IEEE P802.3ae D3.3 Comment Resolutions

Ieee P802.3ae Task Force

Description

P802.3ae Draft 3.3 CommentsCl SC P L Cl SC P L00 # 105 00 # 5Dawe, Piers Agilent Gaither, Justin XilinxComment Type E Comment Status A Comment Type TR Comment Status DBERT is bit error RATIO (most times) - see definitions 1.5 or manufacturer's literature. Loopback support (register 3.0.14) has been (I believe mistakenly) removed in D3.3 for 10GBASE-X PCS MMDs; only the 10GBASE-R MMD is now referred to in this clause. The SuggestedRemedy explicit dependance on 10GBASE-R should be removed and a reference to the relevant 6 times in clause 52, change rate to ratio. Also clauses 30, 47, 49, 50, 53 or their subclause in clause 48 should be added. It will then align with the related register appendices. definitions in the MMD registers for DTE XGXS (5.0.14) and PHY XGXS (4.0.14)Response Response Status C SuggestedRemedyACCEPT IN PRINCIPLE. BERT remains BERT. We assume commenter is referring to Alter the first paragraph of 45.2.3.1.2 to read:BER, not BERT. The PCS shall be placed in a loopback mode of operation when bit 3.0.14 is set to a one. When bit 3.0.14 is set to a one, the PCS shall accept data on the transmit path and return Cl 00 SC P L # 124 it on the receive path. The specific behavior of the 10GBASE-R PCS during loopback is Dawe, Piers Agilent specified in Clause 49.2, and the specific behaviour of the 10GBASE-X PCS duringloopback is specified in Clause 48.3.3. For all other port types when operating at Comment Type E Comment Status A 10Gb/s, the ...

Informations

• Publié par: Fewyar
• Nombre de lectures: 28
• Langue: English

Extrait

P802.3ae Draft 3.3 Comments Cl 00 SC P L # 10 Cl 00 SC P L # 5 Dawe, Piers Agilent Gaither, Justin Xilinx Comment Type E Comment Status A Comment Type TR Comment Status D BERT is bit error RATIO (most times) - see definitions 1.5 or manufacturer's literature. Loopback support (register 3.0.14) has been (I believe mistakenly) removed in D3.3 for SuggestedRemedy  10G theonlyDs;  sonDMi R-M ABESino  tedrrferew hT .esualc siht CP XMM SBG01-ESA t  oneecerelht e vant  e xelpcitid penedance on 10GBASEs R-luoheb dmer edovnd a r aeref 6 times in clause 52, change rate to ratio. Also clauses 30, 47, 49, 50, 53 or their  subclause in clause 48 should be added. It will then align with the related register appendices. definitions  in the MMD registers for DTE XGXS (5.0.14) and PHY XGXS (4.0.14) Response Response Status C SuggestedRemedy ACCEPT IN PRINCIPLE. BERT remains BERT. We assume commenter is referring to  Alter the first paragraph of 45.2.3.1.2 to read: BER, not BERT.  The PCS shall be placed in a loopback mode of operation when bit 3.0.14 is set to a one. e on the d return Cl 00  LSC P # 12 t i tonp ev.htar ehiece pt datap ta hnartnamsti te,ona o  tet scca llahs SCP ehWh 4si0.1.ti3 neb   R-ESABGirud SCPpboo lng isk acs epT eh cebicifor ohavie 10f th Dawe, Piers Agilent  specified in Clause 49.2, and the specific behaviour of the 10GBASE-X PCS rati at Comment Type E Comment Status A kcf nutc Solpoab, the PC10Gb/s gnnehwepo tyt s peerthor pllo roa .3F .8.3se 4Clau in fiediceps si kcabpooglinurd all t shbe  sotiret sibt ihabicplap wnd aleytilanoi ton si  Precise words for MDIO PICS varies  ignored and reads from this bit shall return a value of zero. SuggestedRemedy Response Response Status Z Chief editor please recommend  PROPOSED REJECT. Loopback is performed in 10GBASE-X PMA, not in the PCS. Response Response Status C Cl 00 SC 52 P L # 94 ACCEPT IN PRINCIPLE. All MMD-capable clauses will list MDIO capabilities in the PICS  aw t under "Major capabilities/options". The table entry will be specified as:  D e, Piers Agilen Item = MD Comment Type E Comment Status R Feature = MDIO  Ludicrous tions. Some of these wo Subclause = <applicable to each clause> don't need     h t y o. phenards could be hyphenated (differently) but you  Value/Comment = Registers and interface supported Status = O SuggestedRemedy Support = Yes [ ], No [ ]  Remove the following from hyphenation dictionary: implementa-tion sig-nals sta-tus specifica-tions modula-tion tech-niques jit-ter hap-pen pat-tern inter-val extinc-tion See 47.6.3 as a reference for the required format.  proce-dure fre-quency envi-ronmental opti-cal connec-tions sin-gle Clause 48 and 52 need to add this entry. Response Response Status C REJECT. Withdrawn by commenter.

