need help repairing LSI 9201-16i card

BLinux

cat lover server enthusiast
Jul 7, 2016
2,519
964
113
artofserver.com
i have a LSI 9201-16i HBA controller I need to repair. got it off local CL for $100 and thought I got a nice deal, but turns out Port-A has some physical damage I didn't notice. The card boots up fine and everything and I thought it was good to go. been testing it with a 846A backplane and noticed one slot in the backplane didn't recognize the hard drive. swapped cables and such and eventually realized port-A of the LSI 9201-16i wasn't right. took a closer look at the card and noticed some broken capacitors. see picture at C3, C4, C5, and C12.

LSI9201-16i-damage.jpg

i blasted off an email to LSI/Broadcom/Avago, but not sure if they will be willing to tell me the specifications of these components. 3ware use to be open about sharing that kind of information, but not sure about LSI and the larger Broadcom/Avago now.

anyone here have an internal contact at LSI that might be able to help? Or, anyone done this repair before and know what those components are?
 

sullivan

New Member
Mar 27, 2016
24
16
3
Based the placement near the connector, it is very likely these are AC-coupling capacitors on the TX+/TX- or RX+/RX- high speed data differential pairs.

I believe the SAS / SATA specs define these as 10 nF (or 0.01 uF, or 10000 pF -- depending on your choice of SI prefixes).

These caps look like 0402 case size to me. Other typical specs for this part would be 10 nF +/- 10% tolerance, 50V, X7R dielectric.

However, it looks like these capacitors were sheared off the board and the surface-mount pads were ripped off. If so, you won't be able to solder down a replacement part.

If the pads are indeed damaged, and if you are feeling adventurous, here's what I recommend:

For each pair where one of the caps is damaged, remove both caps off the board and clean up the pads as much as possible.

LSI has nicely left soldermask off the nearby VIAs to use as manufacturing test points. Cut 2 short pieces of wire and bridge them across the test points to short out the removed caps. Trim the wires right at the test points so you have no stubs. Try to imagine what the wiring would look like if there were no caps and the traces on the PCB just ran straight through.

As noted above -- you need to do this for both sides of the pair, i.e. you either have 2 caps or no caps / 2x shorts.

The basic idea here is that SAS / SATA is specified to run as an AC-coupled interface. However, there are usually AC coupling caps on both ends of the cable. If you short out the caps on one end of the cable, you are still AC-coupled on the other end. You might end up with worse signal integrity but with short cables in an enclosure you probably have plenty of margin.
 
  • Like
Reactions: BLinux