<marcan>
whitequark: so drop resistors on the top layer aren't happening, but I have enough space for vias to drop Vio down between the regulators and the output capacitance/feedback network
<marcan>
so we could put protection/drop resistors on the rear
<whitequark>
marcan: sgtm
<marcan>
of course they could be 0Ω-ed if unnecessary
<marcan>
0603 I guess?
<whitequark>
yep
<marcan>
kk
<whitequark>
that's 122 mW at 350 mA
<marcan>
yeah, should be fine-ish
<marcan>
it's just about 1/8W
<marcan>
they make them in 1/4W at that size
<marcan>
ideally they will be lower anyway
<marcan>
I just don't know how low we can go yet
<_whitenotifier-9>
[GlasgowEmbedded/Glasgow] marcan pushed 1 commit to master [+0/-0/±3] https://git.io/fjsZl
<_whitenotifier-9>
[GlasgowEmbedded/Glasgow] marcan 255e8c5 - revC1: add series current limiting resistors to Vio
<marcan>
figured out what's wrong with the silkscreen. the footprint is definitely bugged.
<marcan>
looks like the converter I used generates a self-intersecting polygon, which is not allowed
<marcan>
the problem is that it just randomly picks bridge lines between inner holes and the outer perimeter to join them, but does not check whether those bridge lines intersect
<marcan>
it only checks if they intersect holes, not other such bridge lines within the solid area
* marcan
writes a horrible O(n²) abomination of a hack to try to fix it
<marcan>
yeah, that seems to have worked
<marcan>
definitely not upstreamable to svg2mod as-is, but good enough to unglitch Hana :p
<marcan>
I seriously need to rewrite that from scratch
<_whitenotifier-9>
[GlasgowEmbedded/Glasgow] marcan pushed 1 commit to master [+0/-0/±2] https://git.io/fjslL
<marcan>
the algorithm that svg2mod uses for hole inserting is just entirely the wrong way to go around doing it; with that hack it's good enough for this instance but it's not general (it's slow as fuck and not guaranteed to find a solution)
<marcan>
oh well, worry about that later :p
<_whitenotifier-9>
[GlasgowEmbedded/Glasgow] marcan pushed 1 commit to master [+0/-0/±2] https://git.io/fjsld
<_whitenotifier-9>
[GlasgowEmbedded/Glasgow] marcan c5ff15b - revC1: double up protection resistors on the rear side
<marcan>
force pushing to fix a snafu
<_whitenotifier-9>
[GlasgowEmbedded/Glasgow] marcan pushed 3 commits to master [+0/-0/±5] https://git.io/fjslx
<_whitenotifier-9>
[GlasgowEmbedded/Glasgow] marcan a820803 - revC1: more minor adjustments
jevinskie has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<cr1901_modern>
whitequark: I understand not wanting to desolder the microcontroller. I have one from a dead ST-412 (that I've long since gotten rid of) I could dump instead. Guessing the firmware between the two would be very similar.
<cr1901_modern>
It's not going to be doing too much anyway; the original ST-506 didn't have a microcontroller. It was a glorified faster, larger floppy drive.
<whitequark>
yep
<whitequark>
I do kinda wonder
<whitequark>
cr1901_modern: i mean
<whitequark>
right now i do have an IR preheater
<whitequark>
so i *could* probably desolder it super super carefully
<cr1901_modern>
you are certainly in a position to desolder and resolder it without damaging it :). IIRC, the micro on the ST-412 was socketed
<whitequark>
cr1901_modern: it's very tricky with the old PCBs
<whitequark>
it's really on the margins of my skill
<whitequark>
they delaminate way too easily
<whitequark>
i mean
<whitequark>
if it does delaminate i can still fix it, of course.
<whitequark>
but that doesn't count as "without damaging"
<cr1901_modern>
mmm, true. I had a Data I/O EPROM programmer from the 80's. Now, it wasn't working when I got it so I ultimated scrapped it for parts. But I found most of the parts very difficult to remove.
<cr1901_modern>
I only managed to really save the 41256 dram and the 8088 inside.
<whitequark>
i mean
<whitequark>
i'd say there's a 80% chance i could get it out intact
<whitequark>
maybe more if i try to be really patient
<whitequark>
but i have serious hatred for old PCB laminate.
<cr1901_modern>
What about one of those _powered_ vacuum pump desoldering tools?
<whitequark>
it's just so... shitty
<whitequark>
i don't have a decent one
<whitequark>
and tbh
<whitequark>
they tend to tear off pads
<whitequark>
not with mechanical force but with heat
<whitequark>
i'd say the chances with that one are more like 40% than with hot air and IR preheat
* cr1901_modern
wonders where he put his 6500/1
<whitequark>
then again once it works with glasgow i have no real use for it
<whitequark>
i dont actually need an st412 hard drive for any intents or purposes
<cr1901_modern>
haha right. If someone files a bug report, since st412 support will be a clearly in high demand applet, you can just buy a new one :P
<whitequark>
loooool
<whitequark>
i mean it's basically just shugart
<whitequark>
not sure if i should even make a separate applet for it
<whitequark>
or rename the existing one to memory-shugart
<cr1901_modern>
I _believe_ (don't quote me on this) that both the hard disk variant and floppy variant of the protocol came from Shugart- it just so happens that the hard drive version was first used on Seagate's first product
<cr1901_modern>
(Oh and Al Shugart also founded Seagate).
<whitequark>
mmm
<cr1901_modern>
Probably okay to keep them under one applet and if you support various hard disk track formats, perhaps make each one a submodule you can "plug in" to the main applet