[SI-LIST] Power Integrity effects on Signal Integrity slides from my IEEE ED/CAS chapter presentation

Discussion:

"Cosmin Iorga" (Redacted sender "ci42775" for DMARC)

2018-11-01 15:44:35 UTC

For who is interested I have posted the slides from my IEEE Buenaventura ED/CAS chapter presentation "Power Integrity Effects on Signal Integrity in FPGA Systems" on the downloads page at http://www.piscanner.com

Regards,
Cosmin

------------------------------------------------------------------
To unsubscribe from si-list:
si-list-***@freelists.org with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-***@freelists.org with 'help' in the Subject field

List forum is accessible at:
http://tech.groups.yahoo.com/group/si-list

List archives are viewable at:
http://www.freelists.org/archives/si-list

Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu

Filip, Cristian

2018-11-01 16:18:48 UTC

Permalink

Great presentation Cosmin!
Thank you,

Cristian

-----Original Message-----
From: si-list-***@freelists.org [mailto:si-list-***@freelists.org] On Behalf Of Cosmin Iorga
Sent: Thursday, November 1, 2018 9:45 AM
To: si-***@freelists.org
Subject: [SI-LIST] Power Integrity effects on Signal Integrity slides from my IEEE ED/CAS chapter presentation

For who is interested I have posted the slides from my IEEE Buenaventura ED/CAS chapter presentation "Power Integrity Effects on Signal Integrity in FPGA Systems" on the downloads page at http://www.piscanner.com

Regards,
Cosmin

------------------------------------------------------------------
To unsubscribe from si-list:
si-list-***@freelists.org with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-***@freelists.org with 'help' in the Subject field

List forum is accessible at:
http://tech.groups.yahoo.com/group/si-list

List archives are viewable at:
http://www.freelists.org/archives/si-list

Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu

------------------------------------------------------------------
To unsubscribe from si-list:
si-list-***@freelists.org with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-***@freelists.org with 'help' in the Subject field

List forum is accessible at:
http://tech.groups.yahoo.com/group/si-list

List archives are viewable at:
http://www.freelists.org/archives/si-list

Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu

Hermann Ruckerbauer

2018-11-01 17:19:33 UTC

Permalink

Hello experts,

since a long time i struggled with a question and always was afraid to
ask ...

I see always Rise/fall time as relevant parameter in specs or for
Bandwidth calculation.
But I would expect, that the real relevant parameter would be dV/dt, so
the slewrate.

e. g. for a signal with 100ps Risetime and 1000mV Swing (just talk about
0-100% to simplify discussion) I get a much steeper slope with higher
dV/dt vs. a Signal with 100ps risetime and 100mV swing. Therefore I
would assume the signal with 1V swing to have higher frequency
components vs. the Signal with 100mV swing .. but the risetime is the
same...

Can anybody enlighten me, where my mistake is in looking to this topic ?

Thanks a lot ..

Hermann
--
Our next events (contact us for details):
Seminar: Open the Black Box of Memory
October 24/25 in Munich

Seminar: Mastering Signal and Power Integrity Design Challenges
November 7th in Winnersh (UK)
November 27th in Munich (Germany)

Webinar: Analysis and Verification of DDR3/DDR4 Interfaces
November: 13th (SignalIntegrityJournal)

EKH - EyeKnowHow
Signal Quality - Made in Bavaria
Hermann Ruckerbauer
www.EyeKnowHow.de
***@EyeKnowHow.de
Itzlinger Strasse 21a
94469 Deggendorf
Tel.: +49 (0)991 / 29 69 29 05
Mobile: +49 (0)176 / 787 787 77
Fax: +49 (0)3212 / 121 9008

------------------------------------------------------------------
To unsubscribe from si-list:
si-list-***@freelists.org with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-***@freelists.org with 'help' in the Subject field

List forum is accessible at:
http://tech.groups.yahoo.com/group/si-list

List archives are viewable at:
http://www.freelists.org/archives/si-list

Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu

Tom Dagostino

2018-11-01 17:51:32 UTC

Permalink

Herman

Remember back to your early signal analysis days. The spectrum of an ideal square wave has frequency contents of F, 3F, 5F, 7F, etc. And the amplitude of each harmonic decreases by 1/harmonic number if I remember correctly. This frequency constant remains constant for any square wave but the amplitude varies with the amplitude of the square wave. So a square wave with 1V amplitude will have 10x the amplitude of the 1st, 3rd, 5th, etc. harmonic than the same risetime signal of 100mV.

An ideal impulse response has a flat frequency content DC to infinity but an impulse with a defined width has an upper -3dB point. The step with a non-zero risetime has an upper -3dB frequency of Tr/0.35 for a Gaussian risetime, etc. These -3dB points are related to the risetime not the amplitude so a 350 psec step of any amplitude will have the same -3dB point but the amplitudes will vary with step/impulse amplitude.

The slew rate tells you how much current you will need to charge a given capacitance, I = C*dV/dT.

Tom Dagostino
971-279-5325
***@teraspeedlabs.com

Teraspeed Labs
9999 SW Wilshire Street
Suite 102
Portland, OR 97225

-----Original Message-----
From: si-list-***@freelists.org [mailto:si-list-***@freelists.org] On Behalf Of Hermann Ruckerbauer
Sent: Thursday, November 01, 2018 10:20 AM
To: si-***@freelists.org
Subject: [SI-LIST] tRise vs. slew Rate

Hello experts,

since a long time i struggled with a question and always was afraid to
ask ...

I see always Rise/fall time as relevant parameter in specs or for
Bandwidth calculation.
But I would expect, that the real relevant parameter would be dV/dt, so
the slewrate.

e. g. for a signal with 100ps Risetime and 1000mV Swing (just talk about
0-100% to simplify discussion) I get a much steeper slope with higher
dV/dt vs. a Signal with 100ps risetime and 100mV swing. Therefore I
would assume the signal with 1V swing to have higher frequency
components vs. the Signal with 100mV swing .. but the risetime is the
same...

Can anybody enlighten me, where my mistake is in looking to this topic ?

Thanks a lot ..

Hermann
--
Our next events (contact us for details):
Seminar: Open the Black Box of Memory
October 24/25 in Munich

Seminar: Mastering Signal and Power Integrity Design Challenges
November 7th in Winnersh (UK)
November 27th in Munich (Germany)

Webinar: Analysis and Verification of DDR3/DDR4 Interfaces
November: 13th (SignalIntegrityJournal)

EKH - EyeKnowHow
Signal Quality - Made in Bavaria
Hermann Ruckerbauer
www.EyeKnowHow.de
***@EyeKnowHow.de
Itzlinger Strasse 21a
94469 Deggendorf
Tel.: +49 (0)991 / 29 69 29 05
Mobile: +49 (0)176 / 787 787 77
Fax: +49 (0)3212 / 121 9008

------------------------------------------------------------------
To unsubscribe from si-list:
si-list-***@freelists.org with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-***@freelists.org with 'help' in the Subject field

List forum is accessible at:
http://tech.groups.yahoo.com/group/si-list

List archives are viewable at:
http://www.freelists.org/archives/si-list

Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu

------------------------------------------------------------------
To unsubscribe from si-list:
si-list-***@freelists.org with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-***@freelists.org with 'help' in the Subject field

List forum is accessible at:
http://tech.groups.yahoo.com/group/si-list

List archives are viewable at:
http://www.freelists.org/archives/si-list

Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu

Boris Bakshan

2018-11-02 14:30:03 UTC

Permalink

Hi Hermann,
This is a question I was struggling with as well untill it occured to me -
By definition, the t_rise parameter includes a delta_V, it would be either
20-80% or 10%-90%. If you take a signal with a certain amplitude and a the
t_rise associated with it and then multiply it by "A" (A>1) then you will
aslo be chaning the 10/90% voltage levels.
Slew rate is basically the same thing designating a delta_V /delta_T.
Would you agree?

