|Nigel B||05/10/2016 21:08:52|
|412 forum posts|
Siemens controller....................However they are a very good and robust system.
Awful things. If you think Linux CNC is a bear to get up and running, you would be in for an unpleasant suprise if you tried one of Erlangen's "finest". On the one I am lumbered with, it's a bit of a lottery when you press the start button as to whether or not it will deign to start - or keep going if it does start. "Very good and robust" it is not !
Did you find the link to the Chinese turning control manual you mentioned earlier, John ? I would be interested to have a look. Do you have details of the shortcomings of the Chinese USBCNC boards mentioned a few posts up ?
I have one on order (a whole £17 delivered - what have I done ) to try with the demo software. Apart from the Planet CNC denigrating of them, I couldn't find anything specific elseware. For a mill this seems a reasonable solution - at least it supports tool radius comp - even if turning isn't fully supported yet.
|John Stevenson||05/10/2016 21:50:53|
5068 forum posts
Nigel, Sorry forgot you.
This is the most recent revision and has been altered from Steve Blackmore getting his Orac up and running..
Shortcoming for the USBCNC board was that the Chinese ones were a version or two behind and none of the Chinese ones do lathe.
Can't comment on the other features for mill but dare say they work.
Might be worth looking on the Planet forum for any know bugs
|sam sokolik||06/10/2016 16:44:32|
|25 forum posts|
I have a feeling that linuxcnc has a greater user base than you think...
The linuxcnc forum satistics.. (the forum hasn't been around as long as mach)
Look at the youtube videos sorted by upload date. Lots of activity. As of now - someone posted a video within the last 24 hours.
I think more than 100 people are using the advanced features of linuxcnc (it isn't as hard as people make it out)
|John Stevenson||06/10/2016 17:13:31|
5068 forum posts
|No Sam you misunderstood. There are probably thousands using Linux but possibly only a hundred or so able to delve inside and change the code. |
Again I'm not bashing Linux, far from it but why can't the masses have access to an easily set up program with a workable GUI. None of the screens I have been pointed to are viable screens to put on a machine to sell to a begineer.
By your own admission it's not easy to incorporate path pilot with Linux, even though path pilot IS Linux. Does that mean the Tormach guys are better than the Linux guys ?
Because they have done it.
|sam sokolik||06/10/2016 20:16:20|
|25 forum posts|
Couple of things...
1) I have not changed the source code of linuxcnc for any of my machines conversions. I used the integrated ladder and the configuration files. (that says a heck of a lot about the flexibility of linuxcnc).
2) I don't understand why 'none of the gui's' are usable by beginners. They obviously work. It really is your opinion. (and I am sure others). The axis gui that you so hate is all I use. What is it that you don't like about it? it resets the estop, turns on the machine. homes, loads gcode, edits tool table, touches off the coordinate system, runs the gcode and so on. (I think it is a clean interface - but that is my opinion)
3) I am not understanding your last point either. Tormach took linuxcnc and to use a geeky term - 'forked it' They then modified it to their liking. (Because it is open source) They added stuff and removed stuff modified stuff. The side effect of this is the changes they made make hard to bring their gui back into linuxcnc. This is good from Tormach standpoint. They for obvious reasons want to keep their gui out of the main stream for as long as they can. There has been hints that someone is working on integrating tormachs gui back to linuxcnc but who knows how long that will take. The bad thing for tormach is because they based their work on linuxcnc - any changes built upon it needs to be released (which they have). So it is probably only a matter of time before someone works it out.
|John Stevenson||06/10/2016 22:34:59|
5068 forum posts
I'm sure it's not me.
I usually go visit someone who wants a machine converting and tell them what their options are and what it will cost.
I then tell them what is available as regards software. Up to a few weeks ago this was Mach3, Linux and Eddings CNC.
They look at the options and usually choose Mach 3 without any prompting. Possibly because they have seen it before ?
