What is my problem here?

morrows_end
Posts: 12
Joined: Tue Jun 26, 2012 2:45 am

What is my problem here?

Post by morrows_end » Wed Aug 21, 2013 6:22 pm

I'm using eagle to design the PCB, exporting it using pcb-gcode (latest version) Then running the autoleveller form autoleveller.co.uk and here are my results.

https://dl.dropboxusercontent.com/u/6405154/issues.png

I expected it to be a little gnarly, but this is completely unusable. I only ran 1 of the 5 layers of isolation cuts it was going to do. I admit my z depth is a bit too deep (using a 45 degree engraver bit) but as you can see, in most cases, my Y (up and down) axis seems to be going exactly back into the previous cut as opposed to making a new cut next to it. On my X axis (left to right), only one trace seems to be OK. My traces are 0.01".

I had previously gotten rid of some wobble in my Z axis by replacing the precision washers in my V-wheels and that helped reduce my wobble substantially.

My belts are tight (in fact my Y belt is tighter than my X belt, so I would have assumed the Y would have been fine and the error would be with my X).

I am running my engraver at 0.035" depth (supposedly, I don't trust that it is) at a speed of 3000RPM.

Anyone have any ideas? I am in need of this PCB for a prototype pretty much ASAP. I suppose if all else fails, i can make a thru-hole PCB to test functionality, but I need SMT for sure.

veng1
Posts: 250
Joined: Fri Nov 30, 2012 12:09 pm

Re: What is my problem here?

Post by veng1 » Wed Aug 21, 2013 6:55 pm

I can't say for sure what you problem is but I'll offer a couple of suggestions that you can either use or ignore.

One reason for using traces that are smaller than the pads they connect to is to avoid a heat trap that draws the solder away when machine soldering either reflow or wave. In this case, for a prototype, I'd assume you are going to hand solder or at least hand touch it up. So I,d suggest making the traces bigger up to the point where the design rule checker complains that there isn't enough clearance. I then see if I can move some of the offending traces around until the design passes the checker. Personally, I then raise the design rules until I've got the maximum spacing and maximum trace width possible, even on production boards.

If I'd run this board on one of my T-Tech circuit board mills, my first guess would be that the depth of the bit is set too low. The T-Tech software allows an isolation step that outline the pads and traces. If the isolation step uses too large a value, you can get everything that is outlined too small. Does the software you are using have that option?

If you look at the tool path in a gcode visualizer does it appear to have the trace sizes you are getting?

morrows_end
Posts: 12
Joined: Tue Jun 26, 2012 2:45 am

Re: What is my problem here?

Post by morrows_end » Wed Aug 21, 2013 7:04 pm

Thanks for the reply. I plan on hand-soldering this one.
veng1 wrote:If you look at the tool path in a gcode visualizer does it appear to have the trace sizes you are getting?
The single pass I got looks like the width of all 5 passes in the simulator. I had attempted at a z Depth of 0.07" which was default when I installed PCB-gcode. I broke two expensive bits and cussed a lot, then I changed it to half that (0.035") and ran this board. The depth doesn't look as though it changed at all between the two tests. What depth do you normally run your boards at?

cvoinescu
Posts: 4442
Joined: Thu Jul 19, 2012 6:50 pm
Location: Camberley, UK
Contact:

Re: What is my problem here?

Post by cvoinescu » Wed Aug 21, 2013 7:15 pm

The traces are way too thin for this method. It seems to me that the program (G-code generator) assumed your bit would cut a very thin groove, but the reality is different. There's plenty of clearance in the opposite direction, so maybe you can convince it that your tool makes a wider groove. I think you may be milling too deeply, too. I'm not sure what your workflow entails, specifically, how you adjust the Z position, so I can't give advice on that.

Head over here and compare notes with conoral11 -- his is the best PCB milling I've seen around here.
Proud owner of ShapeOko #709, eShapeOko #0, and of store.amberspyglass.co.uk

veng1
Posts: 250
Joined: Fri Nov 30, 2012 12:09 pm

Re: What is my problem here?

Post by veng1 » Wed Aug 21, 2013 8:32 pm

I've always started out just barely touching the surface, running the program to see what it looks like and then lowering the bit.

However, the T-Tech machine doesn't have Z-axis control. It has a foot and a thumb screw that you adjust by hand, so you really want to take it down slowly. So the recommendation from the manufacturer is to mill a test line away from the area of interest and measure how wide it is. I use an optical reticule to measure the width.

morrows_end
Posts: 12
Joined: Tue Jun 26, 2012 2:45 am

Re: What is my problem here?

Post by morrows_end » Wed Aug 21, 2013 11:38 pm

I've been having trouble figuring out how to get my Z depths correct, you can see my questions on the pcb-gcode forum about that...

