SPI-2 driver test conditions

RMoore at corp.adaptec.com RMoore at corp.adaptec.com
Tue Apr 30 13:34:00 PDT 1996

* From the SCSI Reflector, posted by:
* RMoore at corp.adaptec.com
Form: Reply
Header: Adaptec
Text: (46 lines follow)

Thanks for your response, and I'm glad to see you did the analysis for 
attenuation. But, are you sure of your new minimums? If I translate it back 
to currents, using Vb = 114 mV and Rt = 54.5 ohms from the test circuit, I 

	Ia >= 6.2 mA + .47 In
	In >= .47 Ia

These intersect at a minimum point around (In = 3.75 mA, Ia = 8.0 mA). If I 
use your spreadsheet for this case, and set the terminator at maximum 
voltage and resistance, with maximum cable loading (85 ohms) and 16 nodes 
leaking 20 uA each, I get a first transition from negation to assertion that 
goes to only 132 mV (measured at the load). I'm not sure whether this is 
enough; previously I was going for 150 mV after the first step. The 
difference was, attenuation previously wasn't modeled; but it was budgeted 
for (130 mV was the voltage level we were trying to guarantee at the 
receiver, so the attenuation budget was 20 mV). What number did you use for 
attenuation factor? I think we need to compare any other assumptions to make 
sure we're on the same track. Here are mine:

DC voltage levels (perhaps these numbers could be revisited, for starters):
	Va >= 130 mV
	Vn <= -130 mV
AC voltage levels (first transition, add 20 mV attenuation budget):
	V+ >= 150 mV
	V- <= -150 mV
Cable impedance:
	85 ohms <= ZL <= Z0
	110 ohms <= Z0 <= 135 ohms
Termination (these are new numbers from the Milpitas SPI-2 WG meeting):
	100 mV <= Vb <= 125 mV
	100 ohms <= Rt <= 110 ohms
Leakage and number of nodes:
	-20 uA <= IL <= 20 uA
	2 <= N <= 16
And finally, for the 4:1 maximum signal ratio at the receiver:
	Va <= 4 * |V-|
	|Vn| <= 4 * V+

Note that after I limited In to 11 mA the 4:1 was already guaranteed by the 
other constraints, hence the third pair of test conditions could be dropped.

	Richard Moore
	Adaptec Irvine Technology Center
Original text: (88 lines follow)
>From owner-scsi at Symbios.COM, on 4/30/96 4:47 PM:
To: scsi at Symbios.COM (scsi)

* From the SCSI Reflector, posted by:
* Kevin Gingerich <S=Kevin_Gingerich%S=Gingerich%G=Kevin%TI at mcimail.com>

  To  SCSI Reflector   X400
>From  Kevin Gingerich  GING
/pn=scsi/mbx1=scsi at symbios.com/ems=internet/admd=mci/c=us/
Subj  SPI-2 driver test conditions
">" my comments. /Kevin/
* From the SCSI Reflector, posted by:
* RMoore at corp.adaptec.com
Form: Memo
Text: (32 lines follow)
Following up my proposal at the Milpitas SPI-2 meeting:
I recalculated the test conditions under the new assumptions for terminator
characteristics, i.e., Rt ranges from 100 to 110 ohms and Vb from 100 to 125
I also added another limit as Bill Ham suggested to enforce asymmetric
driver strengths, by limiting In to 11 mA.
The new test conditions I am proposing for Section 10.1.1, Table 10, are as
follows. Replace the first condition:
        416 mV <= Va <= 706 mV
        249 mV + 0.3 * |Vn| <= Va <= 706 mV
> I calculate that this to be 170 mV + 0.47 * |Vn| <= Va <= 650 mV
>My derivation is rather lengthy so, I'll just give the result here as the
>constraints on the assertion current as a function of the negation current
>(In), cable impedance (Zo), the reflection coefficient at the load (pL),
>attenuation factor (A), termination impedance (Zt), and the bias voltage
>(Vos). The basis is to achieve at least 25% of the steady-state voltage on 
>first transition to the opposite state at any point on the bus.
>From steady-state negated to assertion;
> (In(1.25 Zt/2 - (1+pL)AZo/2) + 1.25Vos)/((1+pL)AZo/2) <= Ia
>From steady-state asserted to negated;
> Ia <= (In(1+pL)AZo/2 + 1.25Vos)/(1.25 Zt/2 - (1+pL)AZo/2)
>This can be converted to the form you give by inserting the test circuit
>values, rearranging terms, and multiplying through by 55 ohms to get 
>(If you try it, watch your sign convention.)
Replace the second condition:
        381 mV <= |Vn| <= 977 mV
        255 mV + 0.3 * Va <= |Vn| <= 714 mV
> and this should be 170 mV + 0.47 * Va < |Vn| <= 700 mV.
>I got the new maximums from trial and error and the spreadsheet model. The
>constraint was to keep the differential voltages less than +/-1 V including
>reflected waves.
Remove the third condition:
        -277 mV < |Va| - |Vn| < 181 mV
It turns out that this condition is not necessary because restricting the
maximum In to 11 mA (or |Vn| <= 714 mV) keeps the voltage ratio at the
receiver within the 4:1 we are looking for.
I'd like to point out that my derivation assumes current mode drivers. Are
these test conditions valid for voltage mode? Unfortunately I haven't had a
chance to study that question.
>Either voltage- or current-mode drivers must deliver the first-step 
>required. The difference occurs in the polarity and magnitude of subsequent
>reflected waves. Within the defined bus parameters, the reflections should 
>cause problems.

Use Proportional Font: true
Previous From: owner-scsi at Symbios.COM
Previous To: scsi at Symbios.COM (scsi)
Original to: scsi at Symbios.COM (scsi)
Attachment Count: 0

X-NVL-Content-Type: UNKNOWN
X-NVL-Content-Typename: UNKNOWN
X-NVL-Content-Charset: X-IBM-437
X-NVL-Content-Filename: ATTRIBS.BND
X-NVL-Content-Transfer-Encoding: X-UUENCODE

begin 777 ATTRIBS.BND
M````````3W)I9VEN86P@=&5X=*D+1 at H`````````````_`(#`*D+]@("``(`
M````5&EM97, at 3F5W(%)O;6%N````````````````````````````./\`````
M#@4!``\%*@4!`"L%<@4!`',%B at 4!`(L%I at 4!`*<%PP4!`,0%V at 4!`-L%Z`4!
M`.D%)P8!`"@&-P8!`#@&1P8!`$@&2`8!`$D&X08!`.(&X at 8!`.,&\08!`/(&


More information about the T10 mailing list