Hi, Arpad, This is an interesting point. Here are my $0.02 of ideas: 1) An SI tool might need to extrapolate outside the table range when setting the DC operating point, since the simulator can try any point in the entire [-Vcc,2*Vcc] range while trying to find the operating point. 2) A signal can "double" on reflection. For example, an I/O operating at 1.2V with a 0.8V swing would need to cover from 1.2-(2*0.8) to 1.2+(2*0.8), or -0.4V to 2.8V. This might be a difficult rule for the parser to check. Best regards, Lynne -----Original Message----- From: owner-ibis-users@eda.org [mailto:owner-ibis-users@eda.org] On Behalf Of Aubrey_Sparkman@dell.com Sent: Tuesday, July 12, 2005 2:26 PM To: arpad.muranyi@intel.com; ibis-users@eda.org; ibis@eda.org Subject: RE: [IBIS-Users] Re: IV curve sweep range I would like to see the sweep range linked to the allowed or expected signal swing. If the buffer can be used in a "less than perfectly terminated" topology where rising or falling overshoot outside 0 to Vcc can happen, then the sweep should be REQUIRED to cover that range. If the signal swing will always be well within 0 to Vcc, then a reduced sweep range could be allowed. my $.02 Aubrey Sparkman Enterprise Engineering Signal Integrity Team Dell, Inc. Aubrey_Sparkman@Dell.com (512) 723-3592 -----Original Message----- From: owner-ibis-users@eda.org [mailto:owner-ibis-users@eda.org] On Behalf Of Muranyi, Arpad Sent: Tuesday, July 12, 2005 3:45 PM To: ibis-users@eda.org; ibis@eda.org Subject: [IBIS-Users] Re: IV curve sweep range Hello everyone, I hope you don't mind that I brought this discussion to the IBIS and IBIS users reflectors, but I wanted to save Mike a bunch of forwarding these messages to and from me (since I don't want to sign up to the IBIS quality list at the moment). Plus the subject ultimately is a spec related one, so it may not hurt to discuss it in these lists. I agree with Tom, that those original IBIS ranges are still useful. The point I was trying to make is that IBIS REQUIRES those ranges, period. True, the parser doesn't flag models now which do not obey these requirements of the IBIS specification, but technically these models are out of spec. Also true, the model maker can always do a smaller range, and then add two points on the two ends, just to satisfy the spec requirements, but if the table has 100 points already, the model make will have to figure out which other two points they have to remove before they can add the points at the ends. Forcing the model maker to generate models for the -Vcc to 2Vcc range may reduce the quality of the model, because many points may be used up in those areas of the curves which never get used, and at times it may not be possible to generate those points due to convergence issues. The comment from Bob about the current limiting at 1A or 10A seems to be a related, but different issue. I would like to see that the IBIS specification is fixed so that it doesn't REQUIRE the -Vcc to 2Vcc range. But I agree, this may be a difficult one, because if we just eliminate the rule, we may start getting a bunch of useless models, if people don't know what ranges they should provide. So what should the spec say then? Spell out some STRONGLY RECOMMENDED rules? Thanks, Arpad ================================================================== -----Original Message----- From: ibis-quality-bounce@freelists.org [mailto:ibis-quality-bounce@freelists.org] On Behalf Of Robert Haller Sent: Tuesday, July 12, 2005 1:03 PM To: ibis-quality@freelists.org Subject: [ibis-quality] Re: IV curve sweep range Tom, Michael Agreed that sweeping over the entire range makes sense in come cases and it is excessive in others. But in order to avoid excess currents and meet the spec in its presently approved state, the committee reached consensus that limiting the current at something resonable (1Amp) below the hard ibis check ceiling of 10 Amps seemed acceptable. We want to minimize warnings from the parser and most devices (that people are simulating with IBIS) operate below an 1 amp. The point we were trying to make was when sweep terminated devices (i.e. ODT) it is important to sweep across the entire 'legal' range to avoid simulators interpolating or extrapolating. I have persoannly always