it is as this will make things so much easier.

    I did take a look at Galils videos on tuning, and I must say itís evident that they do know what they are doing. And one thing that looks like a good thing is the dual encoder option. And since I actually have some good rotary encoders available I think I'll try to use that at some point.


    At the same time I know the mill worked quite well with the old controller and no encoder on the servos. So I'm wondering what I'm doing wrong.

    Iím also unsure of whether itís possible to use dual encoder feedback on only one axis and not on the othersÖ. but I hope it is as Iím not going to attempt to put a rotary encoder on the Y-axis. That servo is way too difficult to put an encoder on due to space and accessibility issues. That is unless I take of the cover on the back and attach it on there. But Iím not too sure if that is a good idea, opening up the motor and potentially contaminate the inside of the motor.

    I will mount an encoder on my x-axis as that is very easily accessible to find out if itís making a difference and if itís worth the effort of looking in to it on the other axis as well.


    I might also benefit from tensioning up the drive belts as they have been on there for a long time. I even consider replacing them. But I have no experience in how much tension is correct.
    From the vibrations I experience when trying to tune my system I expect I might need to do something with this. If that is the case I donít look forward to doing the table lifting axis (Y in Maho terms, but I will change that to Z now). Not very good access under thereÖÖ and brake to prevent the table from falling down and everythingÖ..

    Funny though, the manual for the mill states ďas no stretching is happening during operation, a previously adjusted belt will never need to be adjustedĒ (translated from Swedish). And when looking at it there are no provisions for tightening them. So does that mean I wonít benefit from adjusting them? I find that hard to believe as high speed short movements will most certainly induce vibrations. I have literally felt that on my body during servo tuning. I didnít particularly like it when it was trying to move the table up and down so fast the machine was making my house shakeÖ.
    But to my luck the overcurrent protection in the drive kicked inn after a few seconds. But the truth is I wasnít too happy about that either. But better trip that than burn the drive. Better still though would be that the software picks up on the fact that it is now resonating considerably and not trying to crank the values up that highÖ. This was on one of the axis with original interpolator so it wasnít too high resolution either.

    I will call Galil about this. They have a very good support phone, I think I need to call them and ask them about it.

    But the mill was running fine before I started this process. So now I would most of all like to have the basic movements working.
    Truth being told I didnít expect to have problems with this. So I hope Iíll be able to sort it and move on very soon.

    And I need to stop rambling nowÖ..

    Anders

  7. #86
    Join Date
    Jun 2009
    Location
    Norway, Vestfold
    Posts
    181
    Post Thanks / Like
    Likes (Given)
    5
    Likes (Received)
    5

    Default

    Quote Originally Posted by Burgs View Post
    hi Anders
    One question I have is on my Philips scales they are 12VDC but I see the EXE boxes are 5VDC will they still work or have I missed something?
    It would be so nice if someone Knew some thing about technical side of the Philips scales.

    Best Regards
    Mark
    Mark,

    Looking at your schematics I notice you have different encoders than I have. Yours are PE 2520/10. I didn't expect that as most other things are similar on our control systems. And I'm not able to find much useful information on them with Google either....
    It seems your key here is the old controller with the interface card. Without that or info on how that works you could have problems yes.

    Or alternatively you can possibly get this interface card as a spare part. I know the card on the controller it's interfacing in to is accepting normal TTL quadrature input. Not sure if it accepts other signals as well though.

    It could help if you are able to take a photo of one of the heads and see if there are other markings or a sign on there that will make it possible to search for them online.

    What I would have done is to identify the power in to one of the encoders and then try to look at the outputs with my oscilloscope. But you might not have that, or even know how to use it?

    I found these encoders in the first page of the spare parts list in your schematics and it is:

    19A5, measuring syst. separator, MAHO-Scan RS422, Manuf: M&D, Ident: 27.074329

    From this it's possible it is only a signal separator feeding the signals on to the controller acting only as some form of galvanic separator. If it is then it should be possible to use the signals coming from your encoders directly and you don't need an interpolator like the EXE boxes.

