[GreenKeys] John Nagle's USB <-> 60 mA loop board

[John Nagle]

> On Wed, Feb 21, 2018 at 3:24 AM, Bruce Phillips 
>>
>> I was planning to build a few in the very near future.......I'm in
>> Australia, but postage should not be too much for a small item like that if
>> there is any interest. I was very conscious not to tread on anyone's toes.
>> I planned to change the layout of the pcb to as much smd as possible, but
>> that was just my preference for my own assembly. In any case I will have
>> probably 20 pcb's made, so don't mind sharing them around.
>>
>> Regards,
>> Bruce Phillips
>> Australia

    Feel free to use the design. That's why I put all the files on Github.

    Blank boards are the easy part, and cheap. I have mine made by
Seeed, in Shenzhen.  It's the assembly that's hard.

    Notes:

    - An all-SMD design would be nice, but there are some problems.
      The CP2102 USB to serial converter is packaged in a QFN28 package.
      Those mount via pads under the part, 0.4mm apart.
      While it's possible to hand solder such parts, the technique
      is complicated. See it done:

      https://www.youtube.com/watch?v=lhOKSqipwqg
      https://www.youtube.com/watch?v=ljyGMOi5JPo

      Data sheet:
https://www.silabs.com/documents/public/data-sheets/cp2102n-datasheet.pdf

      That's why the daughter board for the USB to serial converter.
      I'd like to get rid of that, and if you're going for assembly
      using a solder paste stencil and pick and place machine,
      go for it.

    - C1 and C2, the 1uf ceramic capacitors, are through-hole
      because the new multilayer surface mount large-value
      ceramic capacitors (MLCC) on't work. Their capacitance
      decreases with voltage, down to about 20% of normal
      value. They're OK for filters but not for energy storage,
      which is how they're used here. Watch this tutorial on
      MLCC capacitors.

 
https://www.digikey.com/en/ptm/m/murata-electronics-north-america/high-capacitance-hi-cap-mlcc/tutorial 
(see slide 7).

   - The layout around the switch controller U1 is critical. Trace
     lengths must be very short in that area, or it won't work at all.
     See the data sheet.

http://cds.linear.com/docs/en/datasheet/3750fa.pdf

     This is a switching power supply operating around 300KHz with
     2ns rise times. Keeping those spikes out of the computer that's
     powering the thing, or generating RF noise, requires
     tight layout and various bypass caps and ferrite beads on the
     board. Steve Garrison says the thing is RF-quiet; he's not
     hearing it on his receivers or seeing noise on a spectrum
     analyzer.

   - The board could be made shorter. There's a price break if you
     get all dimensions below 100mm.  I just used that size because
     there's a nice case available for 125mm x 75mm boards.

It took seven revisions of the board to get this working. It's
non-trivial.  If you redesign the board, be prepared to debug it.

				John Nagle