Monthly Archives: September 2014

Proxxon MF 70 and LinuxCNC – Part VI

I’ve been playing around with FreeCAD (learning with video tutorials on YouTube), PyCAM, LinuxCNC Image to Gcode, OpenSCAM,  etc. We were planning on routing a 2D shape, and I thought Image to G Code would be helpful. I was wrong. It’s actually for milling contours (constant stepover?), but not for routing a stencil.


Image to G Code on LinuxCNC 2.6

Since it was taking ages to process an image on the CNC PC (800 MHz P III + 512 MB RAM + Trident Blade 3D 8MB), I did all this on my actual desktop PC (i5 2.8GHz overclocked to 3.8 GHz + 1600 Ghz 8 GB RAM + NVIDIA GTX 660 Ti) running LinuxCNC 2.6 from a USB disk. So I could try the image with different parameters quickly. Well, Max Baser Jittter was so much better (around 5000 ns) than it is on the CNC PC (around 50000 ns), while everything was running so smoothly as opposed to the flickering windows and late responses of the CNC PC. So, I’m not really sure about those “a faster PC doesn’t necessarily mean you’ll get faster RTAI response” comments.

OpenSCAM in action. The result is not really what I need.

OpenSCAM in action. The result is not really what I need.

I’m thinking about installing LinuxCNC 2.6 on a different partition on my actual desktop PC just for the kicks.

Still no real work on the CNC. I’ll either shallow carve this shape above in brass for a test, or create a DXF to route it in 2D.


Proxxon MF 70 and LinuxCNC – Part V

I got a 20GB 7200RPM IBM PATA HDD last week, and it works good. I also upgraded the ram to 512MB. I successfully installed LinuxCNC 2.6 this time, but the Maximum Base Jitter was too high. Eventually, I installed 2.5, and all is fine now. Friday, my friend Pamir and I managed to do out first cut. I think it turned out alright, considering we didn’t have a suitable bit, and it was out first cut. Motors are 1/8 stepping.

We used a Dremel Tungsten Carbide Ball-end Cutter. We just scaled the program to fit it on a piece of wood, lowered the feed speed to %30 or so, and set the spindle speed to the lowest. Since the ball end mill was 2mm, it was a pretty rough cut for this scale. Since it was a one pass cut (a drop cut instead of a contour cut), and our cutter could not travel deep, we had to rescale the Z axis to allow for cuts 2mm deep max. Drivers set to 1/8 step with no limit on motor current or decay. Crappy video again thanks to shooting accidentally in high speed (low res, dark picture, no sound). I changed video speed to normal and faster on ocassion.


1mm,2mm, wood, metal, etc.

I got some nice bits the day after, yet I still don’t have a nice shape to mill. Gotta learn CAD/CAM software (for Linux) so I can start creating my own parts. I also got some  air-drying clay. Hopefuly, I’ll try all that tomorrow. The computer was acting weird the last time I checked, so much for free stuff…

Proxxon MF 70 and LinuxCNC – Part IV

Great, the 40 GB Seagate Barracuda PATA HDD is dead. I had too many bad sectors while I was trying to install LinuxCNC, anyway. I’ll have to get another PATA HDD and some more RAM’s tomorrow.

In the meantime, I opened one of the unused HY-DIV168N-3.5A Stepper Motor Drivers to figure out the capacitor value at pin 7 (Cosc) of the TB6560AHQ in it.



It is not a 1000 pF like it is in those Blue PCB single board drivers, as stated in a text here, but it’s a 330 pF. This should give us 130 Khz oscillating frequency, and a minimum clock pulse width of 30 us, according to the datasheet.


Where the 330pF capacitor is…

In the Stepper Configuration Wizard…

Step Time: How long the step pulse is on in nano seconds (Mark of Pulse Width -> 30 us = 30000 ns).

Step Space: Minimum time between step pulses in nano seconds (Space of Pulse Width or Period -> 1000000 / 130000 = 7692.31 ns).

6N137 optocoupler delay is negligible, if we round up the Step Space to 7800 ns. I don’t know if I need the speed gained (500 ns to 50 ns) by changing the diodes to ultra-fast ones yet (also stated in the same text). Let’s roll that up to 8500 ns.

I have no idea about Direction Hold or Direction Setup values, some people set it to 20000.

I’ll test all of these with my DSO tomorrow.

I may change that to 100pf for 3 times the speed, but the datasheet states that it was not tested in production. Still a future plan for experimental purposes, maybe then it could be worth changing the diodes, too.

Table travel values should be…

X = 134 mm
Y = 48 mm (crap)
Z = 80 mm

I have been using the drivers at half step setting. I think, I may move up to 1/16 with these settings, probably after setting motor speeds at the axis configuration screens.

I’ll have to shop for some milling bits and brass on Saturday 🙂

Proxxon MF 70 and LinuxCNC – Part III

Finally tested the setup last night with temporary motor cable connections.

The X axis motor has problems at 2:34 and 3:10 (does anyone know the reason, check out the motor sound going crazy), and the table wobbles a bit. I didn’t exceed the table limits. Haven’t touched the gibs yet, and forgot to oil the X axis of this long forgotten mill. I’m also suspicious about the motor timings I made up.

I should get the timing capacitor value on the PCB, and refer to the TB6560 datasheet (page 10).

I think I’ll finalize the hardware tonight and start working on the calibration, etc. for a first run with the spindle on.

Proxxon MF 70 and LinuxCNC – Part I

Today my friend Onur and I started working on our little CNC project. Since he currently has no use for it, Onur offered to leave his MF 70 in my workshop indefinetely, so we can convert it to a small CNC and use it for manufacturing small parts .


MF 70 with motors mounted, standing on the electric box which will house the electronics.

Continue reading