    This would be possible to figure out if you have access to an oscilloscope. It would be possible without as well, and could be tested with simpler devices as well. I can help you with that as well if needed, but would be much better done on the phone or zkype or similar.

    DMG or Mahodoctor might help you with details on what that card is though. Try to get info from them on what this car is.

    I didn't realize you have different encoders on there. And as you say, they use 12V as well. Not only 5V as the ones I have.

    I hope you are able to figure it out. I can try to be of assistance if you need me to. Too bad you're on the other side of the globe now then.... although I wouldn't mind taking a trip to Australia ... if only it weren't that expensive....

    Anders

  8. #87
    Join Date
    Aug 2007
    Location
    Holmen, Wi
    Posts
    194
    Post Thanks / Like
    Likes (Given)
    1
    Likes (Received)
    14

    Default

    There are (and I have) many machines setup this way with no issues.

    here is a quote from the mesa owner..

    Quote Originally Posted by PCW
    I don't think I have ever seen a controller that has differential _outputs_
    We do have the option of isolated analog outputs but they are seldom needed
    (only if there is a severe ground loop issue)
    Good analog drives have differential inputs
    and this is what provides common mode rejection.

    even asah in this same thread...

    Maybe asah could give info on how his servo drives are connected.. It is sad that you discounted linuxcnc on its TP grounds... (ihavenofish might be coming back around to linuxcnc.. maybe) There was a nice discussion on ******* where I think the issues he was having where figured out. (plus I hope he test the new trajectory planner when he gets going) there are at least 3 people that I know running mahos on linuxcnc...

    sam

    Quote Originally Posted by akb1212 View Post

    Anyway, I took a look at the Mesa hardware used for LinuxCNC to see how they do it. And to my disappointment they also use single end signaling only for the servo signals. In the manual they state that normal analog servo drives use differential inputs to avoid ground loops. And then go on to say you should use the ground pin adjacent to the output as the other input. Not Shure how that will avoid problems like this (I expect it won't..).

    Anders

  9. #88
    Join Date
    Jun 2006
    Location
    Munich / Germany
    Posts
    2,338
    Post Thanks / Like
    Likes (Given)
    342
    Likes (Received)
    1068

    Default

    A few notes:

    The servo amp *should* have isolating receivers. But sometimes/often, they only allow a swing of a few volts. Have a closer look at the amp.
    I *completely* opto-isolated all INs and OUTs from my control. For the analog signal, I used an ISO124 (the better ones were too expensive, but I can adjust the minimal offset with a line driver following the ISO124.

    You should hand tune the PID-loop. First set I and D to zero and use a P-value where the drives do not oscillate. Then reference the mill and write a little program like this:
    F500
    G1 X 0.0
    G1 X 10.0
    G1 X 0.0
    G1 X 5.0
    ...

    With that, tune your PID. It should not make any difference what resolution you have on the feedback.

    I don't remember what Maho you have, but mine (MH 700C) has a belt tensioner for the vertical axis too.

    And just some clarification:
    I often hear, that Maho somehow must have confused the axis names. Truth is, that they are mostly configured as horizontal mills, but now often used as vertical mills.


    Edit:
    If you want to know how to tension the Z-belt, ask!


    Nick

  10. #89
    Join Date
    Aug 2010
    Country
    AUSTRALIA
    Posts
    82
    Post Thanks / Like
    Likes (Given)
    1
    Likes (Received)
    3

    Default

    Hi
    The electrical engineers gave me an old oscilloscope at work, I will bring it and see what I get? I will just to learn how to drive it.
    Unfortunately do not have the original controller anymore.

