SFP to FC schematic

[guru]

Hi,

I'm looking for a schematic for an SFP to FC (40-pin SCA connector) adapter.

Any help much appreciated


Guru

[DR-Kiev]

Hi,

I'm looking for a schematic for an SFP to FC (40-pin SCA connector) adapter.

Any help much appreciated


Guru

If you need restore data from drive , you can try to adopt SCA80 PCB instead of Fibre Channel one , sure if we are talking about Seagates.

Or, are doing some necro researches ?

[guru]

Nothing data recovery related. Need test FC drives without any changes to their PCBS

[drc]

Use a T-Card?

[Vulcan]

@guru:

Need test FC drives without any changes to their PCBS

I've done this a couple of different ways:

* small (2/3 drive) FC JBOD chassis, with GBIC socket (yes, it was old!). You could use a copper or optical GBIC in the JBOD, to match whatever is in the host.

or

* little T-card PCBs which came from Seagate USA ages (many years!) ago, which plug onto the SCA-40 drive connector and have separate power, TX & Rx connectors for the FC signal.

This company sells similar PCBs & cabling:

http://www.sca40.com/products.html

You would need to use (and power) an MIA if you use optical-only HBAs. However if your HBA has a GBIC socket, you can use a copper GBIC and cable direct to those little PCBs, so there is no optical component involved (that was my main setup, but it was all bought/made long ago - I only saw that company linked above recently and haven't used them myself yet).

Hope that gives some ideas

@drc:

[drc]

I've gotten stuff from these guys. Good quality, and relatively cheap (though their website seems to be having speed issues currently):

http://store.ckcomputersystems.com/inde ... cPath=1_67

[guru]

Hi

Yes I have seen the t-cards. I need to make a custom PCB for our mass testing needs. If I had the schematics or Gerber files I could just CUT \ PASTE what I need. Making the board from scratch will take me a while and I don't have the bandwidth. Thanks anyway guys, I will keep looking.

G

[Rainbow]

Reviving this old thread because I need to test (SCA-40) FC drives (EMC).

Looks like it's not necessary as hard&expensive. Someone is selling old FC controllers (HHBA-5101) modded by adding a cable with SCA-40 connector:
https://www.ebay.com/itm/314245319246

Maybe I can get a SCA-40 interposer board from a dead EMC drive, solder a cable to pins 24, 25 (+IN1, -IN1) and 30, 31 (+OUT1, -OUT1) and connect the other end to the GBIC or SFP on a FC HBA card? I have some extra QLE2460 HBAs...

The pinouts are here: https://en.wikipedia.org/wiki/Fibre_Cha ... _interface

Attachments

[Rainbow]

The EMC interposer board 303-095-002B provides 5V supply to drive (older drives without this board are 12V-only) using MP1583DN step-down converter and also short-circuit protection using NIS5112D1 electronic fuse and 5A fuse. It also contains READY and FAULT LEDs.

The input and output SCA-40 connectors are wired 1:1 on most pins, except:
Pin 3 on the input side is not connected with other +12V pins but used as pull-up voltage for NIS5112D1 enable input.
Pin 8 on the input side is connected to NIS5112D1 enable input.
Pins 19, 20, 40 are not connected on the input side. They are connected to the onboard 5V regulator on the output side.

So connect GND, short input pins 2+3 together, connect 12V there (or to the fuse on the bottom side) and ground pins 9, 10 to enable spin-up.

Seagate drives will spin up but FAULT LED is ON.
Some HGST drives will spin up without FAULT LED.
But some other HGST drives will not spin up, FAULT LED is blinking some error code (1 long, maybe 5-6 short blinks).

I don't have FC connected yet. Hopefully the FAULT LED behavior is caused by the lack of FC link.

Attachments

[Rainbow]

Removed SFF from QLE2460 and connected HDD +IN1/-IN1 to TD+/TD- (SFF pin 9/10) and +OUT1/-OUT1 to RD+/RD- (SFF pin 5/4). Used twisted pairs from an ethernet cable.

