Homebrew PCBs

I have an idea for making boards using Electrical Discharge Machining (EDM) that I would like to bounce off you guys.

The idea is to use a needle on a CNC x-y table (picture pen plotter with the pen replaced with the needle).  The needle hovers just above the copper board a few thousandths, maybe .003" – .006".  Paint the copper with automotive spray paint.  Apply a high-voltage/high-current to the needle and ground the board.  The spark between the needle and copper will burn and evaporate away the paint.  The board could be submerged in oil or some dielectric to keep oxygen out.  The artwork drawn would be like that of a milled board where the copper around the traces is removed or all copper could be removed with multiple back and forth scans over the whole board.

What do you guys think?  Would acceptable trace spacing and widths be achievable?

Having operated a EDM for over 13 years, its my experience that this won't work. A EDM doesn't work in this way. First the material to be eroded has to be metallic, which paint isn't. Second, for every amount of metal removed from the fork piece around 25% of that amount is removed from the electrode/needle. Third, the EDM is a slow process compared to etching. 
 
Many people have suggested EDM over the years for making a PCB, but not been successful. Many did not understand the EDM process.
 
cary
 
 

--- On Wed, 1/5/11, Jeff <jeff.heiss@...> wrote:

From: Jeff <jeff.heiss@...>
Subject: [Homebrew_PCBs] Electrical Discharge Machining plus painted board plus CNC x-y table
To: Homebrew_PCBs@yahoogroups.com
Date: Wednesday, January 5, 2011, 2:17 AM


  



I have an idea for making boards using Electrical Discharge Machining (EDM) that I would like to bounce off you guys.

The idea is to use a needle on a CNC x-y table (picture pen plotter with the pen replaced with the needle). The needle hovers just above the copper board a few thousandths, maybe .003" – .006". Paint the copper with automotive spray paint. Apply a high-voltage/high-current to the needle and ground the board. The spark between the needle and copper will burn and evaporate away the paint. The board could be submerged in oil or some dielectric to keep oxygen out. The artwork drawn would be like that of a milled board where the copper around the traces is removed or all copper could be removed with multiple back and forth scans over the whole board.

What do you guys think? Would acceptable trace spacing and widths be achievable?








[Non-text portions of this message have been removed]

I was suggesting not to remove the copper, but to remove the paint.  After the paint is removed, place the board in etchant and etch.  The copper will remain where the paint was not removed leaving the traces.  Or is this what you meant when you said it would not work?

--- In Homebrew_PCBs@yahoogroups.com, "Jeff" <jeff.heiss@...> wrote:

>
> I have an idea for making boards using Electrical Discharge Machining (EDM) that I would like to bounce off you guys.
> 
> The idea is to use a needle on a CNC x-y table (picture pen plotter with the pen replaced with the needle).  The needle hovers just above the copper board a few thousandths, maybe .003" – .006".  Paint the copper with automotive spray paint.  Apply a high-voltage/high-current to the needle and ground the board.  The spark between the needle and copper will burn and evaporate away the paint.  The board could be submerged in oil or some dielectric to keep oxygen out.  The artwork drawn would be like that of a milled board where the copper around the traces is removed or all copper could be removed with multiple back and forth scans over the whole board.
> 
> What do you guys think?  Would acceptable trace spacing and widths be achievable?
>

--- In Homebrew_PCBs@yahoogroups.com, "Jeff" <jeff.heiss@...> wrote:
>
> I was suggesting not to remove the copper, but to remove the paint.  After the paint is removed, place the board in etchant and etch.  The copper will remain where the paint was not removed leaving the traces.  Or is this what you meant when you said it would not work?
> 