On Thu, Nov 1, 2018, 19:20 Hermann Ruckerbauer <

Post by Hermann Ruckerbauer
Hello experts,
since a long time i struggled with a question and always was afraid to
ask ...
I see always Rise/fall time as relevant parameter in specs or for
Bandwidth calculation.
But I would expect, that the real relevant parameter would be dV/dt, so
the slewrate.
e. g. for a signal with 100ps Risetime and 1000mV Swing (just talk about
0-100% to simplify discussion) I get a much steeper slope with higher
dV/dt vs. a Signal with 100ps risetime and 100mV swing. Therefore I
would assume the signal with 1V swing to have higher frequency
components vs. the Signal with 100mV swing .. but the risetime is the
same...
Can anybody enlighten me, where my mistake is in looking to this topic ?
Thanks a lot ..
Hermann
--
Seminar: Open the Black Box of Memory
October 24/25 in Munich
Seminar: Mastering Signal and Power Integrity Design Challenges
November 7th in Winnersh (UK)
November 27th in Munich (Germany)
Webinar: Analysis and Verification of DDR3/DDR4 Interfaces
November: 13th (SignalIntegrityJournal)
EKH - EyeKnowHow
Signal Quality - Made in Bavaria
Hermann Ruckerbauer
www.EyeKnowHow.de
Itzlinger Strasse 21a
94469 Deggendorf
Tel.: +49 (0)991 / 29 69 29 05
Mobile: +49 (0)176 / 787 787 77
Fax: +49 (0)3212 / 121 9008
------------------------------------------------------------------
http://www.freelists.org/webpage/si-list
http://tech.groups.yahoo.com/group/si-list
http://www.freelists.org/archives/si-list
http://www.qsl.net/wb6tpu

------------------------------------------------------------------
To unsubscribe from si-list:
si-list-***@freelists.org with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-***@freelists.org with 'help' in the Subject field

List forum is accessible at:
http://tech.groups.yahoo.com/group/si-list

List archives are viewable at:
http://www.freelists.org/archives/si-list

Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu

Mark Marlett

2018-11-07 01:26:45 UTC

Permalink

Hermann,
A good derivation and history of the rise time to BW is given in this well
written article reference below.
https://interferencetechnology.com/rise-time-vs-bandwidth-and-applications/
I hope this reference helps.

Best Regards,
Mark Marlett
(Sr. Principal System Engineer - Inphi Corporation)

On Thu, Nov 1, 2018 at 10:21 AM Hermann Ruckerbauer <

David Banas

2018-11-07 14:22:00 UTC

Permalink

Gosh, Mark, I have to take exception to your description of this article as “well written”.
It seems to be full of mistakes, syntactical, conceptual, and structural.
I’ve included some examples below, moving through the article from beginning to end.

Regards,
-db

In Par. 1:
- “maximum” -> “minimum”
- Use of the verbiage, “-3dB bandwidth,” indicates an incomplete conceptual understanding of the relationship between system bandwidth and rise-time. A clearer understanding of this relationship would’ve been indicated by the use of, “bandwidth,” alone.

In Par. 2:
- Sentence 1 implies an assumption that the signal starts from zero, as opposed to some Vmin.
Certainly forgivable, but, again, indicative of an incomplete conceptual understanding.
- The language in the rest of this paragraph, along with the poorly labeled figures, is too inexact for an engineering article. For example:
- Sentence 2 should be stricken.
- The final clause of sentence 3 is both unnecessary and incorrect. (By making the unfortunate choice of a high-pass filter as his example, the author has inadvertently depicted a case of infinite bandwidth.)
- Fig. 1 shows a step response, not an impulse response, as claimed in the accompanying verbiage.
- Several unfortunate word choices convey an incomplete understanding. Examples:
- “of” -> “to” in figure captions.
- “response of a signal” -> “response of a system”.

In Eq. 1:
- The equation is incomplete, as the frequency units have been given, but not the time units.

In “How does the relationship arise?”:
- The author is being extremely misleading in Par. 1, by insinuating that Fourier and/or Gaussian analysis produce “errors”. It would be more correct to note that these different methods of relating rise time to bandwidth produce different outcomes. If anything, Gaussian analysis should be used as the normative reference, since it carries the full force and authority of the Central Limit Theorem behind it.

In Eq. 10:
- This equation is inconsistent with itself. The “1” in the fraction should be “0.1”.
- In the immediately following sentence, the parenthetical text is misleading and incorrect.

In the paragraph following Eq. 12:
- “power…absorbed by the reactive element…” - No! Power is never absorbed by a purely reactive element. The power is absorbed by the resistor.

The two equations (and accompanying verbiage) following Eq. 16 never should’ve made it past the technical editor. When this line of reasoning culminates with, “Note that the imaginary…because phase has no relevance…,” there can be no doubt left that the author has only a marginal understanding of the physics he’s describing.

Following Eq. 18 the author begins, “Alternatively, the above analysis could easily be achieved…” Right! So, why not strip out all of the ridiculously convoluted stuff you wrote above and just leave us with this nice short concise explanation?!

One final nit: to write such an article without even making mention of the beautifully symmetric nature of the Gaussian Fourier transform pair, and the resulting ease that symmetry lends to establishing the relationship between system bandwidth and rise-time, is suspect.
To fail to note that the Gaussian approach carries behind it the full weight and authority of the Central Limit Theorem is criminal.

Post by Boris Bakshan
Hermann,
A good derivation and history of the rise time to BW is given in this well
written article reference below.
https://interferencetechnology.com/rise-time-vs-bandwidth-and-applications/
I hope this reference helps.
Best Regards,
Mark Marlett
(Sr. Principal System Engineer - Inphi Corporation)
On Thu, Nov 1, 2018 at 10:21 AM Hermann Ruckerbauer <

------------------------------------------------------------------
http://www.freelists.org/webpage/si-list
http://tech.groups.yahoo.com/group/si-list
http://www.freelists.org/archives/si-list
http://www.qsl.net/wb6tpu

Alfred P. Neves

2018-11-07 14:50:23 UTC

Permalink

Great post David!, I agree that the Central Limit Theorem plays a big part of analyzing complex digital systems but I also see systems, especially backplanes (we are designing a lot of test backplanes and compliance analysis now), that are dominated by via-connector-via discontinuities so there is a lot of resonance in the system appearing at quarter and half wave freq intervals based on connector and via spacings. So, instead of and dominant pole dominating the system response (that practical application goes back to 1Meg probes and 30pf input cap) there is a combination of resonances, two of them being very significant. This emails is a bit off the original thread, sorry!

The windowing from transforming bandwidth limited frequency domain also plagues time domain fidelity, we like to analyze that using Single Bit Response for NRZ. The added benefit is I can analyze the impact with EQ in ADS including DFE, CTLE; itâs pulse response analysis on steroids!

We have a structure on our CMP-32 Channel modeling platform that has precision stubs that relate fast and slow TDR resolutions (30psec and 10psec rise approximately) to frequency domain data (see Via paper we did with Tek and SiSoft on our website, I can send out a copy if your interested also). We used that in a paper with Tektronix for 70GHz via design some time back and corresponded time domain to frequency before analyzing via behavior.

The big problem I am working on now (test fixtures with good SI to 70GHz) is time domain TDR excites higher order modes versus lower frequency VNA data. So even if you PERFECTLY window (that isnât possible, but lets pretend) and transform the data you are still not going to get good correlation between the really fast TDR and 40-50GHz VNA data. The march to 70GHz stellar SI is going to be really hard!

An example of a practical problem is ELF 1.85mm edge launch design to 70GHz.

