Topics

doing a 2 sided board #mill

John Ferguson
 

On 5/19/20 9:44 AM, john wrote:
I finally have to face doing a double-sided board.  I have a surface-mount barometric sensor which wants little pads and also the Through-hole Molex jack for connecting the result to my micro-controller.

It appears that I need to toggle "top" on the setup, and then cut and drill the bottom first, then flip the board and do the top.

It also appears that the first thing the bottom-drill gcode does is drill the origin.  This makes it possible to reset the origin on the flipped board.  it also looks as if the top drill routine doesn't do this, so one would always want to etch and drill the bottom first, then etch the top and mill the perimeter.

I've arrayed four of these llitle boards onto a small work board with enough space between them that the work-piece will support each remaining board as its perimeter is cut out.  I provide my perimeters with tabs to continue to support them and then saw them apart

Does all of this sound right?

John

Harvey White
 

Another alternative is to take thinner single sided board (0.030 thick board), make the top and bottom separately, and then epoxy them together using the guide holes you make.  I use push in map pins which are not flexible, have a relatively tight fit to keep wobbling down, and then use a 1 hour epoxy.  All the epoxy mixes take 24 hours to fully cure, and 5 minutes can be a bit short, as can 15.

I'd cover the top and bottom of the board with masking tape before epoxy, since it keeps epoxy fingerprints off the copper.

Harvey

On 5/19/2020 9:50 AM, John Ferguson via groups.io wrote:

On 5/19/20 9:44 AM, john wrote:
I finally have to face doing a double-sided board.  I have a surface-mount barometric sensor which wants little pads and also the Through-hole Molex jack for connecting the result to my micro-controller.

It appears that I need to toggle "top" on the setup, and then cut and drill the bottom first, then flip the board and do the top.

It also appears that the first thing the bottom-drill gcode does is drill the origin.  This makes it possible to reset the origin on the flipped board.  it also looks as if the top drill routine doesn't do this, so one would always want to etch and drill the bottom first, then etch the top and mill the perimeter.

I've arrayed four of these llitle boards onto a small work board with enough space between them that the work-piece will support each remaining board as its perimeter is cut out.  I provide my perimeters with tabs to continue to support them and then saw them apart

Does all of this sound right?

John



John Johnson
 

You could put one part on the other side of the board.
I.e. the surface mount part could go on the "bottom".


Regards,
John

On 19 May 2020, at 9:50, John Ferguson via groups.io wrote:

On 5/19/20 9:44 AM, john wrote:

I finally have to face doing a double-sided board.  I have a surface-mount barometric sensor which wants little pads and also the Through-hole Molex jack for connecting the result to my micro-controller.

It appears that I need to toggle "top" on the setup, and then cut and drill the bottom first, then flip the board and do the top.

It also appears that the first thing the bottom-drill gcode does is drill the origin.  This makes it possible to reset the origin on the flipped board.  it also looks as if the top drill routine doesn't do this, so one would always want to etch and drill the bottom first, then etch the top and mill the perimeter.

I've arrayed four of these llitle boards onto a small work board with enough space between them that the work-piece will support each remaining board as its perimeter is cut out.  I provide my perimeters with tabs to continue to support them and then saw them apart

Does all of this sound right?

John

John Ferguson
 

Hi Harvey,

That's a neat idea.  I have a fixture for mounting boards which I'm cutting which should assure reliable registration.But of course I'll find out soon enough.

john

On 5/19/20 11:24 AM, Harvey White wrote:
Another alternative is to take thinner single sided board (0.030 thick board), make the top and bottom separately, and then epoxy them together using the guide holes you make.  I use push in map pins which are not flexible, have a relatively tight fit to keep wobbling down, and then use a 1 hour epoxy.  All the epoxy mixes take 24 hours to fully cure, and 5 minutes can be a bit short, as can 15.

I'd cover the top and bottom of the board with masking tape before epoxy, since it keeps epoxy fingerprints off the copper.

Harvey


On 5/19/2020 9:50 AM, John Ferguson via groups.io wrote:

On 5/19/20 9:44 AM, john wrote:
I finally have to face doing a double-sided board.  I have a surface-mount barometric sensor which wants little pads and also the Through-hole Molex jack for connecting the result to my micro-controller.

It appears that I need to toggle "top" on the setup, and then cut and drill the bottom first, then flip the board and do the top.

It also appears that the first thing the bottom-drill gcode does is drill the origin.  This makes it possible to reset the origin on the flipped board.  it also looks as if the top drill routine doesn't do this, so one would always want to etch and drill the bottom first, then etch the top and mill the perimeter.

I've arrayed four of these llitle boards onto a small work board with enough space between them that the work-piece will support each remaining board as its perimeter is cut out.  I provide my perimeters with tabs to continue to support them and then saw them apart

Does all of this sound right?

John




John Ferguson
 

I thought of this at first, but the sensor is tall and the space they are going into is quite constrained.

besides, I ought to learn how to do this.

thanks,


john

On 5/19/20 11:30 AM, John Johnson wrote:

You could put one part on the other side of the board.
I.e. the surface mount part could go on the "bottom".


Regards,
John

On 19 May 2020, at 9:50, John Ferguson via groups.io wrote:

On 5/19/20 9:44 AM, john wrote:

I finally have to face doing a double-sided board.  I have a surface-mount barometric sensor which wants little pads and also the Through-hole Molex jack for connecting the result to my micro-controller.

It appears that I need to toggle "top" on the setup, and then cut and drill the bottom first, then flip the board and do the top.

