[Gta04-owner] [Tinkerphones] Status of GTA04A5 production +++ has started +++ video available
H. Nikolaus Schaller
hns at goldelico.com
Sat Oct 8 09:20:26 CEST 2016
> Am 08.10.2016 um 08:39 schrieb dalo9fz at gmail.com:
> Hi Nikolaus (and others),
> About the shorts on the DM3730, i asked some profs in university about it.
> Maybe i'm late.
> One prof told me (rephrasing):
> There are 3 things to check but considering the setup described, the
> people (you) know what they are doing. It is not a bad thing to remind
> them of it.
> It *could* be a simple problem and not an exotic one.
> Here are some things one could check to troubleshoot this problem.
That is something to check!
> 1. Heat and warping:
> It could be that the board itself warps enough to have misplacement of
> the chip:
> Do you have gold inlays on the traces (between solder mask)?
> if not, this could help.
> Also, the board needs to be super flat, the inlays must have the
> correct height,
> or warping might take place on the board.
That is something we can't exclude completely, but the board is small
so that it is quite stiff compared to "big" boards.
Otherwise it has the same characteristics as the previous GTA04A3 and A4
> 2. Misplacement/alignment:
> It could be that the pick and place machine needs to be calibrated a
> can you verify that the placing of the chip is exact ?
> Also it could be that the translation of the gerber messes with the
> size of traces etc..
> can you verify that the board has the exact measurments like in the
They have tested this by pasting the PCB and placing the chips and then
putting them into X-Ray before soldering (fortunately they have bought their
own X-Ray machine two months ago). All 4 cips in the panel were placed
Potential gerber displacement is covered by their Pick&Place machines by
a camera system. So it looks onto the PCB and the Chip and fine-tunes the
position on the fly.
> 3. Solder paste:
> Can you verify that the solder paste is distributed evenly ?
> enough paste and not too much either.
> maybe change solderpaste?
Hm. We did make one experiment without solder paste but they were dipped
into flux. And did have the shorts as well. So I think it is not strongly
influenced by solder paste, but there could be a general influence by the
amount/type of flux applied. Either through solder paste or dipping.
Unfortunately, no documentation exists for these details from GTA04A3
or A4 production to compare to.
> I understand that these might not really be answers you are waiting for,
> but sharing is caring :)
I think this is a problem without simple (textbook like) answers.
Unfortunately for all of us who are eagerly waiting for success, it has
become more a research project...
Doing experiments, forming theories, doing more experiments, updating
theories... But budget is very limited.
For the next experiment (likely to happen next week) I have discussed 4
different ways for preparing the PCB. One is without any solder paste and
flux (to see if the DM3730 itself warps too much). Others are with partial
solder paste and again others with placing a tiny snipped of kapton tape
under the chip.
The last one did already work in my manual lab setup - and X-Ray shows that
there are no shorts any more. But soldering the PoP RAM/NAND chip makes
problems in this setup. This all points towards a warping effect of the
DM3730 chip itself.
By my manual rework, I have currently one board (without 3G module) which
is almost working (1GB RAM is there, display works, USB works, etc.) and
which we can use for fine-tuning the kernel (device tree) and U-Boot.
Unfortunately I wasn't able to get any reaction from the WLAN and Bluetooth
modules yet. And the IrDA module is not properly turned off and interferes
with the RS232 console. This might be pinmux related (maybe wrongly set up
> No reply required.
"sharing (thoughts) is caring :)"
> Thank you so much for al your (the whole team and list) efforts.
Thanks for the help!
More information about the Gta04-owner