- Al

Products for the Signal Integrity Practitioner

Alfred P. Neves
Chief Technology Officer

Office: 503-679-2429
www.wildrivertech.com
2015 Best In Design&Test Finalist

Gosh, Mark, I have to take exception to your description of this article as âwell writtenâ.
It seems to be full of mistakes, syntactical, conceptual, and structural.
Iâve included some examples below, moving through the article from beginning to end.
Regards,
-db
- âmaximumâ -> âminimumâ
- Use of the verbiage, â-3dB bandwidth,â indicates an incomplete conceptual understanding of the relationship between system bandwidth and rise-time. A clearer understanding of this relationship wouldâve been indicated by the use of, âbandwidth,â alone.
- Sentence 1 implies an assumption that the signal starts from zero, as opposed to some Vmin.
Certainly forgivable, but, again, indicative of an incomplete conceptual understanding.
- Sentence 2 should be stricken.
- The final clause of sentence 3 is both unnecessary and incorrect. (By making the unfortunate choice of a high-pass filter as his example, the author has inadvertently depicted a case of infinite bandwidth.)
- Fig. 1 shows a step response, not an impulse response, as claimed in the accompanying verbiage.
- âofâ -> âtoâ in figure captions.
- âresponse of a signalâ -> âresponse of a systemâ.
- The equation is incomplete, as the frequency units have been given, but not the time units.
- The author is being extremely misleading in Par. 1, by insinuating that Fourier and/or Gaussian analysis produce âerrorsâ. It would be more correct to note that these different methods of relating rise time to bandwidth produce different outcomes. If anything, Gaussian analysis should be used as the normative reference, since it carries the full force and authority of the Central Limit Theorem behind it.
- This equation is inconsistent with itself. The â1â in the fraction should be â0.1â.
- In the immediately following sentence, the parenthetical text is misleading and incorrect.
- âpowerâŠabsorbed by the reactive elementâŠâ - No! Power is never absorbed by a purely reactive element. The power is absorbed by the resistor.
The two equations (and accompanying verbiage) following Eq. 16 never shouldâve made it past the technical editor. When this line of reasoning culminates with, âNote that the imaginaryâŠbecause phase has no relevanceâŠ,â there can be no doubt left that the author has only a marginal understanding of the physics heâs describing.
Following Eq. 18 the author begins, âAlternatively, the above analysis could easily be achievedâŠâ Right! So, why not strip out all of the ridiculously convoluted stuff you wrote above and just leave us with this nice short concise explanation?!
One final nit: to write such an article without even making mention of the beautifully symmetric nature of the Gaussian Fourier transform pair, and the resulting ease that symmetry lends to establishing the relationship between system bandwidth and rise-time, is suspect.
To fail to note that the Gaussian approach carries behind it the full weight and authority of the Central Limit Theorem is criminal.

Dennis Han

2018-11-07 15:19:52 UTC

Permalink

I agree with David. Also, the article in Wikipedia [1]is better and it
shows the Gaussian response (~0.34/tr), the 1-pole low-pass RC filter
response (~0.35/tr), and others for using 10-90% rise/fall time. Most ICs
are specified for 20-80% rise/fall time so a conversion to 10-90% is needed;
conversion factors for various responses are in Howard Johnson and Martin
Graham's first book, table B.1. They also make a convincing argument that
Gaussian is the response that should be used most of the time.

Dennis

On 11/7/2018 8:22 AM, David Banas wrote:
Gosh, Mark, I have to take exception to your description of this article as
“well written”. It seems to be full of mistakes, syntactical,
conceptual,and structural. I’ve included some examples below, moving
through the article from beginning to end. Regards, -db In Par. 1: -
“maximum” -> “minimum” - Use of the verbiage, “-3dB bandwidth,”
indicates an incomplete conceptual understanding of the relationship between
system bandwidth and rise-time. A clearer understanding of this relationship
would’ve been indicated by the use of, “bandwidth,” alone. In Par. 2:
-Sentence 1 implies an assumption that the signal starts from zero, as
opposed to some Vmin. Certainly forgivable, but, again, indicative of an
incomplete conceptual understanding. - The language in the rest of this
paragraph, along with the poorly labeled figures, is too inexact for an
engineering article. For example: - Sentence 2 should be stricken. - The
final clause of sentence 3 is both unnecessary and incorrect. (By making the
unfortunate choice of a high-pass filter as his example, the author has
inadvertently depicted a case of infinite bandwidth.) - Fig. 1 shows a step
response, not an impulse response, as claimed in the accompanying verbiage.
-Several unfortunate word choices convey an incomplete understanding.
Examples: - “of” -> “to” in figure captions. - “response of a
signal” -> “response of a system”. In Eq. 1: - The equation is
incomplete, as the frequency units have been given, but not the time units.
In “How does the relationship arise?”: - The author is being extremely
misleading in Par. 1, by insinuating that Fourier and/or Gaussian analysis
produce “errors”. It would be more correct to note that these different
methods of relating rise time to bandwidth produce different outcomes. If
anything, Gaussian analysis should be used as the normative reference, since
it carries the full force and authority of the Central Limit Theorem behind
it. In Eq. 10: - This equation is inconsistent with itself. The “1” in
the fraction should be “0.1”. - In the immediately following sentence,
the parenthetical text is misleading and incorrect. In the paragraph
following Eq. 12: - “power…absorbed by the reactive element…” - No!
Power is never absorbed by a purely reactive element. The power is absorbed
by the resistor. The two equations (and accompanying verbiage) following Eq.
16 never should’ve made it past the technical editor. When this line of
reasoning culminates with, “Note that the imaginary…because phase has no
relevance…,” there can be no doubt left that the author has only a
marginal understanding of the physics he’s describing. Following Eq. 18
theauthor begins, “Alternatively, the above analysis could easily be
achieved…” Right! So, why not strip out all of the ridiculously
convoluted stuff you wrote above and just leave us with this nice short
concise explanation?! One final nit: to write such an article without even
making mention of the beautifully symmetric nature of the Gaussian Fourier
transform pair, and the resulting ease that symmetry lends to establishing
the relationship between system bandwidth and rise-time, is suspect. To fail
to note that the Gaussian approach carries behind it the full weight and
authority of the Central Limit Theorem is criminal. On Nov 6, 2018, at 5:26
PM, Mark Marlett <***@gmail.com>[2] wrote: Hermann, A good
derivation and history of the rise time to BW is given in this well written
article reference below.
https://interferencetechnology.com/rise-time-vs-bandwidth-and
-applications/[3]I hope this reference helps. Best Regards, Mark Marlett
(Sr.Principal System Engineer - Inphi Corporation) On Thu, Nov 1, 2018 at
10:21 AM Hermann Ruckerbauer <***@eyeknowhow.de[4]> wrote:
Hello experts, since a long time i struggled with a question and always was
afraid to ask ... I see always Rise/fall time as relevant parameter in specs
or for Bandwidth calculation. But I would expect, that the real relevant
parameter would be dV/dt, so the slewrate. e. g. for a signal with 100ps
Risetime and 1000mV Swing (just talk about 0-100% to simplify discussion) I
get a much steeper slope with higher dV/dt vs. a Signal with 100ps risetime
and 100mV swing. Therefore I would assume the signal with 1V swing to have
higher frequency components vs. the Signal with 100mV swing .. but the
risetime is the same... Can anybody enlighten me, where my mistake is in
looking to this topic ? Thanks a lot .. Hermann -- Our next events (contact
us for details): Seminar: Open the Black Box of Memory October 24/25 in
Munich Seminar: Mastering Signal and Power Integrity Design Challenges
November 7th in Winnersh (UK) November 27th in Munich (Germany) Webinar:
Analysis and Verification of DDR3/DDR4 Interfaces November: 13th
(SignalIntegrityJournal) EKH - EyeKnowHow Signal Quality - Made in Bavaria
Hermann Ruckerbauer www.EyeKnowHow.de[5]
***@EyeKnowHow.de[6]Itzlinger Strasse 21a 94469 Deggendorf
Tel.: +49 (0)991 / 29 69 29 05 Mobile: +49 (0)176 / 787 787 77 Fax: +49
(0)3212 / 121 9008
------------------------------------------------------------------ To
unsubscribe from si-list: si-list-***@freelists.org[7] with
'unsubscribe'in the Subject field or to administer your membership from a
webpage, go to: http://www.freelists.org/webpage/si-list[8] For help:
si-list-***@freelists.org[9] with 'help' in the Subject field List forum
is accessible at: http://tech.groups.yahoo.com/group/si-list[10] List
archives are viewable at: http://www.freelists.org/archives/si-list[11] Old
(prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu[12]
------------------------------------------------------------------ To
unsubscribe from si-list: si-list-***@freelists.org[13] with
'unsubscribe' in the Subject field or to administer your membership from a
web page, go to: http://www.freelists.org/webpage/si-list[14] For help:
si-list-***@freelists.org[15] with 'help' in the Subject field List
forumis accessible at: http://tech.groups.yahoo.com/group/si-list[16] List
archives are viewable at: http://www.freelists.org/archives/si-list[17] Old
(prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu[18]
------------------------------------------------------------------ To
unsubscribe from si-list: si-list-***@freelists.org[19] with
'unsubscribe' in the Subject field or to administer your membership from a
web page, go to: http://www.freelists.org/webpage/si-list[20] For help:
si-list-***@freelists.org[21] with 'help' in the Subject field List
forumis accessible at: http://tech.groups.yahoo.com/group/si-list[22] List
archives are viewable at: http://www.freelists.org/archives/si-list[23] Old
(prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu[24]