had an issue with the excessivly wide range and would support a bird that reduces it to something reasonable. The rub is getting consensus in Open forum on what is reasonable with all of the exceptions.... my $.02 Bob Tom Dagostino wrote: >There are still a lot non busses that people need to simulate, not all of >them will be terminated. In addition some people simulate open connectors >which may have non terminated lines by definition. So, yes there are places >where it does not make sense to do the -Vcc to 2*Vcc range but there are >cases still where it makes sense. > >Tom Dagostino >Teraspeed(R) Labs >13610 SW Harness Lane >Beaverton, OR 97008 >503-430-1065 >tom@teraspeed.com >www.teraspeed.com > >Teraspeed Consulting Group LLC >121 North River Drive >Narragansett, RI 02882 >401-284-1827 > >-----Original Message----- >From: ibis-quality-bounce@freelists.org >[mailto:ibis-quality-bounce@freelists.org]On Behalf Of Mirmak, Michael >Sent: Tuesday, July 12, 2005 12:00 PM >To: ibis-quality@freelists.org >Subject: [ibis-quality] IV curve sweep range > > >Forwarded from Arpad Muranyi... > >- MM > >-----Original Message----- >From: Muranyi, Arpad >Sent: Tuesday, July 12, 2005 11:20 AM >To: ibis-quality-bounce@freelists.org >Cc: Mirmak, Michael >Subject: IV curve sweep range > >Hello everyone, > >I am not monitoring this email reflector, and I don't plan on >subscribing, but Mike Mirmak forwarded me a message regarding the IV >curve sweep range discussion, to which I would like to add a few >comments. I don't know what exactly is being proposed, I am just going >to mention my views here. > >The problem I see is that IBIS requires a sweep range of -Vcc to 2*Vcc. >This comes from the days when most buffers switched rail to rail and >most buses were unterminated. The reason for this range stems from the >T-line reflection theory, that the signal can double at the end of the >line. If we had a strong driver, which could drive close to the rail, >then the doubling of this signal could span a range close to -Vcc or >+2*Vcc assuming no clamping effects at the end of the line. > >The problem is that most modern buses and drivers are not operating like >this any more. For one, the signal swing doesn't go rail-to-rail any >more, and second, most buses are terminated. >Even if we ignore the termination question, the reduced-swing signaling >alone warrants a reconsideration of the IBIS rule. > >One of my favorite examples is GTL. The signal goes from a low level >between 0-0.5V to a high level of 1.5V. If I took 0-1.5, even if I >doubled the signal swing on the top and bottom, I would get a range of >-1.5V to 3.0V. Aside from the fact that the bus is terminated and we >will never see a doubling at the end of the T-line, compare this range >with what IBIS would require if this buffer was powered by 3.3V: it >would be -3.6V to 6.6V, almost double! This clearly doesn't make sense. > >I would propose that we should write a BIRD on this, and change the IBIS >specification so that it would not REQUIRE the range of -Vcc to 2*Vcc. >I am not sure what it should say yet, but I would definitely NOT favor >that the parser should be changed to check the ranges in the IBIS files >against the IBIS rules. I would much prefer to have a rule that is >based on signaling, though I have to admit that this may be somewhat >more complicated to spell out. > >Please comment, > >Arpad >=========================================================== |------------------------------------------------------------------ |For help or to subscribe/unsubscribe, email majordomo@eda.org |with just the appropriate command message(s) in the body: | | help | subscribe ibis <optional e-mail address, if different> | subscribe ibis-users <optional e-mail address, if different> | unsubscribe ibis <optional e-mail address, if different> | unsubscribe ibis-users <optional e-mail address, if different> | |or email a written request to ibis-request@eda.org. | |IBIS reflector archives exist under: | | http://www.eda.org/pub/ibis/email_archive/ Recent | http://www.eda.org/pub/ibis/users_archive/ Recent | http://www.eda.org/pub/ibis/email/ E-mail since 1993Received on Tue Jul 12 22:20:21 2005
This archive was generated by hypermail 2.1.8 : Tue Jul 12 2005 - 22:20:28 PDT