Test status with bonded/globbed/balled carrier 20141009 ------------------------------------------------------------- Packaged PGA chip ================= Works as is. The carrier did not fit in the testsocket ========================================= One short side had to much glob outside the carrier. The side was grinded. The globbed surface is not flat. At the place with too much glob is also a hill. A piece of bicycle tube was placed between the chip and the pressure plate of the socket, this was to make the pressure more equal over the surface. Trying without this made the carrier to tilt in the socket. Can not test the connectivity in the test socket, the pressure was roughly the same as with shorted carrier (as felt by "hand") Input voltages to the test socket board ======================================= AdcAnalog power was fluctuating. A capacitor was installed on the protoLV to even it out. Bug on carrier ============== Instead of termination has the sampling clock a filter that according to simulations kills the clock. The SALTRO needs the sampling clock to work. Tried workaround: There is an auxilliary clock input for the digital part that can be enabled, which we think should make it possible to read/write registers. After some investigation and better understanding of the test board could we enable the auxilliary clock on the testboard: To get clocks from the testboard was it needed to use the GTL clock path (used to set the clock amplitude to 1.5V) instead of using the resistor divider on the testboard. The resistor divider killed the clocks from the testboard, for no known reason. Short circuits ============== There is a short between: 1) RefP and AdcaAnalog power 2) ScanMode and data line 29 (BD29 which is bit 0 of the chip address) workaround: RefP must be set to 1.5V via the DAC (nominal value 1.25V). ScanMode and BD29 was cut and set to ground via a resistor on the testsocketboard. This means that BD29 is also grounded, and the chip must be addressed as 0,2,4 or 6. We think ScanMode and ScanEnable are active high, and was therefore always set by BD29 which is normally high, i.e. the SALTRO was in a scan mode, which we do not know how it works. Further tests done to determine the place of the short circuit: The short is NOT on the test board (measured) the test socket board (measured) Carrier board as measured on a globbed carrier without chip. Suspect it is the bonding wires that are shorted. The short circuits are between neighbouring chip pads. Pickups ======= ScanMode and ScanEnable have pickup from the data bus on the testsocketboard. May this disturb the chip? DOLO_EN (determines the ALTRO data bus direction), which is set by the SALTRO, have noise that coincides with the write strobe. DOLO_EN should only be set during data readout or reading a register, NOT during write. It is not there when the carrier is not in the socket. Else of unknown origin. Currents ======== The currents drawn by the different powers as measured by the ADC on the protoLV are: By the chip: AdcAPower 100 mA AdcDPower 0 mA PasaPower 120 mA DigPower 0 mA IOpower 30 mA By the testsocketboard: I2CPower 0 mA A3V3Power 90 mA Total 310 mA which also measured from the power supply. Note that AdcDPower (digital) and DigPower do not draw any current. Damages ======= Have taken out the carrier from the test socket to examine socket and carrier. Need better equipment to be able to inspect possible damages to the carrier/balls/socket in detail. Though comparing pictures (location see below): IMG_92.BMP ballside before grinding IMG_98.BMP ballside after taken out of test socket may indicate damage to the balls. Questions asked to Massimiliano =============================== 1. There are a few things about the clock distribution we do not understand, e.g. why the resistor divider do not work for us. 2. Is it enough to have the auxilliary clock to read/write registers 3. How does ScanMode/ScanEnable work No answer yet. Summary ======= We can not get any contact with the SALTRO on the carrier. Could one grind the carrier to cut the capacitor to the sampling clock? Have no further ideas about what to do with current carrier. Pictures in http://www.hep.lu.se/eudet/pictures/testsocketboard/chiptests-20141009/ IMG_87.BMP balls on carrier IMG_88.BMP balls on carrier IMG_89.BMP balls on carrier IMG_90.BMP balls on carrier IMG_91.BMP globside before grinding IMG_92.BMP ballside before grinding IMG_93.BMP placed in socket before grinding IMG_94.BMP placed in socket after grinding IMG_95.BMP test board IMG_98.BMP ballside after taken out of test socket IMG_99.BMP short side with hill on glob IMG_100.BMP short side with hill on glob IMG_101.BMP other short side IMG_102.BMP bicycle tube after removing from socket IMG_103.BMP bicycle tube after removing from socket saltrotest-allon.jpg Settings as shown in the control user interface