No one yet has chosen LinuxCNC, when I ask they seem confused as to how to operate it. A beginner expects buttons to press which frankly are too well hidden in Linux CNC. So you say it's a clean interface but that may be the problem, it's too clean ? However we both have our views and I dare say nothing will change this and I certainly don't want to change yours but I would like the option of offering a version of Linux that the average beginner can use.
This is liable to change though in the very near future in a way not envisaged.
Two weeks ago I was approached by one of the university's at Oxford [ contrary to popular view there are 100's based in Oxford ] Over the years I have done a lot of business with them buying up old Denfords and Boxford CNC's when they have upgraded to newer. I still have 8 or 9 left in the hayloft.
So I was expecting to be asked to quote for clearing out some old non working machines but had a very pleasant surprise when the guy explained that yes they had some, in fact they had also been stock piling them as he knew the time would come when it would pay to upgrade instead of buying new.
I then explained that we could offer the Chinese stand alone controller and this guy wanted to know more and it looks as if end of November, beginning of December it's 10 days down in Oxford to convert 15 machines.
Now these controllers are ARM based, complete stand alone, no PC, no monitor, no keyboard or mouse but they use a version of Debian, which is Linux.
So first Tormach get it working and now the Chinese has it working for less than two Mesa cards.
3637 forum posts
Oxford university has to get someone in from Notts to set up a computer They never were a patch on CMBRIDGE
|John Stevenson||07/10/2016 00:06:27|
5068 forum posts
No Oxford has millions of people who can use computers and finger tips. they just lack people who can use hands
|Rod Ashton||07/10/2016 08:15:25|
|186 forum posts|
Nigel - I hope you will keep thread readers informed about your experience with your £17 investment. Would make a very interesting post I think!
|138 forum posts|
Don't forget that without the 69.00 Euro license fee Nigel is restricted to 25 lines of Gcode.
You say that 5i25/7i76 is an over the top combination and go on to mention that a 7i92 is all that you need, yet this is the combination most mentioned anywhere I look for information on what Mesa boards to use.
I have re-read the description and manual of the 7i92 and yes, it would appear that it can be used stand-alone parallel port replacement - it would still need some form of breakout board even if it is a basic $4.00 Chinese one, since most home-brew conversions aren't parallel port ready
This also re-affirms my earlier point about it being difficult for someone new to work out what Mesa cards they need - no guidance on Mesa's own site, the LinuxCNC Wiki is amazingly vague on the subject, merely mentioning a large number of possibilities that should work and doesn't even mention the 7i92 card!
If I specifically search for the 7i92 card, I see very few posts about it, and many of those I do find are from the LinuxCNC forum which seem to be cries for help with using one - hardly a great advert.
To buy one I would need to order from the US, not necessarily an issue, but Mesa quote 6-8 week lead times, EUsurplus have no stock, and I have yet to find a UK supplier.
I like the idea of this board in that it de-couples from the required PC, making some form of laptop more of an option, or a mini PC that can fit behind the monitor.
One thing I am not so keen on with the Mesa breakout boards is that their screw terminals are too close to each other to use crimped ferrules, and pretty awful from a strain relief point of view - 90 degree plug in screw terminals would be a better option in that regard.
Edited By Zebethyal on 07/10/2016 09:33:53
2465 forum posts
Well those are the wrong crimps. The approved terminations would be "bootlace ferrule" crimps such as this from RS, although you can get them from CPC and other outlets. IIRC, you want the 1mm2 size (red) for those terminals, which fit easily into the space. Whoever terminated those wires used the wrong crimps. You will find pics on the LinuxCNC site if you have time to kill. You can also prewire the terminals and then plug them into the headers of course.
These terminal blocks are designed to take stranded wire directly but if you choose to do so, do NOT solder tin them first. Solder suffers from "cold creep" when under pressure, resulting in intermittent connections after time. Not a fire hazard in this instance but it can be if you are talking about higher voltages and currents.
