[SI-LIST] [Fwd: Re: Re: Rocket I/O]
- From: Vinayak AGRAWAL <vinayak.agrawal@xxxxxx>
- To: si-list@xxxxxxxxxxxxx
- Date: Thu, 09 Sep 2004 14:33:55 +0530
Hi
Most of the SerDes solutions available in the market that make 8B/10B
available also make a bypass feature available in case you don't want to
use the SerDes's 8B/10B.
In that case long transitionless runs are possible and they can
certainly cause the CDR, whether PLL/DLL or even digital (algorithmic
selection based), to loose lock.
Another thing is that CDR clock rate needs to be only an integer
multiple of half the baud rate (and integer fraction of baud rate is
also possible if multiple phases of the clock are available). I don't
know how RocketIO implements the CDR but let's say that in the scenario
Greg stated the CDR's internal clock-frequency was 620MHz (equal to the
max Baud rate). I don't see any problem with data recovery from a data
sequence that behaves as a 4MHz clock as long as the max run-length spec
is adhered to (which it is) and the time period between any two
transitions in the data stream is an integer multiple of 1/620 microseconds.
Greg's doubt probably stems from a confusion in run-length in data
transmission and run-length in training sequence. Most of the CDRs that
can tolerate run-lengths this long will require that initially a
training sequence be sent to them. And the training sequence will
certainly not have this sort of a large run-length.
As for frequency mismatch, if it is a PLL based CDR, and if the PLL is
of the second order or higher, there will be no frequency mistmatch
between the recovered clock and data-rate once lock has been achieved.
First-order systems will have trouble achieving lock in case the
frequency offset between data-rate and initial clock-frequency (may be
of a reference clock) is too high. If lock is achieved there'll be a
static phase offset depending on the magnitude of frequency offset and
that'll degrade the jitter-tolerance and max run length. That is why
8B10B will be able to tolerate more frequency offset.
I hope it was helpful
Vinayak
Mike Brown wrote:
>I'm not a Rocket I/O expert; it's been years since I looked at it. But,
>something does not add up.
>
>If the system uses 8b-10b encoding (per Weston), there should be no
>(electrical) reason for limiting the unencoded run length. The encoding will
>limit the run length to 5 or less regardless of the transistions in the input
>stream.
>
>It sounds like the quoted run length spec (Greg) is for a device which does
>not incorporate RLL encoding.
>
>If the system transmits unencoded data (or data encoded with a long run-length
>code) then the limitation on the run length is established by the data
>recovery PLL(DLL?) which must lock onto the bit stream transitions, and the
>permissible frequency error between the transmit and receive references.
>
>I'd expect that the 75 bit run length carries perhaps a 25 PPM max frequency
>error, whereas an 8b-10b encoded stream probably permits 100 PPM or greater
>error. These tolerance numbers are from my experience with a vendor's IP
>SerDes spec.
>
>Mike
>
>Beal, Weston wrote:
>
>
>
>> Rocket I/O uses 8b10b encoding so your stream of 75 ones and 75 zeros will
>> be encoded to runs of 5 and 3 ones and zeros. Now you're back closer to 620
>> Mbps.
>>
>>
>>-----Original Message-----
>>From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On
>>Behalf Of Inchauspe, Greg E
>>Sent: Wednesday, September 08, 2004 2:45 PM
>>To: Sig Integ Group
>>Subject: [SI-LIST] Rocket I/O
>>
>> The data sheet for Rocket I/O specifies a minimum data rate of
>> 620 Mbps.
>>
>>
>It also specifies a maximum of 75 transition less data bits. If I transmit a
>pattern with
>75 ones then 75 zeros repeating NRZ at 620 Mbps, this is equivalent to a 4 MHz
>square wave.
>Does this mean the Rocket I/O can do clock and data recovery on a 4 MHz square
>wave signal?
>
>
>> thanks,
>> Greg
>>
>>------------------------------------------------------------------
>>To unsubscribe from si-list:
>>si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
>>
>>or to administer your membership from a web page, go to:
>>//www.freelists.org/webpage/si-list
>>
>>For help:
>>si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
>>
>>List FAQ wiki page is located at:
>> http://si-list.org/wiki/wiki.pl?Si-List_FAQ
>>
>>List technical documents are available at:
>> http://www.si-list.org
>>
>>List archives are viewable at:
>> //www.freelists.org/archives/si-list
>>or at our remote archives:
>> http://groups.yahoo.com/group/si-list/messages
>>Old (prior to June 6, 2001) list archives are viewable at:
>> http://www.qsl.net/wb6tpu
>>
>>
>>
>>
>>
>
>
>------------------------------------------------------------------
>To unsubscribe from si-list:
>si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
>
>or to administer your membership from a web page, go to:
>//www.freelists.org/webpage/si-list
>
>For help:
>si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
>
>List FAQ wiki page is located at:
> http://si-list.org/wiki/wiki.pl?Si-List_FAQ
>
>List technical documents are available at:
> http://www.si-list.org
>
>List archives are viewable at:
> //www.freelists.org/archives/si-list
>or at our remote archives:
> http://groups.yahoo.com/group/si-list/messages
>Old (prior to June 6, 2001) list archives are viewable at:
> http://www.qsl.net/wb6tpu
>
>
>
>
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List FAQ wiki page is located at:
http://si-list.org/wiki/wiki.pl?Si-List_FAQ
List technical documents are available at:
http://www.si-list.org
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
- » [SI-LIST] [Fwd: Re: Re: Rocket I/O]
