I was really worried about my new ADC (ADS 1115) when the ports behaved very strangely, but pretty soon I realised that they were working fine; they were just misidentified in the silk screen on the board. They were not merely reversed, but scrambled:
A3, A2, A1, A0 on the edge of the board turned out actually to be
A0, A3, A2, A1
OK, no harm done, just a few moments of head scratching. I am pleased with the device otherwise, it gives me a good stable non-jittery reading for a pot (and soon for other sensors). The Due's analog ports are apparently notoriously noisy, so this is a cheap fix.
But I wondered in passing, how common is it for silk screen to be wrong? Potentially, I suppose it might lead to blowing stuff up if the vcc and ground labels are in the wrong places!
So just idle curiosity: what's the usual incidence of silk screen booboos? This was a cheap Chinese part from Ebay ... I'm suspecting that one of the reasons these parts are so cheap is because they are "seconds" from the production line. If a little paint in the wrong place is its only problem, I suppose I should feel lucky
On products designed with care? Incidence is very low.
On el-cheapo crap from china? Pretty common. It's very unusual to see, like you mentioned, Vcc/Gnd swapped or something, but other pins being mislabeled isn't rare. I've also seen words misspelled or mirrored (I'm sure we've all seen those crappy 433mhz transmitters with the DATA pin labeled "ATAD")
This is not a production problem, but a design problem - they screwed up the silkscreen layer, and didn't notice, or didn't notice until they had the boards back and were too cheap to pitch them and respin it with the correct markings. It is really easy to screw up the silkscreen - I've botched that on at least a third of the board I designed. Hell, I once ordered 600 boards (with the extra cost ENIG surface treatment) and then noticed when I got them back that there was no silk on one side because I didn't include tSilk (only tNames and tPlace)... On another one I labeled QFN-20 as QFN-25. Still got some of both kicking around - I've been selling them at buy-one-get-one-free and everyone still picks the ones with the correct silk
There are hundreds of these "Breakout boards" or "Electronic Bricks" etc. and their build and silkscreen quality vary a lot.
I'm fortunate to have a partner in China who has a good relationship with one of the better Arduino parts designers (Usually called "Keyes" or "KeyesStudio"). Their design work and silkscreen printing is usually very good. Example:
I have worked with them on an Arduino UNO compatible variant that has 3-pin connectors on every I/O, I2C connector etc. and a nice 5V 2A switching power supply. They have made thousands of these for me with my silkscreen artwork, and custom Logos for one of my large customers. I can send them a photoshopped image showing what I want and they will make a good silkscreen. They even dropped my wife's cute Robot logo on it:
OK I have been warned. I'll remember that unexpected behaviour might not be hardware failure, but misleading silk! I thought it was kind of funny that they couldn't even get the traces in order, i.e. neither 0123 nor 3210.
I thought it was kind of funny that they couldn't even get the traces in order, i.e. neither 0123 nor 3210.
A lot of break-out boards are made with "simplified" routing, which means that the the pins are connected to the chip in whatever order is most convenient for the PCB design, rather than anything logically consistent...
