Arduino project with stepper motors

.

Ian,

Please check your messages; when you have a moment.

MichaelG.

For small steppers the 'Polulu' board (developed for repraps is excellent and easy to control.

Neil

Hi Neil , they are ok for the small reprap nema14 motors but not for nema23 motors which would be needed on a lathe or mill , i use the nema23 on my repScrap which started as a reprap and is now evolved into a full metal constructed cnc engraver, I use the cheap chinese driver board as it uses the toshiba td656070 (4a@50v rated) on a huge heat sink. Only down side on reprap was they got a bad press as they need a much longer step pulse duration of 30ns than the pololu 5ns and most people could not get them to work on stock reprap, one oscilloscope and little pulse stretch and they are fine.

cheers ian

I don't think you mean ns! Microseconds perhaps...

I'm quite taken by the Chillipeppr / TinyG system. Chillipeppr is a browser based GUI and the TinyG is a motion controller that apparently has some clever look ahead functions. Nice example of some of the neat developments coming out of the CNC printer / router world. Possibly a little piddly for you, Ian but interesting to follow.

I'm going with the Mesa 5i25 / 7i76 combination and DMM Tech servo motors driven by LinuxCNC. The Mesa stuff is remarkably good value for money when you think how much a parallel port breakout board costs and then stop to compare the specs.

Looking forward to resuming this and getting it all running when we finally move into a house of our own again. Obviously I'm eyeing up houses with unfeasibly large garages / workshops with that in mind...

Murray

Edited By Muzzer on 17/07/2015 22:40:02

Hi John  yep it should say micro , looked back at data sheet for td6560 it's tw (clk)min 30 micro sec, so pololu is min 5 micro.sec

Edited By ian slaney on 18/07/2015 00:17:26

Ooh Dam tech they do look good but too expensive for me to think about.

Hi Muzzer

I also used Mesa/DMM/linuxcnc on my mill and it works a treat. Brilliant actually.

I did make a stupid oversight initially where I didn't fully protect one of the servos from the suds (was under bellows and part shrouded) and the encoder went haywire which cost me dearly. Obviously the DMM motors are not waterproof, nor would I expect at the price, so I added additional shrouding and sealant. That not going to happen again.

So would highly recommend

Cheers Will

Hi Will

Very interesting. And useful feedback on the coolant ingress - I will learn from your experience!

Did you use the 7i76 (stepper) or 7i77 (servo) card? I'd already bought my Mesa stuff before I changed my mind and bought the DMM motors instead of buying 2 more Leadshine closed loop steppers. I'm prepared to get a 7i77 if that would be best but as you know, the DMM system can take a step/dir input and my completed Z axis already uses one of the Leadshine motors.

I'd love to see some pictures of your machine - anything you can point me at? Mine's a Bridgeport clone BTW.

Although there are some pics in my albums claiming to be views of the X and Y axes, in fact I changed from direct drive of the ballscrews to 2.5:1 reduction toothed belt drives when I redesigned them for the DMM servos due to their different (better) torque-speed characteristics. Actually a much neater result and the majority of the parts are now finished.

I must add some photos of the real articles to my albums but the existing CAD view of the Z axis is a fair representation of the finished assembly.

Murray

<Nerd mode>

The A4988 (which is the chip on Polulu) has minimum high & low times of 1uS, so it's actually 2 usec.

</nerd mode>

Neil

Hi Muzzer

First the warning - I'm a hammer yielding mechanical engineer; the finer points of CNC control are certainly beyond me, so I arrived at a working solution by good luck and perseverance rather than skill !!

So to answer the MESA question, I don't use either to control the motors. I used the DMM controller board for the motors and the emergency stop and fault functions, which seemed an easy first step, plus a Mesa 5i20 inside an old HP Pentium computer feeding a couple of MESA 7i37's for the rest of the i/os.

I suspect using the DMM board, which uses DIR and STEP is doubling up on some of the capabilities of the 5i20 and probably compromising the performance, but it works really well for me, so I don't really understand it so have not tried to optimise it - let sleeping CNC machines lie is my adage.

The base machine was a Defiance Innovations small high speed CNC mill, an American company that went under in the early naughties, so no support and next to zero information is available. It was a capable machine originally, fitted with some good stuff, but I ended up replacing the motors, all the control system and even the spindle (field burned out, so perhaps one day I will have it rewound and get auto tool change back) plus totally rewiring. OK, I admit, it was mostly for the fun of it.

The photo below shows the machine (note this morning I removed one of the bellows for repair as it had become torn and I am paranoid about keeping motors dry now).

The nice thing about the Defiance is it's an accurately built machine with proper ball screws, slides and lubrication and even with the refit it is cutting very accurately. I run it quite slowly though, with the maximum velocity at 500mm/min and generally cut no more than 350, which means that so far I have been able to run it off a 13A domestic socket without having to put a bolt back in the plug fuse holder ***

Linux was OK to set up and once I'd got my head around how the program was structured, the I/Os were straight forward to add. You are more than welcome to have set up files if you want them.

Using the machine is easy - I also use Fusion 360 to do the CAM bit and it writes G code in almost perfect LinuxCNC without any code manipulation, unlike another big CAM system that I previously had access to. I never leave it running alone, but It's got to the stage where I trust the code and I'm so confident I don't normally do a dry run of new code, just let it go straight into cutting metal. You watch, later today, having made that claim, I'll plough a big jagged channel into the bed ...

Lastly, I am aware that I still have not got adequate fault protection and it's surprising how many IOs are required for a half decent set-up. Obviously the DMM board has a watchdog and other fault protection, but I still need to be more clever on the spindle side. If this stops rotating for any non obvious reason, I will be caught out and the Chinese spindle does not have sensors as did the original. I am starting to work out what signals I need from the VFD to catch more faults, which I will sort shortly.

Cheers Will

*** DEAR SAFETY ELVES, ONLY JOKING ABOUT THE BOLT, HONEST

Hi Will. You seem to be very self-effacing. Either way, that's a nice machine whether a metal basher or not! Handy that you were able to find a proper CNC machine to start from rather than converting a manual machine like me. I have to say, I have struggled to bottom out all that HAL configuration stuff, so when I get to that stage I may benefit from copying your homework.

The link from the 5i20 / 5i25 to the 7i76 etc daughterboard allows for higher bandwidth comms and of course the 7i76 contains a blistering fast FPGA to do the final high speed signal generation. So your implementation doesn't make the best use of your Mesa gear but it's clearly not an issue. The DMM Tech breakout board looks a cut above the normal ebay offerings. Have you connected up the spindle PWM control? Sounds as if a tacho signal would be a sensible additional input so that you can detect a locked spindle.

One of the main reasons for buying the Mesa system was to have enough IO to connect up all the limit switches etc. Although LinuxCNC seems to be pretty good at knowing where the travel limits are, there's a lot at stake if it (or I) got it wrong.

If you have a spare moment, could you PM me a link or email address so I can look at your HAL and config files? The (increasingly) grey cells struggle with this stuff and I would like to understand how they all fit together rather than hitting and hoping, which is what will happen otherwise. Some of the material on the LinuxCNC site is rather obscure and difficult to follow.

Murray