--- Links ---
1 https://en.wikipedia.org/wiki/Rise_time
2 mailto:***@gmail.com
3 https://interferencetechnology.com/rise-time-vs-bandwidth-and-applications/
4 mailto:***@eyeknowhow.de
5 http://www.EyeKnowHow.de
6 mailto:***@EyeKnowHow.de
7 mailto:si-list-***@freelists.org
8 http://www.freelists.org/webpage/si-list
9 mailto:si-list-***@freelists.org
10 http://tech.groups.yahoo.com/group/si-list
11 http://www.freelists.org/archives/si-list
12 http://www.qsl.net/wb6tpu
13 mailto:si-list-***@freelists.org
14 http://www.freelists.org/webpage/si-list
15 mailto:si-list-***@freelists.org
16 http://tech.groups.yahoo.com/group/si-list
17 http://www.freelists.org/archives/si-list
18 http://www.qsl.net/wb6tpu
19 mailto:si-list-***@freelists.org
20 http://www.freelists.org/webpage/si-list
21 mailto:si-list-***@freelists.org
22 http://tech.groups.yahoo.com/group/si-list
23 http://www.freelists.org/archives/si-list
24 http://www.qsl.net/wb6tpu
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-***@freelists.org with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-***@freelists.org with 'help' in the Subject field

List forum is accessible at:
http://tech.groups.yahoo.com/group/si-list

List archives are viewable at:
http://www.freelists.org/archives/si-list

Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu

Alfred P. Neves

2018-11-07 15:34:38 UTC

Permalink

Dennis,
I donât agree. You first identify the lumped equivalence of the system then apply the appropriate rule of thumb for time-freq conversion estimate.

A via design that has large pads and too large anti pads will be C-L-C and not gaussian, for example - it will resonates. A very high impedance scope probe and 20pf stray capacitance is probably R-C (.35/Tr for 10-90%). A long lossy transmission line will be RLGC.

A well behaved system with a lot of poles spaced frequency with gaussian pdf will be gaussian.

- Al

Products for the Signal Integrity Practitioner

Alfred P. Neves
Chief Technology Officer

Office: 503-679-2429
www.wildrivertech.com
2015 Best In Design&Test Finalist

I agree with David. Also, the article in Wikipedia [1]is better and it
shows the Gaussian response (~0.34/tr), the 1-pole low-pass RC filter
response (~0.35/tr), and others for using 10-90% rise/fall time. Most ICs
are specified for 20-80% rise/fall time so a conversion to 10-90% is needed;
conversion factors for various responses are in Howard Johnson and Martin
Graham's first book, table B.1. They also make a convincing argument that
Gaussian is the response that should be used most of the time.
Dennis
Gosh, Mark, I have to take exception to your description of this article as
âwell writtenâ. It seems to be full of mistakes, syntactical,
conceptual,and structural. Iâve included some examples below, moving
through the article from beginning to end. Regards, -db In Par. 1: -
âmaximumâ -> âminimumâ - Use of the verbiage, â-3dB bandwidth,â
indicates an incomplete conceptual understanding of the relationship between
system bandwidth and rise-time. A clearer understanding of this relationship
-Sentence 1 implies an assumption that the signal starts from zero, as
opposed to some Vmin. Certainly forgivable, but, again, indicative of an
incomplete conceptual understanding. - The language in the rest of this
paragraph, along with the poorly labeled figures, is too inexact for an
engineering article. For example: - Sentence 2 should be stricken. - The
final clause of sentence 3 is both unnecessary and incorrect. (By making the
unfortunate choice of a high-pass filter as his example, the author has
inadvertently depicted a case of infinite bandwidth.) - Fig. 1 shows a step
response, not an impulse response, as claimed in the accompanying verbiage.
-Several unfortunate word choices convey an incomplete understanding.
Examples: - âofâ -> âtoâ in figure captions. - âresponse of a
signalâ -> âresponse of a systemâ. In Eq. 1: - The equation is
incomplete, as the frequency units have been given, but not the time units.
In âHow does the relationship arise?â: - The author is being extremely
misleading in Par. 1, by insinuating that Fourier and/or Gaussian analysis
produce âerrorsâ. It would be more correct to note that these different
methods of relating rise time to bandwidth produce different outcomes. If
anything, Gaussian analysis should be used as the normative reference, since
it carries the full force and authority of the Central Limit Theorem behind
it. In Eq. 10: - This equation is inconsistent with itself. The â1â in
the fraction should be â0.1â. - In the immediately following sentence,
the parenthetical text is misleading and incorrect. In the paragraph
following Eq. 12: - âpowerâŠabsorbed by the reactive elementâŠâ - No!
Power is never absorbed by a purely reactive element. The power is absorbed
by the resistor. The two equations (and accompanying verbiage) following Eq.
16 never shouldâve made it past the technical editor. When this line of
reasoning culminates with, âNote that the imaginaryâŠbecause phase has no
relevanceâŠ,â there can be no doubt left that the author has only a
marginal understanding of the physics heâs describing. Following Eq. 18
theauthor begins, âAlternatively, the above analysis could easily be
achievedâŠâ Right! So, why not strip out all of the ridiculously
convoluted stuff you wrote above and just leave us with this nice short
concise explanation?! One final nit: to write such an article without even
making mention of the beautifully symmetric nature of the Gaussian Fourier
transform pair, and the resulting ease that symmetry lends to establishing
the relationship between system bandwidth and rise-time, is suspect. To fail
to note that the Gaussian approach carries behind it the full weight and
authority of the Central Limit Theorem is criminal. On Nov 6, 2018, at 5:26
derivation and history of the rise time to BW is given in this well written
article reference below.
https://interferencetechnology.com/rise-time-vs-bandwidth-and <https://interferencetechnology.com/rise-time-vs-bandwidth-and>
-applications/[3]I hope this reference helps. Best Regards, Mark Marlett
(Sr.Principal System Engineer - Inphi Corporation) On Thu, Nov 1, 2018 at
Hello experts, since a long time i struggled with a question and always was
afraid to ask ... I see always Rise/fall time as relevant parameter in specs
or for Bandwidth calculation. But I would expect, that the real relevant
parameter would be dV/dt, so the slewrate. e. g. for a signal with 100ps
Risetime and 1000mV Swing (just talk about 0-100% to simplify discussion) I
get a much steeper slope with higher dV/dt vs. a Signal with 100ps risetime
and 100mV swing. Therefore I would assume the signal with 1V swing to have
higher frequency components vs. the Signal with 100mV swing .. but the
risetime is the same... Can anybody enlighten me, where my mistake is in
looking to this topic ? Thanks a lot .. Hermann -- Our next events (contact
us for details): Seminar: Open the Black Box of Memory October 24/25 in
Munich Seminar: Mastering Signal and Power Integrity Design Challenges
Analysis and Verification of DDR3/DDR4 Interfaces November: 13th
(SignalIntegrityJournal) EKH - EyeKnowHow Signal Quality - Made in Bavaria
Hermann Ruckerbauer www.EyeKnowHow.de <http://www.eyeknowhow.de/>[5]
Tel.: +49 (0)991 / 29 69 29 05 Mobile: +49 (0)176 / 787 787 77 Fax: +49
(0)3212 / 121 9008
------------------------------------------------------------------ To
'unsubscribe'in the Subject field or to administer your membership from a
is accessible at: http://tech.groups.yahoo.com/group/si-list[10 <http://tech.groups.yahoo.com/group/si-list[10>] List
archives are viewable at: http://www.freelists.org/archives/si-list[11 <http://www.freelists.org/archives/si-list[11>] Old
http://www.qsl.net/wb6tpu[12 <http://www.qsl.net/wb6tpu[12>]
------------------------------------------------------------------ To
'unsubscribe' in the Subject field or to administer your membership from a
forumis accessible at: http://tech.groups.yahoo.com/group/si-list[16 <http://tech.groups.yahoo.com/group/si-list[16>] List
archives are viewable at: http://www.freelists.org/archives/si-list[17 <http://www.freelists.org/archives/si-list[17>] Old
http://www.qsl.net/wb6tpu[18 <http://www.qsl.net/wb6tpu[18>]
------------------------------------------------------------------ To
'unsubscribe' in the Subject field or to administer your membership from a
forumis accessible at: http://tech.groups.yahoo.com/group/si-list[22 <http://tech.groups.yahoo.com/group/si-list[22>] List
archives are viewable at: http://www.freelists.org/archives/si-list[23 <http://www.freelists.org/archives/si-list[23>] Old
http://www.qsl.net/wb6tpu[24 <http://www.qsl.net/wb6tpu[24>]
--- Links ---
1 https://en.wikipedia.org/wiki/Rise_time <https://en.wikipedia.org/wiki/Rise_time>
3 https://interferencetechnology.com/rise-time-vs-bandwidth-and-applications/ <https://interferencetechnology.com/rise-time-vs-bandwidth-and-applications/>
5 http://www.EyeKnowHow.de <http://www.eyeknowhow.de/>
8 http://www.freelists.org/webpage/si-list <http://www.freelists.org/webpage/si-list>
10 http://tech.groups.yahoo.com/group/si-list <http://tech.groups.yahoo.com/group/si-list>
11 http://www.freelists.org/archives/si-list <http://www.freelists.org/archives/si-list>
12 http://www.qsl.net/wb6tpu <http://www.qsl.net/wb6tpu>
14 http://www.freelists.org/webpage/si-list <http://www.freelists.org/webpage/si-list>
16 http://tech.groups.yahoo.com/group/si-list <http://tech.groups.yahoo.com/group/si-list>
17 http://www.freelists.org/archives/si-list <http://www.freelists.org/archives/si-list>
18 http://www.qsl.net/wb6tpu <http://www.qsl.net/wb6tpu>
20 http://www.freelists.org/webpage/si-list <http://www.freelists.org/webpage/si-list>
22 http://tech.groups.yahoo.com/group/si-list <http://tech.groups.yahoo.com/group/si-list>
23 http://www.freelists.org/archives/si-list <http://www.freelists.org/archives/si-list>
24 http://www.qsl.net/wb6tpu <http://www.qsl.net/wb6tpu>
------------------------------------------------------------------
http://www.freelists.org/webpage/si-list
http://tech.groups.yahoo.com/group/si-list
http://www.freelists.org/archives/si-list
http://www.qsl.net/wb6tpu