TYPE: TR/technical required T/technical E/editorial COMMENT STATUS: D/dispatched A/accepted R/rejected SORT ORDER: Clause, Page, Line, Subclause RESPONSE STATUS: O/open W/written C/closed U/unsatisfied Z/withdrawn

Page 1 of 30 Cl 00 SC 52

P802.3ae Draft 3.3 Comments Cl 45 SC 45.4.1 P 238 L 32 #  2 Cl 45 SC Table 45-28 P 190 L # 13 Turner, Ed Lattice Semiconductor Dawe, Piers Agilent Comment Type T Comment Status A Comment Type E Comment Status A Since positive currents are defined to flow into the circuit, IOH currents are normally Table 45-11 is spread over two pages n M e A g X at I i M ve U .  M T   h v e a   l I u O e H  o   f s  - p 4 ec m if A ic .  a L t i i k o e n  w in is  t e a ,  b t l h e e     4 I 5 O - L 6   2 s   p o e n  c l i i f n ic e a   t 4 i 6 o   n a   t o  p n  a li g n e e     2 4 3 8 8  s  s h h o o ul u d ld     h h a a v v e e  a     a  SuggestedRemedy minimum value of +4mA. (At 0.2V for consistency with the VOL specification range.) The  Make the right hand column much wider and with a minimum of formatting work it should fit EC7 and EC8 values in table 45.5.5.14 should be adjusted accordingly. The specifications  on the page on VI, VIH and VIL in table 45-62 on lines 36 - 40 at page 238 are  i a n ll i u n the format normally  Response Response Status C f d o e u vi n c d e  i .  n I   n d  t e a v b ic le e     4 s 5 p - e 6 c 2 if   i h c o at w io e n v s e , r    t w h h e e   r o e b   j r e e c q t  u i i s r   e to m  h e a n v ts e     a s r p e e   c p i u fi t c  o at n i   o t n h s e     on p  M t  D s I ig O n  r a e ls c   e t i o v  t e h rs e ,    not  ACCEPT. the input signals to them. That can be changed in several ways. The smallest departure Cl 45 SC Table 45-28 P 205 L 38 #  13 from the present specifications would probably be to remove the VI spec and reword the definitions of VIH and VIL:  Dawe, Piers Agilent VIH Range of input voltages that 0.84V 1.5V Comment Type E  StatusCo t A  must be interpreted as a logic 1. mmen VIL Range of input voltages that -0.3V 0.36V  Table 45-28 formatting             must be interpreted as a logic 0. SuggestedRemedy SuggestedRemedy Make the right hand column much wider   Change IOH to -4mA MAX, change IOL to +4mA MIN (@0.2V), update EC7, EC8.  StatusResponse Response C Remove VI line. Change VIH to 0.84V (MIN), 1.5V (MAX)  ACCEPT. Change VIL to -0.3V (MIN), 0.36V (MAX) Response Response Status C Cl 45 SC Table 45-49 P 225 L  13 11 # ACCEPT. Dawe, Piers Agilent Comment Type E Comment Status A Cl 45 SC Table 45-11 P 205 L 38 #  132 Table 45-49 formatting Dawe, Piers Agilent SuggestedR ed em y Comment Type E Comment Status A Make the table full width, make the "Name" and "Description" columns wider   Table 45-28 formatting  StatusResponse Response C SuggestedRemedy ACCEPT. Make the right hand column much wider Response Response Status C ACCEPT.