As far as I know, you can only buy Mesa boards from 2 outlets in the US, plus Eurospares in Portugal. Availability seems to be hit and miss to say the least. I guess I was lucky.
|J A Harvey-Smith||07/10/2016 11:30:26|
|8 forum posts|
A very interesting thread, thanks to all contributors.
In my workshop LinuxCNC is the clear leader at the moment but yes, I do find it tedious getting the setup right and the keyboard and mouse are a pain. I ditched Mach3 the last time the KX3 mill caused grief, wasted a little time playing with Arduinos and Pis, but produced some seriously capable hardware using cheap ARM Cortex boards using the free mbed.org online compiler, along with Altera FPGAs. However, this was all for academic interest as I was never going to write software to create a full CNC controller, I don't have the years left and no desire to reinvent the wheel.
I work in an engineering workshop of a heritage railway. Last year we got a "new" (which in heritage railway parlance means post WW II) Harrison 500 CNC lathe. Lovely !! - for a week until the electronics failed. I'll steer clear of railway politics, but a year on it ain't fixed. The computer part of the control electronics was a DOS pc with obsolete proprietary plugin boards, this no longer talks to the considerably older servo drives which take +/- 10V signals to control axis velocity, position encoders feeding back to computer to close the loop. The spindle and axis encoders are good, the servos work (as proved using AA cell as signal source). I would not waste my time trying to persuade management to let me loose on the machine with a LinuxCNC system - nobody but me would ever use it (or feel confident with it), but I'll be reading through the manual for the 990TDB which (unless I've missed something obvious) just might be sellable to them.
If anyone can spot why this won't work, please advise. Thanks again.
2465 forum posts
Unless I've misread the manuals I've got for these, they are intended to generate step / dir position outputs rather than analogue servo (+/-10V) velocity outputs. They tend to have one or two analogue outputs as well but these are for single or dual spindle speed control (ie to drive a VFD), rather than axis position control.
If you want to use a modern controller with analogue servos, I believe you will need a "positioner". This takes the step / dir demand information from the controller and the positional feedback from the encoder to generate an analogue drive (error) signal for the servos. There is a PID controller in there typically. The servo motor will probably incorporate a velocity sensor as well as an incremental position encoder but this typically provides feedforward info into the servo drive, not the positioner.
Somebody may correct me but that's how I understand it t the moment.
I suspect your Harrison has just converted itself into a very large piece of scrap!
Edited By Muzzer on 07/10/2016 11:57:01
|Jat Grewal||07/10/2016 11:56:37|
|10 forum posts|
Been reading everyones comments and its such a great feedback.
Keeping things simple has been the key issue and also getting rid of the computer based OS. I have gone through the same issues while making the CNC machines for hobby.
And thats why I started making a non OS controller that is only made to run a CNC machine. Having custom built hardware means full control of hardware design and custom software that for the hardware.
I will appreciate everyones views and comments as I would like to keep adding more features to the design.
|138 forum posts||
I agree with you entirely, and those are the ones I use, albeit from a far eastern source - the picture was an example to show the limited space.
With SmoothStepper being available from a number of UK sources, including some on the right hand side of the board, I would say that unfortunately this is another point against LinuxCNC.
Looks like if I decide on the MESA route it will be another holiday purchase to bring back in the suitcase, if I can acutally find some one with stock
|John Stevenson||07/10/2016 12:58:25|
5068 forum posts
Jat, Definitely an interesting concept but feel that you are at a halfway house stage and not meeting goals.
You say no PC needed but you need everything else like monitor etc.
If you went to a stand alone controller you wouldn't even need that.
But if you kept the PC then you could have your CAD?CAM system of choice on the same PC.
I'll grant that with a stand alone you don't get a CAD CAM but you don't have the clutter.