Dennis Han

2018-11-07 16:18:43 UTC

Permalink

Al,

I don't see the disagreement. Your last statement agreed with me. A
channel's response includes all the things you mentioned resulting in a
Gaussian response.

Dennis

Post by Alfred P. Neves
Dennis,
I donâ€™t agree. You first identify the lumped equivalence of the system then apply the appropriate rule of thumb for time-freq conversion estimate.
A via design that has large pads and too large anti pads will be C-L-C and not gaussian, for example - it will resonates. A very high impedance scope probe and 20pf stray capacitance is probably R-C (.35/Tr for 10-90%). A long lossy transmission line will be RLGC.
A well behaved system with a lot of poles spaced frequency with gaussian pdf will be gaussian.
- Al
Products for the Signal Integrity Practitioner
Alfred P. Neves
Chief Technology Officer
Office: 503-679-2429
www.wildrivertech.com
2015 Best In Design&Test Finalist

I agree with David. Also, the article in Wikipedia [1]is better and it
shows the Gaussian response (~0.34/tr), the 1-pole low-pass RC filter
response (~0.35/tr), and others for using 10-90% rise/fall time. Most ICs
are specified for 20-80% rise/fall time so a conversion to 10-90% is needed;
conversion factors for various responses are in Howard Johnson and Martin
Graham's first book, table B.1. They also make a convincing argument that
Gaussian is the response that should be used most of the time.
Dennis
Gosh, Mark, I have to take exception to your description of this article as
â€œwell writtenâ€. It seems to be full of mistakes, syntactical,
conceptual,and structural. Iâ€™ve included some examples below, moving
through the article from beginning to end. Regards, -db In Par. 1: -
â€œmaximumâ€ -> â€œminimumâ€ - Use of the verbiage, â€œ-3dB bandwidth,â€
indicates an incomplete conceptual understanding of the relationship between
system bandwidth and rise-time. A clearer understanding of this relationship
-Sentence 1 implies an assumption that the signal starts from zero, as
opposed to some Vmin. Certainly forgivable, but, again, indicative of an
incomplete conceptual understanding. - The language in the rest of this
paragraph, along with the poorly labeled figures, is too inexact for an
engineering article. For example: - Sentence 2 should be stricken. - The
final clause of sentence 3 is both unnecessary and incorrect. (By making the
unfortunate choice of a high-pass filter as his example, the author has
inadvertently depicted a case of infinite bandwidth.) - Fig. 1 shows a step
response, not an impulse response, as claimed in the accompanying verbiage.
-Several unfortunate word choices convey an incomplete understanding.
Examples: - â€œofâ€ -> â€œtoâ€ in figure captions. - â€œresponse of a
signalâ€ -> â€œresponse of a systemâ€. In Eq. 1: - The equation is
incomplete, as the frequency units have been given, but not the time units.
In â€œHow does the relationship arise?â€: - The author is being extremely
misleading in Par. 1, by insinuating that Fourier and/or Gaussian analysis
produce â€œerrorsâ€. It would be more correct to note that these different
methods of relating rise time to bandwidth produce different outcomes. If
anything, Gaussian analysis should be used as the normative reference, since
it carries the full force and authority of the Central Limit Theorem behind
it. In Eq. 10: - This equation is inconsistent with itself. The â€œ1â€ in
the fraction should be â€œ0.1â€. - In the immediately following sentence,
the parenthetical text is misleading and incorrect. In the paragraph
following Eq. 12: - â€œpowerâ€¦absorbed by the reactive elementâ€¦â€ - No!
Power is never absorbed by a purely reactive element. The power is absorbed
by the resistor. The two equations (and accompanying verbiage) following Eq.
16 never shouldâ€™ve made it past the technical editor. When this line of
reasoning culminates with, â€œNote that the imaginaryâ€¦because phase has no
relevanceâ€¦,â€ there can be no doubt left that the author has only a
marginal understanding of the physics heâ€™s describing. Following Eq. 18
theauthor begins, â€œAlternatively, the above analysis could easily be
achievedâ€¦â€ Right! So, why not strip out all of the ridiculously
convoluted stuff you wrote above and just leave us with this nice short
concise explanation?! One final nit: to write such an article without even
making mention of the beautifully symmetric nature of the Gaussian Fourier
transform pair, and the resulting ease that symmetry lends to establishing
the relationship between system bandwidth and rise-time, is suspect. To fail
to note that the Gaussian approach carries behind it the full weight and
authority of the Central Limit Theorem is criminal. On Nov 6, 2018, at 5:26
derivation and history of the rise time to BW is given in this well written
article reference below.
https://interferencetechnology.com/rise-time-vs-bandwidth-and <https://interferencetechnology.com/rise-time-vs-bandwidth-and>
-applications/[3]I hope this reference helps. Best Regards, Mark Marlett
(Sr.Principal System Engineer - Inphi Corporation) On Thu, Nov 1, 2018 at
Hello experts, since a long time i struggled with a question and always was
afraid to ask ... I see always Rise/fall time as relevant parameter in specs
or for Bandwidth calculation. But I would expect, that the real relevant
parameter would be dV/dt, so the slewrate. e. g. for a signal with 100ps
Risetime and 1000mV Swing (just talk about 0-100% to simplify discussion) I
get a much steeper slope with higher dV/dt vs. a Signal with 100ps risetime
and 100mV swing. Therefore I would assume the signal with 1V swing to have
higher frequency components vs. the Signal with 100mV swing .. but the
risetime is the same... Can anybody enlighten me, where my mistake is in
looking to this topic ? Thanks a lot .. Hermann -- Our next events (contact
us for details): Seminar: Open the Black Box of Memory October 24/25 in
Munich Seminar: Mastering Signal and Power Integrity Design Challenges
Analysis and Verification of DDR3/DDR4 Interfaces November: 13th
(SignalIntegrityJournal) EKH - EyeKnowHow Signal Quality - Made in Bavaria
Hermann Ruckerbauer www.EyeKnowHow.de <http://www.eyeknowhow.de/>[5]
Tel.: +49 (0)991 / 29 69 29 05 Mobile: +49 (0)176 / 787 787 77 Fax: +49
(0)3212 / 121 9008
------------------------------------------------------------------ To
'unsubscribe'in the Subject field or to administer your membership from a
is accessible at: http://tech.groups.yahoo.com/group/si-list[10 <http://tech.groups.yahoo.com/group/si-list[10>] List
archives are viewable at: http://www.freelists.org/archives/si-list[11 <http://www.freelists.org/archives/si-list[11>] Old
http://www.qsl.net/wb6tpu[12 <http://www.qsl.net/wb6tpu[12>]
------------------------------------------------------------------ To
'unsubscribe' in the Subject field or to administer your membership from a
forumis accessible at: http://tech.groups.yahoo.com/group/si-list[16 <http://tech.groups.yahoo.com/group/si-list[16>] List
archives are viewable at: http://www.freelists.org/archives/si-list[17 <http://www.freelists.org/archives/si-list[17>] Old
http://www.qsl.net/wb6tpu[18 <http://www.qsl.net/wb6tpu[18>]
------------------------------------------------------------------ To
'unsubscribe' in the Subject field or to administer your membership from a
forumis accessible at: http://tech.groups.yahoo.com/group/si-list[22 <http://tech.groups.yahoo.com/group/si-list[22>] List
archives are viewable at: http://www.freelists.org/archives/si-list[23 <http://www.freelists.org/archives/si-list[23>] Old
http://www.qsl.net/wb6tpu[24 <http://www.qsl.net/wb6tpu[24>]
--- Links ---
1 https://en.wikipedia.org/wiki/Rise_time <https://en.wikipedia.org/wiki/Rise_time>
3 https://interferencetechnology.com/rise-time-vs-bandwidth-and-applications/ <https://interferencetechnology.com/rise-time-vs-bandwidth-and-applications/>
5 http://www.EyeKnowHow.de <http://www.eyeknowhow.de/>
8 http://www.freelists.org/webpage/si-list <http://www.freelists.org/webpage/si-list>
10 http://tech.groups.yahoo.com/group/si-list <http://tech.groups.yahoo.com/group/si-list>
11 http://www.freelists.org/archives/si-list <http://www.freelists.org/archives/si-list>
12 http://www.qsl.net/wb6tpu <http://www.qsl.net/wb6tpu>
14 http://www.freelists.org/webpage/si-list <http://www.freelists.org/webpage/si-list>
16 http://tech.groups.yahoo.com/group/si-list <http://tech.groups.yahoo.com/group/si-list>
17 http://www.freelists.org/archives/si-list <http://www.freelists.org/archives/si-list>
18 http://www.qsl.net/wb6tpu <http://www.qsl.net/wb6tpu>
20 http://www.freelists.org/webpage/si-list <http://www.freelists.org/webpage/si-list>
22 http://tech.groups.yahoo.com/group/si-list <http://tech.groups.yahoo.com/group/si-list>
23 http://www.freelists.org/archives/si-list <http://www.freelists.org/archives/si-list>
24 http://www.qsl.net/wb6tpu <http://www.qsl.net/wb6tpu>
------------------------------------------------------------------
http://www.freelists.org/webpage/si-list
http://tech.groups.yahoo.com/group/si-list
http://www.freelists.org/archives/si-list
http://www.qsl.net/wb6tpu

