SPI-4 Where is the Beef? Receiver Driver specifications

Aloisi, Paul paul_aloisi at ti.com
Sun May 21 10:02:38 PDT 2000


* From the T10 Reflector (t10 at t10.org), posted by:
* "Aloisi, Paul" <paul_aloisi at ti.com>
*
Richard,

The weak driver must remain above the 320 mV. The spread sheet shows the
amount of fallback and moves the strong driver higher in the chart. The
spread sheet shaded area is below the 320 mV drive with the weak driver.

00-227r3 addes 2 lines that are the signal if no fall back is used with the
maximum losses. When the worst case driver asymmetry occurs the signal is
-20 to -30 mV - you don't reach the zero crossing. The asymmetry is about
30%, this either reduces or adds to the fall back by nominal fall back. The
asymmetry is based on P channel versus N channel in a process. 

The proposals to date did not address asymmetry effects on the signal, only
the perfect driver and imperical data. If you use the standards cable and
backplane losses, there is no zero crossing at the receiver. If asymmetry is
cut in half, then there is at least a zero crossing but only about 25 mV
signal for the receiver.

Thank you,
Paul


> -----Original Message-----
> From:	Richard Moore [SMTP:r_moore at qlc.com]
> Sent:	Saturday, May 20, 2000 1:51 PM
> To:	'Aloisi, Paul '; ''SCSI Reflector' '
> Subject:	RE: SPI-4 Where is the Beef? Receiver Driver specifications
> 
> Paul,
> 
> There is an as-yet unconsidered question of what a 60-78% reduced drive
> strength really means. At least I don't remember anyone addressing this
> in the working group meetings.
> 
> If we pick some level for the strong drive between 320 mV and 800 mV,
> this level includes the roughly 100 mV of bias. Asymmetry is built in
> to the drivers to overcome this bias. Thus a 500 mV driver might be
> 400 mV negation and 600 mV assertion when the bias is removed. This might
> correspond to 12 mA assertion current and 8 mA negation current (I forget
> what resistance value was used in the Annex A test circuit so I picked 50
> ohms for this example).
> 
> If we elect 75% reduced drive strength does this mean we reduce the
> nominal signal voltage from 500 mV to 375 mV (475mV or 9.5 mA assertion
> and 275 mV or 5.5 mA negation), or that we reduce the drive current from
> 12 mA and 8 mA to 9 mA (450 mV) and 6 mA (300 mV)? I suspect the former
> since this maintains the same asymmetry with respect to the 100 mV offset.
> 
>  -- Richard Moore
>     QLogic Corp.
> 
> 
> -----Original Message-----
> From: Aloisi, Paul
> To: 'SCSI Reflector'
> Sent: 5/18/00 3:44 AM
> Subject: SPI-4 Where is the Beef? Receiver Driver specifications
> 
> * From the T10 Reflector (t10 at t10.org), posted by:
> * "Aloisi, Paul" <paul_aloisi at ti.com>
> *
> The presentations on the SPI-4 driver and receiver specifications to
> data
> have not gone through the calculations to present the tolerances allowed
> and
> the cable loss allowed. A lot of the data has been imperial with Ideal
> drivers. We have a lot of TBDs left in the SPI-4 document. We have a lot
> of
> designers well into designs that do no have all the right numbers.
> 
> 00-227r2 uses the tolerances as they have been define to date for the
> driver. The driver asymmetry has not been mentioned in any of the
> presentations the current chart and equations are what we have to base
> numbers on. V=(0.69*Vx)+50 and meet the minimum 320 mV and the maximum
> 800
> mV, where V is the drive in one direction and Vx is the drive in the
> other
> direction.
> 
> The agreement on  drive system from Tuesday is a strong driver that does
> not
> exceed the 800 mV and a weak driver used after the first transition
> reducing
> the level to 60 - 78% of the strong driver that does not go below the
> minimum 320 mV.
> 
> The cable loss has been defined by the cable working group for the
> media,
> backplanes are not well defines. Cables attenuation define with
> backplanes
> which are loaded give a signal that is 60% of the driving signal. 
> 
> The tolerance for the terminators and connectors are an additional 10%
> loss.
> 
> The calculation of the tolerances using the driver asymmetry, Precomp
> drive
> as defined, the cable loss and the other system losses leaves a isolated
> 0
> or 1 at the zero crossing. There is no eye opening. Receivers must
> consider
> a valid signal a transition from the fall back level to 0 mV as a valid
> transition.
> 
> 
> Document 00-239r0 defines the assumptions and the proposed table A1 that
> should be used for design.
> 
> There is no meetings until July 11 for SPI-4 to make any changes, this
> is
> too late for most designs. Designs in progress should use the current
> data
> available based on 00-227r2 and 00-239r0 to insure interoperability for
> SPI-4
> 
> Thank you,
> Paul
> 
> 
> 
> Paul D. Aloisi                                  Phone: 603-429-8687
> Texas Instruments   MS: 84            FAX: 603-429-8963 
> 7 Continental Blvd                          Mobile: 603-930-4104
> Merrimack, NH 03054                   Email: paul_aloisi at ti.com
>                                                       Web site:
> www.ti.com
> 
> *
> * 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 mailing list