TYPE: TR/technical required T/technical E/editorial COMMENT STATUS: D/dispatched A/accepted R/rejected SORT ORDER: Clause, Page, Line, Subclause RESPONSE STATUS: O/open W/written C/closed U/unsatisfied Z/withdrawn

Page 2 of 30 Cl 45 SC Table 45-4

P802.3ae Draft 3.3 Comments Cl 46 SC 46.1.6 P 264 L 47 #  90 Cl 46 SC 46.3.4.3 P 277 L 49 #  93 Robert Grow Intel Foulds, Chris Intel Comment Type E Comment Status A Comment Type T Comment Status A Loopback is a popular feature, and there has been some confusion if loopback at the Condition (b) is not correct. If a fault_sequence is not received within a period of 128 XGMII is possible. columns, the link_fault variable will actually be cleared. SuggestedRemedy SuggestedRemedy Add a note to 46.1.6"Note -- XGMII signals are specified such that transmit signals may be 1. Leave condition (a) the same. connected to receive signals to create a loopback path. To do this, TXD<0> is connected 2. Change condition (b) to the following: to RXD<0> ... TXD<31> to RXD<31>, TXC<0> to RXC<0> ... TXC<3> to RXC<3>, and b) Without any intervening fault_sequences of a different value, and TXCLK to RXCLK. Such a loopback does not test the Link Fault Signaling state machine, 3. Add condition (c) which says the following: nor any of the error handling functions of the receive RS." c) Without any intervening period of 128 columns not containing a fault_sequence Response Response Status C  StatusResponse Response C ACCEPT. ACCEPT IN PRINCIPLE. Cl 46 SC 46.3.4.2 P 276 L 30 #  37 Replace lettered list with: Tim Warland Quake Technologies  Four fault_sequences containing the same fault value have been rec ived with each pair of e Comment Type E Comment Status A fault sequences separated by less than 128 columns and no intervening fault_sequences In the subclause title, the word Variables is mis-spelled  of a different fault value. SuggestedRemedy [Make grammar match with preceding paragraph.] Change Varibles to Variables Cl 46 SC 46.4 P 279 L 37 #  6 Response Response Status C Gaither, Justin Xilinx ACCEPT. Comment Type TR Comment Status A Cl 46 SC 46.3.4.3 P 277 L 39 The setup and hold time requirements do not properly constraint duty cycle variations of Robert Grow Intel  the XGMII clocks Comment Type E Comment Status A SuggestedRemedy Missin exit term from state NEW FAULT TYPE  Specifiy a 40/60 % duty cyle requirement for the TX_CLK and RX_CLK in Figure 46-12. g _ _ d Response Response Status C Sugg R e e s s t t e or d e R e e x m it e co y  IN PRINCIPLE. ACCEPT  FAULT TYPEndition of UCT from state NEW _ _  Add lines to Figure 46-12 to both high and low clock periods from active threshold of Response Response Status C certainty region for tpwmin ACCEPT. Add  to imbeded table a row consisting of tpwmin with a value of 2.5ns at the driver

# 91

TYPE: TR/technical required T/technical E/editorial COMMENT STATUS: D/dispatched A/accepted R/rejected SORT ORDER: Clause, Page, Line, Subclause Page 3 of 30 RESPONSE STATUS: O/open W/written C/closed U/unsatisfied Z/withdrawn Cl 46 SC 46.4

