<atk>
when random capacitors and squiggly traces make the difference between broken and working, you know you're dealing with high speed electronics or RF stuff
<Joerg-Neo900>
hehe, yes
<atk>
Especially interesting are those high end oscilloscopes which look like someone went into paint and drew random shapes and then made a PCB out of it.
<Joerg-Neo900>
this capacitor though wasn't exactly random. I even calculated the capacitance of the chewing gum paper somebody stuck on top of the uSD card which evidently fixed the issue ;-P
<atk>
I wonder how someone came up with the idea of putting chewing gum wrapper on the SD card
<Joerg-Neo900>
we also tested the effect of the increased capacitive burden on clock slew rate
<Joerg-Neo900>
(idea) I don't know, our users were *very* innovative
<wpwrak>
paper = "shielding" :)
ahycka has quit [Ping timeout: 240 seconds]
<Joerg-Neo900>
was a very nasty heissenbug too since it only hit those who actually used their uSD and same time GPS
<Joerg-Neo900>
we tested GPS as nauseum and didn't find a problem since none of the engineers had a uSD in their test devices
<wpwrak>
wasn't it that the GPS antenna was too close to uSD ?
<Joerg-Neo900>
wpwrak: I don't know if the mere distance ever been considered
<Joerg-Neo900>
can't be physical distance to the patch antenna since that one was at very top end of device
<Joerg-Neo900>
fsckd up EMC
<Joerg-Neo900>
the clock simply was allowed to radiate I guess, nobody checked what *exactly* was the failure operation mode
<Joerg-Neo900>
maybe a trace was too close to the clock trace, maybe (my fav) the card holder simply acted as antenna
<Joerg-Neo900>
but yeah, you can see the GPS antenna plug next to the uSD holder
<wpwrak>
yup. rolling out the red carpet for trouble ;)
<Joerg-Neo900>
who knows which sort of aerial the gndplane forms with the sleeve shielding of the antenna cable, creating waves propagating along the cable surface to the whole patch antenne and there jumping directly to the antenna patch and into the amp#
<Joerg-Neo900>
never forget: *ALL* antennas are dipoles
<Joerg-Neo900>
so wiggling with the GNd potential of the patch LNA antenna is exactly same as sening a signal to the antenna patch on top of it
<Joerg-Neo900>
sending
<Joerg-Neo900>
and a wire shielding gaze running in parallel to the gnd plane it's connected to forms a nice loop antenna by itself
<Joerg-Neo900>
s/gaze/braid/
<Joerg-Neo900>
reducing the clock driver strength and damping the slew rate further by a 10pF load effectively cured the issue though
<Joerg-Neo900>
and simply doing "the right thing" and switch off clock as long as uSD had no operations to perform was the final killer of the issue
<Joerg-Neo900>
at least as long as you were not constantly loading map tiles from uSD while using GPS
jonsger has joined #neo900
pagurus` has joined #neo900
ahycka has joined #neo900
<Joerg-Neo900>
wpwrak: D1402 IR-LED VBAT n uF bypass capacitor to GND missing
<Joerg-Neo900>
also we *might* consider a series R in/from gate to R1410, to reduce slew rate of high current
<Joerg-Neo900>
>>In other cases, a maximum slew rate is specified[2] in order to limit the high frequency content present in the signal, thereby preventing such undesirable effects as ringing or radiated EMI<< https://en.wikipedia.org/wiki/Slew_rate
<Joerg-Neo900>
we don't need frequency components >500kHz there
<Joerg-Neo900>
a very rough guestimate by parasitic capacity of gate and 500kHz is absolutely sufficient for size of series R
<Joerg-Neo900>
we have am order of magintude headroom for error to the low side
<Joerg-Neo900>
and no hard limits at all to the higher frequency side
Kabouik_ has joined #neo900
Kabouik_ has quit [Remote host closed the connection]
Kabouik has quit [Read error: Connection reset by peer]
Kabouik has joined #neo900
<ahycka>
Joerg-Neo900, could you make me a branch to push my schematics review? (SPI resistors)
<ahycka>
i can push/pull in master but i cant checkout a new branch
<ahycka>
push the new branch, I mean
chomwitt has quit [Ping timeout: 250 seconds]
<wpwrak>
ahycka: that would be for hellekin (joerg doesn't know much about such details of the git setup). i've notified him.
<ahycka>
i can ask him
<ahycka>
how900?
<how900>
yo ahycka
<ahycka>
hi!
<ahycka>
could you make me a branch to push my schematics review?
<ahycka>
i can push/pull in master but i can push a new branch
<how900>
you have a dedicated branch, let me see... If you ssh git@neo900.org you should have it listed...
<how900>
ahycka: it's proto_v2
<wpwrak>
how900: ah, you're already here, good :)
<ahycka>
PTY allocation request failed on channel 0
<ahycka>
hello ahycka, this is git@newbie running gitolite3 v3.6.6-2-g8620d5f on git 2.10.0
<ahycka>
Connection to neo900.org closed.
<ahycka>
R Wtesting
<ahycka>
R Wee
<ahycka>
i can't see proto v2 xD
pagurus` has quit [Ping timeout: 252 seconds]
<how900>
OK, but if you run: git push -u origin proto_v2 it should work ahycka
<how900>
try this :)
<wpwrak>
how900: is the idea that we'd all work on the proto_v2 branch then ? or would that be ahycka's branch, and we then merge into master ?
<Joerg-Neo900>
the latter
<ahycka>
wpwrak, by the moment i only add 8 resistor in 2 SPI bus
<ahycka>
the best part is that kicad is plain text! O_O
<ahycka>
altium is binary
<ahycka>
so, it's impossible compare 2 versions
<wpwrak>
ahycka: kicad has its share of quirks despite the text, too :)
<Joerg-Neo900>
with kicad and Neo900 you get eeschow :-)
<wpwrak>
ahycka: e.g., if you run ERC, invisible information on all the power symbols changes
<Joerg-Neo900>
not terribly useful for git, but for "compare" it's without any competition
<wpwrak>
ahycka: also, sometimes you redraw a few things, commit, and when you open the sheet again, without making changes, eeschema decides to optimize some wires. so you suddently get a bunch of hard to interpret changes out of the blue. of course, there aren't "real", which eeshow can tell (i.e., no difference)
<wpwrak>
Joerg-Neo900: comparing is especially useful with git ;-)
<Joerg-Neo900>
with git, but not FOR git
<wpwrak>
ah, your visual merge. yeah, that would be quite a project :)
<Joerg-Neo900>
git diff is (c-)sourcecode centric by design
<Joerg-Neo900>
the more so is git merge
<wpwrak>
with kicad, git diff tells you when there's a lot more going on than you expected :)
<wpwrak>
and yes, avoid git merge of kicad files
<ahycka>
wpwrak rocks!! \o/
<Joerg-Neo900>
for very simple changes git merge might work for kicad, but as soon as it gets a tiny bit more involved or kicad messes up, you're lost
<ahycka>
xD
<wpwrak>
:)
Kabouik has quit [Remote host closed the connection]