Rubber stamp PCB?

[Winston]

Much to my horror, I discovered one of my 'standard' PCB footprints
is 40% larger in both dimensions than the part it was intended for
on my latest board.
(It's a 16 pin device with pads on 0.5 mm centers)!
The board is about 70% assembled and is much later than I would prefer.
If I had unlimited time and money, I'd just spin the
artwork with the correct footprint. Lacking either, I want to
patch the existing board.

I've not located a prototype board that has the necessary footprint
terminating in matching pads.

http://www.rubberstampchamp.com/ProductDetail.aspx?productid=HS-
ROUND-3-4INCH-ARTWORK

Here's what I'm thinking. I can upload a big bitmap of the
correct footprint with exactly the right terminating pads to
these guys and stamp a piece of my 0.005" PCB stock using say
Epson Durabrite yellow ala:
http://www.pabr.org/pcbprt/pcbprt.en.html

Then I can stamp and etch as many adapter boards as I want.
(I need one [1] adapter board.)
Mount the device to the adapter then solder the terminating
pads to the huge pads of the incorrect footprint.

What do you think of this idea?

--Winston

 

[mike]

Much to my horror, I discovered one of my 'standard' PCB footprints
is 40% larger in both dimensions than the part it was intended for
on my latest board.
(It's a 16 pin device with pads on 0.5 mm centers)!
The board is about 70% assembled and is much later than I would prefer.
If I had unlimited time and money, I'd just spin the
artwork with the correct footprint. Lacking either, I want to
patch the existing board.

I've not located a prototype board that has the necessary footprint
terminating in matching pads.

http://www.rubberstampchamp.com/ProductDetail.aspx?productid=HS-
ROUND-3-4INCH-ARTWORK

Here's what I'm thinking. I can upload a big bitmap of the
correct footprint with exactly the right terminating pads to
these guys and stamp a piece of my 0.005" PCB stock using say
Epson Durabrite yellow ala:
http://www.pabr.org/pcbprt/pcbprt.en.html

Then I can stamp and etch as many adapter boards as I want.
(I need one [1] adapter board.)
Mount the device to the adapter then solder the terminating
pads to the huge pads of the incorrect footprint.

What do you think of this idea?

--Winston

I've done considerable experimentation over the years.
I liked a flatbed pen plotter best, but with .005" resolution,
mine ran out of gas years ago.

I switched to toner transfer.
I use a laserjet 4MP at 600DPI with real HP toner.
I use a gerber viewer and have it print with
a scale factor of .987. The printout isn't exactly 1:1.

The secret is in the paper. Hardest part is to get
the paper off the copper without taking the traces.
Shiny photo-paper works best. There's lots of discussion
of this on the web.

I augment the etch with an eyedropper. Getting uniform etch
is difficult if you just stick it in the etchant. The
plastic traces aren't as robust as one might like.
There's not a whole lot of room for overetch with narrow traces.

Takes some experimentation, but turnaround is quick and cheap.

I'd worry about ink thickness with a rubber stamp.
I don't know anything about the ink suggested, but the stuff
that worked best in the pen plotter was so nasty that they
couldn't sell it in the USA. I got one plot session out of a pen
before the plastic dissolved.

 

[Adrian Nievergelt]

Much to my horror, I discovered one of my 'standard' PCB footprints
is 40% larger in both dimensions than the part it was intended for
on my latest board.
(It's a 16 pin device with pads on 0.5 mm centers)!
The board is about 70% assembled and is much later than I would prefer.
If I had unlimited time and money, I'd just spin the
artwork with the correct footprint. Lacking either, I want to
patch the existing board.

I've not located a prototype board that has the necessary footprint
terminating in matching pads.

http://www.rubberstampchamp.com/ProductDetail.aspx?productid=HS-
ROUND-3-4INCH-ARTWORK

Here's what I'm thinking. I can upload a big bitmap of the
correct footprint with exactly the right terminating pads to
these guys and stamp a piece of my 0.005" PCB stock using say
Epson Durabrite yellow ala:
http://www.pabr.org/pcbprt/pcbprt.en.html

Then I can stamp and etch as many adapter boards as I want.
(I need one [1] adapter board.)
Mount the device to the adapter then solder the terminating
pads to the huge pads of the incorrect footprint.