    Regards
    Mark

  11. #90
    Join Date
    Aug 2010
    Country
    AUSTRALIA
    Posts
    82
    Post Thanks / Like
    Likes (Given)
    1
    Likes (Received)
    3

    Default

    Hi Anders
    My thoughts are the encoders are sending out sine single as the drawing on pg98 nothing the wiring or numbers are different to the ROD and RON rotary encoders.
    Ov wh, RSEL bl, +12v br, +12 -, Ov -, PALM vi, 500 wh, MARKRN pn, MARKR ga, 590 yl,AREA BU, ALARM rd, 500N br like you say they go into the 19A5 measuring system separator?
    Best Regards
    Mark

  12. #91
    Join Date
    Jun 2009
    Location
    Norway, Vestfold
    Posts
    181
    Post Thanks / Like
    Likes (Given)
    5
    Likes (Received)
    5

    Default

    Quote Originally Posted by Burgs View Post
    Hi Anders
    My thoughts are the encoders are sending out sine single as the drawing on pg98 nothing the wiring or numbers are different to the ROD and RON rotary encoders.
    Ov wh, RSEL bl, +12v br, +12 -, Ov -, PALM vi, 500 wh, MARKRN pn, MARKR ga, 590 yl,AREA BU, ALARM rd, 500N br like you say they go into the 19A5 measuring system separator?
    Best Regards
    Mark
    Mark,

    I still believe the interface card to be only a simple interface card. This is a print out of page 98 and 99 of your manual aligned on top of each other with the signal paths highlighted:

    encoder-signal-route.jpg
    As you can see the naming of the different signals are identical on both. I can confirm that the inputs are working as TTL inputs for the encoders with "500" and "500N" (marked in green) as one differential input, "590" and "590N" (marked in yellow) as another and "MARKER" and "MARKWERN" (marked in blue) as index. Although there are 2 wires that are switched on my mill, but I believe this to be as mistake. I don't remember exactly as I don't have access to my notes here, but it is possibly the two marked by the pink arrow at the top in the drawing. Or come to think of it I think it's actually input 4 and 15.... or something like that....
    This is one of the things I wanted to confirm anyway. Is the index pulse is wired correctly and will the machine is able to home properly as it is wired now.
    I'm still a bit stuck in tuning the axis yet (and lots of other things to do......) so I havenít had time to look in to this in detail yet.

    Anyway, I hope the picture comes through with enough detail so you can see my marking properly. I'll send you the original scan Mark, so I know you get all the details you need.

    What you do now then is to connect 12V power to the 2 connectors highlighted. These wires are marked as 0,5mm2. This is indicating that they are power to the unit. Then you use the scope you got from your work and try to monitor the outputs marked. You need to move the axis while you are doing this in order to get readings. And you should take care not to short the outputs (they should be able to handle it, but there is no point in taking chances here).

    By doing this you should be able to find out what kind of signals is coming out of this.

    I was able to determine the resolution I got when measuring. But that require the axis to be under control of a controller capable of moving it at a steady and known speed.

    If the signals coming out of your encoders are something other than TTL level digital signals (like 1Vpp sine or something) you will need to find out more. But if you find it has something like TTL level signals you could try to connect it to your Galil card.

    BTW, you do know that all the two letter markings on the wires are markings of the color of the wire, to make it possible to follow like I have highlighted?
    On page 99 on your manual these markings are in fact missing, thus making it more difficult to follow. So it's not 100% sure this markings I have made are correct. All I had to go on was the naming of the signals that seems to be correct. It also fits with what I know to be working with the EXE boxes as far as what signals is going in to the controller (but I have now rerouted to the Galil card).

    I hope you have Galil terminated wires or something BTW. Those 3 row D-sub connectors used on there is a ****** pain in the *** to solder on. In particular when you try to correct mistakes and switch around on cables..... not DIY friendly IMHO. MESA has a better solution there.... in my opinion anyway.....

    Anyway, I hope you are able to figure it out. I do still have some hope you'll be able to use your encoders without too much fuss.

    And let us know how it goes!

    Anders

  13. #92
    Join Date
    Aug 2010
    Country
    AUSTRALIA
    Posts
    82
    Post Thanks / Like
    Likes (Given)
    1
    Likes (Received)
    3

    Default

    Hi Anders
    That sounds good, all I have to do now is learn how to drive the scope?
    I will also have to undo the covers again so I can get at the X axis drive not sure how I will go spinning it by hand at a steady rate as you suggest though.
    I will leave the covers off this time.
    I assume that I will need to connect the scope up to Va2 and Va2 with the bar over the top to get a reading and then do the same for the Va1s?

