Controller PCB problem solved
1/24/2012 11:07 AM EST
The year was 1996 and it was my second year with this Hard Disk Drive company. I had joined them as an EMC Engineer to provide EMC support to the PCB design engineers. Within a year I was being called to help on almost every noise issue with the disk drives.
Our off-shore drive assembly plant reported a problem with the controller PCB, which was causing significant yield loss. A few suspect cards were sent to me to investigate. I was also informed that they had replaced all the modules (Processor, DRAM, DSP, Controller, Flash, etc.), and were suspecting the problem with the card fabrication process. I started sniffing the good and bad cards with my close-field EMI probe, and found out that the bad cards showed 40 MHz noise at the back side of the card. I traced the source of the noise to a couple of SRAMs, which were on the back of the card and had not been replaced by the assembly plant during troubleshooting. Inspection of the SRAMs showed that the SRAM lot number for the bad cards was different. The SRAM supplier company was contacted, and they informed us that they had been shipping die-shrink parts without obtaining our approval.
Once the problem was diagnosed, the use of die-shrink parts was stopped. Our commodity group got involved in the How and Why of the issue, which continued for a very long time.
Ravender Ajmani is the Principal Engineer at Hitachi Global Storage Technologies.