GTA04b7/Neo900

Issue 518: PCB affecting issues collection

Reported by Nikolaus Schaller, Nov 23, 2013

upper board:

* we don't have thought about the (different!) C-springs and 
contacts
* there is a small "wire" line above the button contacts
following the dome sheet contour (at least partially)
* there appears to be a rectangular piece of metal soldered on a 
smple GND pad right nead the display connector. It's function isn't 
clear at all (magnet???)

lower board:

* there should be more clearance around the "hot" ends of 
the C-springs
* the position of the C-spring between the lockswitch and speaker 
appears to be wrong
* we have 2 VIAs in 2 pads of the SIM reader - should be avoided

Comment 1 by Joerg Reisenweber, Nov 23, 2013

The strange thing under the display connctor is labeled 
"X2000" in component placement, is not showing up in 
schematics, and magnetic properties of any kind couldn't be found in 
it.
I however found a microscopic "2D-barcode" on it which 
looks like laser-marking. I guess that's the main purpose. We 
probably can safely ignore it.
Sugestion: close ticket.

Comment 2 by Joerg Reisenweber, Nov 23, 2013

sorry of course NOT close ticket since there are several other 
points in it

Comment 3 by Nikolaus Schaller, Nov 24, 2013

More observarions:

* LOWER: PCB snaps nicely into snap-fits and cutout between + and - 
buttons is very good :)
* but mounting holes are not exact enough
* LOWER: mounting holes for USB connector are muchtoo big (radius 
vs. diameter?)
* LOWER: position of ON-Key appears to be weird compared to button 
on case, but fits exactly to PCB
* LOWER: Camera hole appears still to be deplaced by 0.1-0.2mm and 
needs to be drilled in the corners before milling
* LOWER: Pogo-Pins are not well centered where they should be, but 
should work
* LOWER: pushbuttons appear to fit :)
* UPPER: mounting holes do not fit to the display assembly
* LOWER: SD reader and SIM reader footprint appears to fit :)
LOWER: SIM reader position is very good (SD-card not in final 
position - needs another PCB)
* LOWER: 3 of the 4 spring contacts of the battery bay metal do not 
connect to GND because they contact the shield cages on the N900
* LOWER: detector button has wrong footprint (too wide!) - 
calculation mistake when converting mechanical drawing to footprint 
coordinates
* LOWER: we might want to find an OTG socket variant from the 
047589-0001 that has 2 additional pegs like the 047590-0001 -- but 
that may not be necessary at all

Comment 4 by Nikolaus Schaller, Nov 24, 2013

I have soldered some components to one board. This gives new 
observations:

* LOWER: position of USB socket is ca. 0.3-0.5mm off
* LOWER: position of SIM-Reader could be moved by 0.5mm
* LOWER: position of battery connector is almost perfect :) but 
could be retracted a little to reduce risk of short circuit by metal 
of battery bay (but it seems to be exactly the same as on the N900)
* LOWER: SD-reader appears to be operatable if soldered where it is! 
So we might think about avoiding another PCB (but we need it anways 
for the camera lense sensor)
* LOWER: +/-/on pushbuttons work perfectly :)
* LOWER: camera button is unclear if the middle level is already 
pressed (needs to be tested)

Comment 5 by Nikolaus Schaller, Nov 26, 2013

SO far this all was only mechanical and optical observations.
Now, I have taken the Ohmmeter to check important electrical 
functionality before we risk time and expenses to solder more 
components.
The result looks encouraging (and the found problems are not 
show-stopping)