    Regards
    Mark

  14. #93
    Join Date
    Jun 2009
    Location
    Norway, Vestfold
    Posts
    181
    Post Thanks / Like
    Likes (Given)
    5
    Likes (Received)
    5

    Default

    Hi Mark,

    Using an oscilloscope can seem like a challenge yes. But for what you need it shouldn't be to hard. All you need is to look at the shape of the signals coming out of your scales so see if they are sine or square shaped. And then try to get an idea of the levels.

    I hope you got some good probes with you when you got that scope? If not you should ask to borrow 2 from the one who gave you the scope. Alternatively you can buy some cheap online. But that will take time to get.....

    BTW, all you really need to do to measure is to find the connector coming from the connector that used to go in to the separator card. You don't need to physically measure on the scale itself. So you shouldn't need to remove the covers. Not to do this measurement anyway.

    And what you do is to (always) connect shield of the scope to 0V, and then put the probe to one signal after the other. But only one at a time. If you use 2 probes at the same time you should be able to see 2 signals at the same time. This way you can verify if they are opposing (like 500 and 500N should be). You should also be able to verify they are 90 degree off (like 500 and 590 should be). This last information is also of some importance because this will tell if the signals are quadrature encoded or not.

    The easiest axis to measure on would probably be the Y-axis (Z-axis in Maho terms), the ram. I expect you are able to reach the servo on the back there. You might have to climb on or in to your mill to reach it, but I expect it to be the easiest one to reach. On this axis the servo motor is also moving with the ram, and if you move it some of the way out you should be able to have room in there to work.

    What I did was to make an adaptor that has two fingers that is spaced to enter two opposing holes in the axis clutch mechanism (the big belt sprocket on the screw). I used a block of wood because of availability then and there (and havenít bothered to make a new and better one later). So then you have a piece of wood or steel with two screws or something that will fit in to the belt sprocket (which the clutch is built in to). Then from the other side you put in a third screw so it sticks out the other way from the plate. This should fit on to a standard battery drill or something. The middle arm (screw) should preferably be located in the center of the thing to make it turn straight.

    With this I was able to move the Z and X axis quite efficiently, but not the Y-axis. The belt and sprockets on this axis is under the machine and isn't possible to move with this device. I also used this when cleaning the ball screws on x and z.

    Anyway, for now you really don't need to have accurate speed. All you need now is to move it and verify the shape of the signals coming out. If they are 0-5V TTL (square wave) you are good. They could also be 12V signals given they use 12V to power the devices. The separator card might be a 12V to 5V converter as well. We should be able to know this from what you find.

    Pictures of the screen is a good thing to help here. And be sure to get details like how the knobs are adjusted as well. The time base isnít to important, but the level is. This is what will tell us what level the signals have. And anything from 1V sin to 12V is what we expect.

    You mention you had a friend/colleague who was an electronics engineer. This would probably be a good time to ask him if he could help you measure. He should have a basic knowledge on how to make the oscilloscope show what kind of signal the encoder is putting out. You will probably need someone to help you with this anyway. One person to turn the axis and one to measure.

    I won't be home this weekend (I will be in a cabin in the mountains). So I would be online until next week. In the mean time I hope you are able to get some measurements done.

    I hope to have the time to get back to the other topics being discussed here as well. I really appreciate everybodyís interest in this discussion.

    Anders

  15. #94
    Join Date
    Aug 2010
    Country
    AUSTRALIA
    Posts
    82
    Post Thanks / Like
    Likes (Given)
    1
    Likes (Received)
    3

    Default

    Hi ALL
    Thanks for that Anders have a good weekend, I also have to go to Sydney tomorrow and work on my classic motorbikes on Sunday so will not have a chance to test the outputs.
    I have Googled some info on oscilloscopes plus also spoke to the electrical engineers but they keep going over head when at this stage I only need the basics,
    I have taken on board with the Z axis and yes this should be the easiest axis to try.
    Looking forward to testing.
    regards
    Mark

  16. #95
    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

    Maybe asah could give info on how his servo drives are connected.. It is sad that you discounted linuxcnc on its TP grounds... (ihavenofish might be coming back around to linuxcnc.. maybe) There was a nice discussion on ******* where I think the issues he was having where figured out. (plus I hope he test the new trajectory planner when he gets going) there are at least 3 people that I know running mahos on linuxcnc...

    sam
    Itís true I made the decision to try other solutions on grounds of the TP yes. But I already had most of the hardware, so I have kept an eye open to see if there is any progress in that regard. And Iím pleased to see that there have been! As you know it looks like Ihavenofish also consider coming back to LinuxCNC.