What do you think of this idea?

--Winston

I've used two prototyping methods in the past. One is to use
prelaminated PCB material that i illuminated with a simple uv tube. As
mask i used normal transparencies that were printed on a
thermosublimation printer, as they put down a pretty decently thick film
of ink with little sidewallroughness.
The other method is actually cooler, involving a milling system (doesn't
need to be super fancy LPKF ones), where you would just do isolation
milling. The cool thing about this one is that once you take your board
out it's even already got holes. Obviously no conductive vias possible
with both methods.

 

[Nico Coesel]

Much to my horror, I discovered one of my 'standard' PCB footprints
is 40% larger in both dimensions than the part it was intended for
on my latest board.
(It's a 16 pin device with pads on 0.5 mm centers)!
The board is about 70% assembled and is much later than I would prefer.
If I had unlimited time and money, I'd just spin the
artwork with the correct footprint. =A0Lacking either, I want to
patch the existing board.

I've not located a prototype board that has the necessary footprint
terminating in matching pads.

http://www.rubberstampchamp.com/ProductDetail.aspx?productid=3DHS-
ROUND-3-4INCH-ARTWORK

Here's what I'm thinking. =A0I can upload a big bitmap of the
correct footprint with exactly the right terminating pads to
these guys and stamp a piece of my 0.005" PCB stock using say
Epson Durabrite yellow ala:http://www.pabr.org/pcbprt/pcbprt.en.html

Then I can stamp and etch as many adapter boards as I want.
(I need one [1] adapter board.)
Mount the device to the adapter then solder the terminating
pads to the huge pads of the incorrect footprint.

What do you think of this idea?

--Winston

think its more trouble than getting a real pcb made and less likely to
work

My idea exactly. When I have a small PCB made I order two panels. The
PCB house I order from has a special prototyping tarif where you get
two boards. Because of the start-up costs the price for a small PCB is
the same as a larger one (say the size of a euro card). So for around
€100 I get 10 to 30 PCBs. I estimate that a simple adapter board is so
small you can easely fit 30 pieces in one panel.

 

[Ian Malcolm]

Much to my horror, I discovered one of my 'standard' PCB footprints
is 40% larger in both dimensions than the part it was intended for
on my latest board.
(It's a 16 pin device with pads on 0.5 mm centers)!
The board is about 70% assembled and is much later than I would
prefer. If I had unlimited time and money, I'd just spin the
artwork with the correct footprint.  Lacking either, I want to
patch the existing board.

I've not located a prototype board that has the necessary footprint
terminating in matching pads.

http://www.rubberstampchamp.com/ProductDetail.aspx?productid=HS-
ROUND-3-4INCH-ARTWORK

Here's what I'm thinking.  I can upload a big bitmap of the
correct footprint with exactly the right terminating pads to
these guys and stamp a piece of my 0.005" PCB stock using say
Epson Durabrite yellow ala:http://www.pabr.org/pcbprt/pcbprt.en.html

Then I can stamp and etch as many adapter boards as I want.
(I need one [1] adapter board.)
Mount the device to the adapter then solder the terminating
pads to the huge pads of the incorrect footprint.

What do you think of this idea?

--Winston

think its more trouble than getting a real pcb made and less likely to
work

try Press-n-Peel if you just need a quick simple pcb

-Lasse

For a one-off, if this is a leaded part (not J leaded) just glue it to
the board and hand wire it. 0.5 mm is tricky to hand wire, but far
from impossible IF you use fine enough solid tinned wire (e.g. a single
strand taken from some stranded wire) and have an iron with a fine
enough bit. No coffee that morning!

 

[Rich Webb]

For a one-off, if this is a leaded part (not J leaded) just glue it to
the board and hand wire it. 0.5 mm is tricky to hand wire, but far
from impossible IF you use fine enough solid tinned wire (e.g. a single
strand taken from some stranded wire) and have an iron with a fine
enough bit. No coffee that morning!

This is probably the best chance, especially since the wiring should be
all straight out from the pin to the pad. At least, I'd try this first.

For the other approaches:

0.5 mm is probably beyond the capability of a rubber stamp to lay down
with sufficient accuracy or to leave a uniform resist layer on bare
copper.

