Impedance of a capacitor is Z = 1/(2 x pi x f x C) Capacitor feeds a 50 ohm transmission line Capacitor and transmission line form a voltage divider Vout = Vin x (50/(Zcap(f)+50) charging time constant of a capacitor t = 1/RC = 1/(50*C) These equations along with the spectrum of your serial bit stream tell you everything you need to know. As eric bogatin says: "Do the math." Scott McMorrow Consultant - R&D 16 Stormy Brook Rd Falmouth, ME 04105 (401) 284-1827 Business http://www.teraspeed.com On Thu, Jan 8, 2015 at 10:05 AM, Zheng Edison <edisonzheng@xxxxxxxx> wrote: > Hi Scott, > in most applications,the value of 0402 ac coupling cap is 100nf,my > question is why 100nf?is 220nf ok?What is the difference? > Edison > > --- Original Message --- > > From: "Scott McMorrow" <scott@xxxxxxxxxxxxx> > Sent: January 8, 2015 8:39 PM > To: "Carson Au" <carson.au@xxxxxxxxx> > Cc: edisonzheng@xxxxxxxx, "Boris Bakshan" <bbakshan@xxxxxxxxx>, > andrew@xxxxxxxxxxxxxxxxxx, si-list@xxxxxxxxxxxxx > Subject: [SI-LIST] Re: 100nF AC coupling caps > > Carson, > If you have a 2D simulator available, as a first approximation take your > capacitor pads and join them together into one trace that is 20 mil wide > and 60 mil long, and insert into your 50 ohm controlled impedance trace > that is somewhere around 4 or 5 mil wide. Perform a simulation with your > traces, connectors and packages in the circuit and see what happens. > > As for "free", engineering is a non-recurring expense, and as such is free > with regards to manufacturing costs. Of course, it is not free in terms of > capital expense if you do not have the appropriate tools at hand. That is > why I gave a very simple rule of thumb, which is to relieve the plane under > the capacitors ...with the caveat ... if it is possible to do so without > disrupting other signals, and if there are two planes directly under the > top layer. IF there is only one plane, then the problem just got much more > difficult. If the 2nd plane is a power plane, just provide a ground fill in > the region underneath the capacitors. These are really simple things to do > for "free margin", without the time and expense of using a $100K full wave > 3D field solver. > > best regards, > > Scott > > > > > > > > Scott McMorrow > Consultant - R&D > 16 Stormy Brook Rd > Falmouth, ME 04105 > (401) 284-1827 Business > http://www.teraspeed.com > > On Thu, Jan 8, 2015 at 6:58 AM, Carson Au <carson.au@xxxxxxxxx> wrote: > > > Hi Scott, > > > > I do not doubt that impedance compensation of the pad will improve > channel > > margins. However, compensating for this is not quite free margin - there > is > > considerable difficulty in implementing this on lower end EDA tools (let > > alone working out the appropriate compensation using a 3D field solver > > tool). The ground copper removal could also disturb the return path of > > signals on the opposite side of this plane. > > > > I guess my question is - when is pad compensation critical? What does it > > depend on? > > > > For example, a typical pad on an 0402 component could be 0.6mm in length. > > Wavelength of the nyquist frequency of a 10Gb channel on a PCB would be > > approximately 33.4mm (200ps/6ps/mm) (I acknowledge that higher frequency > > content would exist, but this is just a quick estimation). Is the > impedance > > discontinuity of 0.6mm in comparison to the nyquist wavelength > > insignificant? > > > > Regards, > > Carson > > > > On Thu, Jan 8, 2015 at 10:38 PM, Scott McMorrow <scott@xxxxxxxxxxxxx> > > wrote: > > > >> Carson > >> > >> I'm not sure what you are asking. We can certainly document that > >> compensation of capacitor transitions improves performance in Serdes > >> channels. That is not up for debate any more than tight impedance > control > >> is. Whether there is a specific design study that says a system failed > due > >> to just a cap ... who knows. Systems fail for all sorts of reasons, and > >> most of those reasons are the interaction between reflections across > >> multiple components. Optimization is free margin. Once done, the > margin > >> is always there and costs nothing to implement. Optimization of > structures > >> always provides better system performance. > >> > >> regards > >> > >> Scott > >> > >> > >> > >> > >> > >> > >> Scott McMorrow > >> Consultant - R&D > >> 16 Stormy Brook Rd > >> Falmouth, ME 04105 > >> (401) 284-1827 Business > >> http://www.teraspeed.com > >> > >> On Thu, Jan 8, 2015 at 6:30 AM, Carson Au <carson.au@xxxxxxxxx> wrote: > >> > >>> Are there studies/papers on the practical effect of not compensating > for > >>> the pad impedance in SERDES channel design? Has there been any > documented > >>> channel failures due to not compensating for these pad impedances? > >>> > >>> On Thu, Jan 8, 2015 at 9:53 PM, Zheng Edison <edisonzheng@xxxxxxxx> > >>> wrote: > >>> > >>>> Hi,experts > >>>> What