Hi, >> It did not go well. Upon powering up the boards without a bitstream loaded >> (FPGA idle with pre-configuration weak pull-up resistors on all I/Os) the >> current was at the usual value (0.6A). Then, after about 30s it jumped to
>> more than 1A with a short circuit to ground appearing on the 3.3V and 1.8V >> supplies. Once the board developed the short circuits, they did not go >> away by power-cycling the board and the 3.3V and 1.8V voltages could not >> be restored. The problem was exactly the same on the two boards. >> >> I am quite puzzled about explaining the cause of this issue. Possible >> tracks: * connecting P[15:8] to the weak pull up resistors of the FPGA >> causes the video chip to fail and short circuit 3.3V and 1.8V to ground? >> how come? * solder-related problems, e.g. tin whiskers? >> * chip overheated during rework? >> * bad/counterfeit chip installed? (like the IR sensors...) To eliminate the last 3 reasons, please don't use #1 #3 #4 to suffer from 1V2 injected potential reason. Here can be a quick test: a), Using another new video decoder in #2, #5 or #6. Not sure which one you have now. b), If a) still gets the same status, high probably not the last 3 reasons caused. Otherwise, my rework procedures or sourcing are no good. c), If yes the last 3 reasons, in order to find root cause(sourcing or over-heated), pls use another new video decoder and rework it at #1 or # 4. then the answer will be discovered. Nevertheless, there are some info which are shared first to try to help: 1), Those two M1s are # 1 & #4 which their each current branch has been measured [1] before video input patched and without fpga configuration porting. The measure is over 2 minute at least. Even they have "ever" been injected 1V2 supply only during initial test. The #3 also "ever" but without test all current branches. So did that injection test make damages inside Spartan-6 already? We didn't know exactly then. 2), The total current(measured at DC plug cable side) of #1 & #4 after video input patched is still around 0.6A. I didn't measure current over 30s when 5V power on, so no chance to meet over-1A condition. 3), The patched I used new ADV7181B with the same lot as first 6 (#1~#6) pcs run. The datacode is 0941. Sample ordered them from Answer Technology Co. Ltd [2] [3]. 4), The two M1s patched I measured each adjacent pins(64pins) to check if short existed before powered up. Should be No tin whiskers existed. so b) is needed to be checked/clarified. But not sure if you have available time. If not 3 reasons, we may look into more checks as below: 5), After power-on 5V, is there any codes running inside fpga started between power up after 50ms and before 30s? >> The video input chip seems to be the cause of the problem - I have removed >> it from one board which had developed the short circuits, and all the >> voltages and currents went back to normal and the rest of the board works. This can be realized that the routed traces to fpga is at high impedance status when you removed. It is the same conditions as decoder set by default value at 8 bit mode in P[15:8] and the one we met were at Three-state on P[7:0] without patched. Then all went back and well are reasonable. So the next more checks on like Florian's email said. 6), > How about a test-mode sequence of the video-chip that you enter because of its > input states? (I did not look at its datasheet). http://www.analog.com/static/imported-files/data_sheets/ADV7181B.pdf 7), There's a question: The digital core of the ADV7181B can be shut down by using a pin (PWRDN) and a bit (PWRDN). U21's pin30 PWRDN_N is always at HIGH.The PDBP controls which of the two has the higher priority. By default, the pin (PWRDN) is given priority. This allows the user to have the ADV7181B powered down by default. So if you set PDBP bit to 1 immediately though I2C after configuring fpga I/O firstly, we can know if over current high again. --> this can use # 1 or #4 to test. Hope this help. Tks, Adam [1] http://en.qi-hardware.com/wiki/Milkymist_One_RC1_Reports#Measurement_of_Voltage.2FCurrent_during_Standby.2FQuiescent_Mode [2] http://www.analog.com/salesdir/countries.asp?regionID=2&countryID=205 [3] http://www.anstek.com.tw/ _______________________________________________ http://lists.milkymist.org/listinfo.cgi/devel-milkymist.org IRC: #milkym...@freenode Webchat: www.milkymist.org/irc.html Wiki: www.milkymist.org/wiki