Press-n-Peel Blue <http://www.techniks.com/> could probably do it but it
requires some finesse with fine pitch artwork as well as a laser printer
(not ink jet) that uses a compatible toner. Helps to have a hot-roller
laminator, too; much better reproducibility than a clothes iron. I have
done 10/10 rule boards with P-n-P but other than the glory of doing it,
it's a P-in-the-A. Pro tip: bare copper isn't the sweetest thing for
soldering, especially fine pitches. MG Chemicals has a "liquid tin"
solution that can be used to lay down a thin tin layer which makes the
soldering easier. More shelf-stable, too, as compared to the dry mix
tinning powder.
<http://www.mgchemicals.com/products/prototyping-and-circuit-repair/prototyping/liquid-tin-421/>

There are carrier boards that might be adopted, e.g., over at
<http://www.beldynsys.com/snapapart.htm>.

But I'd bite the bullet and get a "real" PCB fabricated. There's a time
versus cost trade-off, of course. A double sided, silk-screened, solder
masked prototype from Sunstone could be in your hands fairly quickly or
go with SeeedStudio for a really low cost ($1/board for 5cm x 5cm) but a
longer wait.

<http://www.sunstone.com/>
<http://www.seeedstudio.com/depot/fusion-pcb-service-p-835.html?cPath=185>

 

[petrus bitbyter]

Much to my horror, I discovered one of my 'standard' PCB footprints
is 40% larger in both dimensions than the part it was intended for
on my latest board.
(It's a 16 pin device with pads on 0.5 mm centers)!
The board is about 70% assembled and is much later than I would prefer.
If I had unlimited time and money, I'd just spin the
artwork with the correct footprint. Lacking either, I want to
patch the existing board.

I've not located a prototype board that has the necessary footprint
terminating in matching pads.

http://www.rubberstampchamp.com/ProductDetail.aspx?productid=HS-
ROUND-3-4INCH-ARTWORK

Here's what I'm thinking. I can upload a big bitmap of the
correct footprint with exactly the right terminating pads to
these guys and stamp a piece of my 0.005" PCB stock using say
Epson Durabrite yellow ala:
http://www.pabr.org/pcbprt/pcbprt.en.html

Then I can stamp and etch as many adapter boards as I want.
(I need one [1] adapter board.)
Mount the device to the adapter then solder the terminating
pads to the huge pads of the incorrect footprint.

What do you think of this idea?

--Winston

Seems too risky. Best way is a professional double sided PCB with the actual
components footprint on top and the main PCB footprint on the bottom. If you
place the vias in the middle of the pads of the bottom and (let) mill the
outer half of them, you can solder them pretty easely.

BTW What's the component you're talking about or at least what's its
package?

petrus bitbyter

 

[Winston]

On Sun, 12 Aug 2012 00:22:42 -0700, mike wrote:


(Toner Transfer)

Takes some experimentation, but turnaround is quick and cheap.

OK. Thanks for your thoughts, Mike.

I'd worry about ink thickness with a rubber stamp. I don't know anything
about the ink suggested, but the stuff that worked best in the pen
plotter was so nasty that they couldn't sell it in the USA. I got one
plot session out of a pen before the plastic dissolved.

Whoa.

--Winston

 

[Winston]

On Sun, 12 Aug 2012 02:07:41 -0700, Adrian Nievergelt wrote:

(...)

I've used two prototyping methods in the past. One is to use
prelaminated PCB material that i illuminated with a simple uv tube.

I tried that and actually made a usable PCB.
Then I learned all about 'actinic keratosis'.

As mask i used normal transparencies that were printed on a
thermosublimation printer, as they put down a pretty decently thick film
of ink with little sidewallroughness. The other method is actually
cooler, involving a milling system (doesn't need to be super fancy LPKF
ones), where you would just do isolation milling.

I used my cnc mini-mill to make a couple double-sided boards that way.
Very time consuming but *much* better quality than anything I have
tried, before or since. (With the exception of PCB houses of course.)

The cool thing about this one is that once you take your board out
it's even already got holes.

Properly centered, I might add.

Obviously no conductive vias possible with both methods.

Axial component leads and wires placed through holes work well.

Thanks, Adrian.

--Winston

 

[Winston]

(...)

OK. Thanks!