P802.3ae Draft 3.3 Comments Cl 46 SC Figure 46.9 P 277 L  92 842 # Cl 47 SC 47.3.3.4 P 336 L  9900 34 # Foulds, Chris Intel Baumer, Howard Broadcom Corp. Comment Type T Comment Status R Comment Type TR Comment Status A XAUI (D3.1) NC - Done The Link Fault Signaling State Machine is not complete. The main problem is that there is Differential return loss specified as as a flat responce of 10dB from 100MHz to 2.5GHz is no transition for the condition where a fault sequence is not received (!fault_sequence) and unrealistic and cannot be met with pratical and reasonable designs and packages. Also the _ common mode return loss pe cations exculdes pure differential designs, that is a pu col cnt is <= 127. s cifi re SuggestedRemedy tui ssb  nolterus a  midffre100hoerminatiential tevah0 a w no llimon  rde cdBmoom senilhe tins l nai . Add if-else conditions to COUNT and FAULT states for incrementing col_cnt or seq cnt. Sugg pr e e s f t e e r d ab R le de d  ssign since it keeps all current g _ Add transition conditions for !fault_seq*col_cnt<128.See included file showing proposed eme y state diagram. Specify the driver output differential return loss with a nonflat response and remove the Response Response Status C urn losstial retidffrenetli  n al alsure ncda sh e  tupepmirevituo  "Dread:to rion irtpedcseN wtn .mereuieq rsslon ruter edom nommoc 5M.281 7 nd aHzter  bet 10dthanmo1  Brf zot00HM REJECT. Column count is defined as incrementing continously unless reset, so no reduce 20dB per decade from 781.25MHz to 2.5GHz". The last sentence in this paragraph change is required to increment the counter.  will then need to read: "The reference impedance for differential return loss measurements is 100ohms. Table 47-1 in subclause 47.3.3 on page 334 will need to be updated with " these redefined return loss specifications. Response Response Status C ACCEPT IN PRINCIPLE. Resolution reached during the formal Clause 47 comment review session was to recognize the problem and accept the suggested remedy as an interim solution. A phone call to the commenter discussing the suggested remedy indicated that a preferrable remedy was to instead specify a driver output differential return loss with a flat response of 5.5 dB but that this was not optimized either. Additional subsequent discussion between multiple parties indicated a preferrable solution as follows. The preferrable solution is also illustrated in a graph in the Clause 47 status report: Specify the driver output differential return loss with a nonflat response and remove the common mode return loss requirement. New description to read: "Driver output impedance shall result in a differential return loss better than 10dB from 100MHz to 781.25MHz, reduce 20dB per decade from 781.25MHz to 3.5GHz and 20dB per decade from 3.5GHz on". The last sentence in this paragraph reads: "The reference impedance for differential return loss measurements is 100ohms." Table 47-1 in subclause 47.3.3 is updated to reflect the new return loss specifications. PICS item E4 in 47.6.4.3 is updated to reflect the new return loss specifications. Cl 47 SC 47.3.3.4 P 336 L  3 34 # Gaither, Justin Xilinx Comment Type TR Comment Status A Done I support comment #99007 SuggestedRemedy Follow the remedy in #99007 Response Response Status C ACCEPT.

TYPE: TR/technical required T/technical E/editorial COMMENT STATUS: D/dispatched A/accepted R/rejected SORT ORDER: Clause, Page, Line, Subclause Page 4 of 30 RESPONSE STATUS: O/open W/written C/closed U/unsatisfied Z/withdrawn Cl 47 SC 47.3.3.4