My workflow is EagleCAD 6 --> pcb-gcode (latest) --> Autoleveller.jar (from autoleveller.co.uk) --> linuxCNC.

in PCB-gcode, I set my depths to be 0.035" below the surface of the PCB. The autoleveller program goes through and makes a grid of probe readings to set my Z heights in the pcb-gcode output based on the actual heights is measures. This leaves me with a new gcode file that starts out by taking the test readings, then begins to mill using the new Z heights it has measured to tell it where the top of the PCB is.

Other than that, I think that maybe my bit is going too deep (for an unknown reason) and then causing the Z axis to pitch forward or backward a bit since it can't cut as fast as it needs to. Additionally, I noticed that when I tell the machine to be at a position, then apply a little force with my finger to the tip of the bit, the loose end of the gantry moves slightly. I immediately put in an order for a dual motor setup for my machine. Since I only have 3 NEMA 17s, I'll use the NEMA 17 from my Z axis on the loose end of my gantry, an replace it with a NEMA 23 I have laying around. (I'm using unipolar steppers).

jasonharper
Posts: 11
Joined: Mon Jun 17, 2013 2:06 am
Location: Oklahoma

Re: What is my problem here?

Post by jasonharper » Thu Aug 22, 2013 12:03 am

0.035" is over half the thickness of typical PCB stock! That's insanely deep for an engraving pass. If your Z probing is perfectly accurate, then 0.002" engraving would be sufficient for 1 oz. copper cladding.
ShapeOko #1743F, double X axis, 5mm Y drive shaft, DW660 spindle

cvoinescu
Posts: 4442
Joined: Thu Jul 19, 2012 6:50 pm
Location: Camberley, UK
Contact:

Re: What is my problem here?

Post by cvoinescu » Thu Aug 22, 2013 6:12 am

morrows_end wrote:Additionally, I noticed that when I tell the machine to be at a position, then apply a little force with my finger to the tip of the bit, the loose end of the gantry moves slightly. I immediately put in an order for a dual motor setup for my machine. Since I only have 3 NEMA 17s, I'll use the NEMA 17 from my Z axis on the loose end of my gantry, an replace it with a NEMA 23 I have laying around. (I'm using unipolar steppers).
You won't go far without a dual Y of some sort. That explains some of the artifacts I saw, which puzzled me.
I'm not sure how you're going to mount a NEMA23 on Z, and that's the place where you need it least. Why not put it on X instead? Or buy a NEMA17, it's not much more expensive than all the parts you need to put a NEMA23 on either the Z or X axis.
Proud owner of ShapeOko #709, eShapeOko #0, and of store.amberspyglass.co.uk

wlanfox
Posts: 74
Joined: Sat Oct 20, 2012 10:49 am

Re: What is my problem here?

Post by wlanfox » Thu Aug 22, 2013 5:49 pm

Other things to consider,

1) getting the height of the tool to pcb to be consistence across your PCB is a big problem specially at copper thickness (consider backlash), as mention before in this thread, fatter traces will alleviate the variation in height form your machine. Also consider milling a MDF surface first to be the base for your PCB work. By milling MDF this will have a consistence height for the tool to pcb.
2) Double gantry is a must too, most people are ok having 0.5mm error for wood and other work, for PCBs this is a disaster, you need a sturdy machine, dual drive is a must.
3) consider end-mill as oppose to V shape bit, I have used a 0.5mm end-mill successfully before, with end mill height is less of a factor.

Search on the PCB section for other example and experiences from many of us, who have done this before... for example:

http://www.shapeoko.com/forum/viewtopic.php?f=28&t=1039
http://www.shapeoko.com/forum/viewtopic.php?f=28&t=1045

morrows_end
Posts: 12
Joined: Tue Jun 26, 2012 2:45 am

Re: What is my problem here?

Post by morrows_end » Fri Aug 23, 2013 3:58 pm

Thanks for all the tips guys! here's my plan:

1. Make the traces wider.

2. Decrease the depth to 0.002" or so. I didn't even think about it, but yea, 0.0035 IS half of the 0.006-ish PCB material. duh! I must have misplaced a 0 when I was thinking about it. I'm used to metric too, so I'm not used to thinking in mils.

3. I *could* just get two drive drive pulleys for NEMA 23s and make those my Y axis. I didn't think of that and just looked for a quick kit to buy that could solve my problem so I got the NEMS 23 z-axis kit. I will probably end up reaplacing all my NEMA 17s with 23s when my projects get to the point that my machine required more force. I just got my Z axis conversion kit in the mail, so I'll start replacing that first though.

4. Endmills to the rescue! Engraver bits have varying widths depending on depth and just to eliminate any errors this will cause, I can simply get a tiny endmill with a consistent diameter.

5. Even though I am using an autoleveller program, for best results I should mill a flat MDF platform to attach PCBs to for milling. That should reduce the possible sources of error even more.

Thanks again guys!

Post Reply