0191-2024-01-31.md

31 Jan 2024

Previous journal:	Next journal:
0190-2024-01-17.md	0192-2024-02-02.md

It's been another couple of weeks since my last udpate, but I am still busy with other happenings and a number of hardware projects.

So far, these are just rough notes of things I need to document (or work out) of late.

Soldering QFNs and GFMPW0 bring-up

TBC:

• QFN breakout boards
• RP2040 QFN56 0.4mm
• RP2040 test boards assembled by JLCPCB
• Soldering steps for GFMPW0 QFN64 0.5mm
• Tigard
• GFMPW1 acceptance
• GFMPW0 chips from Tholin
• Guides/tips for GFMPW0 bring-up
• TT3p5 testing guide and upython code

GFMPW0

Work out the correct pinout

References:

• Caravel M.2 QFN card -- GFMPW0 has same pinout (?) as sky130 chips, except all positive voltage pins are +5V instead of varied power domains.
• Caravel development board version 6B for GF180 5V -- used for GFMPW0 and probably also GFMPW1?
• With the two above, we can map QFN pins to actual termination in the dev board, because I think the same M.2 QFN card is used for mounting both sky130 and gf180 chips.
• Could this be a Caravel QFN64 footprint? It says 'Test socket'
• Not so important, but this confirms pin arrangement on the package top and bottom views.
• WLCSP pin descriptions for sky130
• Caravel QFN64 bond diagram showing top-view package pin descriptions
• The SVG file under the 'Die arrangement and pads' heading is not correctly zoomed, but if you right-click and save that image, then open in (say) Adobe Illustrator, you can see a lot of extra info about the pins, including the secondary functions of mprj_io[37:32]
• Tim Edwards explains flash2

TODO: Check with others who have GFMPW0 chips, whether they have used them on their own PCB designs.

Here's my attempt to create a nice diagram of the Caravel QFN-64 pinouts:

TT03p5 demo

• Include lots of photos in guide to show what's expected of oscilloscope, etc.

• Schematic:

  • Clock/reset sources
  • ui_in buttons and pulldowns
  • uo_out VGA signals, inc. RGB111 and double-up for RGB222
  • Speaker (voice coil) vs. piezo (speaker not buzzer) vs. line-out levels

  Here's a draft schematic:

  

• Audio out:

  • Assume max 4mA for GPIO source/sink, tho TT04 board assumes 20mA. RISK: With 4mA plus existing hardwired LED 20mA, we're over the limit.
  • For 8Ω speaker, min resistance should be 820Ω so 1kΩ is probably safest.
  • For piezo speaker, ~560Ω would probably be safe. Easier just to go with 1kΩ anyway.
  • For line level (i.e. line out, or 'aux'):
    • Target is ~1.4V, and input impedance is perhaps 10kΩ...
    • ...so we could go with 10kΩ in series to get ~1.65V or two of them (20kΩ) to get ~1.1V.
    • OR better to use a low-impedance voltage divider? Say, 2x1k high side, 1k low side, or 1k:470R (~1.055V).
  • General designs:
    • Good: 10kΩ
    • Better: 1kΩ:470Ω divider or even just a pot (no less than 1k!) => ~1.055V
    • Best: R in series with pot and filter caps (both high- and low-pass)

• Should be possible to hear LZC outputs?

• Test unregistered outputs