* +/-/on buttons have electrical contact
* !!! camera shutter is always slightly pressed (first contact level 
is always closed)
* speaker contact points appear to fit - can see 7.7 Ohms
* !!! lock switch does not activate (wrong position? bad soldering 
because of wrong footprint?)
* headset connector appears to give contact
* pin order on test points (pcb inner to PCB edge): 2 (right) - 4 
(gnd1) - 1 (tip = left) - 4 (gnd2) - 3 (mic) - 4 (gnd3)
* two of the three "GND" are not really grounded but 
activated by the insertion switch
* if one looks into the hole we can see a bigger and a smaller 
contact right at the beginning of the hole. The smaller one is 
connected to the last (gnd3) and the bigger one to the first (gnd1)
* battery has contact (can see 4V on the test points on the top 
layer)
* footprint orientation of the PHS8 appears to be ok (not mirrored)
* battery power can be seen on the correct LGA pads (Figure 32: 
PHS8-E top view: Pad assignments)
* GPS signal can be traced from LNA to 0R (alternative to U.FL 
socket)
* pin assignment measured through [USB host socket - USB-microUSB 
cable - board - LGA pads] appears to match with 
http://commons.wikimedia.org/wiki/File:USB.svg
* LGA IGT connected to on button
* SIM Card-Detect input at module is GNDed when SIM tray is closed 
(with/without card)
* SIM reader pins connected to LGA pads compared according to 
http://gsmserver.com/pinout/sim.php (needs to solder series-R's in 3 
signal wires)
* !!! Vibramotor case solder pad is not grounded

Comment 6 by Nikolaus Schaller, Nov 27, 2013

Now as we have built one more complete board, the following new 
observations can be added:

* trace width of VBATT may not be wide everywhere (check near C1005
* we need some keepout left and right of the SIM reader because the 
battery bay metal is flanged in that area and touches the PCB. 
Unfortunately we have placed two capacitors there but it is not a 
problem for evaluation
* R12 is not connected to the U.FL antenna socket (bug in 
Schematics)
* LED names are swapped
* Anodes of status LEDs should be driven by VBATT and not VBUS so 
that we can see them in CPU suspend/off state (can be fixed during 
prototyping)
* we need to make sure that we order EVQ-P40B3M and not EVQ-P4HB3B. 
The latter has 2.5N force the first one has 1.0N (bug in BOM 
database)

Comment 7 by Nikolaus Schaller, Nov 28, 2013

CRITICAL!!!

U$3 (the B2B connector near the camera hole) has wrong orientation 
(rotated by 180 degrees)!!! The other one appears to be ok.

Therefore the Lower and the Upper boards do not match and since they 
are (partially) wired, this may lead to serious damage if a battery 
is inserted and/or USB is connected.

So NEVER connect the first prototypes (GTA04b7V0) and power them up. 
Do it just for unpowered mechanical tests and demos.

Comment 8 by Nikolaus Schaller, Dec 17, 2013

Most issues have been worked into the GTA04b7V1 design.
Still open are (and most can only be checked by printing the V1 
boards):

* there is a small "wire" line above the button contacts
following the dome sheet contour (at least partially)
* mounting holes have been moved around, but still may not exactly 
fit
* camera button has been moved by 0.2mm but may still have some 
ofset
* LOWER: we might want to find an OTG socket variant from that has 2 
additional pegs -- but that may not be necessary at all
* LOWER: SD-reader appears to be operatable if soldered where it is! 
So we might think about avoiding another PCB (but we need it anways 
for the camera lense sensor)
* we need to make sure that we order EVQ-P40B3M and not EVQ-P4HB3B. 
The latter has 2.5N force the first one has 1.0N (bug in BOM 
database)