Unfortunately, it does not work. With a HGST drive, I get even LOOP UP at 1Gbps but the drive itself is not detected:

Code:

Jun 29 13:41:43 debian kernel: qla2xxx [0000:00:00.0]-0005: : QLogic Fibre Channel HBA Driver: 10.02.07.900-k.
Jun 29 13:41:43 debian kernel: qla2xxx [0000:03:00.0]-001d: : Found an ISP2432 irq 16 iobase 0x00000000ebd5a2bd.
Jun 29 13:41:43 debian kernel: qla2xxx [0000:03:00.0]-ffff:8: FC4 priority set to FCP
Jun 29 13:41:43 debian kernel: qla2xxx 0000:03:00.0: firmware: direct-loading firmware ql2400_fw.bin
Jun 29 13:41:43 debian kernel: scsi host8: qla2xxx
Jun 29 13:41:43 debian kernel: qla2xxx [0000:03:00.0]-00fb:8: QLogic QLE2460 - PCI-Express Single Channel 4Gb Fibre Channel HBA.
Jun 29 13:41:43 debian kernel: qla2xxx [0000:03:00.0]-00fc:8: ISP2432: PCIe (2.5GT/s x4) @ 0000:03:00.0 hdma- host#=8 fw=8.07.00 (9496).
Jun 29 13:41:43 debian kernel: qla2xxx [0000:03:00.0]-500a:8: LOOP UP detected (1 Gbps).


Seagate flashes the FAULT LED and loop is down:

Code:

Jun 29 13:51:32 debian kernel: qla2xxx [0000:03:00.0]-8038:8: Cable is unplugged...

[Rainbow]

Got LOOP UP with Seagate too, the SCA connector was probably not inserted enough before. But drive not detected, though...

[Rainbow]

Success!

According to HGST Ultrastar 15K600 (FC-AL) Hard Disk Drive Specification, SEL_ID of 0x7F (which is the default as there are pull-ups on all SEL_IDx pins) "forces the drive to obtain a soft address". Don't know what is a soft address but it's not enough for the drive to be detected.

It also says:

These 7 signals define 128 possible values and are directly translated into an 8 bit hard AL_PA via the table below. Only AL_PA's with neutral disparity are valid values.

Neutral disparity means that the byte must have 4 zero and 4 one bits.
Is grounding one SEL_IDx pin enough to get a valid AL_PA? Yes, e.g. SEL_ID=0x77 means valid AL_PA 0x17 (0b00010111).

So grounded pin 15 (SEL_ID3) and it works!

Code:

[   14.671121] qla2xxx [0000:00:00.0]-0005: : QLogic Fibre Channel HBA Driver: 10.02.07.900-k.
[   14.671331] qla2xxx [0000:03:00.0]-001d: : Found an ISP2432 irq 16 iobase 0x000000002f9de308.
[   14.769764] qla2xxx [0000:03:00.0]-ffff:8: FC4 priority set to FCP
[   15.091067] qla2xxx 0000:03:00.0: firmware: direct-loading firmware ql2400_fw.bin
[   15.414448] scsi host8: qla2xxx
[   15.418576] qla2xxx [0000:03:00.0]-00fb:8: QLogic QLE2460 - PCI-Express Single Channel 4Gb Fibre Channel HBA.
[   15.418597] qla2xxx [0000:03:00.0]-00fc:8: ISP2432: PCIe (2.5GT/s x4) @ 0000:03:00.0 hdma- host#=8 fw=8.07.00 (9496).
[   16.339800] qla2xxx [0000:03:00.0]-500a:8: LOOP UP detected (1 Gbps).
[   16.422600] scsi 8:0:0:0: Direct-Access     SEAGATE  STT14685 CLAR146 6308 PQ: 0 ANSI: 3
[   16.426931] sd 8:0:0:0: Attached scsi generic sg1 type 0
[   16.427065] sd 8:0:0:0: Power-on or device reset occurred
[   16.428181] sd 8:0:0:0: [sdb] Unsupported sector size 520.
[   16.428196] sd 8:0:0:0: [sdb] 0 512-byte logical blocks: (0 B/0 B)
[   16.428201] sd 8:0:0:0: [sdb] 520-byte physical blocks
[   16.429211] sd 8:0:0:0: [sdb] Write Protect is off
[   16.429224] sd 8:0:0:0: [sdb] Mode Sense: a7 00 10 08
[   16.430776] sd 8:0:0:0: [sdb] Write cache: disabled, read cache: enabled, supports DPO and FUA
[   16.431621] sd 8:0:0:0: [sdb] Attached SCSI disk


Hot-swapped a HGST drive:

Code:

