Questions about SAS coefficients' configuration

Yupengfei (B) yu.yupengfei at hisilicon.com
Tue Mar 24 04:03:53 PDT 2015


Formatted message: <a href="http://www.t10.org/cgi-bin/ac.pl?t=r&f=r1503240_f.htm">HTML-formatted message</a>
Attachment #1: <a href="http://www.t10.org/cgi-bin/ac.pl?t=r&f=r1503240_image001.jpg">image001.jpg</a>

Hi Harvey,
Here are our understandings about the figure “Minimum and maximum
coefficient ranges at maximum peak to peak voltage”:
(1)   |C1|+C2+|C3|<=1, thus C2<=1-|C1|-|C3|, and “C2 at maximum”
represents C2=1-|C1|-|C3| in the figure, which means Vp-p is always
constrained by C2=1-|C1|-|C3|.
This is how I understand your words “Keep in mind that this figure is
defined with C2 at maximum.”. Right?
(2)   The specifications of “Table 43 – Transmitter device signal output
characteristics for trained 12Gb/s at ET, IT, and CT” imply 0>=C1>=-0.2 and
0>=C3>=-0.35. Besides, C2-|C1|-|C3|>=X, (X=VMAmin/2K0, refer Table 48).
The purple line is drawn according to the above 3 constraints, and dash or
solid just represents the legal possible real differential peak to peak
output voltage “at maximum peak to peak voltage”.
The red line represents the points with possible minimum non transition bit
differential output voltage (i.e. VMA) “at maximum peak to peak voltage”,
especially Point x located at the left bottom corner.
So, we can see,
(a)   The Point x generates the minimum VMA within the red line, so VMAx>80mV
should be satisfied.
(b)   If 850mV<=Vpp_noeq<=1200mV is meet, VMAx>80mV is satisfied too. Because
Vpp_noeq*(C2-|C1|-|C3|)>=850mV*(0.55-0.15-0.3)=85mV>80mV.
Finally, 2 conclusions can be made from the figure in my opinion.
(A)   A Transmitter should guarantee the coefficient ranges located on red
dash line, so that Rpre=1.66 and Rpost=3.33 in Table 43 will be obtained.
(B)   A Transmitter should guarantee 850mV<=Vpp_noeq<=1200mV, so that
VMA>80mV in Table 43 will be obtained when (A) is satisfied.
Of course, both (A) and (B) are “at maximum peak to peak voltage”.
(3)   We agree with your answer “Since my answer to 2a was no we need to
address question 2b that starts ‘If yes,’. We wanted to provide the
receiver the ability to adjust C2 and lower the amplitude. We have placed the
responsibility of maintaining the VMA >= 80 mV on the transmitter. We believe
this provides the required constraints.”, but I think “maintaining
VMA>=80mV” is implemented by constraining C2-|C1|-|C3| which is equivalent
to VMA/2K0 just like 1 is equivalent to Vpp_noeq.
(4)   The specification 850mV<=Vpp_noeq<=1200mV is defined only when “C2 at
maximum”. So, when C2 is reduced, the differential peak to peak output
voltage does not need to satisfy 850mV<=Vpp_noeq<=1200mV, right?
(5)   I think a SAS3.0 compliant Transmitter shall meet the following
constraints according to (1)-(4), right?
a)	  |C1|+C2+|C3|<=1
b)	 |C1|<=0.2
c)	  |C3|<=0.35
d)	 C2-|C1|-|C3|>=X, X=VMAmin/2K0.
(6)   “Table 48 – 12Gb/s reference transmitter device characteristics at
ET” gives a different coefficient range C1>=-0.15 and C3>=-0.3, which does
not match with Table 43. Why?
We are expecting your comments about our understandings above.
Thank you!
Pengfei Yu
//-------------------------
Hi Pengfei,
The figure you are asking about was the subject of a lot of discussion during
the specification development. It has been the subject of additional
discussion following the release of this document as well. Let me try and
provide some insights that hopefully provides you and everyone else a
consistent interpretation.
I start with the title of the figure "Minimum and maximum coefficient ranges
at maximum peak to peak voltage". When writing the specification we realized
the range the coefficients are capable of adjusting to on the transmitter
impacted the amount of equalization available to the over all budget. In
order to assure compatibility we needed to have some way to require all SAS-3
compliant transmitters to support a minimum specified range of the
coefficients. Due to the interdependence between coefficient range and
amplitude we attempted to specify the range of the coefficients when the
amplitude (C2) was at maximum.
Maximum amplitude is limited by the 1200 mV differential peak to peak number
covered elsewhere in the specification. This limit is written into the
specification to protect receivers both now and in the future from being over
driven. Because it is a shall not exceed number all transmitters are required
to stay below this value when the coefficient C2 responds with Maximum to an
Increment command. The allowed voltage range for a transmitter with C2 at
maximum is 1200 mV to 850 mV. See SPL3r07 section 5.14 for details on the phy
