<Mangy_Dog>
hi guys going back on an old convo when iw as dealing with hdmi issues..... The pi4 HD or WIfi thing has come up recently... And it got me wondering.... So HDMI might be interfearing with the wifi signal... in what way I dont know as ive not got one or tested it... But at 1440p and 4k what is the bandwidth speeds? around 2.4ghz? within the known range of wifi to cause jamming?
<Mangy_Dog>
the thing is id have imagened the pi4 board is 4 layers and there would have been adiquate shielding with ground planes over the signal lines.
<KotCzarny>
its not about bandwidth, but clocks used, most likely
<Mangy_Dog>
Not to mention maybe put the wifi chip and its antenna on the other side of the board
<Mangy_Dog>
yeah sorry i meant clock
<KotCzarny>
and it's electronics 101
<Mangy_Dog>
or do you think the fualt might be within the broadcom chip itself?
<KotCzarny>
if you transmit/generate signal at particular frequency, it's getting propagated into space
<KotCzarny>
it's not the fault
<KotCzarny>
it's physics
<KotCzarny>
in the old times we had spread spectrum option in bios for pci ;)
<Mangy_Dog>
but what are the clock rates for 1440p and 4k?
<KotCzarny>
because pci was interfering with memory buses
<Mangy_Dog>
heh
<mru>
pcie also has spread spectrum
<Mangy_Dog>
i think my the old days i kinda remember seeing the options in bios, but having NO idea what it did
<KotCzarny>
mru, but nowadays there is no option to toggle it anymore :/
<mru>
sometimes there is
<Mangy_Dog>
well if it just works :p
<Mangy_Dog>
no need to toggle
<Mangy_Dog>
e
<KotCzarny>
it has it's quirks
<Mangy_Dog>
oh
<KotCzarny>
but as long it's spreading inside tolerated margins, sure.
<KotCzarny>
it works.
<KotCzarny>
;)
<Mangy_Dog>
one reason why im asking is, though tbh i wont be runinng at HD or 4k..... there will be a pi4 version of myhandheld when they do the CM modules.... and i need to take this into considdereation when i do my board design
<Mangy_Dog>
esp if they offload wifi off teh cm
<KotCzarny>
that's also the reason chips and parts of the boards are shielded with metal
matthias_bgg has quit [Read error: Connection reset by peer]
<KotCzarny>
but not in the cheapoland
<KotCzarny>
;)
<Mangy_Dog>
thought those were heat spreaders :p
<KotCzarny>
nope.
<Mangy_Dog>
ill likely still do 2 layer boards
matthias_bgg has joined #linux-sunxi
<Mangy_Dog>
but if i need to... I might do some creative adhoc shilding
<mru>
why 2 layers? 4 is much easier
<Mangy_Dog>
put some ground pads and then stick some copper tape over
<KotCzarny>
main reason was to contain/protect from emi
<Mangy_Dog>
cost
<KotCzarny>
you should really dig into 'shielding' topic
<KotCzarny>
emi shielding, that is
<Mangy_Dog>
nods
<Mangy_Dog>
i got away with it with my HDMI
<mru>
a 4 layer board will help
<Mangy_Dog>
but i think its a just thing
<Mangy_Dog>
"just"
<Mangy_Dog>
i mean matching the pairs helped tons :p
<mru>
you want to minimise the loop area of any high-frequency signal path
<Mangy_Dog>
but yeah would be useful to have shielding
<Mangy_Dog>
what would you considder high frequincy?
<mru>
and a 4-layer board lets you put a ground plane at a small depth below the signal traces
<Mangy_Dog>
jlc slap on a 32$ engineering fee for 4 layers
<KotCzarny>
welcome to cheapo-land
<Mangy_Dog>
i have to stay cheap
<Mangy_Dog>
im poor
<KotCzarny>
would you like it fast, good or cheap. (choose 2)
<Mangy_Dog>
oh good
<KotCzarny>
it's like using old plastics and other trash in home furnace
<Mangy_Dog>
loads for me and i get tons of network issues :p
<Mangy_Dog>
so yeah googling around it does indeed appear qualcom will only sell chips direct to large manufactureres and not sell them on the open market for us plebs to build hardware wtih
<Mangy_Dog>
So really my only option, if i was to build a sbc from scratch.... rockchip?
<Mangy_Dog>
as long i could learn enought o build a custom linux build for it :p
<fALSO>
;-)
<fALSO>
its not that hard
<Mangy_Dog>
tbh the hardware would be the easy bit
<Mangy_Dog>
its the software thats harder for me :p
<fALSO>
well at least the bootloader and kernel is easy
<fALSO>
;-)
<megi>
finished compiling mesa, panforst works better than before on H6
<fALSO>
=)
<megi>
it can run some browser based webglk demos for a while
<megi>
but slowly :)
<Mangy_Dog>
its why im looking at pi4 for the new board tbh.... the vast majorityof the hardworks been done
<megi>
I still managed to crash the board, but that may be because of non-mainline GPU OPP table I got from somewhere previously without any testing/verification
<KotCzarny>
blow some cold air and repeat?