Alfred P. Neves

2018-11-07 17:12:54 UTC

Permalink

No it doesnât, most big systems are dominated by something(s), like a via-connector-via field where the poles are not spaced with Gaussian-like pdf. Typically you are de-embedding something or youâre doing a full path system simulation - you have to 1st identify the system (RC, RL, CLC, LCL, RLCG, etc) . An interesting experiment is to limit the number of Rational compact model fit to some order n (n is order of polynomial) that is not large. Might be an interesting experiment using Simbeor. There is a TDA Systems poster laying around that describes this.
In the case of the system simulation you may have Gaussian pdf of poles in the system, but not necessarily and not most likely. A system with large SI artifacts does not Gaussian distributed poles, but maybe have several poles for Sinusoidal or Uniform like pdf. A system with significant resonance does not have Gaussian pulse fidelity. The Central Limit theorem is a abstract construct (the limit of number of poles with Gaussian pdf ââ>infinity).

For the de-embedded case of a small structure I have never experienced Gaussian pulse behavior unless it was a Time-Domain Gaussian Filter.

Products for the Signal Integrity Practitioner

Alfred P. Neves
Chief Technology Officer

Office: 503-679-2429
www.wildrivertech.com
2015 Best In Design&Test Finalist

Al,
I don't see the disagreement. Your last statement agreed with me. A
channel's response includes all the things you mentioned resulting in a
Gaussian response.
Dennis

Post by Alfred P. Neves
Dennis,
I donÃ¢â¬â¢t agree. You first identify the lumped equivalence of the system then apply the appropriate rule of thumb for time-freq conversion estimate.
A via design that has large pads and too large anti pads will be C-L-C and not gaussian, for example - it will resonates. A very high impedance scope probe and 20pf stray capacitance is probably R-C (.35/Tr for 10-90%). A long lossy transmission line will be RLGC.
A well behaved system with a lot of poles spaced frequency with gaussian pdf will be gaussian.
- Al
Products for the Signal Integrity Practitioner
Alfred P. Neves
Chief Technology Officer
Office: 503-679-2429
www.wildrivertech.com
2015 Best In Design&Test Finalist