    Anyway, I have to admit it wasnít only the TP that made me look elsewhere. The work involved in configuring the machine I want is also quite daunting. In particular having to make the logics involved in the ATC is somewhat scary. And put together it made me look in to other solutions. But also the inverse kinetics for 5 axis is making me unsure.

    The truth is though that I have come to learn that the other solutions I was looking in to have its own issues. So when I found out the TP issue in LinuxCNC now seems to be sorted I decided to consider looking at it in parallel with the Camsoft/Galil solution.

    Another issue that also helped me to decide to put more effort in to it is the fact that there seems to be some proper progress in making a new version that is using a modern version of Ubuntu. I know there are arguments of not using a new one, and the ones used are LTS. Nevertheless the 10.04 desktop is already discontinued (one year ago) and only the server version is maintained.

    There have been numerous discussions in forums previously about this. All I can say is that regardless of the fact that the version used at present might be ďgood enoughĒ, and possibly even better suited in some technical way it feels somewhat ďwrongĒ. The truth is Iíd like to be able to use a more ďpresentĒ version of Ubuntu.
    After all I will need to dig in to the new OS, and the feeling of not being able to do that in a current version makes it all feel a little less satisfying. I would like to consider the experience I get from using the OS valuable for use in other ways as well. And for a lot of users this will be their first contact with Linux.
    Iím just saying that in order to attract new users forcing them to use an old and somewhat outdated version of Linux might not be the best idea. I donít pretend to know much about this here, so take this as thoughts from a potential user only. And this isnít from a purely technical point of view. But psychologically it doesnít feel as much worth the effort when you are forced to use old versions of the OS. Iím just putting it out there.

    AnywayÖ.. What I have done now is to set up 2 computers.

    One with the Galil card in it, and making all the connectors I need to interface the mill in that one box. This will be used with Camsoft primarily, but could also have Mach3 installed.

    The other have the same computer hardware (somewhat different CPU though) and Mesa hardware. This has the same interfaces to connect to the mill.

    Since I broke the EXE converter in my 432 controller I havenít been able to connect that one back. And the first ones I bought didnít have the same resolution. So I couldnít use those on the 423 controller. But now I have bought 3 off them with the same resolution and hope to be able to connect it back now. I will install these permanently in the mill now so no matter what controller I decide to use I get normal quadrature signals from all encoders.

    So my mill will in fact be able to be run with 3 different controllers. 2 new computer based ones and the original. This makes it possible to do some direct comparisons.

    Not sure if I want to spend the money to get a full version for Mach3. But it should be possible to test it in the trial version anyway. Not sure itís worth the effort though. Iím not using the I/O on the Galil (which isnít enough by farÖ so I had to buy PCI cards for that, and I donít think Mach3 support themÖ.).

    BTW, I see there are new user interfaces for LinuxCNC now as well. That is a good thing, and also a contribution to me to want to come back to it.

    Things are really starting to look good now. All I need all I need to do now is wait for version 2.6 to come out. For now I donít feel like installing the developer version with everything involved. So I can only hope the new version is coming out soon.

    The todo-2.6 list looks long, but I have no idea if this will take a long time. Anyone here dare make an estimate on when it might be out?

    Anders

  17. #96
    Join Date
    Jun 2006
    Location
    Munich / Germany
    Posts
    2,338
    Post Thanks / Like
    Likes (Given)
    342
    Likes (Received)
    1068

    Default

    Re the outdated OS:
    I don't need the latest and greatest, I need something stable. I don't care if the OS supports twitter, chatter, facebook or arseTV. It just has to run my mill.
    I also don't care that my Haas lathe has an OS from 2009. She just runs.

