Hi, * The whole M1rc3 run[1] has been almost finished, a total yield ratio 88pcs / 90pcs. (97.8%) is came out. Remaining two boards are still under investigation.
While initial test of first board, we ran into darn[2] of never booted up about solutions from reset IC and diode since an unqualified and unverified with strict reviews after rc2 run we made. Then thanks to Werner and Sebastien's helps, this post-design verification has been implemented by h/w patch called 'fix2b' [3]. * Classifications of failure[4]: - No VGA synchronized signals out :: ratio 27 / 90 caused by unqualified Schmitt-Trigger IC( u19 or u20 ), a whole rework of replacing them on 90sets is done. - Impedance/Current failure :: 9 / 90 caused by: 1 / 9: unqualified Schmitt-Trigger IC( u7 ), a whole rework of replacing u7 on 90 sets is done. 4 / 9: unqualified usb transceiver ( u16 or u17 ), rework of replacing failed one is done. 1 / 9: by C30 or C36 ( not sure which one caused it since I removed them at moment ) C0402 footprint, we'll need to review this footprint later. 1 / 9: by C154, C0805 footprint; the gap between two pads probably is thinner, this is to be changed for sure. 1 / 9: by C104, C0805 footprint; this part is nearby area of rework of reset circuit, and a solder splash on C104, is a victim of rework. 1 / 9: Impedance of TP2 to GND is 380 ohm that differs from normal 628 ohm. Quiescent current is still stays 0.54A in normal. Works well after overnight rendering. - Flash failure :: 11 / 90 Due to twice significant h/w patches( reworks ) called 'fix2' and 'fix2b', we have got total 11 pcs in fail from them. After used a version of reflash script with 'lockflash', they are all fixed. see [5] in details. This's especially thanks to Werner's found.[6] - VGA DDC :: 6 / 90 Records here indicates that (female to male) expendables connected to VGA Monitor's bounded cable are prepared beforehand, so that frequently replug-in and replug-out connections with m1 boards. - Audio :: 2 / 90 by L1 is a 0402 package but footprint is 0603, needs to change 0402 footprint size. - Ethernet ARP :: 1 / 90 Indicates that just plugged cable was contacted well though. - Video I2C :: 1 / 90 The VIDEOIN_SDA doesn't still stay steadily at high level. Then used heat air to blow AB17(VIDEO_SDA) ball and passed. It's probably caused by that it's a flux stocked/mixed with volatile cleaner surrounding ball area while several times of reworks nearby 'reset circuit' to cause it. A infrastructure of clean way needs to be settle well. - MIDI :: 12 / 90 10 / 12: unqualified Schmitt-Trigger IC( u7 ), a whole rework of replacing u7 on 90 sets is done. 1 / 12: 0x7d, after reflashed successfully with './reflash_m1.sh --rc3' version "2011-10-14" with locked bit, then TP31 works @ 3.3V correctly.It's weird but still checked its level and worked well after overnight rendering test. 1 / 12: 0x46, failed by a 'foreign body' which mixed and composed of unqualified process while printing soldering balls, see[7]; after reflow by bga rework station then became 3.3V level works well, see[8]. - USB :: 22 / 90 by unqualified usb transceiver, replaced the failure one and pass. - Rendering :: 9 / 90 this recorded failure result( i.e. nor-corruption issue ) that in first round of "fix2" rework was not effective. After rework of "fix2b" patch, accept it and face it as pass. - CRC :: 1 / 90 After applied fix2b circuit patch, solved. * Improvements in topics next: - A design verification being approved by enough relevant experiments. - A specified and well-done temperature profile to be measured and best that monitoring the reflow process in personally appearance. This is try to do the smt works good at the beginning. - Sourcing source picked must change to boom system or famous manufacturers. - Reduce reworks as possible. If need, think it firstly before rework and also make sure an infrastructure of clean way settled well on both factory and my site. * Known issues of M1rc3 main board run, see[9] - After then a nor corruption problem was discovered. A M1pre-rc4 proposed reset circuit[10][11] are came up. - Video-In codec will be switched to 7181C in rc4. - USB power switch can be added[12], Werner's test results of "USB current limit vs. capacitative loads"[13] - replaced L19 with a short connection by soldering on all run3 boards to solve the intermittent loss of video input. - some topics like LEDs and DVI-I idea have been brought up. * Some improvements[14] to be prepared before rc4 about assembling and packing: With packing instructions[15], M1rc3 they've been assembled with acrylic cases and packed with accessories. - Incoming parts inspections :: cases, cleanliness of shield sheet, if existed that big spacers without screw threads, insertion of retailer box. - work out an efficient way to clean such "small bits and pieces" on surface of rc3 top case after tearing off films. If some details I didn't bring up with regardless here, they should have been recorded in [1]. I'll watch it out. [1] http://en.qi-hardware.com/wiki/Milkymist_One_run_3_schedule [2] http://en.qi-hardware.com/wiki/Milkymist_One_RC3_Known_Issues#darn_of_never_booted_up [3] http://en.qi-hardware.com/wiki/Milkymist_One_run_3_schedule#Hardware_Patches_of_Boot_up [4] http://en.qi-hardware.com/wiki/Milkymist_One_run_3_schedule#Classifications_of_Failure [5] http://en.qi-hardware.com/wiki/Milkymist_One_run_3_schedule#Test_Results [6] http://lists.milkymist.org/pipermail/devel-milkymist.org/2011-October/001951.html [7] http://en.qi-hardware.com/wiki/File:M1rc3_0x46_AB21_ball.png [8] http://en.qi-hardware.com/wiki/File:M1rc3_0x46_AB21_ball_after_reflow.png [9] http://en.qi-hardware.com/wiki/Milkymist_One_RC3_Known_Issues [10] http://lists.milkymist.org/pipermail/devel-milkymist.org/2011-December/002335.html [11] http://lists.milkymist.org/pipermail/devel-milkymist.org/2011-December/002487.html [12] http://lists.milkymist.org/pipermail/devel-milkymist.org/2011-December/002334.html [13] http://lists.milkymist.org/pipermail/devel-milkymist.org/2011-December/002502.html [14] http://en.qi-hardware.com/wiki/Milkymist_One_run_3_schedule#Picture_Gallary_is_collected_between_RC3_and_Pre-RC4 [15] http://en.qi-hardware.com/wiki/Milkymist_One_accessories#packing_instructions Thanks, - adam
_______________________________________________ http://lists.milkymist.org/listinfo.cgi/devel-milkymist.org IRC: #milkymist@Freenode