[  963.903534] qla2xxx [0000:03:00.0]-500b:8: LOOP DOWN detected (2 5 0 0).
[  989.745628] qla2xxx [0000:03:00.0]-500a:8: LOOP UP detected (1 Gbps).
[  990.278043] scsi 8:0:1:0: Direct-Access     HITACHI  HUS15606 CLAR600 C7A0 PQ: 0 ANSI: 6
[  990.357682] sd 8:0:1:0: Attached scsi generic sg2 type 0
[  990.368566] sd 8:0:1:0: Power-on or device reset occurred
[  990.381653] sd 8:0:1:0: [sdc] Spinning up disk...
[  991.394130] .................ready
[ 1007.842477] sd 8:0:1:0: [sdc] Unsupported sector size 520.
[ 1007.842494] sd 8:0:1:0: [sdc] 0 512-byte logical blocks: (0 B/0 B)
[ 1007.842499] sd 8:0:1:0: [sdc] 520-byte physical blocks
[ 1007.842781] sd 8:0:1:0: [sdc] Write Protect is off
[ 1007.842787] sd 8:0:1:0: [sdc] Mode Sense: d3 00 10 08
[ 1007.843181] sd 8:0:1:0: [sdc] Write cache: disabled, read cache: enabled, supports DPO and FUA
[ 1007.861423] sd 8:0:1:0: [sdc] Attached SCSI disk


There are still 4 remaining HGST drives that only blink the FAULT LED. Maybe they're really faulty?

[Rainbow]

Some photos.
I should probably use some better cable (shielded) for data wires. This seems to work but might fail with big data transfers.

Attachments

[Rainbow]

The 4 non-working HGST drives are NOT faulty. They spin up and detect in the real thing (CX3)
Every other drive worked - including the same HGST model - weird.

[Rainbow]

Looks like the 4 non-working drives don't support 1Gbps rate.
Grounded pin 18 (DEVCTRL1) and they now work!

Code:

[   14.637525] qla2xxx [0000:00:00.0]-0005: : QLogic Fibre Channel HBA Driver: 10.02.07.900-k.
[   14.637763] qla2xxx [0000:03:00.0]-001d: : Found an ISP2432 irq 16 iobase 0x000000002c3cd9cc.
[   14.710832] qla2xxx [0000:03:00.0]-ffff:8: FC4 priority set to FCP
[   14.877718] qla2xxx 0000:03:00.0: firmware: direct-loading firmware ql2400_fw.bin
[   15.198260] scsi host8: qla2xxx
[   15.201000] qla2xxx [0000:03:00.0]-00fb:8: QLogic QLE2460 - PCI-Express Single Channel 4Gb Fibre Channel HBA.
[   15.201025] qla2xxx [0000:03:00.0]-00fc:8: ISP2432: PCIe (2.5GT/s x4) @ 0000:03:00.0 hdma- host#=8 fw=8.07.00 (9496).
[   15.457323] qla2xxx [0000:03:00.0]-500a:8: LOOP UP detected (4 Gbps).
[   16.229535] scsi 8:0:0:0: Direct-Access     HITACHI  HUS15606 CLAR600 C3A8 PQ: 0 ANSI: 6
[   16.231208] sd 8:0:0:0: Attached scsi generic sg1 type 0
[   16.231478] sd 8:0:0:0: Power-on or device reset occurred
[   16.231838] sd 8:0:0:0: [sdb] Unsupported sector size 520.
[   16.231856] sd 8:0:0:0: [sdb] 0 512-byte logical blocks: (0 B/0 B)
[   16.231861] sd 8:0:0:0: [sdb] 520-byte physical blocks
[   16.232050] sd 8:0:0:0: [sdb] Write Protect is off
[   16.232055] sd 8:0:0:0: [sdb] Mode Sense: d3 00 10 08
[   16.232347] sd 8:0:0:0: [sdb] Write cache: disabled, read cache: enabled, supports DPO and FUA
[   16.233563] sd 8:0:0:0: [sdb] Attached SCSI disk


[mr44er]

FC can do multipath like SAS and your drives show different firmware, where it maybe has enabled only one of the two ports by modepage policy. If that is the one you have no contact, it won't work.
I could imagine connecting one initiator to the same both ports of the same target would do damage and it tells this by detecting the loop.

I never had such stuff in my hands, but I can also imagine this soft address is similar to SAS: https://storage.microsemi.com/en-us/sup ... dress.html

If connection is right the next problem will be reading the 520b sector size.