I agree with David. Also, the article in Wikipedia [1]is better and it
shows the Gaussian response (~0.34/tr), the 1-pole low-pass RC filter
response (~0.35/tr), and others for using 10-90% rise/fall time. Most ICs
are specified for 20-80% rise/fall time so a conversion to 10-90% is needed;
conversion factors for various responses are in Howard Johnson and Martin
Graham's first book, table B.1. They also make a convincing argument that
Gaussian is the response that should be used most of the time.
Dennis
Gosh, Mark, I have to take exception to your description of this article as
Ã¢â¬Åwell writtenÃ¢â¬Â. It seems to be full of mistakes, syntactical,
conceptual,and structural. IÃ¢â¬â¢ve included some examples below, moving
through the article from beginning to end. Regards, -db In Par. 1: -
Ã¢â¬ÅmaximumÃ¢â¬Â -> Ã¢â¬ÅminimumÃ¢â¬Â - Use of the verbiage, Ã¢â¬Å-3dB bandwidth,Ã¢â¬Â
indicates an incomplete conceptual understanding of the relationship between
system bandwidth and rise-time. A clearer understanding of this relationship
-Sentence 1 implies an assumption that the signal starts from zero, as
opposed to some Vmin. Certainly forgivable, but, again, indicative of an
incomplete conceptual understanding. - The language in the rest of this
paragraph, along with the poorly labeled figures, is too inexact for an
engineering article. For example: - Sentence 2 should be stricken. - The
final clause of sentence 3 is both unnecessary and incorrect. (By making the
unfortunate choice of a high-pass filter as his example, the author has
inadvertently depicted a case of infinite bandwidth.) - Fig. 1 shows a step
response, not an impulse response, as claimed in the accompanying verbiage.
-Several unfortunate word choices convey an incomplete understanding.
Examples: - Ã¢â¬ÅofÃ¢â¬Â -> Ã¢â¬ÅtoÃ¢â¬Â in figure captions. - Ã¢â¬Åresponse of a
signalÃ¢â¬Â -> Ã¢â¬Åresponse of a systemÃ¢â¬Â. In Eq. 1: - The equation is
incomplete, as the frequency units have been given, but not the time units.
In Ã¢â¬ÅHow does the relationship arise?Ã¢â¬Â: - The author is being extremely
misleading in Par. 1, by insinuating that Fourier and/or Gaussian analysis
produce Ã¢â¬ÅerrorsÃ¢â¬Â. It would be more correct to note that these different
methods of relating rise time to bandwidth produce different outcomes. If
anything, Gaussian analysis should be used as the normative reference, since
it carries the full force and authority of the Central Limit Theorem behind
it. In Eq. 10: - This equation is inconsistent with itself. The Ã¢â¬Å1Ã¢â¬Â in
the fraction should be Ã¢â¬Å0.1Ã¢â¬Â. - In the immediately following sentence,
the parenthetical text is misleading and incorrect. In the paragraph
following Eq. 12: - Ã¢â¬ÅpowerÃ¢â¬ÂŠabsorbed by the reactive elementÃ¢â¬ÂŠÃ¢â¬Â - No!
Power is never absorbed by a purely reactive element. The power is absorbed
by the resistor. The two equations (and accompanying verbiage) following Eq.
16 never shouldÃ¢â¬â¢ve made it past the technical editor. When this line of
reasoning culminates with, Ã¢â¬ÅNote that the imaginaryÃ¢â¬ÂŠbecause phase has no
relevanceÃ¢â¬ÂŠ,Ã¢â¬Â there can be no doubt left that the author has only a
marginal understanding of the physics heÃ¢â¬â¢s describing. Following Eq. 18
theauthor begins, Ã¢â¬ÅAlternatively, the above analysis could easily be
achievedÃ¢â¬ÂŠÃ¢â¬Â Right! So, why not strip out all of the ridiculously
convoluted stuff you wrote above and just leave us with this nice short
concise explanation?! One final nit: to write such an article without even
making mention of the beautifully symmetric nature of the Gaussian Fourier
transform pair, and the resulting ease that symmetry lends to establishing
the relationship between system bandwidth and rise-time, is suspect. To fail
to note that the Gaussian approach carries behind it the full weight and
authority of the Central Limit Theorem is criminal. On Nov 6, 2018, at 5:26
derivation and history of the rise time to BW is given in this well written
article reference below.
https://interferencetechnology.com/rise-time-vs-bandwidth-and <https://interferencetechnology.com/rise-time-vs-bandwidth-and>
-applications/[3]I hope this reference helps. Best Regards, Mark Marlett
(Sr.Principal System Engineer - Inphi Corporation) On Thu, Nov 1, 2018 at
Hello experts, since a long time i struggled with a question and always was
afraid to ask ... I see always Rise/fall time as relevant parameter in specs
or for Bandwidth calculation. But I would expect, that the real relevant
parameter would be dV/dt, so the slewrate. e. g. for a signal with 100ps
Risetime and 1000mV Swing (just talk about 0-100% to simplify discussion) I
get a much steeper slope with higher dV/dt vs. a Signal with 100ps risetime
and 100mV swing. Therefore I would assume the signal with 1V swing to have
higher frequency components vs. the Signal with 100mV swing .. but the
risetime is the same... Can anybody enlighten me, where my mistake is in
looking to this topic ? Thanks a lot .. Hermann -- Our next events (contact
us for details): Seminar: Open the Black Box of Memory October 24/25 in
Munich Seminar: Mastering Signal and Power Integrity Design Challenges
Analysis and Verification of DDR3/DDR4 Interfaces November: 13th
(SignalIntegrityJournal) EKH - EyeKnowHow Signal Quality - Made in Bavaria
Hermann Ruckerbauer www.EyeKnowHow.de <http://www.eyeknowhow.de/>[5]
Tel.: +49 (0)991 / 29 69 29 05 Mobile: +49 (0)176 / 787 787 77 Fax: +49
(0)3212 / 121 9008
------------------------------------------------------------------ To
'unsubscribe'in the Subject field or to administer your membership from a
is accessible at: http://tech.groups.yahoo.com/group/si-list[10 <http://tech.groups.yahoo.com/group/si-list[10>] List
archives are viewable at: http://www.freelists.org/archives/si-list[11 <http://www.freelists.org/archives/si-list[11>] Old
http://www.qsl.net/wb6tpu[12 <http://www.qsl.net/wb6tpu[12>]
------------------------------------------------------------------ To
'unsubscribe' in the Subject field or to administer your membership from a
forumis accessible at: http://tech.groups.yahoo.com/group/si-list[16 <http://tech.groups.yahoo.com/group/si-list[16>] List
archives are viewable at: http://www.freelists.org/archives/si-list[17 <http://www.freelists.org/archives/si-list[17>] Old
http://www.qsl.net/wb6tpu[18 <http://www.qsl.net/wb6tpu[18>]
------------------------------------------------------------------ To
'unsubscribe' in the Subject field or to administer your membership from a
forumis accessible at: http://tech.groups.yahoo.com/group/si-list[22 <http://tech.groups.yahoo.com/group/si-list[22>] List
archives are viewable at: http://www.freelists.org/archives/si-list[23 <http://www.freelists.org/archives/si-list[23>] Old
http://www.qsl.net/wb6tpu[24 <http://www.qsl.net/wb6tpu[24>]
--- Links ---
1 https://en.wikipedia.org/wiki/Rise_time <https://en.wikipedia.org/wiki/Rise_time>
3 https://interferencetechnology.com/rise-time-vs-bandwidth-and-applications/ <https://interferencetechnology.com/rise-time-vs-bandwidth-and-applications/>
5 http://www.EyeKnowHow.de <http://www.eyeknowhow.de/>
8 http://www.freelists.org/webpage/si-list <http://www.freelists.org/webpage/si-list>
10 http://tech.groups.yahoo.com/group/si-list <http://tech.groups.yahoo.com/group/si-list>
11 http://www.freelists.org/archives/si-list <http://www.freelists.org/archives/si-list>
12 http://www.qsl.net/wb6tpu <http://www.qsl.net/wb6tpu>
14 http://www.freelists.org/webpage/si-list <http://www.freelists.org/webpage/si-list>
16 http://tech.groups.yahoo.com/group/si-list <http://tech.groups.yahoo.com/group/si-list>
17 http://www.freelists.org/archives/si-list <http://www.freelists.org/archives/si-list>
18 http://www.qsl.net/wb6tpu <http://www.qsl.net/wb6tpu>
20 http://www.freelists.org/webpage/si-list <http://www.freelists.org/webpage/si-list>
22 http://tech.groups.yahoo.com/group/si-list <http://tech.groups.yahoo.com/group/si-list>
23 http://www.freelists.org/archives/si-list <http://www.freelists.org/archives/si-list>
24 http://www.qsl.net/wb6tpu <http://www.qsl.net/wb6tpu>
------------------------------------------------------------------
http://www.freelists.org/webpage/si-list
http://tech.groups.yahoo.com/group/si-list
http://www.freelists.org/archives/si-list
http://www.qsl.net/wb6tpu

"Dan Bostan" (Redacted sender "dbostan" for DMARC)

2018-11-01 17:37:57 UTC

Permalink

D/CAS chapter presentation "Power Integrity Effects on Signal Integrity in FPGA Systems" on the downloads page at http://www.piscanner.com
Regards,
Cosmin

------------------------------------------------------------------
To unsubscribe from si-list:
si-list-***@freelists.org with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-***@freelists.org with 'help' in the Subject field

List forumÂ is accessible at:
Â Â Â Â Â Â Â http://tech.groups.yahoo.com/group/si-list

List archives are viewable at:Â Â
Â Â Â Â Â Â http://www.freelists.org/archives/si-list

Old (prior to June 6, 2001) list archives are viewable at:
Â Â Â Â Â Â http://www.qsl.net/wb6tpu
Â

------------------------------------------------------------------
To unsubscribe from si-list:
si-list-***@freelists.org with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-***@freelists.org with 'help' in the Subject field

List forum is accessible at:
http://tech.groups.yahoo.com/group/si-list

List archives are viewable at:
http://www.freelists.org/archives/si-list

Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu

e***@ericbogatin.com

2018-11-07 17:16:02 UTC

Permalink

Hi guys- I am late to the party as usual. I have a few comments
about the question of rise time and bandwidth in the recent
discussions.

At the risk of adding some more gasoline to this bonfire, here is
a link to a piece I did on this same question of using a 1 pole
response to show the relationship between rise time and
bandwidth:
https://www.signalintegrityjournal.com/blogs/12-si-pi-fundamental
s/post/853-back-to-basics-bandwidth-and-rise-time

An important message in this piece is the mis-use of the terms
rise time and bandwidth. They are figures of merit based on
interpretations of the signals with assumptions. Are these two
terms good figures of merit? Wait for it.."it depends". This is
why it's important to understand the assumptions behind the
relationship.

If you worry about should it be BW = 0.5/RT or = 0.35/RT, you may
not want to use a simple figure of merit like rise time and
bandwidth. It you have significant resonances in your spectrum,
the term bandwidth may not be useful. You may need the entire
spectrum- maybe there are other figures of merit that are more
useful, but you should define these figures of merit carefully.

I watch my students and young engineers use scopes. After they
press the auto scale button, and I slap their hand, the next
button they press is the measurement button to use the built-in
analysis tools to extract a specific measurement from an
acquisition, like the average, or the peak to peak or the rise
time.

While all these terms can be calculated analytically from an
acquisition window of measurements, how we interpret these
figures of merit depend a lot on how well the signal matches some
assumptions.

What could be wrong with an average value? If the signal is a
sine wave and we have n cycles + a fraction of a cycle, the
average of the acquisition window will vary depending on how much
of a fraction we have left. We will always calculate an average,
but this is not always the DC component of the sine wave.

What could be wrong with a peak to peak value? If our signal has
noise that gets through the aliasing filter, we just get a few
measurements that are outside the signal levels and artificially
affect the peak values measured in the acquisition buffer.

How could rise time be off? Is your measurement set for
calculating the 10-90 or the 20-80? Many scopes don't tell you
the algorithm. How does it get the initial steady state and the
final steady state levels from which to calculate the 0% and 100%
levels? What if there is overshoot or ringing? What is the rise
time? What if the signal is coming through a lossy interconnect-
there will be a long tail- it will give a huge 10-90 rise time,
and a much smaller 20-80 rise time. Which one is relevant? Or
even useful?