<KotCzarny>
;)
<megi>
I blow air all the time
<KotCzarny>
i mean put a fan into the work
<megi>
:)
nexgen has joined #linux-sunxi
dgp has joined #linux-sunxi
AneoX has quit [Remote host closed the connection]
AneoX has joined #linux-sunxi
kilobyte_ch has quit [Quit: WeeChat 2.3]
jernej has quit [Remote host closed the connection]
rra has left #linux-sunxi [#linux-sunxi]
jernej has joined #linux-sunxi
airstorm has quit [Quit: airstorm]
rra has joined #linux-sunxi
AneoX_ has joined #linux-sunxi
AneoX has quit [Ping timeout: 268 seconds]
rra has left #linux-sunxi [#linux-sunxi]
Gerwin_J has joined #linux-sunxi
gsz has joined #linux-sunxi
<wens>
megi: no mention of H6 in that report?
JohnDoe1 has joined #linux-sunxi
<megi>
wens: no mention, but in #panfrost irc channel there are mentions they're testing on PineH64, which is H6
<wens>
regarding rpi4 wifi + 1440p HDMI issue, IIRC the pixel clock for 1440p60 is 241.5 MHz, and the TMDS clock is 2.415 GHz, which is between WiFi channels 1 and 2... so ...
<wens>
Mangy_Dog: ^
<wens>
montjoie: sh*t how? :p
OnkelUlla has quit [Ping timeout: 246 seconds]
OnkelUlla has joined #linux-sunxi
<diego71>
10th armonic is over the wifi ch1 :P
sunilmohan has quit [Ping timeout: 252 seconds]
<wens>
don't need the 10th harmonic. the TMDS clock is enough :p
<wens>
plus the pixel clock never leaves the chip; the TMDS clock does
selfbg has quit [Remote host closed the connection]
sunilmohan has joined #linux-sunxi
sunilmohan has quit [Changing host]
sunilmohan has joined #linux-sunxi
<mru>
the tmds clock is the pixel clock
<mru>
for bit rates up to 3.4 Gbps
<megi>
wens: thanks for asking for HSYNC/VSYNC clarification on the mailing list
<wens>
mru: thought there was some 8/10 encoding stuff going on?
<mru>
yes, for cycle of the tmds clock, each data channel transmits a 10-bit word
<mru>
which encodes an 8-bit value
tllim has joined #linux-sunxi
reinforce has quit [Quit: Leaving.]
<wens>
oh, I thought the clock line rate was the pixel rate * 10
<mru>
the bit rate is, yes
<mru>
the receiver is expected to be able to count to 10
<mru>
for hdmi 2.0 links with >3.4 Gbps there are 40 bits per tmds clock
RichardG867_ is now known as RichardG867
<montjoie>
wens: I need to add rpi4 on CI, no uboot USB, no uboot net, my only way is to code some kexec fun
<montjoie>
and when you saw the official advice "put your board vertically for fixing heat problem"...
<mru>
which rpi was it that crashed if you flashed a bright light at it?
tl_lim has quit [Remote host closed the connection]
tl_lim has joined #linux-sunxi
<willmore>
The answer is right there in the name. :) TMDS. Transition (bit changes from low/high or high/low) Minimized Differential Signaling. Part of the encoding is to reduce the number of transitions in the signal. This has the intention of reducing emissions from the encoded signal.
<willmore>
Using a full speed clock has no benefits, really, since you have to do phase differential (clock skew) compensation and equalization anyway, you can add in the PLL multiplication effectively for free. You basically put in an early/late gate error signal back to the PLL which has the effect of aligning the recovered clock with the data.
<willmore>
Or at least that's what we learned in the signal transmission class I took back in the late 90's taught me. Who knows what they do 20 years later.
<KotCzarny>
30 years.
<willmore>
Just over 20 years.
<KotCzarny>
:)
<willmore>
2019-1998 = 21
<KotCzarny>
almost 2020
<willmore>
I'm old enough already, KotCzarny, no reason to make me older.
<willmore>
Well other than the not dying method of getting older.
<KotCzarny>
:)
<KotCzarny>
there are kids that dont know what a floppy disc is
<anarsoul>
KotCzarny: have you used a floppy disc in last 10-12 years?
<KotCzarny>
i think yes
<anarsoul>
my laptop from 2007 didn't have a floppy drive, thus no floppy discs. My PC that I used before that had one, but I don't think that I used it much either
<KotCzarny>
i think i used it for some firmware updates
<KotCzarny>
(before i've learnt about virtual floppy in grub)
* willmore
was just sorting some stuff in the basement yesterday and came across a box of 5.25" and 3.5" discs.
<willmore>
Plus some QIC tapes.
<anarsoul>
don't throw them away :)
<willmore>
I won't. I have working drives for them as well. I can read anything up to 525MB in QIC. And DLT up to 20GiB. Got an 8mm drive as well, think it was the 2.5GB version. Also a few DAT drives. Oh, and a pair of 8" floppy drives!
<willmore>
I need to get my old computer lab up and running again.
<willmore>
Found the newer model C64 and 2600 in a corner as well.