Cary was responding to both possibilities -- if you want to remove copper directly, EDM *could* do it, but 1) it would be slow, and 2) you would have to compensate for the erosion of the electrode -- ie, you can't just move X and Y, but you would also have to move Z, all while determining whether the spark gap is too small (shorting) or too long -- and if too long, in which axis it is too long. I'll add two other points to what Cary said: 3) EDM produces a certain amount of overcut, meaning that your electrode will have to be *smaller* than the desired cut you are trying to achieve. If you want your spacing to be 10 mils, you'll need to have your electrode be around 6 mils in diameter -- and don't forget that it can't just be the point of the electrode that size; it has to be enough length of the electrode to allow for the gradual erosion of the electrode. And the electrode needs to be stiff enough not to deform if/when it contacts the board in the process of fine-tuning the spark gap. 4) You would have to program your cuts in such a way that you always maintain the ground on the material to be removed up until the very last spark. IOW, you could not clear out an area after cutting its outline, because the area would then be isolated from the ground. So -- all of that to say that using EDM to remove copper from the board is theoretically possible, but it would be a poor choice.

If you want to remove paint, rather than copper directly, EDM cannot do it at all, since EDM depends on generating a spark between the electrode and the conductive metal to be removed. I suppose you could get around that by using a conductive paint, but then 1-3 above would apply, and you would add the new problem of electrically distinguishing between the paint and the copper -- I don't think there is any way you could spark all of the paint off without also sparking off at least some of the copper, and in any case, until the copper is also gone, your EDM would not be able to tell that it was finished with that spot. You could try to get around that by having it only spend a certain amount of time in each spot, but note that that is not a simple matter when the control circuitry is sensing the spark gap and adjusting the electrode to account for wear as well as for how much cutting has occurred.

Whew! Just trying to describe all the problems of using EDM for this purpose has made me tired.

To remove paint you just need to up the voltage - say 10kV or more. This 
could be generated by a simple car ignition coil driven from an 
oscillator. I have no idea how much paint would be removed on each 
spark. It would definitely be a case of trial and error.

Many years ago I used an engine cylinder pressure measuring machine that 
used this principle. It had an electrode that moved depending on the 
pressure from a compressed gas supply. When the cylinder pressure 
equaled the gas pressure it fired a spark, making a mark on a paper roll 
that rotated at crank speed. By varying the gas pressure you could map 
the cylinder pressures over the whole combustion cycle.

Les

On 05/01/2011 18:05, Andrew wrote

> If you want to remove paint, rather than copper directly, EDM cannot do it at all, since EDM depends on generating a spark between the electrode and the conductive metal to be removed. I suppose you could get around that by using a conductive paint, but then 1-3 above would apply, and you would add the new problem of electrically distinguishing between the paint and the copper -- I don't think there is any way you could spark all of the paint off without also sparking off at least some of the copper, and in any case, until the copper is also gone, your EDM would not be able to tell that it was finished with that spot. You could try to get around that by having it only spend a certain amount of time in each spot, but note that that is not a simple matter when the control circuitry is sensing the spark gap and adjusting the electrode to account for wear as well as for how much cutting has occurred.
>
> Whew! Just trying to describe all the problems of using EDM for this purpose has made me tired.

On Wed, Jan 5, 2011 at 1:23 PM, Les Newell <les.newell@...> wrote:
>
>
>
> To remove paint you just need to up the voltage - say 10kV or more. This
> could be generated by a simple car ignition coil driven from an
> oscillator. I have no idea how much paint would be removed on each
> spark. It would definitely be a case of trial and error.
>

Having spent a great deal of time dealing with HV, commonly over a 75
KV, I can tell you that you are likely to have a huge problem with the
selectivity of the spark that breaks the paint/insulation.  Like
lightning, it seeks the easiest path to complete the circuit and with
any variation in the thickness of the paint at all it will tend to
wander around where you're trying to target based on a number of
variables like the thickness, air flow, the electrode shape/wear, and
so forth.  Then there's the issue of current supplied: you must supply
enough current to vaporize paint, not knock tiny holes in it as it
breaks down, but actually remove a significant amount of it.  The more
current expended, the larger the diameter of the spark, which
accentuates your inaccuracy in targeting a specific point.  The
process at best would be inaccurate and slow since if you go too fast
your electrode will heat significantly, pre-ionizing the air within
the gap and adding another variable to the problem; go fast enough and
you get an arc-welder :-)

It's an interesting idea but I believe the amount of time to even get
any kind of result (The setup for routing pulsed HV down to a point
within a plotter would be hazardous to the electronics as it is) is
somewhat daunting.  I would really think skipping the HV part and
doing the resist-scratch-removal process would be faster, easier, and
far more precise.

