On Thu, 24 May 2007 15:26:15 +0100, "Symon" <symon_brewer@hotmail.com> wrote:>"John_H" <newsgroup@johnhandwork.com> wrote in message >news:cUg5i.9206$ix.312@trndny01... >> stefan.elmsted@gmail.com wrote: >>> Hi >> >> On the transmitter, you want a 100 ohm to 173 ohm impedance match so the >> transmitter sees 100 ohm but the transmission line sees 173 ohm. You'll >> need a differential termination on the transmitter side of this network >> and two series resistors to the ribbon cable. The signal amplitude will >> again be reduced. >> >You don't _need_ to match both transmitter and receiver. One or the other is >good enough, provided the path between the driver and the cable has the same >impedance as the cable, or this path is short. Cf. ECL logic, low output >impedance, but can drive a properly terminated diff. pair. LVDS outputs are >matched to the line to get a belt 'n' braces approach to reduce reflections, >but it's not necessary to match the transmitter to the line. > >Here's an app note which describes the output structures. >http://www.maxim-ic.com/appnotes.cfm/an_pk/291 > >HTH, Syms. >National's CMOS structure is different, http://www.national.com/appinfo/lvds/files/lvds_ch1.pdf more of a real current source. I'd guess that the Maxim is the oddball here. The National and Fairchild transmitters I've played with will go rail-to-rail if not terminated. I haven't tried an unterminated Xilinx lvds driver; their receivers are pretty good r-r comparators. John
LVDS termination scheme to nonstandard ribbon cable
Started by ●May 24, 2007
Reply by ●May 24, 20072007-05-24
Reply by ●May 24, 20072007-05-24
"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in message news:0ppb531bd38oj99t3dne4fei8n4lrotm00@4ax.com...> > National's CMOS structure is different, > > http://www.national.com/appinfo/lvds/files/lvds_ch1.pdf > > more of a real current source. I'd guess that the Maxim is the oddball > here. The National and Fairchild transmitters I've played with will go > rail-to-rail if not terminated. I haven't tried an unterminated Xilinx > lvds driver; their receivers are pretty good r-r comparators. > > JohnThe FPGA structures appear to be what are important to this conversation. There are transmitters that are true current sources both with and without internal 100 ohm parallel terminations. There are transmitters that present voltage drivers with (series) source impedances that roughly match the current-source approach. For anyone designing with LVDS, appropriate reading of the data sheet information on the transmitter is important. For a proper LVDS connection, the transmitter needs to appear to be a 100 ohm differential source. Different manufacturers are happy to deviate slightly from the originally proposed driver structure to present something with equivalent characteristics when properly configured whether this means plug&play, an external parallel termination, or a 3-resistor network to "look" like the equivalent source. Without the 100 ohm equivalent transmit impedance, any reflections from the receiver or other impedance mismatches will reflect back toward the receiver rather than be absorbed at the source. - John_H
Reply by ●May 24, 20072007-05-24
John, The LVDS standard specifies that the transmitter has a 100 ohm resistive termination to absorb reflections. Austin
Reply by ●May 24, 20072007-05-24
"austin" <austin@xilinx.com> wrote in message news:f34sfg$62n1@cnn.xilinx.com...> John [Larkin], > > The LVDS standard specifies that the transmitter has a 100 ohm resistive > termination to absorb reflections. > > AustinSome LVDS transmitters have no (design) impedance. This is what I [John_H] was referring to in another post. Those transmitters that are implemented as current sources require that the transmitter have the differential impedance required to absorb the reflections from the LVDS line and receiver. The transmitter data should always be referenced to make sure the impedance is integrated into the transmitter, is required externally, or if the driver needs to be "transformed" to an equivalent source such as the Xilinx BLVDS. - John_H
Reply by ●May 25, 20072007-05-25
On Thu, 24 May 2007 12:54:13 -0700, "John_H" <newsgroup@johnhandwork.com> wrote:>"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in message >news:0ppb531bd38oj99t3dne4fei8n4lrotm00@4ax.com... >> >> National's CMOS structure is different, >> >> http://www.national.com/appinfo/lvds/files/lvds_ch1.pdf >> >> more of a real current source. I'd guess that the Maxim is the oddball >> here. The National and Fairchild transmitters I've played with will go >> rail-to-rail if not terminated. I haven't tried an unterminated Xilinx >> lvds driver; their receivers are pretty good r-r comparators. >> >> John > >The FPGA structures appear to be what are important to this conversation. >There are transmitters that are true current sources both with and without >internal 100 ohm parallel terminations. There are transmitters that present >voltage drivers with (series) source impedances that roughly match the >current-source approach. > >For anyone designing with LVDS, appropriate reading of the data sheet >information on the transmitter is important. For a proper LVDS connection, >the transmitter needs to appear to be a 100 ohm differential source.Everywhere I look, I see unterminated transmitters: http://zone.ni.com/devzone/cda/tut/p/id/4441 http://www.interfacebus.com/Design_Connector_RS644.html http://www.fairchildsemi.com/ms/MS/MS-547.pdf http://www.analog.com/UploadedFiles/Application_Notes/42118600205599850975382134073431740717454123180480718AN586.pdf http://www.ams.aeroflex.com/ProductFiles/Presentations/LVDSOverview9-04.pdf http://spacewire.esa.int/content/TechPapers/documents/SpaceWire%20Standard%20%20ISWS%202003.pdf This makes sense: a 3.5 mA current source (transmitter) drops the proper 350 mV across the 100 ohm receive termination. If the transmitter also terminated in 100 ohms, you'd net half that swing. That's consistant with my observation that unterminated transmitters slew rail-to-rail on both pins.