is the difference between using 100nf caps and using 200nf > caps(or > >>>> some other values) in serdes links?thanks. > >>>> HW > >>>> edison > >>>> > >>>> --- Original Message --- > >>>> > >>>> From: "Scott McMorrow" <scott@xxxxxxxxxxxxx> > >>>> Sent: January 8, 2015 6:20 AM > >>>> To: bbakshan@xxxxxxxxx > >>>> Cc: andrew@xxxxxxxxxxxxxxxxxx, si-list@xxxxxxxxxxxxx > >>>> Subject: [SI-LIST] Re: 100nF AC coupling caps > >>>> > >>>> All, > >>>> For an 0402 capacitor, use pretty much any MLCC and you will be fine. > >>>> In > >>>> the configuration that a DC blocking capacitor is used, the important > >>>> criteria is distance from the PCB to the lowest plate in the > >>>> capacitor. An > >>>> 0402 DC blocking capacitor in a differential configuration can be > >>>> designed > >>>> to be mounted on a PCB and have a return loss of better than -35 dB > out > >>>> to > >>>> 12.5 GHz. Such a capacitor has essentially flat response. I just > >>>> designed > >>>> such a transition last week, and modeled the pcb, traces, pads,planes, > >>>> full > >>>> body of the capacitor and all of it's plates. Having done this a few > >>>> times > >>>> in the past, I can tell you that an 0201 capacitor can be designed to > >>>> have > >>>> about twice the bandwidth, and would definitely work up to 56 Gbps > NRZ. > >>>> > >>>> For guardband against Murphy, I will run 0402 capacitors up to 16 > Gbps, > >>>> and > >>>> switch to 0201 capacitor for 25+ Gbps designs. > >>>> > >>>> As a first approximation, in a board with two planes directly > underneath > >>>> the capacitors, make those two planes ground in the vicinity of the > >>>> capacitors, stitch them with vias, and place a hole in the plane > >>>> adjacent > >>>> to the capacitor that extends under both pads. This will provide some > >>>> amount of compensation, and get you to around -15 dB return loss out > to > >>>> 10 > >>>> GHz. > >>>> > >>>> best regards, > >>>> > >>>> Scott > >>>> > >>>> > >>>> > >>>> > >>>> > >>>> > >>>> > >>>> Scott McMorrow > >>>> Consultant - R&D > >>>> 16 Stormy Brook Rd > >>>> Falmouth, ME 04105 > >>>> (401) 284-1827 Business > >>>> http://www.teraspeed.com > >>>> > >>>> On Wed, Jan 7, 2015 at 4:53 PM, Boris Bakshan <bbakshan@xxxxxxxxx> > >>>> wrote: > >>>> > >>>> > Hi Andrew, > >>>> > For high rate signaling it won't matter that much if you choose 0201 > >>>> caps > >>>> > over 0402. > >>>> > For 0201, the ESL of just the capacitor alone (excluding the > artifacts > >>>> > associated with the mounting on a PCB) is in the order of 400pH. > >>>> > For shape of 0402, it is ~550pH. > >>>> > Remember that the impedance of the bulk capacitance and the series > >>>> > resistance (ESR) are negligible when it comes to high speed > signaling. > >>>> > Furthermore, you will not benefit from reducing the package but > >>>> instead > >>>> > what you should be doing is reducing the parasitic-shunt capacitance > >>>> of the > >>>> > 0402 structure (shape a void area in the plane underneath) and > target > >>>> the > >>>> > effective impedance to match your transmission line. > >>>> > > >>>> > Hope it helped.. > >>>> > Boris Bakshan. > >>>> > > >>>> > > >>>> > On Wed, Jan 7, 2015 at 8:32 PM, Andrew Holme < > >>>> andrew@xxxxxxxxxxxxxxxxxx> > >>>> > wrote: > >>>> > > >>>> > > Can anyone recommend by manufacturer's part number an 0402 size > >>>> 100nF AC > >>>> > > coupling capacitor for use in a DisplayPort Main Link at 5.4 Gbps? > >>>> Many > >>>> > > 0402 size 100nF caps have self-resonant frequencies in the tens of > >>>> MHz > >>>> > > range. We would consider 0201 devices but shy away from 01005. > >>>> The VESA > >>>> > > spec says min 70 nF to max 265 nF is the allowable range of values > >>>> for AC > >>>> > > coupling caps in the main link. I presume such large values are > >>>> required > >>>> > > because of DC wander? > >>>> > > > >>>> > > TIA > >>>> > > Andrew. > >>>> > > > >>>> > > ------------------------------------------------------------------ > >>>> > > 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 forum is accessible at: > >>>> > > http://tech.groups.yahoo.com/group/si-list > >>>> > > > >>>> > > 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 forum is accessible at: > >>>> > http://tech.groups.yahoo.com/group/si-list > >>>> > > >>>> > 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 forum is accessible at: > >>>> http://tech.groups.yahoo.com/group/si-list > >>>> > >>>> 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 forum is accessible at: > >>>> http://tech.groups.yahoo.com/group/si-list > >>>> > >>>> 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 forum is accessible at: > http://tech.groups.yahoo.com/group/si-list > > 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 forum is accessible at: http://tech.groups.yahoo.com/group/si-list 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