To use these figures of merit, like rise time and bandwidth, you
have to be aware of the definitions of how these are calculated
from the response, and whether these assumptions apply to your
situation.

There is ambiguity in each figure of merit we use. They are not
meant for precise sign off, but on helping to speed you to
insight. It's not that figures of merit are wrong, they just have
an appropriate time and place for their use. We should pay
attention to when it is appropriate to use any figure of merit,
and when it is not. They should be used in the case where an OK
answer NOW! Is more important than a good answer late. But, as in
all engineering analysis, be aware of your assumptions.

I'm stepping down from my soapbox now..

--eric

*******************************************************
Dr. Eric Bogatin

Signal Integrity Evangelist, Teledyne LeCroy,
<http://www.teledynelecroy.com/> www.TeledyneLeCroy.com

Dean, <http://www.bethesignal.com/> Signal Integrity Academy

Technical Editor, <https://www.signalintegrityjournal.com/>
Signal Integrity Journal,

Adjunct Prof, ECEE Univ of Colorado, Boulder

Fall 2018: Practical Printed Circuit Board Design and Manufacture

707 Windflower Dr

Longmont, CO 80504 USA

Twitter @beTheSignal
e: <mailto:***@beTheSignal.com> ***@beTheSignal.com

cell: 913-424-4333

<http://www.bethesignal.com/> www.beTheSignal.com

Great resources:

Recorded webinars:
<http://teledynelecroy.com/support/techlib/webcasts.aspx>
http://teledynelecroy.com/support/techlib/webcasts.aspx

Upcoming event:
<http://teledynelecroy.com/events/?capid=106&mid=1041>
http://teledynelecroy.com/events/?capid=106&mid=1041,

<https://click.linksynergy.com/fs-bin/click?id=SHZHWbD4AcI&subid=
&offerid=163217.1&type=10&tmpid=12438&RD_PARM1=http%253A%252F%252
Fwww.informit.com%252Fstore%252Fsignal-and-power-integrity-simpli
fied-9780132349796> Signal and Power Integrity- Simplified

<https://click.linksynergy.com/fs-bin/click?id=SHZHWbD4AcI&subid=
&offerid=163217.1&type=10&tmpid=12438&RD_PARM1=http%253A%252F%252
Fwww.informit.com%252Fstore%252Fprinciples-of-power-integrity-for
-pdn-design-simplified-9780132735551> Principles of PDN Design-
Simplified
******************************************************

------------------------------------------------------------------
To unsubscribe from si-list:
si-list-***@freelists.org with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-***@freelists.org with 'help' in the Subject field

List forum is accessible at:
http://tech.groups.yahoo.com/group/si-list

List archives are viewable at:
http://www.freelists.org/archives/si-list

Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu

Hermann Ruckerbauer

2018-11-12 07:15:09 UTC

Permalink

Hello,
first of all thanks for all who answered!
I was out and still need to go through all answers .. will come back if
I have more questions..

thanks

Hermann
--
Our next events (contact us for details):
Seminar: Mastering Signal and Power Integrity Design Challenges
November 7th in Winnersh (UK)
November 27th in Munich (Germany)

Webinar: Analysis and Verification of DDR3/DDR4 Interfaces
November 13th by Signal Integrity Journal

EKH - EyeKnowHow
Signal Quality - Made in Bavaria
Hermann Ruckerbauer
www.EyeKnowHow.de
***@EyeKnowHow.de
Itzlinger Strasse 21a
94469 Deggendorf
Tel.: +49 (0)991 / 29 69 29 05
Mobile: +49 (0)176 / 787 787 77
Fax: +49 (0)3212 / 121 9008

Post by e***@ericbogatin.com
Hi guys- I am late to the party as usual. I have a few comments
about the question of rise time and bandwidth in the recent
discussions.
At the risk of adding some more gasoline to this bonfire, here is
a link to a piece I did on this same question of using a 1 pole
response to show the relationship between rise time and
https://www.signalintegrityjournal.com/blogs/12-si-pi-fundamental
s/post/853-back-to-basics-bandwidth-and-rise-time
An important message in this piece is the mis-use of the terms
rise time and bandwidth. They are figures of merit based on
interpretations of the signals with assumptions. Are these two
terms good figures of merit? Wait for it.."it depends". This is
why it's important to understand the assumptions behind the
relationship.
If you worry about should it be BW = 0.5/RT or = 0.35/RT, you may
not want to use a simple figure of merit like rise time and
bandwidth. It you have significant resonances in your spectrum,
the term bandwidth may not be useful. You may need the entire
spectrum- maybe there are other figures of merit that are more
useful, but you should define these figures of merit carefully.
I watch my students and young engineers use scopes. After they
press the auto scale button, and I slap their hand, the next
button they press is the measurement button to use the built-in
analysis tools to extract a specific measurement from an
acquisition, like the average, or the peak to peak or the rise
time.
While all these terms can be calculated analytically from an
acquisition window of measurements, how we interpret these
figures of merit depend a lot on how well the signal matches some
assumptions.
What could be wrong with an average value? If the signal is a
sine wave and we have n cycles + a fraction of a cycle, the
average of the acquisition window will vary depending on how much
of a fraction we have left. We will always calculate an average,
but this is not always the DC component of the sine wave.
What could be wrong with a peak to peak value? If our signal has
noise that gets through the aliasing filter, we just get a few
measurements that are outside the signal levels and artificially
affect the peak values measured in the acquisition buffer.
How could rise time be off? Is your measurement set for
calculating the 10-90 or the 20-80? Many scopes don't tell you
the algorithm. How does it get the initial steady state and the
final steady state levels from which to calculate the 0% and 100%
levels? What if there is overshoot or ringing? What is the rise
time? What if the signal is coming through a lossy interconnect-
there will be a long tail- it will give a huge 10-90 rise time,
and a much smaller 20-80 rise time. Which one is relevant? Or
even useful?
To use these figures of merit, like rise time and bandwidth, you
have to be aware of the definitions of how these are calculated
from the response, and whether these assumptions apply to your
situation.
There is ambiguity in each figure of merit we use. They are not
meant for precise sign off, but on helping to speed you to
insight. It's not that figures of merit are wrong, they just have
an appropriate time and place for their use. We should pay
attention to when it is appropriate to use any figure of merit,
and when it is not. They should be used in the case where an OK
answer NOW! Is more important than a good answer late. But, as in
all engineering analysis, be aware of your assumptions.
I'm stepping down from my soapbox now..
--eric
*******************************************************
Dr. Eric Bogatin
Signal Integrity Evangelist, Teledyne LeCroy,
<http://www.teledynelecroy.com/> www.TeledyneLeCroy.com
Dean, <http://www.bethesignal.com/> Signal Integrity Academy
Technical Editor, <https://www.signalintegrityjournal.com/>
Signal Integrity Journal,
Adjunct Prof, ECEE Univ of Colorado, Boulder
Fall 2018: Practical Printed Circuit Board Design and Manufacture
707 Windflower Dr
Longmont, CO 80504 USA
cell: 913-424-4333
<http://www.bethesignal.com/> www.beTheSignal.com
<http://teledynelecroy.com/support/techlib/webcasts.aspx>
http://teledynelecroy.com/support/techlib/webcasts.aspx
<http://teledynelecroy.com/events/?capid=106&mid=1041>
http://teledynelecroy.com/events/?capid=106&mid=1041,
<https://click.linksynergy.com/fs-bin/click?id=SHZHWbD4AcI&subid=
&offerid=163217.1&type=10&tmpid=12438&RD_PARM1=http%253A%252F%252
Fwww.informit.com%252Fstore%252Fsignal-and-power-integrity-simpli
fied-9780132349796> Signal and Power Integrity- Simplified
<https://click.linksynergy.com/fs-bin/click?id=SHZHWbD4AcI&subid=
&offerid=163217.1&type=10&tmpid=12438&RD_PARM1=http%253A%252F%252
Fwww.informit.com%252Fstore%252Fprinciples-of-power-integrity-for
-pdn-design-simplified-9780132735551> Principles of PDN Design-
Simplified
******************************************************
------------------------------------------------------------------
http://www.freelists.org/webpage/si-list
http://tech.groups.yahoo.com/group/si-list
http://www.freelists.org/archives/si-list
http://www.qsl.net/wb6tpu