layer transmitter training coefficient adjustments.
With C2 at maximum "...at maximum peak to peak voltage" the figure places
limits on the C1 and C3 "Minimum and maximum coefficient ranges". There was
concern that as amplitudes reduce we did not want the non-transition bit
(i.e., VMA) amplitude to fall below 80 mV differential peak to peak. The
dashed purple line is intended to communicate that the coefficient range
might be limited by the VMA requirement for lower amplitude signals. Keep in
mind that this figure is defined with C2 at maximum. Therefore we do not need
to add a constraint “C2-|C1|-|C3|>=Fixed Value” because C2 is not a
variable for the purpose of these limit lines.
Now to your questions.
Question 1:
Whether the value of C1 and C3 is the total value surrounding by the purple
solid line when Vpp_noeq = 1200mv?
Answer 1a:
The purple solid line represents the maximum range of C1 and C3 when C2 is at
maximum.
Question 1b:
If so, does it mean the value of C1 and C3 in purple solid line satisfied
equation “VMA>=80mv” when Vpp_noeq=1200mv?
Answer 1b:
If the amplitude of the signal is very close to the 1200 mV amplitude then
the VMA limit should be reached along the lower left diagonal limit line. If
the amplitude of the signal is lower, then the VMA limit will be the limiting
factor causing the maximum range to move in the direction indicated by the
dashed purple line.
Question 2:
Do we need a constraint like “C2-|C1|-|C3|>=Fixed Value” to ensure
“VMA>=80mv” when Vpp_noeq is variable between 850mv and 1200mv?
Answer 2a:
Since the figure applies when C2 is maximum we do not need to add the
constraint because for the purpose of this figure C2 is not a variable. As
soon as C2 is adjusted this figure does not apply. The VMA >= 80 mV always
applies and provides the primary maximum bound for C1 and C3 once C2 has been
reduced.
Question 2b:
If yes, when we constraint “C3>=-0.3” and “C1>=-0.15”(the range
surrounding by the red dotted line shown in Figure 131, which is the minimum
range we should support), we do not need the constraint
“C2-|C1|-|C3|>=Fixed Value” or any other constraint of C2 to ensure the
minimum value of VMA, right?
Answer 2b:
Since my answer to 2a was no we need to address question 2b that starts "If
yes,". We wanted to provide the receiver the ability to adjust C2 and lower
the amplitude. We have placed the responsibility of maintaining the VMA >= 80
mV on the transmitter. We believe this provides the required constraints.
Please let me know if there are additional questions.
Regards,
Harvey Newman
Avago Technologies
On Mon, Mar 16, 2015 at 7:50 PM, Yupengfei (B)
<yu.yupengfei at hisilicon.com> wrote:
Hi All,
I have questions needing your help.
In SAS-3R06’s Figure 131 -- “Minimum and maximum coefficient ranges at
maximum peak to peak voltage”,
it defines the minimum range and maximum range of C1 and C3 with different
Vpp_noeq.
Question 1:
Whether the value of C1 and C3 is the total value surrounding by the purple
solid line when Vpp_noeq = 1200mv?
If so, does it mean the value of C1 and C3 in purple solid line satisfied
equation “VMA>=80mv” when Vpp_noeq=1200mv?
Question 2:
Do we need a constraint like “C2-|C1|-|C3|>=Fixed Value” to ensure
“VMA>=80mv” when Vpp_noeq is variable between 850mv and 1200mv?
If yes, when we constraint “C3>=-0.3” and “C1>=-0.15”(the range
surrounding by the red dotted line shown in Figure 131, which is the minimum
range we should support), we do not need the constraint
“C2-|C1|-|C3|>=Fixed Value” or any other constraint of C2 to ensure the
minimum value of VMA, right?
Best Regards,
Pengfei Yu
________________________________
华为技术有限公司 Huawei Technologies Co., Ltd.
[Company_logo]
Phone:
Fax:
Mobile: 18665362797<tel:18665362797>
Email: yu.yupengfei at hisilicon.com
地址:深圳市龙岗区坂田华为基地 邮编:518129
Huawei Technologies Co., Ltd.
Bantian, Longgang District,Shenzhen 518129, P.R.China
http://www.huawei.com
________________________________
本邮件及其附件含有华为公司的保密信息,仅限于发送给ä¸
Šé¢åœ°å€ä¸­åˆ—出的个人或群组。禁
止任何其他人以任何形式使用(包括但不限于全部或部分åœ
°æ³„露、复制、或散发)本邮件中
的信息。如果您错收了本邮件,请您立即电话或邮件通知å
‘件人并删除本邮件!
This e-mail and its attachments contain confidential information from HUAWEI,
which
is intended only for the person or entity whose address is listed above. Any
use of the
information contained herein in any way (including, but not limited to, total
or partial
disclosure, reproduction, or dissemination) by persons other than the
intended
recipient(s) is prohibited. If you receive this e-mail in error, please
notify the sender by
phone or email immediately and delete it!
--
Harvey Newman
Avago Technologies
(408) 433-4028



More information about the T10 mailing list