Difference between revisions of "MultiSerial Shield"

From Hackstrich
(Stop mask will hopefully fix the text readability. Double sided silkscreen more than doubles the board cost so is not realistic.)
(Changed strategy on bottom text.)
Line 40: Line 40:
 
* Fixed in revision 4: <s>BOM doesn't contain 1x3 header for INT jumper</s>
 
* Fixed in revision 4: <s>BOM doesn't contain 1x3 header for INT jumper</s>
 
* Removed in revision 4: <s>May be able to remove the I2C pullup resistors and use the built-in Arduino pullups</s>
 
* Removed in revision 4: <s>May be able to remove the I2C pullup resistors and use the built-in Arduino pullups</s>
* Changed in revision 4 w/ addition of stop mask (hopefully this should make it readable): <s>Copper text on the bottom is completely unreadable w/ black solder mask</s>
+
* Changed text to stop mask layer in revision 4: <s>Copper text on the bottom is completely unreadable w/ black solder mask</s>
  
 
== Overview ==
 
== Overview ==

Revision as of 02:46, 15 November 2010

MultiSerial Shield
MultiSerial Shield.png
This Page
Function: Provide two RS232 ports with 3.3v or RS232 levels, accessible via I2C.
Key Chips: NXP SC16IS752
External Ports: 2x RS232 (3.3v or RS232 levels), 8x GPIO
Required Pins: A4 (I2C SDA), A5 (I2C SCL)
Optional Pins: D2, D3 (Interrupts)

Project Status

Revision 2 PCBs have been received, first one assembled (using an SC16IS762 because the 752 was out of stock and they're pin-and-software-compatible) and passes CTLR, SERI, FIFO, INTR, and GPIO tests with no major hardware bugs found so far. Quick validation at RS232 levels has been done, but a better unit test needs to be written for this. Revision 3 PCBs (production RC1) received from MyRO and parts received from Digikey. Revision 3 PCBs have a fatal issue (argh), the I2C level shifter pads are too far apart for the current batch of the chip to fit! The last batch had extra-long leads it looks like, but this one does not. I'm going to have to do a Revision 4 of the PCB and eliminate the level shifting on the input all together (as the controller doesn't need it (anymore?) anyway).

Revision 1 PCB Issues

  • Fixed in revision 2: Should connect GND pins on Arduino header together
  • Fixed in revision 2: +5V and GND labels on RS232 port are swapped
  • Fixed in revision 2: Pins 15-28 on controller IC are mirrored! Doh!
  • Fixed in revision 2: Fix needlessly close spacing on traces running between pins for no reason.
  • Fixed in revision 2: SCL33/SDA33 test points/jumper points are needlessly bigger than other pads.
  • Fixed in revision 2: Addresses printed on board are 8 bit, not 7 bit like the Wire library uses.
  • Added in revision 2: Functions for Arduino headers not printed on board, makes debugging harder.
  • Fixed in revision 2: C7 and C12 are needlessly too close to the header, making it hard not to melt the header (if placed first) while adding the caps.
  • Fixed in revision 2: RX level translators/jumpers not designed right. TX and RX both need to be switched between the output of the level translator and the input of the level translator. TX already is, RX is not.
  • Fixed in revision 2: LEDs wired such that they are on for idle state.
  • Added in revision 2: Power LED would be very useful as shield blocks visibility of Arduino's power LED.
  • Added in revision 2: No interrupt capability because /IRQ line not connected.
  • Fixed in revision 2: Channels should be 0 and 1, not 1 and 2 (to match chip and reduce confusion)
  • Fixed in revision 2: LEDs very dim, replace resistors with 150 ohm not 330.
  • Fixed in revision 2: RS232 header pins touch USB header on Arduino, shorting them to ground. Tape over the USB connector helps, but the pins can punch through the tape and cause hours of debugging frustration. Should consider moving the header to the opposite side of the board, even though it means basically rerouting the entire board.

Revision 2 PCB Issues

  • Fixed in revision 3: It doesn't make much sense to send +5v out on the header when the board is all +3.3v. Should change this.
  • Fixed in revision 3: No label on power header VIN pin
  • Fixed in revision 3: The board is 2.05" high, should make it 2" to increase flexibility in panelizing it.

Revision 3 PCB issues

  • Fixed in revision 4 (by removing IC all together): Pads on the I2C level shifter are too far apart, some batches of the IC don't reach the pads!
  • Fixed in revision 4: BOM doesn't match board for LED SMD sizes
  • Fixed in revision 4: BOM doesn't contain 1x3 header for INT jumper
  • Removed in revision 4: May be able to remove the I2C pullup resistors and use the built-in Arduino pullups
  • Changed text to stop mask layer in revision 4: Copper text on the bottom is completely unreadable w/ black solder mask

Overview

The Arduino MultiSerial shield will let the Arduino access two RS232 serial ports via I2C. It will have switchable 3.3v or RS232 levels. This project was initially created as a debugging tool for networks of BlinkieBits, as (at least during BlinkieNet design/prototyping) it would be extremely useful to be able to receive debugging information from a whole network of BlinkieBits, rather than just one at a time.

  • The NXP SC16IS752 chip is used, which gives 2 channels for ~$7. It talks 3.3v RS232, so level shifters are needed (but is 5V tolerant on the I2C side). The RS232 level shifter used is a second-sourced TI MAX3232 in a SOT16 package (~$1.70 qty. 1).
MultiSerial Shield Rev. 2 Schematic