Lists of PCB services by country?

Well the maths is clear --- for someone who makes project boards for themselves regularly, a PCB mill could pay for itself very rapidly indeed.  If that PCB mill also happens to be a 3D printer and a general purpose lightweight CNC all at the same time, it'll pay for itself almost instantly since you'd just be using an existing domestic facility.

I agree that we're not there yet, but things seem to be moving strongly and rapidly in that direction, including attaining the required level of precision.  Multilayer ... now there's a quandary.

Do some Jedi-level OSHW enthusiasts laminate up multiple 1 or 2-sided boards?

Morgaine.

Morgaine Dinova wrote:

Do some Jedi-level OSHW enthusiasts laminate up multiple 1 or 2-sided boards?

It's possible: http://megauk.com/multilayer_press.php  not sure my pockets are deep enough to justify it though.

For home production, even if you do have that miracle machine to mill them, you also have to consider through hole plating which is another part that's not cheap and requires interesting chemicals.

Topologically speaking, can every circuit be routed on 2 layers, given enough real estate?

I'm pretty certain that this is something that I should know, but alas I don't --- perhaps I was asleep that day.  Of course at high speed where track length and impedance mismatch and clock skew are important, using extra real estate to get a complex circuit onto 2 layers may be total anathema.

Nevertheless, in the OSHW world where board area and circuit density are rarely driving factors, it's worth considering the KISS principle, of which sticking to 2 layers is a part.  Multilayer boards are of course efficient in many ways and definitely "cool", but "cool" is not an engineering figure of merit.  Multilayer doesn't aid HW debugging either.

Morgaine.

PS. I notice that Olimex try to get their OSHW boards onto 2 layers whenever possible.

Morgaine Dinova wrote:

 

Topologically speaking, can every circuit be routed on 2 layers, given enough real estate?

Maybe, but it can start to get complex quite quickly.

High pin count devices or BGA's may make it practically impossible to escape all the necessary signals in the available space. Not necessarily topologically impossible, but eventually you can only etch tracks down so far before you hit the limits of your physical process

In a lot of cases, a four layer board will have the inner two layers essentailly being power and ground, leaving only two layers for signals anyway.

There are gerbers available for the minnowboard, it's an 8 layer board and is interesting to look at to see how some of the problems have been solved. There's effectively four power/ground type layers, the inner two of which are split in different ways, likely as a means to route various voltages to different areas, with two ground planes on layers 2 & 7.

selsinork wrote (on laminating PCBs):

 

It's possible: http://megauk.com/multilayer_press.php  not sure my pockets are deep enough to justify it though.

 

For home production, even if you do have that miracle machine to mill them, you also have to consider through hole plating which is another part that's not cheap and requires interesting chemicals.

I wonder sometimes whether the future of OSHW isn't at the thin film level or below, because multilayer boards really aren't all that nice to us --- even dead-bug linking of packages (or of dies?) might be better.

I know there is a future at the nanoscale level since hardware reverts to being more or less software at that point, but we're not there yet.  Perhaps we should accelerate the process.  There would be research involved, but it's not hugely different to the R&D that RepRappers do already, just lower level and needing some more physics.  It's not really a problem, in the sense that it's not magic nor needs CERN-type funding, and with an explorer's mindset would be fun.

Morgaine.

PS. You may also have noticed the recent advances in thin-film flexible circuitry, published in the last week or two.  The realm of possibilities is really widening a lot, and among those many possibilities there is probably a sweet spot for OSHW.

The problem with home brew pcb technology (especially milled but also etched) is the precision required - low cost prototoype services offer between 0.15 and 0.3 mm track and gap design rules. To use 0.5mm bga packages you need to operate with 0.1 or 0.075 mm design rules.

You can't mechanically mill gaps that small with any reasonable machine and you can't get to the required 1/4 design rule precision with low cost machines. So this means that you can't even deal with 0.5mm pitch TQFP packages which are easily hand solderable.

Since all the chips you can buy (well nearly all) expect to be soldered on to a pcb thee OSHW is pretty much bound to be pcb based unless thta changes.

Additive processes for PCBs are in their infancy (to put it mildly) and still have the precision problem to overcome.

One can just about imagine  a printer type approach but you will need to develop solderable inks (or  a solder substitute glue and process) and a layer interconnect technology. Sounds more like CERN than RepRap level funding to me

My work is commercial but I don't find the cost of pcbs (and I buy all my prototypes in the UK) to be an issue - although it is often comparable with, or more than, the cost of all the electronic components on a board - so I don't see a huge commercial demand for other prototype processes.

I can see it's an issue for home brew stuff.

MK

Topologically speaking, can every circuit be routed on 2 layers, given enough real estate?

Interesting question.

'every': no.

'most': yes.

For performance analog circuits, a solid ground plane is a must. For most hardware that needs EMC certification, having a GND plane is a large benefit, as you can route decoupling caps easier and with lower induction, which is good for HF behaviour.

Michael Kellett wrote:

 

Additive processes for PCBs are in their infancy (to put it mildly) and still have the precision problem to overcome.

Around 1990 I saw an Ariel Electronics (California) Circuit Writer in operation.  The Circuit Writer extruded conductive plastic wires onto a substrate, and could cross wires on the same layer.  I don't think the technology got anywhere, but maybe it was just ahead of its time and with newer 3D extrusions this could be done practically.  The patents have probably all expired by now.

Having spent many years in the board assembly industry, I've seen a couple of additive methods used practically. Normally only as some sort of rework to a normal pcb though.  The most common one deposited some sort of rough looking substance onto the pcb then drew lines on top of that in what would be best described as conductive paint.  I never did find out exactly how it was done, and as Michael points out the precision was a problem.

Part of the problem with doing things this way is that while we're still soldering components to the board, whatever the result is it needs to be able to survive a reflow oven and/or a solder wave. Or a hobbiest with a gas torch instead of a soldering iron

Victor Sluiter wrote:

Interesting question.

'every': no.

'most': yes.

I think that just taking 'topologically' and the implied 'as much space as you need' the answer is probably Yes.  Morgaine made the mistake of not also asking for the result to actually work

I'm definately in agreement that doing layouts that either work or meet the relevant regulations is going to be impossible on two layers in various circumstances