As a follow up to the last page, the system is currently (5-Nov-13) at revision C which is basically revision B with a few silly/obvious mistakes taken out.
The real reason for getting a revision C board was due to my abysmal failure to build the PCB using my usual surface mount methods. The smallest pitch I’ve ever done at home is 0.8mm, but mostly 1mm or bigger, now that I remember it a bit of a struggle and required quite a bit of solder wick to remove solder bridges even at 0.8mm. The Current PCB has several devices and connectors down at 0.5mm pitch and my first attempt was a washout due to excessive solder paste with just about every pin on all the 0.5mm devices having solder bridges. (I’m too embarrassed to even post a photo!)
The solder paste was quality lead-free stuff that cost me $50 at element14 rather than my usual $5 e-bay specials, but the solder was not at fault I was. The second attempt was a little better, I used a much finer nozzel on the paste syringe, but it was still a disaster, the 0.5mm stuff really only needs a sniff of solder paste so with the Rev C board I have ordered a solder stencil in order to apply the correct dosage of solder paste before I place the components.
Unfortunately both the WIFI chips soldered perfectly but could not be de-soldered once on the board no matter how hard I tried and as these little buggers were $36 each I was mighty pissed at myself!
I wanted to use seeed studios again as they really are producing quality PCB’s at the moment and delivery times are sub 2-weeks from order to delivery, however the cost of the stencil and delivery would have cost me a minimum of $140 ($80 + cheapest delivery) the delivery charge seemed excessive but the site said the item weighed 1.5Kg which I’m sure is some sort of mistake.
The other guys I’ve heard good things about are a company called smart prototyping, the ones who I got by reflow oven off. These guys do the stencil for $20 and the cost of 10 boards (their minimum) with better plating and the stencil and UPS delivery came to $90
My timescale for the project is currently being revised forward as I struggle with other work, however this is more interesting and I’m still hoping to have a couple of working prototypes for the 2014 vintage (which is approx April 2014 down here in the southern hemisphere)
Anyway, as you can see from the revision C schematics it is a very simple circuit, basically a CPU with a couple of serial interfaces for the WIFI and barcode and a small amount of dedicated logic to read in the manchester encoded 2KBaud RFID signal, the rest is power supply.
Most other MPU/TFT designs seem to use the 16 or 18 bit connection mode to the displays, the 64 pin PIC32’s only have 8 Parallel Master Pins (16-bits require the 100 pin devices) and so I am going with an 8 bit interface to the TFT, logically this will mean updates will be 50% slower than other designs, however in practice it is really not going to matter. I was almost tempted to use SPI and to see if I could squeeze the design down to a 44 pin device, but I’ve enough to do really.
As another exercise though I have created a circuit and PCB layout for the marson 3.3V barcode which does not require the level converters or boost regulator so should be about the minimum component count possible. This design allows for a smaller and thinner final unit which is a bonus and I may get this made in parallel to test, also looking to possibly use a different Wi-Fi card such as the wiznet module which I recently obtained but dont yet have eagle library part for.