Larry

Your idea is interesting.
I use to service gestetner products in my hay-days. They had a machine which used a 24 inch long rotating aluminum drum. A paper copy on the left side was read by a photomultiplier/microscope assembly and the signal was used to drive a high freq power osc. This signal went to a tiny stylus (think it was tungsten) in contact with a stencil. The sparks burned holes through the stencil to make an image silkscreen for a duplicator ( a thingy with 2 4inch rollers that squeezed ink through the stencil and onto the paper)for the youngsters. Sparks (high frequency only) actually do a nice job but 
the mechanics to do this on a flat pwb I think isn't worth the effort considering the cheap imaging devices that can be hacked to do the same.  


--- In Homebrew_PCBs@yahoogroups.com, "Jeff" <jeff.heiss@...> wrote:

>
> I have an idea for making boards using Electrical Discharge Machining (EDM) that I would like to bounce off you guys.
> 
> The idea is to use a needle on a CNC x-y table (picture pen plotter with the pen replaced with the needle).  The needle hovers just above the copper board a few thousandths, maybe .003" – .006".  Paint the copper with automotive spray paint.  Apply a high-voltage/high-current to the needle and ground the board.  The spark between the needle and copper will burn and evaporate away the paint.  The board could be submerged in oil or some dielectric to keep oxygen out.  The artwork drawn would be like that of a milled board where the copper around the traces is removed or all copper could be removed with multiple back and forth scans over the whole board.
> 
> What do you guys think?  Would acceptable trace spacing and widths be achievable?
>

On Thu, 06 Jan 2011 00:42:16 -0000, you wrote:

>Your idea is interesting.
>I use to service gestetner products in my hay-days. They had a machine which used a 24 inch long rotating aluminum drum. A paper copy on the left side was read by a photomultiplier/microscope assembly and the signal was used to drive a high freq power osc. This signal went to a tiny stylus (think it was tungsten) in contact with a stencil. The sparks burned holes through the stencil to make an image silkscreen for a duplicator ( a thingy with 2 4inch rollers that squeezed ink through the stencil and onto the paper)for the youngsters. Sparks (high frequency only) actually do a nice job but 
>the mechanics to do this on a flat pwb I think isn't worth the effort considering the cheap imaging devices that can be hacked to do the same.  
>
>
>
Back in the '60s there was a similar device (made by Fairchild) that mounted 
a photograph  on one end and a plastic sheet on the other side with a hot needle
to make plates for newspaper photographs on ink-based printing presses. They
still had Linotype machines and lead type for text.

There was another takeoff on the same idea for proto pc boards. A 1:1 board drawing
on one side and a flex pcb on the other. The traces were cut by something like
a dremel tool using a little disk/burr. The flex board was then glued to a thicker piece
of fiberglass. Needless to say you didn't run traces between dip socket pins 8-)

Keith Bowers WB4LSJ- Thomasville, NC

A few years ago before I retired I had access to a large (3ft X 4 ft) CNC YAG laser system that I used on occasion to do the same thing. It worked well and I was able to get very fine lines, in fact, too narrow. The laser cut was only about .001-.002" wide, and usually I had to make multiple cuts to get it to etch cleanly. Definitely overkill considering the cost of the syatem, but it was not being used for anything else at the time.

One other problem that needs to be resolved as with any technique like this. You need software to remove the paint between traces. It is much easier to create lines where you want traces. I used AutoCad to draw lines around all connected pads creating "islands" of connected pads. While this worked, the time I spent fiddling with it was not cost effective. You might be able to create a negative image of the PCB traces and find some software that would create a raster scan of all areas between traces. Perhaps if you have MasterCam or some similar program you could create "G code" to accomplish this since it is essentially a pocket routine.

