libera/#neo900/ Thursday, 2019-07-25

dos1houkime: "cloning from returns a pack index error" there's a mirror at
houkimedos1, thanks15:55
dos1norly: ^^^15:55
dos1I'll try to take a look at what's wrong with the repo, but not sure yet when15:57
houkimea bit sad that my branch is actually the most recent.15:59
houkimenobody worked on neo900 after me it seems.15:59
houkime(in ee-full i mean)16:00
norlythanks guys16:01
houkimetrying to figure out where to best place a common mode choke on a differential pair which then goes through a connector and through a cable18:33
houkimeand fo 500 MHz (Neo900/N900 CSI) I can't really think of a way the exact placement could make a significant difference18:34
houkimeJoerg-Neo900, do you know of any particular design considerations for that? I searched for several hours today and learned a lot of new things about differential pairs but couldn't really find an answer for placement of common node chokes.18:40
houkime*mode chokes.18:42
houkimeSTI filters application notes for example just say to place them on the both sides of the cable somewhere between a connector and an IC but that's it.18:44
houkimephysics-wise, 500 Mhz has a wavelength of 60 cm and even if we consider an edge to be 10-20 cm it is very huge.18:46
houkimethat said, noise can have a higher frequency than 500 MHz18:48
houkimeDocScrutinizer05, ^^^^19:47
DocScrutinizer05depends on particular design19:50
DocScrutinizer05There must be the repos which our last layouter created19:51
houkimeDocScrutinizer05,  I am speaking about a connection between neo900 front camera (via display connector) and bb-xm19:51
houkimeit wasn't layed out unfortunately neither by me or by metacollin19:52
DocScrutinizer05get a plain ferrite block / bead and place it on top of the diff pair, find the place that works best19:52
DocScrutinizer05for better effect, shield the diff pair with a plane above and below19:54
houkimeDocScrutinizer05, the schematics specifies the exact part number for a choke. ACM2012H. It was actually the placement i was not sure about.19:55
houkimeAs far as physics and documents i found go, the difference between places should be quite negligible.19:56
houkimeat least on +- a few cm scale.19:57
Joerg-Neo900check N90019:58
Joerg-Neo900I guess it's legacy19:58
houkimeJoerg-Neo900, thanks. Haven't though about just checking original pcb.19:59
Joerg-Neo900the purpose is to block common mode signals from reaching chip inputs. Noixe gets picked up on long traces19:59
houkimeJoerg-Neo900, from this logic the ideal placement is near client ICs on both sides. Thoughin this particular case both clients are not on neo900 (one on bb-xm and one on front panel near display)20:02
houkimeand in schematics there is only one CMF per differential lane and not 2.20:03
houkimethe question is then - does N900 frontal camera has a CMF of its own?20:04
houkimeif it does and if there is also a CMF on bb-xm then CMF on neo900 itself shouldn't be needed.20:04
houkimehowever, i doubt that any of the ends actually has CMFs20:05
houkimepins on bb-xm connector which are carrying differential signals for CSI were chosen on an unclear principle.20:06
DocScrutinizer05only one needed when signal direction unidir20:12
DocScrutinizer05the pins prolly where chosen for their function20:13
houkimewhich is... based on bb-xm docs, DVI? And also one stray pin which is on another connector entirely.20:14
houkimethough there might be some multiplexor magic going on...20:16
houkimehowever mux magic is unlikely to create and destroy CMFs on bb-xm.20:17
houkimeSo, if unidir, then logically CMF should be near a receiver.20:19
houkimeAnd there are 2 data lanes in CSI.20:19
houkimeI don't however know how exactly they are directed.20:20
houkimethey might be both from camera to bb-xm, or one to camera and one to bb=xm20:20
houkimeclock-wise, one can suppose that bb-xm is master.20:21
houkimeso for clock lane camera is a receiver.20:22
houkimehowever, in the first scenario where camera has 2 lines to bb-xm it becomes uncontrollable and can operate only in some default mode.20:27
houkimehmm... or not if not only CSI is present. need to check.20:29
houkimeit seems like there is only CSI, one extra external clock signal (for some reason), 2 powerlines and a shutdown signal.20:34
houkimefound the front camera model ST Microelectronics VS6555.20:38
houkime datasheet for it however returns an error 40420:41
houkimebut this works20:43
houkimeaaand it has nothing to do with csi whatsoever. It uses some cryptic CCP 2.- for data which is one clock diff. lane and one data lane.20:56
houkimeCCI which is used for commands is 1 data pin and one clock pin. And that's it.21:08
houkimePerhaps ccp was wrongly identified as CSI on the display connector, but then there is a question: for whom is the second dataline and if it even exists.21:10
houkimeJoerg-Neo900, i couldn't find based in what data the pinout of the neo900 display connector was established.21:11
houkime*based on what data21:11
houkimei found in neo900 feasibility study that "cdp" interface is used by the panel also. (2 lanes + SPI)21:15
DocScrutinizer05I think this is covered in out proto-v2 whitepaper. Or ask wpwrak21:15
houkimeDocScrutinizer05, v2 says nothing about front camera, i checked21:16
DocScrutinizer05I'm really sorry I tried to keeep ALL this in my mind for over 5 yeras. At one point in time some headcrash and trash collection happened21:16
DocScrutinizer05Neo900 mimics a N900, so the pins on BB-xM are same as on N90021:17
DocScrutinizer05and yes, I think this is covered in v2 whitepaper21:18
DocScrutinizer05more generically as I just paraphrased21:19
houkimeI have checked original N900 board. All 3 CMFs are grouped right near display connector. However it doesn't seem to have any ESD protection which neo900 schematic calls for.21:48

Generated by 2.17.0 by Marius Gedminas - find it at!