Homebrew PCBs

As an experiment, I'm designing a home-made single-sided PCB for a
Philips LPC2106 ARM device. It has an LQFP48 package (0.5 mm pin
spacing). I've just made a test PCB (no tracks), there were no shorts
visible between the pads so it looks like my technique is viable for
fine-pitch devices.

I used the rather expensive JetStar Premium film in my Epson Photo 780
inkjet printer, UV exposure onto Economy Fotoboard PCB material, and
FeCl3 etchant.

Although the Philips LPC21xx chips can run at 60 MHz, all the
high-speed stuff is inside the chip, so it should work OK on a simple
single-sided PCB, given adequate decoupling for the 1.8 V and 3.3 V
supplies. They work fine on properly designed double-sided PCBs.

Leon

Hi Leon,


Last year I made a small rectangular single-sided board containing an 
LPC2105, some decoupling, a 1.8V regulator and the crystal and its 
associated caps. I brought all the GPIOs out to the longer edges of the 
board and soldered in some long-pin SIL strips so that the whole 
arrangement is like a DIP48 package. It worked well for prototyping -- 
I've got it plugged into a stripboard, with an MMC card (7.5MHz clock) 
and some other peripherals hanging off it. It seems to work flawlessly 
-- I still use it most days.

I guess the point is that it's entirely feasible to put something like 
what you propose on single-sided FR2 board and run it at a reasonable 
speed. The off-board MMC port runs at 7.5MHz (the cpu's limit given a 
60MHz cclk) without errors -- I guess that's the fastest signal on my board.

My process was photo-etching (Canon i560 for artwork, generic 
transparency sheets, DIY UV box, DIY bubble-etcher with FeCl3 etchant, 
cold-plated the board etc) and hand-soldering with my bargain-basement 
17W Antex iron. It took about an afternoon to build and populate the 
board and bring it up -- I was pleasantly surprised!

I think I recognise you from the LPC group :)


Stuart

0.5mm spacing is no problem, even with TT, but you need at least 600dpi on  
the printer side.

As you say the bigger challenge is the no-PTH thing. In your case you  
should be fine, but some of the QFNs and other cases with exposed thermal  
pad for cooling are problematic. Also, the newer chips that use these  
small cases are sometimes really fast and some have very strict  
layout/decoupling requirements, to the extent that it is recommended how  
to run supply traces, where to put vias and capacitors, and the lot. Some  
even use vias as part of filters for the supply. Those recommendations are  
always assuming you have industrial boards and automatic  
placement/soldering. I can't always folow them but so far everyhthing  
still worked.

BGAs are another thing, i see no way to use them if they have "surrounded"  
contacts. But luckily it looks as if they are not that popular anymore  
with ics up to say 50 pins.

ST

On Tue, 31 Jan 2006 16:34:22 +0100, leon_heller  
<leon.heller@...> wrote:

> As an experiment, I'm designing a home-made single-sided PCB for a
>
> Philips LPC2106 ARM device. It has an LQFP48 package (0.5 mm pin
>
> spacing). I've just made a test PCB (no tracks), there were no shorts
>
> visible between the pads so it looks like my technique is viable for
>
> fine-pitch devices.
>
>
> I used the rather expensive JetStar Premium film in my Epson Photo 780
>
> inkjet printer, UV exposure onto Economy Fotoboard PCB material, and
>
> FeCl3 etchant.
>
>
> Although the Philips LPC21xx chips can run at 60 MHz, all the
>
> high-speed stuff is inside the chip, so it should work OK on a simple
>
> single-sided PCB, given adequate decoupling for the 1.8 V and 3.3 V
>
> supplies. They work fine on properly designed double-sided PCBs.
>
>
> Leon
>