Seems to me like you have a foot in both camps but then don't know which way to step. I did mention that it's expensive to say you still need all the extra's and from looking at the kickstarter project it seems that I'm not the only one thinking this as projected figures are saying you won't make half of the goal.
|J A Harvey-Smith||07/10/2016 13:30:21|
|8 forum posts|
I was briefly clutching at straws, thinking I might donate a new controller for a couple of hundred to make my working day a bit more interesting and productive, but I'm not going to fork out on positioners to talk to 30+ year old obsolete analogue servo drives on top. A proper job could be done replacing servos and motors, and power supply - expensive items to suit this big machine - and a couple of hundred soon turns in to thousands - but not from my pocket. As you suspect, a very large piece of scrap.
|Nigel B||07/10/2016 13:47:57|
|412 forum posts|
The Chinese stand-alone control looks like it would do most of what I would want a turning control to do.
The area clearance & threading cycles seem to use the Fanuc 0T appproach, having the cycle parameters entered over two lines. I was a little amused to see that their diagrams for the cycles (and TNRC) appear to be "borrowed" directly from the genuine Fanuc manuals - that's one approach to "Fanuc compatibility" !
Their implementation of the G31 "Skip" function appears flawed and rather pointless without an equivalent to "Custom Macro B" (parametric programming). No dedicated high speed input(s) for probing & no ability to manipulate data in the program means no automatic tool offset setting or part probing routines & possibly dubious repeatabilty.
I am suprised to see no "Control operational" (or Watchdog) output (same goes for the Massa here). I would not be happy to fit a control that had no form of internal hardware & software monitoring to any machine in an industrial or educational environment (you are a braver man than me if that is what you are proposing, John). I appreciate that these are "open loop" controls (no position feedback from the axes) & may be less likely to run away if a fault developed, but even Mach 3 had the "charge pump" arrangement to drop the drives if there was a computer glitch.
I will report back on the Chinese USB board when it arrives. I had seen the 25 line demo limitation (and the release code cost is more than €69, as 22% Slovenian Vat has to be added to that). While it is, in effect, a GRBL on steroids, it looks like it should be possible to program straightforward 2 1/2 D milling without having to go down the CAM route that GRBL would require, as GRBL doesn't support tool radius comp.
As it stands at the moment, the Massa is too expensive (for me) for what it offers - I look at it on the basis that the functionality is what it ships with now, as promises of future functionality are just that & may or may not come to pass. Going to a subscription model to retain the functionality on a month-by-month basis to fund (and maybe improve) the functionality would kill it completely for me. The genuine Planet CNC 4 axis board with spindle control sub-board & software licence is around £280 inc. Vat delivered from Zapp Automation - my opinion is that when Massa reaches a similar level of functionality, it should be much nearer that pricing than it is at present to be sucessful. SWMBO would take a lot of convincicng to allow me to drop in excess of a week's gross salary on a circuit board !
Haven't looked very deeply at Linuc CNC - both the desktops I tried didn't do very well on the latency / jitter front, so I got no further, though I may try the DAK Engineering TurboCNC Dos solution on one of those.
2465 forum posts
The concept of using an FPGA to generate the drive signals avoids the need for good latency which you would otherwise need if you use the parallel port or USB. That's the basis of the Mesa cards. You'll notice that the Chinese controllers also use FPGAs. Unless you are making a hobby CNC machine, that approach is almost certainly required.
Jat - what have you got in your controller? - a microcontroller and FPGA or just a DSP? I can't make out any part numbers and there aren't many views of the PCBA. I can see a bit of a large QFP device but not much more.
|John Alexander Stewart||07/10/2016 14:48:09|
|658 forum posts|
The last Mesa product I purchased (5i25, 7i76) came from www.mesaus.com, and was shipped internationally as requested and arrived very promptly. No personal interest in the company, just passing along my experience.
(ship date was 29 August 2016, so not too long ago)
Please login to post a reply.
Want the latest issue of Model Engineer or Model Engineers' Workshop? Use our magazine locator links to find your nearest stockist!
Love Model Engineering? Sign up to our emails for the latest news and special offers!
You can contact us by phone, mail or email about the magazines including becoming a contributor, submitting reader's letters or making queries about articles. You can also get in touch about this website, advertising or other general issues.
Click THIS LINK for full contact details.
For subscription issues please see THIS LINK.