Hi all, there wasn't anything new to report because we are waiting for the production line.
> Am 07.11.2016 um 10:32 schrieb H. Nikolaus Schaller <h...@goldelico.com>: > > Anyways there is a very clear pattern. > If we have (almost) no flux there are many unconnected pads. > If we have (too) much flux there are shorts in the middle. > > Therefore the most promising approach appears to reduce the amount of flux in > the middle. > > This either means using a stencil with smaller openings in the middle or > close them completely. But today I got a phone call from production that they have a time slot today to do the next attempt. So I went there immediately (fortunately it is not far away) and we worked 3 hours to produce 4 more complete GTA04A5 boards. Unfortunately, we still have no working PCB :( * one of them have short circuits - but much less than before. * one of them has a rotated memory chip (which can very likely be fixed) * one has a short under the tps65950 power controller. * and one has no shorts to measure. So we placed an µSD card into the one without shorts and connected a battery. But no LED did light up yet. It could be possible that there is something with the GTA battery I took with me or the µSD card. I would have needed an oscilloscope and some RS232 cable to look what the board is really doing when being powered on (are all voltages available, are the oscillators running etc.). Then we did X-Ray and as expected the shorts in the middle of the DM3730 processor are gone :) But this time some boards have visible shorts in the outer ring which we did not have in earlier experiments. So this means we are on the right track, but still not there to risk mass production. The idea is to get another stencil (we did this work with the same stencil as for all attempts before) where we reduce the openings for in the outer ring to have less solder paste there. Again this will need some time. We need to talk to the stencil manufacturer, have the stencil built (this takes the shortest time: 2 days) and then wait for another time slot in the production line. Talking about risk. Well, so far we have spent money for ca. 15 GTA04 boards and only one of them shows activities. But it wasn't populated completely and has solder openings in the outer ring. Therefore I wasn't able to test it completely. But it did boot, makes the LEDs blink, shows X11 on the display. But 14 of these are scrap with low probability that they can ever be repaired. As you know what each working GTA04A5 costs this is already a lot of money that wasn't in the financial plan in this amount. So let me ask for donations to support this experiments which need to be done to really fix production issues. What is your benefit: a) it supports getting the GTA04A5 to a success, b) success of the GTA04A5 means the most free hardware becomes available again, c) the device is well supported by mainline kernel, d) since the Neo900 wants to use exactly the same DM3730CBP chip it will run into the same soldering problems, unless we solve them here. Here is a link to the goldelico shop where you can donate for this project: http://shop.goldelico.com/wiki.php?page=GTA04%3ADonation Thank you, Nikolaus _______________________________________________ Community mailing list Community@openphoenux.org http://lists.goldelico.com/mailman/listinfo.cgi/community http://www.tinkerphones.org
