Assembled board with routing error, RS485 A/B swapped

[invzim]

I guess everyone does this once  .


I've had a 100 of these made, but unfortunately I made a mistake swapping two signals, as the SOIC-8  Eagle part in schematics was numbered 1,2,3,4,5,7,6,8 which I didn't catch before I got the boards back.  This is the A/B rs485 lines.

After facepalming for an hour, I googled this and it seems swapping A/B is a pretty common error, with a possible software fix.  RS485 is differential, so in theory just flipping the bits with an xor should do the trick. 

This is how it looks:
Working board:

Board with A/B swapped:

It indeed looks promising, but flipping the data didn't fix the leading edge, which was still in the wrong direction messing everyting up:

Bummer..

So my list of options is down to:

• Cut the traces and solder on kynar
• Ditch the boards, facepalm and pay up for new ones
• OR, get new boards assembled but without the MCU and manually move these over

I'm crossing my fingers option 3 is cheapish, they should be able to use the same stencil and the MCU is by far the most expensive component. 

What do you guys think, am I too fuzzed about the cutting traces and doing wire rework?

[mikeselectricstuff]

What baudrate are you using ? If not too high you could use a soft UART so you can invert in software.
A few UARTS  have polarity control, but I don't think PIC18 is one of them.

[invzim]

I was actually not connecting the startbit with the waveform, or thinking about it at all - maybe this is fixable  after all
The baud is 115200 (and has to be) - I haven't done bit banging like this before, but will get on the startbit polarity asap. 
thanks!

[invzim]

A few UARTS  have polarity control, but I don't think PIC18 is one of them.

The 18f is INDEED one of them.

Code: [Select]

REGISTER 20-3: BAUDCON: BAUD RATE CONTROL REGISTER
bit 5 RXDTP: Received Data Polarity Select bit
Asynchronous mode:
1 = RX data is inverted
0 = RX data received is not inverted
bit 4 TXCKP: Clock and Data Polarity Select bit
Asynchronous mode:
1 = TX data is inverted
0 = TX data is not inverted

Even had the registers defined in the C18 header files making this a 2 minute fix, what a relief!! 

BAUDCONbits.RXDTP = POLARITY;
BAUDCONbits.TXCKP = POLARITY;

If you like beer, PM me your address and I will see to it that some is delivered to you!

[David_AVD]

Ah, don't we all like a good result!   

[Short Circuit]

Lucky man! Not often that you can cancel a hardware error this easily.

[AndyC_772]

I've done it too, and on a commercial product - but fortunately the UART was implemented in an FPGA, so the fix was just as trivial.

The moral of the story is: not every device / vendor has the same relationship between A, B, +ve and -ve. Just because 'A' is the +ve half of the pair on one chip doesn't mean it'll be the same on some other chip.