    I don't do any updates on the OS of my mill. That's not an experiment, that's a tool.



    Nick

  18. #97
    Join Date
    Jun 2009
    Location
    Norway, Vestfold
    Posts
    181
    Post Thanks / Like
    Likes (Given)
    5
    Likes (Received)
    5

    Default

    Outdated OS:

    Nick, I understand what you are saying. And I would probably do the same as you, leave the computer with the software as it is if it is working as I want. And I wouldn't put effort in to upgrading it if there is no need for that.

    BUT.... Now I face the start of a learning process. I need to learn how to install and configure a new OS. And what I have at hand as far as hardware is pretty much brand new. It has obviously new BIOS and all of that. Who would think it’s a good thing to start with a 5 year old version of the OS then?

    And with a machine like this it's not just a matter of patching it together and think you have it working. I want to be absolutely sure I can trust it to behave.
    Is the only way to make sure of this to also use old computer hardware as well then? Then count me out.

    Can I be sure all parts of the hardware are supported if I install that version of the OS? If not then how can I get help? The version is no longer officially supported (or are you using the server version now?). These ARE valid questions for those among us considering trying LinuxCNC.


    Or can someone say with absolute confidence I will not have any problems at all with the BIOS of my new main board with the present versions of Linux used in LinuxCNC?

    I understand you guys that already have a working system. But consider the ones who are contemplating on starting fresh now. There will be others but me who will be thinking it might be a good idea to wait until a new version of the OS is supported. There were discussions on this way back....

    Besides, I will not even consider installing LinuxCNC without the new TP and the other new updates. So I guess I will look in to using the development branch then....

    BTW, what I really meant to say is that I see there is really good progress now with the new TP, and it’s really looking good. This is also why I decided to start looking in to using it again. So I only hope the new version 2.6 isn’t too far ahead. I would feel much more comfortable being able to install a released version with everything already working, and support for modern day BIOS.

    Anders

  19. #98
    Join Date
    Jun 2006
    Location
    Munich / Germany
    Posts
    2,338
    Post Thanks / Like
    Likes (Given)
    342
    Likes (Received)
    1068

    Default

    Anders, to some extent I do understand your worries.
    Yes, my system is "old".

    In the end, it works out to be "just try it!". If you can boot into the demo CD and LinuxCNC comes up, I think you already won. The cards you will install aren't unpredictable. I'd go MESA only. They do work, they do have the best support with LinuxCNC. And they do work with that old OS.

    I am using an ITX mini Intel board (I'd only use Intel in that case) with an Atom (dual core) processor. You need two cores. The board is/was quite cheap (about 80 €) and you don't need more. I do have a 5i20 and two 7i33 attached to it. Plus some more electronics.
    I think that gives you more than enough IOs for the most complicated tool changer. You'd have to add a component (software) to handle all the states and fault conditions. That might be the most challenging part, depending on how good a programmer you are.

    What you want is something stable and proven, nothing fancy.

    BTW:
    I had the 2.5 installed for some time, but then went back. I simply didn't want something in beta state. I had the 2.5 to try out the new interface (forgot the name; someone from Germany making it).


    Nick

  20. #99
    Join Date
    Aug 2010
    Country
    AUSTRALIA
    Posts
    82
    Post Thanks / Like
    Likes (Given)
    1
    Likes (Received)
    3

    Default

    Hi
    Anders
    Got the bikes sorted so have a bit of time to update.
    I have designed a tool to move the X axis as Z too hard to get at, too many parts to remove.

    To get at the X axis all I have to do is hole saw a hole in line with the lead screw and I will be able to drive with a drill from there.
    I also have sorted the scope out ready to connect.
    I also located the old Philips control it was in the wifes garden shed.
    I am taking a week off come Easter to sort out the scales all going good.
    Best regards
    Mark

  21. #100
    Join Date
    Jun 2009
    Location
    Norway, Vestfold
    Posts
    181
    Post Thanks / Like
    Likes (Given)
    5
    Likes (Received)
    5

    Default

    Quote Originally Posted by Burgs View Post
    Hi
    Anders
    Got the bikes sorted so have a bit of time to update.
    I have designed a tool to move the X axis as Z too hard to get at, too many parts to remove.

