[SI-LIST] Re: SATA eye diagram - how much is too much
- From: "Charles Hill" <chuck@xxxxxxxxxxx>
- To: "T.K. Jeon" <tkjeon@xxxxxxxx>, <si-list@xxxxxxxxxxxxx>
- Date: Tue, 5 Oct 2004 18:24:35 -0600
TK,
Again, the eye diagram is not a compliance test in SATA II. No SATA II
specifications or compliance tests are defined in terms of the "eye". The
spec is released and the mode method is not a part of it. The mode method
was discussed for setting levels for receive tolerance. But the drawback of
it is it is not a statistical measure of a signal which definitely has noise
in the case of transmit level. I was suggesting having the mode method for
setting the receive level may be a good change to the spec since it
encourages test setups with lower noise--but that is only my opinion.
My testing with jitter, both random and deterministic does not add any
amplitude modulation nor amplitude noise. Summing a noise source and a
signal with a resistive splitter has a number of drawbacks.
The SATA II spec removes a lot of ambiguity in amplitude because it
specifically describes the impedance matching conditions. All receive level
tests like OOB threshold are done with a matched source, and with voltage
driving a matched load. All transmit level tests are done driving a matched
load. Of course the actual levels seen in a system are different.
Regards,
Chuck
-----Original Message-----
From: T.K. Jeon [mailto:tkjeon@xxxxxxxx]
Sent: Tuesday, October 05, 2004 4:45 PM
To: chuck@xxxxxxxxxxx; si-list@xxxxxxxxxxxxx
Subject: Re: [SI-LIST] Re: SATA eye diagram - how much is too much
Chuck,
Although the spec doesn't call out the additional amplitude noise, there
appears some noise on the recieve input data because we should add some
jitter intentionally to the data in order to meet the Rx DJ spec, which
is part of the test setup.
I didn't realize that the commitee was thinking about the mode method,
but it sounds very good for the test since we can get rid of some
ambiguity about the amplitude spec like min, max, and OOB threshold,
etc. For example, OOB threshold says 200mVp-p as a spec without
specifying if it is inner eye or outer eye, however, (maybe or maybe
not) the OOB detection test would pass or fail dependant on whether it's
inner eye or outer eye.
Regards,
TK
Charles Hill wrote:
>
> TK,
>
> Just to be clear, the amplitude during the receive tolerance test is set
by
> connection to a HBWS (high BW scope), then the connections are moved over
to
> the receiver under test. When using LBP or COMP patterns you'll find the
> minimum amplitude occurs at certain bits and this is described in section
> 6.4.2.1.1 page 108.
>
> The receive tolerance test doesn't really call out the addition of
amplitude
> noise on the input so the noise on the scope trace will be due to the test
> setup--this noise is an unwanted signal. To make your test setup more
> effective, make the noise floor as low as possible and adjust for 275mV
> minimum amplitude.
>
> I suspect the SATA II spec should specify that the mode method be used for
> measuring amplitude (an averaging on the scope) since this would encourage
> low noise test setups, and good tests because the averaged signal would
> reach minimum right at 275mV and not higher due to test setup noise. The
> difficulty in mode method is whether a pattern trigger is available or
not.
> I've found that my HP86100A triggers right on the LBP data (off a
> splitter)--this was controversial in committee.
>
> Regards,
> Chuck
>
> -----Original Message-----
> From: T.K. Jeon [mailto:tkjeon@xxxxxxxx]
> Sent: Tuesday, October 05, 2004 12:38 PM
> To: chuck@xxxxxxxxxxx
> Subject: Re: [SI-LIST] Re: SATA eye diagram - how much is too much
>
> Charles,
>
> Speaking of SATA II spec, let me beg your opinion for interpreting some
> parameter.
> Tx amplitude needs to be measured statistically for both min and max as
> you said. However, I wonder if Rx amplitude also has to follow the Tx
> way.
> In other words, for Rx jitter tolerance test, the data coming into Rx
> diff'l pair should meet the Rx diff'l input amp spec, which is 275mV and
> 750mV for min and max respectively at Gen2i for example. Then, should I
> set the input amp to 275mVp-p inner eye or to be between min and max by
> statistical measurement like Tx? I think that the former is right way
> for Rx jitter tolerance test because not only will the min amplitude be
> the worst case in terms of Rx sensitivity but also the Rx test needs to
> use different patterns than that of Tx measurement.
> What do you think?
>
> Thanks,
> TK
>
> Charles Hill wrote:
> >
> > Charles,
> >
> > I'll try to answer each of your questions.
> >
> > First to set the context of compliance testing. In SATA II (Serial ATA
II
> > Electrical Specification revision 1.0 26 May 2004), eye diagrams are not
a
> > compliance measurement. From the specification page 101 section 6.4
> > Measurements: "Additional measurement methods for several parameters are
> > aimed at providing quick No-Go testing. These use any valid data
pattern
> > and a Laboratory Load to quickly produce a visual ?picture? of the
> > performance of the unit under test. For example, one measurement method
> > uses Data Eyes to quickly understand jitter. Another uses a mode
> measurement
> > for identifying a potentially complex signal with a single amplitude
> value.