Comment 9 by Joerg Reisenweber, Dec 23, 2013

* volume and power switch: should be twice the operation force. 
* same buttons need to go "deeper": Volume rocker when 
pushed exactly in the middle does nothing on N900 (hits mech 
"axis" support)- on Neo900 LOWER *both* Vol+ *and* Vol- 
get operated when volume rocker gets pushed down at center. 
http://maemo.cloud-7.de/share-service/20131223_009.jpg
Power button feels like it also has too little headroom to switch.
* camera trigger button: PCB bay seems to be significantly less deep 
on LOWER than on N900 - thus cam switch is too high and gets 
operated on first (focus) level in idle position.
* all of the above seems partially related to the LOWER PCB being a 
0.n mm "too high" at the longe edge opposite side to 
vol/pwr buttons. Related also the next point:
* plastic latch middle between Vol and Pwr button is beyond 
mechanical operational range (bends till hits case). buttons and 
generally PCB edge at upper long side is so tight to case that it's 
pretty hard to insert and impossible to remove PCB to/from case. PCB 
needs to get less wide by removing a tad on lower long side. There's 
also a tiny bay where the latch on upper side holds board, this bay 
also seems less deep on Neo900 than on N900. 
http://maemo.cloud-7.de/share-service/20131223_006.jpg
* lockswitch not populated on my device so can't evaluate this one. 
Evaluation pending!
* USB sits too deep (shall sit flush with outer surface of plastic 
case, ends before inside surface though), and sits quite a bit too 
close to lower long edge of board - take care about the latter, it 
will get worse when shortening PCB as of *4/*5 two/three above. 
http://maemo.cloud-7.de/share-service/20131223_010.jpg  
http://maemo.cloud-7.de/share-service/20131223_011.jpg
USB receptacle component MUST have through-hole mech support posts 
(see empty mounting holes in latter URL above)
* SIM tray sits too low on PCB (hits lower edge of apperture in 
steel battery bay) I had to bend the lower vertical steel 
"wall" of the battery bay steel part to make SIMtray fit. 
ATTENTION! This will get affected just like USB one above when 
shortening PCB. 
http://maemo.cloud-7.de/share-service/20131223_001.jpg
* battery contacts need to move 0.5mm direction USB short edge of 
PCB, they are too close to the battery bay steel. 
http://maemo.cloud-7.de/share-service/20131223_003.jpg
* camera apperture/cutout obviously is massively off 
http://maemo.cloud-7.de/share-service/20131223_012.jpg (no parallax 
in that photo!)

Finally: UPPER: should we change shape of UPPEr PCB to be flush with 
*outside* of case? It seems feasible to simply have a nice finisch 
on the PCB edge and make that (part of) our "spacer 
frame". The "hight" above LOWER seems to fit to the 
micron. Would also better support the kbd area to avoid bending of 
PCB (and thus not causing mech pressure/force to high comonents on 
LOWER: modem) when pushing a kbd switch/key. Otherwise we might need 
mech support posts soldered to UPPER and protruding through holes in 
LOWER to the battery bay steel, since modem and other large BGA 
components MUST NOT get mech force bending/pushing them down.

UPPER: don't miss on the Hall-switch above "P" key, for 
detecting slider-open from magnet in display half.

finally a spleeny idea: consider exploiting slider mech metal 
structure for a headsink.

Comment 10 by Joerg Reisenweber, Dec 23, 2013

just a further maybe not so spleeny idea: when UPPER creates part of 
our spacer frame outer surface, we should of course have it plated 
by gold. And when we already have gold there, why not make this a 
few cotact pads for e.g. charger-cradle pogopins to contact there. 
The CRTouch chip's 4 capacitive sensor "buttons" are 
another thing that comes to mind. I suggest to have equaly spaced 
and sized separate gold-pated contacts of maybe 5mm length all 
around the perimeter/edge of UPPER, and use some of them for 
predefined assigned functions like charging, leaving the others for 
our hw-hacker customers by routing them to solderpads nearby and 
maybe even some of them via B2B to LOWER and there to expansion 
connectors.

Comment 11 by Joerg Reisenweber, Dec 23, 2013

HDMI goes there?

Comment 12 by Joerg Reisenweber, Dec 23, 2013

now it takes me away I guess: how about separating above mentioned 
5mm gold pads by LEDs embedded flush into PCB edge? LEDs with 0.8mm 
height and some 1.2mm length would nicely separate the 0.8mm high 
gold pads by 1.3..1.4mm

Comment 13 by Joerg Reisenweber, Dec 23, 2013

>>The "hight" above LOWER seems to fit to the 
micron.