My idea exactly. When I have a small PCB made I order two panels. The
PCB house I order from has a special prototyping tarif where you get two
boards. Because of the start-up costs the price for a small PCB is the
same as a larger one (say the size of a euro card). So for around €100 I
get 10 to 30 PCBs. I estimate that a simple adapter board is so small
you can easely fit 30 pieces in one panel.

I had this board made 'way out of country as well.
Fab was very speedy but the board still took 8 weeks to get to me.

Yikes.

Thanks guys.

--Winston

 

[Winston]

Your original post didn't get here in it's entirety and I don't
know why that is. Thanks for your thoughts.

Sounds like a futile exercise to me, given the thermal conductivity
of the wire and its resulting tendency to part company with the
component when soldering to the pad ~ 1 mm away (or vise versa).

This is probably the best chance, especially since the wiring should be
all straight out from the pin to the pad. At least, I'd try this first.

For the other approaches:

0.5 mm is probably beyond the capability of a rubber stamp to lay down
with sufficient accuracy or to leave a uniform resist layer on bare
copper.

One stamp vendor claims resolution of better than 0.045 mm.
(I agree that it would look very rough.)

Press-n-Peel Blue <http://www.techniks.com/> could probably do it but it
requires some finesse with fine pitch artwork as well as a laser printer
(not ink jet) that uses a compatible toner. Helps to have a hot-roller
laminator, too; much better reproducibility than a clothes iron. I have
done 10/10 rule boards with P-n-P but other than the glory of doing it,
it's a P-in-the-A. Pro tip: bare copper isn't the sweetest thing for
soldering, especially fine pitches. MG Chemicals has a "liquid tin"
solution that can be used to lay down a thin tin layer which makes the
soldering easier.

Yup. I've used that stuff and like it too.

More shelf-stable, too, as compared to the dry mix
tinning powder.
<http://www.mgchemicals.com/products/prototyping-and-circuit-repair/ prototyping/liquid-tin-421/>

There are carrier boards that might be adopted, e.g., over at
<http://www.beldynsys.com/snapapart.htm>.

http://www.beldynsys.com/p532.jpg Say! That'd work!
Sixty smackers! (Thud)

But I'd bite the bullet and get a "real" PCB fabricated. There's a time
versus cost trade-off, of course. A double sided, silk-screened, solder
masked prototype from Sunstone could be in your hands fairly quickly or
go with SeeedStudio for a really low cost ($1/board for 5cm x 5cm) but a
longer wait.

<http://www.sunstone.com/>
<http://www.seeedstudio.com/depot/fusion-pcb-service-p-835.html?

cPath=185>

OK. Thanks, Rich.

--Winston

 

[Winston]

On Sun, 12 Aug 2012 16:03:42 +0200, petrus bitbyter wrote:

(...)

Seems too risky. Best way is a professional double sided PCB with the
actual components footprint on top and the main PCB footprint on the
bottom. If you place the vias in the middle of the pads of the bottom
and (let) mill the outer half of them, you can solder them pretty
easely.

OK, thanks.

BTW What's the component you're talking about or at least what's its
package?

It's the LTC5564IUD in the 16 pin 3 mm QFN package with pads on
0.5 mm centers:
http://www.linear.com/product/LTC5564
http://cds.linear.com/docs/Datasheet/5564fa.pdf

Thanks, Petrus.

--Winston

 

[Winston]

On Sun, 12 Aug 2012 09:54:22 -0700, John Larkin wrote:

(...)

Why not have some real PCBs made? AP Circuits will do a few boards for
under $100, plated through. Put a bunch of patterns on each board, and
shear them up.

I've bought from AP Circuits as well and have been delighted with
their quality. I will have to check back with them now because I
recall they didn't allow milled channels inside a board in the
past, for their inexpensive prototypes.

We did a few thousand of this adapter:

https://dl.dropbox.com/u/53724080/PCBs/Break2.jpg

https://dl.dropbox.com/u/53724080/PCBs/OnBoard.jpg

Kewl! Looks like Petrus and you think similarly.

(Perhaps I will just spin the artwork and send it to AP Circuits.)

Thanks, John.

--Winston

 

[Rich Webb]

Sounds like a futile exercise to me, given the thermal conductivity
of the wire and its resulting tendency to part company with the
component when soldering to the pad ~ 1 mm away (or vise versa).

