Sable-2015

[tony]

I've never milled a PCB before but I'm getting tired of using the toner transfer method. I started looking around for info and found this website. I noticed this 3 axis on ebay http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=190265073328&ssPageName=ADME:X:RTQ:US:1123 for $500. I'm thinking that with a pitch of 1.5mm I should be able to do some fine trace work. Anyone care to comment before I waste my money? I was also looking at the Wolfgang spindles but they aren't currently available on ebay.

[admin]

I would ask about the cost of shipping before commenting, my guess shipping is pretty pricey.

[tony]

The shipping is resonable.

Worst case I may want to do a TQFP100 or TQFP144. I'm not sure whether milling and in particular this mill is capable of doing this. I am able to do it with toner transfer but it is fragile at best.

[admin]

The machine has to be "tight" to do those pads, and a spindle that is really, really good. In all proability you would have to be able to tweak the machine mechanicals to virtually eliminate backlash. Can it be done, yes, will that machine do it out of the box, probably not. But a lot of doing that precise work is setup and technique and if so inclined the machine could be tweaked to do it.

[tony]

I broke down and ordered the unit. I will post feedback when I get it up and running.

[rayellam]

Hi Tony, any feedback on the Sable 2015 yet? im looking at getting one with a few mods to the standard configuration and wondering what your current config is on the machine? can it cut 0.05" SMD PCB's? any picturers of work you have done with this machine?

Thanks in advance
Ray

[tony]

Well after about a year of not doing anything I finally got a board cut on the Sable. In addition to the Sable I am using a Wolfgang Engineering spindle bought from Ebay and 60 deg spade milling bits from Drewtronics.
I also tried a couple of different stepper driver boards. The first was the Hobby CNC board http://www.hobbycnc.com. I had a number of issues with noise on this board and would not recommend it. I then bought an optically isolated driver board from http://www.easy-cnc.com/products.html. This board was just slightly more expensive but I didn't have to assemble it. The one downside is that this board does not have pull up resistors for the limit and home switches so you need to do a little bit of work to get it going. Other than that it works great. I did add a couple of cooling fans since the board was running hot.

I downloaded a demo of Diptrace and exported the example PCB_2 to DXF using the Edge_Top option as explained on their website using an edge width of 0.006 inches. I don't think the 0.006 number is correct since the bit is probably bigger than this but the board was designed with some tight spacing and this was the highest number I could use. This gave me a DXF file that contained outlines of all the traces. I then imported this into the free version of CamBam to generate the G-Code. Somewhere in the translation the units got mistaken and had to change from G21 to G20 for inches. I am using EMC2 for my CNC software. If you are going to use EMC2 for PCB's I recommend using G61 to force stops at exact co-ordinates or the results will be unpredictable and highly dependent on your acceleration values.

This is the view of the board from within Diptrace:

diptrace PCB_2.dip Example.JPG

This is the board that was cut on the Sable CNC. The feed rate was 20mm per second. The board feels like there are some burrs on it. I'm not sure if this is because the feed rate was to fast or the cutters aren't that good. After I took it off I was thinking that I should have tried a second pass to see if it cleaned it up. Next time I guess

Sable 2015 PCB_2_500.JPG

As can be seen in the image the depth of cut in slightly deeper on one side than the other. Using a dial gauge I found that the table on the Sable had about 150um bow in it on the X axis (about two sheets of paper). Given that the copper is about 35.56um at 1oz this is pretty bad. I've seen that some people are machining the surface of the table so that it is flat with respect to the cutting tool but I don't think the Wolfgang spindle is up for the challenge not to mention the small steppers on the Sable. I think they are Nema 17 but I didn't measure them. What I ended up doing was putting the copper clad in the middle of the table to reduce the bow to about 75um.

I guess that the results are not that bad. I was able to cut a board with a TQFP-44 on it with very little setup. I would say that this is about the limit for this machine in it current form though. I will try some more when I get time. This should give an idea of what you can do for $500 plus a spindle of some kind.

[admin]

Looks pretty good for a first attempt. The burrs are normal, clean them up with a scotchbrite pad.

[LukeS]

Tony,

Thanks for sharing your experience. I am new to the CNC world and thinking of getting a Sable 2015. How did you mount the Wolfgang Engineering spindle to the sable 2015? Can the bow in the table be mediated by putting a flat aluminum tooling plate down on top of the table?

[tony]

To mount the Wolfgang spindle I had a friend make an adapter plate out of aluminum for me. However, when I purchased the spindle the seller had an option to make the plate for the same price. He sends out a generic one that will not fit on the Sable.

I've seen a bunch of different ways to overcome the imperfection in the table. Write now I am getting my friend to make be a polished steel table. He seemed to think this might be my best option. I'm not sure when he's going to finish but I'll post pictures when I'm done. I'm glad I took the table off though because there was a slight misalignment with the bearings that was causing one side to bind a little bit. In the end milling a piece of plexiglass bolted to the table may be the be thing. There is also the issue of imperfections in the PCB itself that may play a role (I'm haven't looked into this that much). I've seen some posts on the net about floating spindles that seemed interesting but very costly. I was thinking of using the same idea except to float the PCB on double sided foam tape about 2 mm thick. Then put a nose on the spindle so that the depth the tool cuts is the same it just presses down on the PCB a little bit more in some spots and a little less in others. This would be pretty easy to try out. All I need to do is convince my friend to make a collar to fit around the spindle and fix it in place with a set screw. We're talking fractions of mm in error so it should do the trick as long at the collar doesn't slip.