    To get at the X axis all I have to do is hole saw a hole in line with the lead screw and I will be able to drive with a drill from there.
    I also have sorted the scope out ready to connect.
    I also located the old Philips control it was in the wifes garden shed.
    I am taking a week off come Easter to sort out the scales all going good.
    Best regards
    Mark
    OK!

    I was just about ready to post about this when I checked to see something and reloaded the Maho forum. And I saw your post here. Good thing I checked this then!

    And it's fantastic news that you were able to locate the old controller! That is quite possibly the only way for you to be able to use the scales on your mill. If you look at Nick's post (#25 in this thread) you see he has already made some scope pictures of what the signal from Philips scales looks like. And as he also concludes there is no way it's feasible to make a converter for these signals.

    It also means it will be a lot easier for you to check, and you don't need to cut holes to get at your encoders.

    What he didn't think of was to reuse the interface card already in the controller. I happen to know the encoder inputs on the 432 controller needs normal quadrature signals. At least that is what mine is using. At first I made the assumption that my controller had a translator, and sure, that proved to be correct. So now I know for a fact that this is what the controller is using, and it seems to be the normal way for these controllers to use something else and then put translation interfaces between the scales and the controller. And to my luck the translators to my scales were even readily available at very decent prices on eBay! Yours on the other hand seems like they are almost impossible to get. This is where your old controller comes in. As mentioned I'm quite confident we can assume you also have a translation card in there. It should be easy to check. It is possibly one of the only places to find a translator for your scales. This is why it was so important for you to find the old controller. BTW, the rest of your controller is possibly worth some money, in particular if you have the CPU card with one of the latest software versions. These are almost impossible to find these days, and should fetch a decent price if you sell it. I might also sell mine at one point....

    I assume the translation card is the card in row 4 in your controller, to the right of the 3 I/O cards. That is where it is drawn in your schematics anyway. If it's like on my mill it's a standalone card (meaning it's attached to the controller but only as a place to fit it, all connections are on the front). The scales are connected to short cables coming out in the front of it (it is on my EXE card anyway, and looks like it is on yours as well from the drawings). Then there is another short cable that goes from that card and over to the driver interface cards in row 12 and 13 (one for each scale). On page 111 in the schematics it is the cables that go in to connectors 19X9, 19X10 and 19X11 that you want.

    These connectors have the quadrature outputs you want! These could be interfaced to the Galil encoder inputs directly. All you need to do is connect it up correctly. If you take out the card it should be easy to follow the cables and see where each is going. On mine the card was in fact 3 separate EXE units made as one common circuit board. There should be cables with connectors matching the scale cables also comming out of it. So all you need to do is connect these back and test.

    You will use this card exactly the same way I use the EXE cards.

    It could possibly be a good idea to feed +12V and +5V to verify this first. It would be best to be 100% sure this is how this is working before putting effort in to interfacing it in to the Galil card. Use the drawing I made a few posts back as guide for what cables goes where.

    So this could be the one thing to get you going! For your sake I really hope so!

    I'm almost done with terminating the cables in my LinuxCNC box now. After that I will make some small corrections to the CamSoft box (which is built almost identical so it's possible to interchange them, some small adaptations I have to make first though). I will have a week off during this Easter, and I hope to be able to get some decent progress during that time.

    I will make it possible for LinuxCNC to read the encoders when the other controllers are connected. Have seen that done to check differences between LinuxCNC and Mach3. Now I'd like to do it to check for differences between all my controll systems including the 432 controller. That way I will be able to check for differences between the original controller and the modern PC based ones on a very detailed level, down to tracing minute differences in movements. Then it will be possible to document the differences very detailed! It requires some extra work, but I hope this is worth it. It will also make it possible for me to get detailed information as to how the original controller is tuned and what accelerations it is set up to have and such.

    I also ordered a 8I20 amp to see if it's possible to use that. But that will have to wait..... I want to make it run first.

    So we should hopefully both have some decent progress in the next couple of weeks.

    Anders


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
  •