What I've done in the past when I've needed similar small "jumpers" is
to strip a length of 30 AWG wire-wrap wire and then, with some shaping
of the curve, a piece or two of Kapton tape (high melting point), and
holding my tongue just right, lay the wire so that it bridges the gap
between the pin and pad and stays in place *before* applying solder.
Flux it up and then sort of drag-solder both the pin and pad. Tada!
Doing this to 0.5mm pitch pins is challenging but quite do-able.

 

[Ian Malcolm]

Sounds like a futile exercise to me, given the thermal conductivity
of the wire and its resulting tendency to part company with the
component when soldering to the pad ~ 1 mm away (or vise versa).

Not futile at all. Start with a reasonably long piece of wire, solder
wire to pin, bend it till it lies freely along and in contact with the
pad, tape down across the wire beyond the pad to hold it there and
solder it to the pad. Yes the joint to the pin may reflow, but there
isn't enogh stress in the wire to spring it clear and the tape stops it
going anywhere. After the wire is soldered down, cut off the excess
with a very small scalpel or a small flat-blade jeweller's screwdriver
sharpened to a polished knife edge.

 

[Winston]

On Sun, 12 Aug 2012 19:05:25 +0000, Ian Malcolm wrote:

(...)

Not futile at all. Start with a reasonably long piece of wire, solder
wire to pin, bend it till it lies freely along and in contact with the
pad, tape down across the wire beyond the pad to hold it there and
solder it to the pad. Yes the joint to the pin may reflow, but there
isn't enogh stress in the wire to spring it clear and the tape stops it
going anywhere. After the wire is soldered down, cut off the excess
with a very small scalpel or a small flat-blade jeweller's screwdriver
sharpened to a polished knife edge.

Thanks, Rich and Ian.

You guys have much more faith in my solder-fu than I do.

It's the LTC5564IUD in the 16 pin 3 mm QFN package with pads on 0.5 mm
centers:
http://www.linear.com/product/LTC5564
http://cds.linear.com/docs/Datasheet/5564fa.pdf

Side-glued 30 AWG kynar would get me my .5 mm spacing
with little error but I'd still be worried about capacitive
coupling to the central ground plane, particularly on the
RF input (pin 1). Were this an analog, or even low-freqency
digital issue, I'd consider trying that.

As it is, I think I'll just spin the PCB.

Thanks for your thoughts on this, guys.


--Winston

 

[Ian Malcolm]

On Sun, 12 Aug 2012 19:05:25 +0000, Ian Malcolm wrote:

(...)


Thanks, Rich and Ian.

You guys have much more faith in my solder-fu than I do.

It's the LTC5564IUD in the 16 pin 3 mm QFN package with pads on 0.5 mm
centers:
http://www.linear.com/product/LTC5564
http://cds.linear.com/docs/Datasheet/5564fa.pdf

Side-glued 30 AWG kynar would get me my .5 mm spacing
with little error but I'd still be worried about capacitive
coupling to the central ground plane, particularly on the
RF input (pin 1). Were this an analog, or even low-freqency
digital issue, I'd consider trying that.

As it is, I think I'll just spin the PCB.

Thanks for your thoughts on this, guys.


--Winston

Ah, QFN yechh! No I wouldn't care to do that myself. I DID say any leaded
package (except J lead)!

I see that it has a RF input of between 600 MHz and 15 GHz. Even with a
high precision double sided PTH daugerterboard that's far from being a sure
thing. A respin is probably cheaper and is certainly the best way
forward.

 

[Winston]

On Mon, 13 Aug 2012 10:13:55 +0000, Ian Malcolm wrote:

(...)

Ah, QFN yechh! No I wouldn't care to do that myself. I DID say any
leaded package (except J lead)!

So you did.

I should have played fair and revealed the package type on
my first post. Sorry about that. I wish I'd chosen the
'small', accurate 'QFN' library footprint instead of the
'incredibly enormous', accurate 'QFN' library footprint.

I see that it has a RF input of between 600 MHz and 15 GHz. Even with a
high precision double sided PTH daugerterboard that's far from being a
sure thing. A respin is probably cheaper and is certainly the best way
forward.

While the respin is being fabbed, I can do firmware development on the
present board.

Thanks again.

--Winston