[SI-LIST] Re: LC filters on SDRAM signal lines
- From: Rick Collins <gnuarm.2006@xxxxxxxxx>
- To: "David Carney (Neenah)" <David.Carney@xxxxxxxxxx>
- Date: Fri, 24 Jun 2011 16:16:50 -0400
David,
I don't think you can determine whether any of these filters can
solve your problem until you understand more about your problem. As
others have indicated fast signals are not necessarily a problem in
many similar systems. The problem comes when the fast signals either
radiate noise which could be picked up by an unshielded receiver or
are conducted into your module either through signal lines or the
power distribution system. The solution you propose might help if
the cause is radiated noise, but I think it would be much less likely
to solve a conducted noise problem.
I can't say anything about the affect these filters might have on the
SDRAM operation. That would require a lot more info on the
waveforms, the bus design and the timing margins. The only signal on
the bus that requires clean edges is the clock. The data, control
and address lines only need to be stable at the time they are sampled
by the device and of course you need to control over and undershoot
within the limits of the devices on the bus.
I'm curious about how you determined the source of the noise is the
75 MHz SDRAM bus. Do you think the noise is caused by all of the
signals or possibly just the clock? I suppose if it is broadband you
would suspect signals other than the clock which should have a fairly
simple harmonic content while the other signals will have a broad
spectrum in most cases.
One last word, SI is not a simple issue that lends itself to simple
solutions like, "add a filter". You need to understand the problem
before you can offer a solution.
Regards,
Rick
At 02:08 PM 6/24/2011, David Carney (Neenah) wrote:
>To clarify, our problem is that the GSM module is picking up broadband
>noise whose source is the processor SDRAM bus on the board that the GSM
>module is attached to. The proposal is to filter the SDRAM data lines
>to reduce the high frequency 869-894 MHz + content in these signals
>without affecting the useful frequency content in these 75 MHz signals
>and wrecking the signal integrity or timing on the SDRAM bus. Reducing
>this high frequency content should reduce the amount of noise coupled in
>to the GSM receiver which is currently causing it to have a higher than
>acceptable BER.
>
>This is not necessarily a desirable solution to us since adding all
>these filters is a challenge for space and redoing a lot of routing, but
>it has been suggested that this is one of the most effective solutions.
>I'm trying to figure out how to evaluate whether it will be effective
>and also whether it will not wreck the SDRAM bus SI and timing.
>
>It sounds like your suggestion is to look more closely at the PDN around
>the SDRAM bus before considering filtering the data lines?
>
>Thanks for the response.
>
>
>
>-----Original Message-----
>From: steve weir [mailto:weirsi@xxxxxxxxxx]
>Sent: Friday, June 24, 2011 12:19 PM
>To: David Carney (Neenah)
>Cc: si-list@xxxxxxxxxxxxx
>Subject: Re: [SI-LIST] LC filters on SDRAM signal lines
>
>The point of any filter is to separate signal energy from noise energy.
>
>Before you can begin thinking about a filter you need to determine if
>the offending noise is well outside your signal spectrum. If it is not,
>
>then your only available option is to shield. That is assuming that
>your problem is GSM pick-up on the signal lines. Even though your SDRAM
>
>is only running a 75MHz clock, in order for the timing to be stable you
>will have spectrum at least out to around 400MHz, which is close to half
>
>of the GSM. You don't have much of a window between your required pass
>and stop bands.
>
>The GSM "buzz of death" can get directly into anything that acts as an
>antenna. It also does bad things to PDNs. You need to have a hard look
>
>at your PDN and your layout to see just where it is that the GSM noise
>is coupling in. If you are lucky, you can fix this with some changes to
>
>your bypass network. If you've got a poorly designed stack-up and
>layout where GSM noise is for example directly coupling into your signal
>
>returns, then you will have to address that.
>
>Steve.
>On 6/24/2011 8:47 AM, David Carney (Neenah) wrote:
> > I'm working on an embedded product with a GSM cellular module. The
> > module is picking up broadband noise in the GSM850 band (869-894 MHz)
>at
> > levels of approximately -90dBm to -100dBm from the board it is on.
>The
> > noise is causing BER levels that are too high (fails some tests). The
> > source of the noise has been tracked to a processor SDRAM bus
>interface
> > running at 75 MHz clock frequency. This processor and SDRAM are
> > physically close to the GSM module on the board. One of the many
> > suggestions for mitigating this problem that we are considering is to
> > add LC filters to all of the SDRAM data lines such as the following
> > example parts:
> >
> >
> > Murata NFM18PS105R0J3
> >
>(http://search.murata.co.jp/Ceramy/image/img/PDF/ENG/L0111S0111NFM18PS.p
> > df)
> >
> > TDK MEA1608PH (http://www.tdk.co.jp/tefe02/e9621_mea_signal_1.pdf)
> >
> >
> >
> > These parts are recommended for LCD interfaces. Does anyone have
> > experience using these on SDRAM interfaces? Is this an effective way
>to
> > solve the problem? What considerations do we need to account for in
> > using them? Has anyone modeled these parts for signal integrity
> > simulations? We have been told by Murata that no model is available.
> > What modeling approach did you use?
> >
> >
> >
> > Thanks.
> >
> >
> >
> >
> >
> > David T. Carney P.E.
> >
> > Senior Design Engineer
> >
> > Plexus Engineering Solutions
> >
> > Neenah Design Center
> >
> > 920.751.5646
> >
> >
> >
> >
> > ------------------------------------------------------------------
> > 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 technical documents are available at:
> > http://www.si-list.net
> >
> > List archives are viewable at:
> > //www.freelists.org/archives/si-list
> >
> > Old (prior to June 6, 2001) list archives are viewable at:
> > http://www.qsl.net/wb6tpu
> >
> >
> >
>
>
>--
>Steve Weir
>IPBLOX, LLC
>150 N. Center St. #211
>Reno, NV 89501
>www.ipblox.com
>
>(775) 299-4236 Business
>(866) 675-4630 Toll-free
>(707) 780-1951 Fax
>
>
>------------------------------------------------------------------
>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 technical documents are available at:
> http://www.si-list.net
>
>List archives are viewable at:
> //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-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 technical documents are available at:
http://www.si-list.net
List archives are viewable at:
//www.freelists.org/archives/si-list
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