>Different manufacturers are happy to deviate slightly from the originally >proposed driver structure to present something with equivalent >characteristics when properly configured whether this means plug&play, an >external parallel termination, or a 3-resistor network to "look" like the >equivalent source. Without the 100 ohm equivalent transmit impedance, any >reflections from the receiver or other impedance mismatches will reflect >back toward the receiver rather than be absorbed at the source.But if the receiver terminates properly, there will be no reflections. John
Reply by ●May 25, 20072007-05-25
"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in message news:u4pb535aii2c97o28bpi2812ljapqmh9pb@4ax.com...> On Thu, 24 May 2007 07:23:01 -0700, austin <austin@xilinx.com> wrote: > > > Does an LVDS transmitter have an impedance? The ones I've played with > seem to behave like current sources; unloaded, the diff outputs swing > (slowly!) almost rail-to-rail. That said, presenting the transmitter > with an equivalent 100 ohm net diff load will normalize the swing to > standard levels and speed things up a bit as compared to letting them > see a 173 or whatever diff load. > > John >Hi John, Have a look at the link I posted yesterday. This one:- http://www.maxim-ic.com/appnotes.cfm/an_pk/291 It shows the output structure for LVDS. The (12mA) current source has high impedance, so the output impedance is determined by the resistors show. I think the LVDS transmitters you mention must have some other type of structure. Can you post what they were? HTH, Syms.
Reply by ●May 25, 20072007-05-25
"Symon" <symon_brewer@hotmail.com> wrote in message news:f364ua$6pj$1@aioe.org...> "John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in > message news:u4pb535aii2c97o28bpi2812ljapqmh9pb@4ax.com... >> > Hi John, > Have a look at the link I posted yesterday. This one:- > http://www.maxim-ic.com/appnotes.cfm/an_pk/291 > It shows the output structure for LVDS. The (12mA) current source has high > impedance, so the output impedance is determined by the resistors show. I > think the LVDS transmitters you mention must have some other type of > structure. Can you post what they were? > HTH, Syms. > >Whoops, forget that, I just read the rest of the thread. Cheers, Syms.
Reply by ●May 25, 20072007-05-25
"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in message news:qjkc539kqeipvuab80a6u3ad9vl2aan2sr@4ax.com...> On Thu, 24 May 2007 12:54:13 -0700, "John_H" > <newsgroup@johnhandwork.com> wrote: > > > But if the receiver terminates properly, there will be no reflections. > > John > >Hi John, Right. My point exactly. Also, many thanks for those links, looks like I found, and posted a link to, the only manufacturer that shows LVDS outputs including the resistors. However, if you revisit the pictures in your links, I think you'll see that they paint a simplified picture. If you go to the National link you posted, it's clear that the lower MOSFET isn't just a short to ground when it's on. Otherwise the device would not be meeting the output common mode voltage spec. There's maybe(?) a resistor in there somewhere, and this provides the required output impedance. The Analog device links show a current source top & bottom, which would suggest the Vcm could be anywhere. So, I think we need to view these diagrams as not the full picture. However, I agree with you that it seems most manufacturers aren't all that concerned with output impedance, which is interesting. Cheers, Syms.
Reply by ●May 25, 20072007-05-25
Symon, I agree. (Twice? I must be slacking...) To meet the IEEE/ANSI specification, there is a transmit termination. If not internal, it would be specified as external. If not shown, it still may be present. Diagrams in data sheets are often simplified. Austin
Reply by ●May 26, 20072007-05-26
On Fri, 25 May 2007 09:30:58 +0100, "Symon" <symon_brewer@hotmail.com> wrote:>"John Larkin" <jjlarkin@highNOTlandTHIStechnologyPART.com> wrote in message >news:qjkc539kqeipvuab80a6u3ad9vl2aan2sr@4ax.com... >> On Thu, 24 May 2007 12:54:13 -0700, "John_H" >> <newsgroup@johnhandwork.com> wrote: >> >> >> But if the receiver terminates properly, there will be no reflections. >> >> John >> >> >Hi John, >Right. My point exactly. > >Also, many thanks for those links, looks like I found, and posted a link to, >the only manufacturer that shows LVDS outputs including the resistors. >However, if you revisit the pictures in your links, I think you'll see that >they paint a simplified picture. If you go to the National link you posted, >it's clear that the lower MOSFET isn't just a short to ground when it's on. >Otherwise the device would not be meeting the output common mode voltage >spec. There's maybe(?) a resistor in there somewhere, and this provides the >required output impedance. The Analog device links show a current source top >& bottom, which would suggest the Vcm could be anywhere. > >So, I think we need to view these diagrams as not the full picture. However, >I agree with you that it seems most manufacturers aren't all that concerned >with output impedance, which is interesting. >Cheers, Syms. >Yup, the common-mode setting thing is an obvious omission. John