P802.3ae Draft 3.3 Comments Cl 47 SC 47.3.3.6 P 339 L  9900 3839 # Cl 47 SC 47.4.3 P 295 L 24 #  48 Baumer, Howard Broadcom Corp. Tim Warland Quake Technologies Comment Type TR Comment Status R  Comment TypeXAUI (D3.1) NC - Done T Comment Status R Done The current transmit jitter specification allows for the near end random jitter to be has high The CJPAT pattern is described as a transmit function but it is unclear that it is necessary as 8ps rms and the far end random jitter to be has high as 12.6ps rms. (Since the to include the CJPAT generation function" specification allows Dj=0 and Rj=Tj-Dj(actual) Rj can then equal Tj. For near end Rj=0.35UI=112ps pk-pk which is 8ps rms {112/14}. For the far end Rj=0.55UI=176ps pk-pk Sugg C e h s a t n edRemedy  orC AT which is 12.6ps rms.) This puts an undue burdon on the Receiver to be able to handle this  ge the text to either indicate the mandatory requirement to generate JP large pure random jitter.  A maximum random jitter should be specified. explicitly indi n c t a s t   e is  t  t h h a e t    C C J J P P A A T T  is optio   n a a s l.     d T e e fi xt could read:"The data pattern for jitter  measureme SuggestedRemedy mentmplen thed iIUt  eAXim tarsnontincfuTh "or" t IUAX e timsnarfunction may  p taetnrnnex 48Aned in A TAPeneghT .JC el al  ibe torashr  Add a maximum random jitter specification that is not based on the determinstic jitter and optionally be implemented in the XAUI transmit function. " a se d n d t  t e h n e c   e c  ( o li n n s e tr s a   i 3 n 8 t  - t 3 h 9 a ) t    t m h o e d   i s fi u e m d    t o o f     r th e e a   d R :  j "   T &  h D e j     m ha a s x   i t m o  u b m e    l p e e s a s  k t  t h o a   n p   e th a e k    r T a j. n S d e o c m o   n jit d t   e t r o ,     l d a e s f t i   ned  Response Response Status Z REJECT. Clarification is included in Annex 48A wording, first paragraph: "The patterns as 14 * rms random jitter, shall be less than 0.22UI. The sum of the measured may be implemented at a bit, code-group or frame level… ..". In addition, CJPAT is defined deterministic and measured peak to peak random jitter shall be less than the total jitter".Table 47-1 in subclause 47.3.3 on page 334 will need to be updated with the  as a "data pattern in 47.4.3 and not as a (XAUI) transmit function. maximum random jitter. Cl 47 SC 47.4.3 P 295 L  8 25 # Response Response Status U Lindsay, Tom Stratos Lightwave REJECT. The working group desires further investigation of an appropriate RJ limit. The editor asks that the commentor determine an RJ limit acceptable to the working group and Comment Type E Comment Status A Done then resubmitted this comment.  Clarify what is being measured. As of November 15, 2001, the commenter has provided no new information during the last SuggestedRemedy 5 months justifying a need for a change, and the committee is satisfied with the current Modify last sentence of paragraph to read "Jitter measurement for the transmitter (or for specifications. calibration of a jitter tolerance setup) shall be performed... Cl 47 SC 47 3.4.5 P 342 L 2937 #  9900 Response Response Status C . Baumer, Howard Broadcom Corp.  ACCEPT. Comment Type TR Comment Status R  Cl XAUI (D3.1) NC - Done 47 SC 47.4.3 P 295 L  49 27 # There is no specific random jitter specified for the receiver jitter tolerance. This results in  Tim Warland Quake Technologies the same problem illustrated in my comment #99008. Comment Type T Comment Status R Done SuggestedRemedy There is currently no requirement for a XAUI receiver to accept the XAUI CJPAT and to a A n d d d     t t h h e e     s fo e l n l t o e w n i c n e g    s s e p n e t c a if n yi c n e t     t T o j   :  s " u T b h c e l   a m us a e x   i 4 m 7 u .3 m .   4 p .5 e   a b k e  t t o w   e p e e n a   k t   h ra e  n s d e o n m te  ji nce o d n  f s i p n e e c d i   f a yi s n  1 g  4 D  * j     report errors. The detection of errors would facilitate BER calculations but should not  tter, e  necessarily be a compliance requirement it Res r p m on s s r e  ter,andom j shall be l R e e s s s p t o h n a s n e 0   . S 2 t 2 at U u I. s " U SuggestedRemedy REJECT. See response to #99008.  Two options: Firstly the text could indicate the option not to detect and report errors on the CJPAT. Second option: The text could indicate that the receiver shall monitor and report 10b/8b coding violations in order to accumulate BER statistics data. In either case, the function of the receiver in CJPAT mode must be defined. Response Response Status Z REJECT. Agree with the comment. However, BER monitoring is not a compliance requirement for XAUI.

TYPE: TR/technical required T/technical E/editorial COMMENT STATUS: D/dispatched A/accepted R/rejected SORT ORDER: Clause, Page, Line, Subclause Page 5 of 30 RESPONSE STATUS: O/open W/written C/closed U/unsatisfied Z/withdrawn Cl 47 SC 47.4.3