Bob




      

[Non-text portions of this message have been removed]

Eagle has an ULP for isolation routing. See 
<http://www.brusselsprout.org/PCB-Routing/> for some implementation data.
Mark


At 07:20 AM 1/7/2011, you wrote:

>A few years ago before I retired I had access to a large (3ft X 4 
>ft) CNC YAG laser system that I used on occasion to do the same 
>thing. It worked well and I was able to get very fine lines, in 
>fact, too narrow. The laser cut was only about .001-.002" wide, and 
>usually I had to make multiple cuts to get it to etch cleanly. 
>Definitely overkill considering the cost of the syatem, but it was 
>not being used for anything else at the time.
>
>One other problem that needs to be resolved as with any technique 
>like this. You need software to remove the paint between traces. It 
>is much easier to create lines where you want traces. I used AutoCad 
>to draw lines around all connected pads creating "islands" of 
>connected pads. While this worked, the time I spent fiddling with it 
>was not cost effective. You might be able to create a negative image 
>of the PCB traces and find some software that would create a raster 
>scan of all areas between traces. Perhaps if you have MasterCam or 
>some similar program you could create "G code" to accomplish this 
>since it is essentially a pocket routine.
>
>Bob
>
>
>
>
>
>
>[Non-text portions of this message have been removed]
>
>
>
>------------------------------------
>
>Be sure to visit the group home and check for new Links, Files, and Photos:
>http://groups.yahoo.com/group/Homebrew_PCBsYahoo! Groups Links
>
>
>

How many watts was the YAG laser?

 

Concerning the cut lines from the laser being too narrow, would it be
possible to increase the diameter of the laser surface workspace diameter so
only one pass (or a couple) would be needed?  

 

  _____  

From: Homebrew_PCBs@yahoogroups.com [mailto:Homebrew_PCBs@yahoogroups.com]
On Behalf Of Bob Butcher
Sent: Friday, January 07, 2011 7:20 AM
To: Homebrew_PCBs@yahoogroups.com
Subject: [Homebrew_PCBs] Re: Electrical Discharge Machining plus painted
board plus CNC x-y table

 

  

A few years ago before I retired I had access to a large (3ft X 4 ft) CNC
YAG laser system that I used on occasion to do the same thing. It worked
well and I was able to get very fine lines, in fact, too narrow. The laser
cut was only about .001-.002" wide, and usually I had to make multiple cuts
to get it to etch cleanly. Definitely overkill considering the cost of the
syatem, but it was not being used for anything else at the time.

One other problem that needs to be resolved as with any technique like this.
You need software to remove the paint between traces. It is much easier to
create lines where you want traces. I used AutoCad to draw lines around all
connected pads creating "islands" of connected pads. While this worked, the
time I spent fiddling with it was not cost effective. You might be able to
create a negative image of the PCB traces and find some software that would
create a raster scan of all areas between traces. Perhaps if you have
MasterCam or some similar program you could create "G code" to accomplish
this since it is essentially a pocket routine.

Bob

[Non-text portions of this message have been removed]





[Non-text portions of this message have been removed]

I forget the details, but as I recall it was around 5 watts continuous. The laser was Q-switched and produced much higher peak power pulses, depending on the repetition rate selected. Usually I ran between 40 kHz and 200 kHz. The cutting speed, repetition rate, and spot size had to be considered or it would produce a "dotted line" rather than a clean continuous line of course. 
The system was designed to cut thin (about 1 um) metallic layers deposited on a Kapton substrate. It was capable of removing the metal without cutting the Kapton since the laser energy was transmitted by the Kapton. 
The end product was a PV solar panel using CIGS (copper-indium-gallium-selenide), but that is a whole different topic from what this fourm is about.
 




      

[Non-text portions of this message have been removed]