> > However, none of these No-Go measurement methods may be used for testing
> > compliance to electrical specifications. These measurement methods are
> > valuable for gaining useful information about the performance of the
unit
> > under test which goes beyond specification compliance issues.
> >
> > Both methods produce measurements of electrical performance, however,
the
> > parametric method shall be used for validation of the unit under test to
> the
> > Serial ATA requirements 6.2 while other methods may be used as general
> No-Go
> > tests."
> >
> > Question: "How many violations of the eye does it take before a design
is
> > brought?" SATA II defines both jitter and amplitude statistically. The
> > measurement of maximum amplitude is described in section 6.4.2.2 page
114.
> > Essentially, an equivalent time sampling oscilloscope is used and a
> > histogram is setup in the amplitude direction. The number of samples
that
> > fall beyond the allowable threshold is counted as well as the total
number
> > of samples taken. The relative frequency is computed and if low enough
> the
> > test is passed (see step 7 page 115). This statistical definition
allows
> > for unbounded amplitude noise to be present in the test setup and still
> have
> > a meaningful test. The test is also structured to encourage low noise
> test
> > setups. Further, the impedance match of the unit under test and test
> setup
> > is also described.
> >
> > The minimum amplitude test is described in section 6.4.2.1 page 106.
Once
> > again, the histogram function in the scope is used, placing the
histogram
> > limits in time at 0.45UI and 0.55UI (this is where minimum amplitude
> > matters) and a test equation defines the requirement based on the mean,
> > standard deviation, and number of the voltage samples. This is based on
> > meeting a minimum amplitude with a confidence level of 95%.
> >
> > The jitter test is more complicated due to the industry problem of
> > reproducible jitter measurements. This is dealt with in section 6.4.7.1
> > page 125. Jitter measurement tools intend to predict BER by
extrapolation
> > from sample sizes which are not large enough to be statistically
> significant
> > at BER=1E-12. The accuracy of jitter measurement tools is left to the
> > instrument manufacturers but a standard for comparison is defined by
SATA
> II
> > as a direct measure of total jitter (TJ) at BER=1E-12: "Methods do exist
> to
> > extrapolate TJ on a BERT from time scan values at higher rates. While
> such
> > methods can be used to predict TJ at a desired BER, only a direct
measure
> > all the way to the desired BER shall be used when using the BERT as a
> jitter
> > standard for comparison to extrapolation methods." This is a
statistical
> > measure of jitter and the required confidence interval is specified
(page
> > 40) and an explanation is given in section 6.4.1.4.2 page 104 as 4
errors
> > are allowed for a BER target of 1E-12 with a sample size of 1E13.
> >
> > I'm not sure whether 1 minute of data is enough since I don't know how
> fast
> > your instrument is.
> >
> > "Can eye violations be graded in terms of severity based on the location
> > of the violation (that is : at the outer bounds, or the internal eye
> > trapezoid )?" NO. That is not in SATA II.
> >
> > Regards,
> > Chuck Hill
> >
> > -----Original Message-----
> > From: si-list-bounce@xxxxxxxxxxxxx
> > [mailto:si-list-bounce@xxxxxxxxxxxxx]On Behalf Of Grasso, Charles
> > Sent: Tuesday, October 05, 2004 8:28 AM
> > To: 'si-list@xxxxxxxxxxxxx'
> > Subject: [SI-LIST] SATA eye diagram - how much is too much
> >
> > Greetings!
> >
> > I have a question regarding the interpretation of the eye diagram data
vis
> a
> > vis
> > SATA.
> >
> > How many violations of the eye does it take before a design is brought
to
> a
> > screeching halt? A follow up question:
> >
> > Can eye violations be graded in terms of severity based on the location
> > of the violation (that is : at the outer bounds, or the internal eye
> > trapezoid )?
> >
> > The basis of this question (naturally) is that I have SATA data that
have
> > exactly
> > one 1 violation at the outer bound of the eye at 1 minute of data
capture.
> > Typical go/no go specification will determine that to be a fail. I am
not
> so
> > sure.
> > Any insight would be greatly appreciated.
> >
> > Thanks
> >
> > Best Regards
> > Charles Grasso
> > Senior Compliance Engineer
> > Echostar Communications Corp.
> > Tel: 303-706-5467
> > Fax: 303-799-6222
> > Cell: 303-204-2974
> > Email: charles.grasso@xxxxxxxxxxxx; <mailto:charles.grasso@xxxxxxxxxxxx;
>
> > Email Alternate: chasgrasso@xxxxxxxx <mailto:chasgrasso@xxxxxxxx>
> >
> > ------------------------------------------------------------------
> > 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:
