SAS PHY working group presentation
Elliott, Robert (Server Storage)
Elliott at hp.com
Fri Apr 11 16:57:39 PDT 2008
* From the T10 Reflector (t10 at t10.org), posted by:
* "Elliott, Robert (Server Storage)" <Elliott at hp.com>
This document discusses a letter ballot comment commenting on how SAS-2
cables can be 10 m, while SAS-1.1 transceivers only support 6 m, meaning a
SAS-1.1 transceiver exposed to a SAS-2 cable could have problems)
1. "A SAS connector designed for 6G SAS, needs
to also meet 3G SAS, but the standards are
NOT the same!
- Example impedance has a tolerance at SAS1.1 but
no tolerance at 6G, which do you use? Its possible to
meet the S-parameters at SAS 2.0 with a different
impedance than what's specified in SAS1.1"
There is a letter ballot comment to add tolerances on those specs in SAS-2,
so that might be corrected during letter ballot resolution.
2. "- Can we add a statement saying specifically that
"Connectors designed to work at 6G automatically are
compliant at 3G/1.5G."
SAS-2 is intended to:
- define SAS-2 transmitter and receiver requirements
- define interconnect requirements to support SAS-1.1 phys at 1.5 and 3 Gbps.
These requirements are carried over from SAS-1.1. All SAS-2 interconnects
are not required meet these requirements (e.g. a 10 m cable probably does
- define interconnect requirements to support SAS-2 phys at 1.5, 3, and 6
- not define SAS-1.1 transmitter and receiver requirements (see SAS-1.1 if
that is still of interest)
The last paragraph of 18.104.22.168.3 mentions some rules like that; perhaps more
text of this sort is needed in 5.2.6 or 22.214.171.124.1?
3. "- So how do you know you have a SAS 2.0 TXRX connection? We
would need an EEPROM in the cable perhaps?"
The Mini SAS connector was first proposed for SAS-1.1 in a SAS Physical WG
face-to-face meeting in March 2005. At that same meeting I proposed that we
add some sideband signals to report cable length. For this situation, the
transmitter could detect a long cable and drop to 1.5 Gbps or refuse to
operate altogether rather than run at 3 Gbps unreliably. However, the
proponents of the connector (including the company now submitting the letter
ballot comment) insisted designs were already done and no such change could
be accommodated. The only option left was to borrow a ground pin, which
risks signal integrity.
An option we could still consider at this time is keying. For SAS-2.1 active
cables, proposal 08-052 keys the active cable and connector such that:
- active receptacle, active cable: OK
- active receptacle, passive cable: OK
- passive receptacle, active cable: blocked
- passive receptacle, passive cable: OK
(r2 just uses no key slot on the cable plug and no key on the receptacle;
since previous cables always had key 4, this suffices.)
We could define a separate keys for > 6 m passive cables, yielding:
- SAS-2 receptacle, long passive cable: OK
- SAS-2 receptacle, short passive cable: OK
- SAS-1.1 receptacle, long passive cable: blocked
- SAS-1.1 receptacle, short passive cable: OK
To coexist with the active cable keying, a key slot on the receptacle and a
key on the cable plug would be needed for each function (long passive cable
and active cable).
The 8-wide connectors being discussed reopen the opportunity to include
sidebands. I hope it's not too late this time :-)
4. "- What happens when you have a bad 10M cable that doesn't
reliably connect at 6G SAS 2.0? Bad lane? Flaky wire or
ground? What do we back track to?"
If the phys get to SNW-3, they get to negotiate 6, 3, and/or 1.5 Gbps with
training (using Train-SNW). This is only supported by SAS-2 phys. If they
pick or fall back to 3 or 1.5 Gbps with training, they can and should utilize
their DFE technology.
Rob Elliott, elliott at hp.com
Hewlett-Packard Industry Standard Server Storage Advanced Technology
> -----Original Message-----
> From: owner-t10 at t10.org [mailto:owner-t10 at t10.org] On Behalf
> Of Felton_Mickey at emc.com
> Sent: Wednesday, April 09, 2008 2:23 PM
> To: t10 at t10.org
> Subject: SAS PHY working group presentation
> * From the T10 Reflector (t10 at t10.org), posted by:
> * Felton_Mickey at emc.com
> I have uploaded a document/presentation, 08-188r0 titled: Electrical
> questions on SAS 2.0 to SAS 1.1 support?
> If you have some time I'd like to get some feedback on some of the
> questions and concerns raised in the document.
> * For T10 Reflector information, send a message with
> * 'info t10' (no quotes) in the message body to majordomo at t10.org
* For T10 Reflector information, send a message with
* 'info t10' (no quotes) in the message body to majordomo at t10.org
More information about the T10