<martinayotte>
I've used cassette tapes on Apple ][ in 1980, but quickly purchase my first floppy
* KotCzarny
has never seen 8" floppy
<willmore>
KotCzarny, just like a 5.25, but larger.
<martinayotte>
110KB per floppy side ... :-)
<willmore>
martinayotte, I didn't use them on the ][, but I did use cassettes extensively on the C64.
<martinayotte>
:-) ... I've used round punch, it was working too
<willmore>
martinayotte, no, I meant apple 2 DOS. Quoting wikipedia "Normal storage capacity per disk side was 113.75 KB with Apple DOS 3.2.1 and earlier (256 bytes per sector, 13 sectors per track, 35 tracks per side), or 140 KB with DOS 3.3 and ProDOS (256 bytes per sector, 16 sectors per track, 35 tracks per side)."
<martinayotte>
Later, first PC, a 10MB harddrive formatted in RLL gaves 15MB
<willmore>
MSDOS did a similar 8 to 9 sector transition as well at one point.
<martinayotte>
Ah ! Maybe my memory is failing ...
<willmore>
You lost a few sectors.:)
<martinayotte>
:-P
<KotCzarny>
blow the dust off
<willmore>
Have you tried using a cleaner in there?
<KotCzarny>
it might be the cause
<willmore>
Be careful blowing the dust off, it may just go elsewhere.
<willmore>
better to suck the dust out.
<willmore>
Optical drive in particular should never be cleaned with compressed air.
<KotCzarny>
flexible "memory disks" offered a storage capacity of 80 kilobytes (KB), approximately 3,000 punched cards
<KotCzarny>
imagine punching floppies
<mru>
there was a time when punching a hole in the right place would double the capacity
<martinayotte>
the notch was simple the write-protect
nashpa has quit [Quit: Going away]
<KotCzarny>
DD/HD
<mru>
on the 3.5" floppies, one of the corner holes indicated "high density"
<mru>
I don't recall if the 5 1/4" ones had the same feature
<martinayotte>
no
nashpa has joined #linux-sunxi
<mru>
I never used those much
<mru>
though I did encounter an 8086 based PC with such drives in production use as recently as 2001
<mru>
they were going to replace it with something more modern "when there was time"
<martinayotte>
... after 20 years of duty ... :-P
<mru>
it was running a simple automation script for QA testing
rzerres has joined #linux-sunxi
<willmore>
Oh, I have punch cards, too. No equipment to read/write them, though.
<mru>
a pointy stick?
<willmore>
It's supposed to be a rectangular hole, so it'd have to be a special pointy stick.
<willmore>
And, yes, there was a hole in the 3.5" disc that indicated capacity. There were pattenns for SD,HD, and ED. (720K/800K, 1.44M, and 2.88M respectively)
<martinayotte>
At school, there was a teletype with paper punch tape, 4-20mA for serial connect at 110baud
<willmore>
martinayotte, probably baudot as well?
<martinayotte>
Yes ! In fact there was one Baudot (5bits), and another newer one in ASCII(7bits)
<martinayotte>
If I recall the newer one wasn't 4-20mA, but real RS232
<willmore>
That sounds correct with my memory as well.
* willmore
used to own a huge teleprinter (with optional keyboard) that used baudot over a 4-20mA serial loop. It was part of a satallite delivered news service at a radio station for decades.
<willmore>
I had to give it up on a move. Hurt so bad to recycle it.
<willmore>
But it was huge and heavy.
<martinayotte>
Movers didn't want to touch it ? :-P
<willmore>
Naw, it was when I graduated college and it had been in my parents garage. They were moving to a smaller house since their kids were all gone and I had no way to take it to my little apartment. ;(
<martinayotte>
You have to get rid of old stuff sometime, otherwise moving becomes a pain. I got rid of an old Sony 12 inch video disk player few years ago, it weight was about 25kg
<willmore>
Damn. My neighbor has a collection of laserdiscs. I should ask her if she wants them transcribed some time.
<willmore>
I wonder if she has the star wars one where han shoots first....
<willmore>
IIRC, that's the only media that version was released on outside of theatres.
<martinayotte>
In my case, I had only 2 discs, I've owned that stuff only because I had to develop automation control of those disc players for museums.
gaston1980 has quit [Quit: Konversation terminated!]
<willmore>
That's a cool thing, though. Did you ever play DragonQuest?
<martinayotte>
Nope ! I'm not really a gamer, although I've worked at Ubisoft during 2 years ...
tl_lim has quit [Remote host closed the connection]
<willmore>
It was a laserdisc based game. Think of a game where almost all of the game is cut scenes. :) Every once and a while you'd get to a decision point where user input determines where the story goes from there. Lots of gruesome death scenes. :)
NeuroScr has joined #linux-sunxi
jstefanop has quit [Remote host closed the connection]
jstefanop has joined #linux-sunxi
jstefanop has quit [Ping timeout: 250 seconds]
florian has quit [Ping timeout: 250 seconds]
suprothunderbolt has joined #linux-sunxi
warpme_ has quit [Quit: Connection closed for inactivity]