It also appears that the first thing the bottom-drill gcode does is drill the origin.  This makes it possible to reset the origin on the flipped board.  it also looks as if the top drill routine doesn't do this, so one would always want to etch and drill the bottom first, then etch the top and mill the perimeter.

I've arrayed four of these llitle boards onto a small work board with enough space between them that the work-piece will support each remaining board as its perimeter is cut out.  I provide my perimeters with tabs to continue to support them and then saw them apart

Does all of this sound right?

John

John Ferguson
 

my little two-sided boards came out fine. I did make the via-pads bigger than Eagle would have to get enough area to ease the soldering.

I use 12 mil traces, which worked fine with the pads for the smd device I'm using.

What is the narrowest trace any of you can cut reliably? trench width?

I ask because I expect to move from mostly through-hole devices to SMD for the projects I do.

john

peterg1000
 

For your information I'm using a Stepcraft SC420/2 CNC router with a UCCNC interface and an SC100 motion controller. 

For double sided boards  I place 3mm guide pins well outside the board milling profile and dimension lines.  These are not considered when Eagle calculates board area to apply limits, so nothing is lost on that account. They are of symmetrical in the X direction.

Vias are drilled with a 0.9mm carbide drill and a 1/32" x 1/4"  (approx 0.79mm) rivet used to bridge the two layers.  These brass rivets are used by model engineers and and have a head diameter of 1.2mm (approx .044").  A 0.05" pad is fine for soldering using a hot air gun at 250C and standard 60/40 solder paste on the head side of the board.  On the reverse side the rivets are clipped to protrude by a 1/16" or so and solder paste and hot air treatment repeated.
It pays to clean the rivets in a "clock cleaning solution" first - soldering is faultless with this precaution.

With a good engraving cutter (not a cheapo Chinese !!) and a properly set up CNC machine, I can reliably cut .010" traces with.016" isolation. Engraving depth is set to .004" with the PCB fixed with double sided tape on a milled flat sacrificial board.

It's essential to test each new engraving tool to check minimum isolation width and clean cutting of the copper.  I have a small test programme that engraves a suitable test pattern (picture attached).   Left is the Chinese cutter, right is the "good" one.  Patterns allow checking of CNC machine backlash and single pass isolation.  Large circle diam is 8mm.

John Ferguson
 

Hi Peter,

I've been using bits from Precise Bit which cut without raising the copper edges and in my use produce trace widths as close to what I've asked for (10 or 12 mil) as I can determine with my optical comparator. I secure my workpieces in an hdpe fixture I made with four 6-32 nylon socket head cap screws. the holes in the board are located with an aluminum jig so they are always in the right place.  my first efforts were two small boards to hold smd differential pressure sensor and a molex jack.  I relied on the screw-holes for alignment when I flipped the board to cut the top. The result was close to perfect.

I have a small length (1/2 inch) of .250 drill rod which i roll against the bit to set height. I use the height that I get when the bit just clears the rod, and get very consistent results.

i like the idea of the rivets for the vias.  what are they called so I can buy some?

it looks as thought there is a lower limit to isolation trench width which would control the pitch possible with some micro-controller packages. I need to do some research to see if I really will be boxed in and have to send board designs out to a fab shop.  I've been building up projects with a Teensy microcontroller, buck typ voltage regulator and MicroSD card holder on a small board with a number of molex jacks for handling gps, serial, i2c, vin, ground, etc. I'd like to make board which have the accelerometer, gps, pressure, differential pressure (for airspeed), compass etc chips all on a single board. I'd also like board to be as small as possible. 

I usually use 10mil traces because they can be run between the larger-pin pads on 2.54 pitch DIP packages.  I also tin my board and use a Kester flux pen where I'm going to solder. I use 60/40 very fine solder and a Weller WTCP iron with their finest 700C tip.

I built a reflow oven for when I get to the really small stuff, but so far haven't used it than to check if it worked - it did.

I think I'm going back to Precise-Bits and buying a couple of the finest bits he has to get the smallest isolation channel I can, thinking that this is likely to be the limitation on smallest pin pitch I can cut for.

Does any of these sound nuts to you?

Thanks again for your note.

john


On 5/23/20 8:22 AM, peterg1000 via groups.io wrote:
For your information I'm using a Stepcraft SC420/2 CNC router with a UCCNC interface and an SC100 motion controller. 

For double sided boards  I place 3mm guide pins well outside the board milling profile and dimension lines.  These are not considered when Eagle calculates board area to apply limits, so nothing is lost on that account. They are of symmetrical in the X direction.

Vias are drilled with a 0.9mm carbide drill and a 1/32" x 1/4"  (approx 0.79mm) rivet used to bridge the two layers.  These brass rivets are used by model engineers and and have a head diameter of 1.2mm (approx .044").  A 0.05" pad is fine for soldering using a hot air gun at 250C and standard 60/40 solder paste on the head side of the board.  On the reverse side the rivets are clipped to protrude by a 1/16" or so and solder paste and hot air treatment repeated.
It pays to clean the rivets in a "clock cleaning solution" first - soldering is faultless with this precaution.

With a good engraving cutter (not a cheapo Chinese !!) and a properly set up CNC machine, I can reliably cut .010" traces with.016" isolation. Engraving depth is set to .004" with the PCB fixed with double sided tape on a milled flat sacrificial board.

It's essential to test each new engraving tool to check minimum isolation width and clean cutting of the copper.  I have a small test programme that engraves a suitable test pattern (picture attached).   Left is the Chinese cutter, right is the "good" one.  Patterns allow checking of CNC machine backlash and single pass isolation.  Large circle diam is 8mm.

Attachments: