Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: FW: [802.3ae_Serial] 12 dB return loss for transmitter?



This is all good information, however returning to the original question.  The 12dB return loss for the transmitter is as Ron said a specification on the transmitter, not the cable plant.  It is measured by injecting an optical signal into the Tx and measuring how much light is reflected back.  It was included to keep Interferometric noise to a low level.  This noise is generated when there is a triple path (ie reflection from the Rx which then gets reflected back by the Tx with a single mode source.).  It would be nice to remove the specification, but a significant amount of work was done in the serial Ad Hoc to show that the specification is needed unless the return loss on the Rx is tightened.  

"Christensen, Benny" wrote:

  
-----Original Message-----
From: Christensen, Benny
Sent: 12. februar 2002 10:25
To: 'Rao Tatavarti'
Subject: RE: [802.3ae_Serial] 12 dB return loss for transmitter?
 
Hi RaoYou are quite right about this. But that specification is related to the optical fiber link installed, not the PMD of the 10GE port.Brillouin backscattering is mainly a problem with high launch power (in excess of 10 dBm, ie. when one uses EDFA with up to 30 dBm lauch power). It depends very much on the optical spectrum power density (which can be broadened by chirp of the laser/modulator or low frequency laser current modulation - spreading the spectrum. This technique is coverd by a patent (BT as far as I remember). Rayleigh backscattering is always present an has some effective length of around 20 km and a level of around -60 dB (as recalled from my mind). This is normally not a problem and far less than discrete reflections in the fiber (connectors and splices)Benny

--------------------------------------------
  llllllll   ii     llllllll     llllll
ll                ll           ll      ll
ll    llll   ll   ll    llll   llllllllll
ll      ll   ll   ll      ll   ll      ll
  llllllll   ll     llllllll   ll      ll

GIGA, an Intel company
Benny Christensen, M.Sc.E.E, Ph.D.
Mileparken 22, DK-2740 Skovlunde, Denmark
Tel: +45 7010 1062, Fax: +45 7010 1063
e-mail: benny.christensen@xxxxxxxxx, http://www.giga.dk
 

-----Original Message-----
From: Rao Tatavarti [mailto:rao@xxxxxxxxx]
Sent: 12. februar 2002 00:53
To: Christensen, Benny; Ron Miller; Lindsay, Tom; Serial reflector 802.3ae (E-mail)
Subject: RE: [802.3ae_Serial] 12 dB return loss for transmitter?
 
Hi all,Optical Return Loss (ORL) according to ITU  is a optical spec specified at the Transmitter. This denotes the returned power from the entire fibre including both discrete reflections and distributed backscattering such as Brillouin or Rayleigh scattering. ITU has no  specifications for target  link distances of 10km  as in 10GE spec. The closest is the I-64.2  or the SR-1 in GR253 which has a value of 14dB.Another  specification that is  specified along with ORL is a max discrete reflectance specification for the optical path between S and R, which in the ITU case is -27dB max. I am not aware of 10GE equivalent spec for this parameter. Rao Tatavarti. Ph.D
Ceyba Inc.,
450 March Road,
Kanata, Ontario K2K 3K2
Email: rao@xxxxxxxxx
Tel: 613 599-5797 ext 5083
Fax: 613 599-1247
-----Original Message-----
From: Christensen, Benny [mailto:benny.christensen@xxxxxxxxx]
Sent: Friday, February 08, 2002 3:33 AM
To: 'Ron Miller'; 'Lindsay, Tom'; Serial reflector 802.3ae (E-mail)
Subject: RE: [802.3ae_Serial] 12 dB return loss for transmitter?
 
Hi allMy perception is, that this is an optical specification and as such it is the optical return loss (measured from the fiber side).Just like SDH/SONET specifications, 10GE is an interface specification (at optical connections points S (=TP2) and R (=TP3)), and (optical) properties of the TX (jitter, rise/fall, eye mask etc.) and RX (sensitivity, tolerance etc.), ie. things that can be measured from the outside of the modules. All 3 standards does not cover internal components specifications like the optical modulators S11 (electrical return loss) nor the MUX (driver) rise/fall time, output swing or S22.  These things are 'implementation specific' and as such it is unimportant, as long as the functionality and specifications are fulfilled at point S and R. From a module vendors point of view, it will of course be nice to have some standard definitions for the interfaces at a component level, but I think it is out of scope for the 10GE standard. XSBI/OIF/MSA interface (where instantiated) are covered in the same way. Benny

--------------------------------------------
  llllllll   ii     llllllll     llllll
ll                ll           ll      ll
ll    llll   ll   ll    llll   llllllllll
ll      ll   ll   ll      ll   ll      ll
  llllllll   ll     llllllll   ll      ll

GIGA, an Intel company
Benny Christensen, M.Sc.E.E, Ph.D.
Mileparken 22, DK-2740 Skovlunde, Denmark
Tel: +45 7010 1062, Fax: +45 7010 1063
e-mail: benny.christensen@xxxxxxxxx, http://www.giga.dk
 

-----Original Message-----
From: Ron Miller [mailto:rmiller@xxxxxxxxxxx]
Sent: 8. februar 2002 06:44
To: 'Lindsay, Tom'; Ron Miller; Serial reflector 802.3ae (E-mail)
Subject: RE: [802.3ae_Serial] 12 dB return loss for transmitter?
 
Hi TomI believe that this is the electrical connection to the electro/optical module.In the case of the optical transmitter(optical output) the same logic is appropriate.  However,it is typically possible to get 30 or more db return loss.  also in the optical arena the source lasercan often be pulled in frequency by the reflected wave from the load, which may be a more importantsource of degradation than the third pass signal combining with the incident signal at the receiver end.Insertion loss S21 is from input  to output.  Return loss is reflected power and at the sending end is usuallyreferred to as S11, and at the receiving end S22.  All parameters are complex numbers.Good luckRon
-----Original Message-----
From: Lindsay, Tom [mailto:tlindsay@xxxxxxxxxxxxxxxxxxxx]
Sent: Thursday, February 07, 2002 4:04 PM
To: Ron Miller; Serial reflector 802.3ae (E-mail)
Subject: RE: [802.3ae_Serial] 12 dB return loss for transmitter?
 
Does everyone agree with Ron?Ron, do you understand that this is an optical spec?Thanks, Tom
-----Original Message-----
From: Ron Miller [mailto:rmiller@xxxxxxxxxxx]
Sent: Tuesday, February 05, 2002 10:40 PM
To: Lindsay, Tom; Serial reflector 802.3ae (E-mail)
Subject: RE: [802.3ae_Serial] 12 dB return loss for transmitter?
 
TomThis is the return loss looking back into the transmitter.  Typically it is measured with thedrive circuit powered.  It is an AC measurement, and may vary according to the signal thatthe transmitter is sending out.The value of this measurement is in calculating or simulating the third pass signal, which is thecombination of the driving signal together with the reflection from the destination, coming back to thesource and reflecting again.  The total reflection from the source termination and the load terminationshould be at least 20 db total for +/-10 percent ripple in the rails of the eye pattern.  So, 12 db ateach end would be 24 db or about +/ -6 percent ripple.  So, you normally want to do better.  20 db isa very safe number at both ends resulting in 40 db or only about +/-1% ripple in the rails.Ron miller
-----Original Message-----
From: Lindsay, Tom [mailto:tlindsay@xxxxxxxxxxxxxxxxxxxx]
Sent: Tuesday, February 05, 2002 9:24 PM
To: Serial reflector 802.3ae (E-mail)
Subject: [802.3ae_Serial] 12 dB return loss for transmitter?
 
Dave Kabal, Dubravko and I were discussing the 12 dB return loss spec in Table 52-12.What is it? Is this return loss looking back into the transmitter port from outside, or is it tolerance to return loss of the cable plant and receiver combination?If the former, what is the technical reasoning? If the latter, it should at least be renamed or more preferably removed.Thanks, Tom
Stratos NW
425/672-8035 x105