Actually not correct. It's even better, we seem to have enough 
distance between LOWER and UPPER to glue a rubber seal with one side 
coated with stickyfilm to the bottom side of UPPER. A 2mm wide, 
maybe 0.2mm thick stripe all around the circumference, to sit 
between UPPER and case plastic

Comment 14 by Nikolaus Schaller, Jan 6, 2014

Latest status (based on V1 design):

* same buttons need to go "deeper": Volume rocker when 
pushed exactly in the middle does nothing on N900 (hits mech 
"axis" support)- on Neo900 LOWER *both* Vol+ *and* Vol- 
get operated when volume rocker gets pushed down at center. 
http://maemo.cloud-7.de/share-service/20131223_009.jpg
Power button feels like it also has too little headroom to switch.

Ok. This is incremental fine-tuning on the PCB contour (which also 
depends on the precision of the production process).

Has to be improved for V2.

* camera trigger button: PCB bay seems to be significantly less deep 

on LOWER than on N900 - thus cam switch is too high and gets 
operated on first (focus) level in idle position.

This has been addressed in V1 - let's evaluate the result.

* all of the above seems partially related to the LOWER PCB being a 
0.n mm "too high" at the longe edge opposite side to 
vol/pwr buttons. Related also the next point:

That needs further clarification. For my prototype it fits exactly - 
do the plastic cases have production tolerances?

* plastic latch middle between Vol and Pwr button is beyond 
mechanical operational range (bends till hits case). buttons and 
generally PCB edge at upper long side is so tight to case that it's 
pretty hard to insert and impossible to remove PCB to/from case. PCB 

needs to get less wide by removing a tad on lower long side. There's 

also a tiny bay where the latch on upper side holds board, this bay 
also seems less deep on Neo900 than on N900. 
http://maemo.cloud-7.de/share-service/20131223_006.jpg

Ok, needs to be looked at for V2.

* lockswitch not populated on my device so can't evaluate this one. 
Evaluation pending!

Should be possible with V1 (there was a bug in the footprint of the 
lockswitch so that it was not possible to solder it to the PCB).

* USB sits too deep (shall sit flush with outer surface of plastic 
case, ends before inside surface though), and sits quite a bit too 
close to lower long edge of board - take care about the latter, it 
will get worse when shortening PCB as of *4/*5 two/three above. 
http://maemo.cloud-7.de/share-service/20131223_010.jpg  
http://maemo.cloud-7.de/share-service/20131223_011.jpg

This has been addressed in V1 - let's evaluate the result.

USB receptacle component MUST have through-hole mech support posts 
(see empty mounting holes in latter URL above)

The USB socket we currently have mounted is the Molex 47590-0001
We just need the part number of the variant "with pegs".

* SIM tray sits too low on PCB (hits lower edge of apperture in 
steel battery bay) I had to bend the lower vertical steel 
"wall" of the battery bay steel part to make SIMtray fit. 
ATTENTION! This will get affected just like USB one above when 
shortening PCB.
http://maemo.cloud-7.de/share-service/20131223_001.jpg

This has been addressed in V1 - let's evaluate the result.

* battery contacts need to move 0.5mm direction USB short edge of 
PCB, they are too close to the battery bay steel. 
http://maemo.cloud-7.de/share-service/20131223_003.jpg

This has been addressed in V1 - let's evaluate the result.

* camera apperture/cutout obviously is massively off 
http://maemo.cloud-7.de/share-service/20131223_012.jpg (no parallax 
in that photo!)

This has been addressed in V1 - let's evaluate the result.
Summary: PCB affecting issues collection
Relations: is related to 544, 545

Comment 15 by Joerg Reisenweber, Feb 2, 2014

re comment #9, 4th *:
>* all of the above seems partially related to the LOWER PCB 
being a 0.n mm "too high" at the long edge opposite side 
to vol/pwr buttons.

n=0.28mm
Original: 55.72mm high
Neo900: 56.00mm high
Measured maximum width at both ends of long sides.
This is not yet considering any resections particularly in long 
sides of PCB, which are partially missing in Neo900 PCB.
http://maemo.cloud-7.de/Neo900/media/pic+vid/PCB-shape/DSCF0160.JPG 
and other photos same dir.

Created: 7 months 19 days ago by Nikolaus Schaller

Updated: 5 months 10 days ago

Status: New

Followed by: 1 person

Labels:
Type:Defect
Priority:Critical