[Milkymist-devel] M1r4 KiCad schematics, 1:1 review

[Werner Almesberger]

I compared the AD schematics with the KiCad schematics and generally
looked for any remaining issues. I only found relatively minor things.
Adam, congratulations are in order again :-)

Here is what I found, bugs - suggestions for improvements, and
questions:

Various:

- POWER (and few other places): resistor tolerances are not always
  indicated. Several resistor values look like 1%. It would be good
  to explicitly state this. (Unless we make 1% our default
  tolerance.)

- POWER (R62): I think 0.1W will be sufficient :-)

- FPGA_BANK0 (near U22A): VGA_SCL is labeled as input to the FPGA.

  In I2C, both SDA and SCL are bidirectional:
  http://en.wikipedia.org/wiki/I²C#Clock_stretching_using_SCL

  boards/milkymist-one/rtl/system.v calls it vga_sdc and declares it
  as an output.

  So in the schematics it should be either an output or, better,
  bidirectional.

  If you make it bidirectional, also the label on DVI-ISingle
  should be updated. Interesting that ERC doesn't complain about
  both uses of the label being inputs.

- NOR_FLASH (U9, pin 43): according to the data sheet, it should
  be labeled VCCQ, not VCC. The AD schematics are correct.

- AUDIO (MK1): do we want to put a part number (MCE-100, it seems)
  in the schematics ?

- AUDIO (near AC97_SYNC): seems that TP6, TP7, and TP8 were lost
  in translation.

- VideoIn (C211): 10nF should be just 10n

- USBExtPorts1 (U16, U17, etc.): you changed pin name VIF to
  VIN. Was this intentional ?

Text/comments:

- DRAM (C157... and C162...): since you mention for C240...  and
  C248... what they are used for, maybe you want to do the same
  for the other two groups of caps as well ?

- MiscControl (J7, J8): in the AD schematics, you mention which
  DMX port is RX and which is TX. I think it would be nice to
  have this here, too.

- MiscControl (J9, J10): here, it would be useful if the numbers
  of the pins on the connector would be indicated, like you did
  on the DMX connectors, since the connector pin numbering
  doesn't follow the symbol/footprint numbering.

  Alternatively, maybe just change the order of the pins in the
  symbol, e.g., 4-5-3-2-1 instead of 3-5-2-4-1, which seems a
  little weird anyway.

- MISC (D3): typo in the LED name. Should be D23.

Style:

- POWER (TP34): it's very well hidde. How about separating it
  from J11, similar to the POWERED-GND subcircuit in the lower
  left corner ?

- Header4: AUDIO (J23, J25, ...), MISC (J5): it's a bit ugly to
  have the pin numbers twice, so I would set number or name to
  invisible.

- VideoIn (near V5...V7, R151...153): there we have cross-style
  junctions. Maybe move GND under V5 and R151, similar to the
  way you've done it with C208...C211, etc.

- VideoIn (near U21 pin 46): there seems to be enough space that
  you could avoid the "hook".

- VideoIn (near U21 pin 32): this "hook" looks particularly ugly.
  Maybe make the ground point straight down and left-justify
  the VIDEOIN_AGND text ?

- MiscControl (U6): the squiggle on the LED looks weird. Real
  arrows or half-arrows (like on the LEDs) would be better.

- USBExtPorts1 (C257...C260): did you notice that, if you swapped
  the two capacitor groups, you could avoid crossing the wires
  twice between U26 and J16.

  Similar with USBExtPorts1 and the AD schematics.

Spacing:

- FPGA_BANK2 (R161): the text (component reference and value) is
  further away from the symbol than usual

- AUDIO (near V24): the text is very close to AUDIO_AGND and
  also the two text lines touch each other.

- VideoIn (text under C208, C215): text touches the junctions

- USBExtPorts1 (text under V3 and U26), USBExtPorts2 (V10, U29),
  USBIntPorts (V22): the text lines touch each other

- MISC (under the LED matrix): text lines touch each other

Backport:

- U22 is called XC6SLX45-2FGG484C in the KiCad schematics but
  XC6SLX45-FGG484C in the AD schematics. It seems that the KiCad
  schematics are more accurate in this case. Shouldn't the AD
  schematics be updated, too ?

Curiosity:

- DRAM (C245): the C240...C245 group has 5 caps while C248...C252
  has only 4. Why that difference ?

We also have a few cases of wires carrying negated signals still
being called something_N in labels, e.g., ETH_INT_N. Not sure if
you want to unify them now or if you prefer staying with the old
names.

- Werner