[ibis-macro] Re: Init Return Impulse models with Redriver
- From: Scott McMorrow <scott@xxxxxxxxxxxxx>
- To: fangyi_rao@xxxxxxxxxxx
- Date: Fri, 20 Jun 2014 13:50:54 -0400
thank you. I'll take some time to read this.
Have you compared these results to physical measurements with low and high
amplitude inputs?
Scott
Scott McMorrow
Teraspeed® Consulting - A XXXXXX company
16 Stormy Brook Rd
Falmouth, ME 04105
(401) 284-1827 Business
http://www.teraspeed.com
More to come ... soon.
On Fri, Jun 20, 2014 at 1:43 PM, <fangyi_rao@xxxxxxxxxxx> wrote:
> Hi Scott,
>
>
>
> Nonlinearity in redriver is modeled by redriver Rx GetWave and/or redriver
> Tx GetWave. In Figs. 4 & 5 of the link below you can find some result of an
> AMI redriver model with a limiting amplifier.
>
>
>
>
> http://signal-integrity.tm.agilent.com/wp-content/uploads/2013/02/8-TA4_Paper_End-to-end_Link_Analysis.pdf
>
>
>
> Fangyi
>
>
>
> *From:* ibis-macro-bounce@xxxxxxxxxxxxx [mailto:
> ibis-macro-bounce@xxxxxxxxxxxxx] *On Behalf Of *Scott McMorrow
> *Sent:* Friday, June 20, 2014 7:29 AM
> *To:* RAO,FANGYI (A-USA,ex1)
> *Cc:* D.Shah@xxxxxx; Walter Katz; IBIS-ATM
>
> *Subject:* [ibis-macro] Re: Init Return Impulse models with Redriver
>
>
>
> Fangyi
>
>
>
> For a redriver, how is non-linearity in the receiver/driver handled. To
> me the simplest useful case would be a limiting amplifier.
>
>
>
> Scott
>
>
>
>
>
> Scott McMorrow
> Teraspeed® Consulting - A XXXXXX company
>
> 16 Stormy Brook Rd
> Falmouth, ME 04105
>
> (401) 284-1827 Business
>
> http://www.teraspeed.com
>
>
>
> More to come ... soon.
>
>
>
> On Fri, Mar 14, 2014 at 1:20 PM, <fangyi_rao@xxxxxxxxxxx> wrote:
>
> Please keep in mind that if one of the redriver Rx and Tx has GetWave,
> then the entire impulse of the 1st channel must be removed from the final
> impulse by deconvolution, which is known to be unreliable for long
> impulses. Not to mention the complexities associated with cascaded
> repeaters and with crosstalk.
>
>
>
> Regards,
>
> Fangyi
>
>
>
> *From:* ibis-macro-bounce@xxxxxxxxxxxxx [mailto:
> ibis-macro-bounce@xxxxxxxxxxxxx] *On Behalf Of *Darshan Shah
> *Sent:* Friday, March 14, 2014 9:51 AM
> *To:* Walter Katz; ibis-macro@xxxxxxxxxxxxx
> *Subject:* [ibis-macro] Re: Init Return Impulse models with Redriver
>
>
>
> Sure, we can work on this.
>
>
>
> -Darshan
>
>
>
> *From:* Walter Katz [mailto:wkatz@xxxxxxxxxx <wkatz@xxxxxxxxxx>]
> *Sent:* Friday, March 14, 2014 7:07 AM
> *To:* Darshan Shah; ibis-macro@xxxxxxxxxxxxx
> *Subject:* RE: [ibis-macro] Init Return Impulse models with Redriver
>
>
>
> Darshan,
>
>
>
> You are correct, the spec is broken for Redrivers. Would you like to work
> with me on creating a BIRD that will correct this?
>
>
>
> Walter
>
>
>
> *From:* ibis-macro-bounce@xxxxxxxxxxxxx [
> mailto:ibis-macro-bounce@xxxxxxxxxxxxx <ibis-macro-bounce@xxxxxxxxxxxxx>] *On
> Behalf Of *Darshan Shah
> *Sent:* Thursday, March 13, 2014 6:50 PM
> *To:* ibis-macro@xxxxxxxxxxxxx
> *Subject:* [ibis-macro] Init Return Impulse models with Redriver
>
>
>
> Hi,
>
>
>
> I have following setup for my simulation
>
>
>
> TX AMI model (Init_Return_Impulse = True, Getwave = False) à Upstream
> Channel à Repeater RX (redriver and not retimer) AMI model
> (Init_Return_Impulse = True, Getwave = False) à Repeater TX (redriver and
> not retimer) AMI model (Init_Return_Impulse = True, Getwave = False) à
> Downstream Channel à RX AMI model (Init_Return_Impulse = True, Getwave =
> False)
>
>
>
> I have Init Return Impulse model for all the 4 serdes and no Getwave model.
>
> Also, RX (not repeater RX but RX at extreme right) has feature of auto
> tuning.
>
>
>
> Looking into all the 9 steps of Repeater simulation (Page 224 of
> ver6_0.pdf), it looks like there is no effect on RX Auto Adaptation
> algorithm due to upstream channel.
>
> The way it works is Upstream channel IR is calculated (convolution of
> upstream channel + TX serdes + Repeater RX serdes IR) which is upto step 3
> of spec. and then Downstream Channel IR is calculated (convolution of
> downstream channel + Repeater TX serdes + RX serdes IR) which is upto step
> 6 of spec.
>
>
>
> It looks like RX serdes eq. parameters are auto tuned based on IR
> presented to RX serdes input which is IR of (downstream channel + Repeater
> TX serdes) and hence effect of upstream channel, TX serdes and Repeater RX
> serdes has no impact on RX adaptation algorithm which is incorrect.
>
>
>
> Please correct if I am wrong.
>
> Any help will be really appreciated.
>
>
>
> Thanks,
>
> Darshan
>
>
>
Other related posts:
