Andrey I've tried this kind of extraction also. The curves looked much more CMOS current source like than you are showing for the pulldown. It still looks like your feedback loop is in control of the driver and it is not happy with Vcm. Tom Dagostino Teraspeed Labs 9999 SW Wilshire St. Suite 102 Portland, OR 97225 USA 971-279-5325 Office 971-279-5326 FAX 503-430-1065 Cell tom@teraspeed.com www.teraspeed.com Teraspeed Consulting Group LLC 16 Stormy Brook Road Falmouth, ME 04105 401-284-1827 -----Original Message----- From: Babintsev Andrey-B13285 [mailto:B13285@freescale.com] Sent: Wednesday, September 18, 2013 12:57 PM To: tom@teraspeed.com; 'Muranyi, Arpad'; ibis-users@eda.org Subject: RE: [IBIS-Users] {Spam?} {Filename?} LVDS IBIS model - simulation issue Arpad, Tom, Thank you for your responses. I used 50 Ohm because it is MLB interface. I called it LVDS to make it clear that I was speaking about low swing differential signal. I inverted the curve polarity on the plot and made both of them ground relative to make it convenient to compare source and sink currents. In the IBIS models they were put with correct polarity, pullUp - Vdd relative, in accordance with IBIS standard and IBIS cookbook. While DC sweep on the non-inverted output was done for IV curves extraction, the inverted output was controlled by the voltage dependent voltage source v=2*Vos-v(padp), Vos ~ 1.25V (depending on conditions). Actually the issue was observed for slow and typ cases while for fast case the models are in good agreement with spice models (Vmax, Vmin, Trise, Tfall, Duty cycle matched reference waveforms). Regards, Andrey ________________________________________ From: owner-ibis-users@eda.org [owner-ibis-users@eda.org] on behalf of Tom Dagostino [tom@teraspeed.com] Sent: Wednesday, September 18, 2013 10:39 PM To: 'Muranyi, Arpad'; ibis-users@eda.org Subject: RE: [IBIS-Users] {Spam?} {Filename?} LVDS IBIS model - simulation issue Andrey I've modeled a lot of LVDS buffers in the past and this is definitely the strangest looking pulldown IV curve I've ever seen. And as Arpad says it does not at all look like a reasonable result. First, why are you loading your LVDS buffer with a 50 Ohm differential termination? Most I have seen are looking for 100Ohms. Have you inverted the pulldown's polarity? I would also expect to see the pulldown curve to have positive current between ground and Vdd. LVDS buffers have a feedback loop in them so stabilize the Vcm of the buffer. Usually this only affect the pullup IV curve but sometimes only the pulldown. This makes extracting the IV characteristics of the controlled device difficult. The feedback is fighting the curve tracer that is measuring the IV characteristics. While you are doing the IV sweep you are by definition changing Vcm and the feedback is trying to correct for this. BTW, Ip = -In. One is sourcing current and the other sinking. The sourcing should have negative currents and the sinking positive. Tom Dagostino Teraspeed Labs 9999 SW Wilshire St. Suite 102 Portland, OR 97225 USA 971-279-5325 Office 971-279-5326 FAX 503-430-1065 Cell tom@teraspeed.com<mailto:tom@teraspeed.com> www.teraspeed.com<http://www.teraspeed.com/> Teraspeed Consulting Group LLC 16 Stormy Brook Road Falmouth, ME 04105 401-284-1827 From: owner-ibis-users@eda.org [mailto:owner-ibis-users@eda.org] On Behalf Of Muranyi, Arpad Sent: Wednesday, September 18, 2013 8:31 AM To: ibis-users@eda.org Subject: RE: [IBIS-Users] {Spam?} {Filename?} LVDS IBIS model - simulation issue Andrey, I am not sure that your PD I-V curve is correct. First of all, it should have a positive sign, but its shape doesn't look right either. Can you describe how you generated these I-V curves? Thanks, Arpad ======================================================== From: owner-ibis-users@eda.org<mailto:owner-ibis-users@eda.org> [mailto:owner-ibis-users@eda.org] On Behalf Of Andrey Babintsev Sent: Wednesday, September 18, 2013 9:38 AM To: ibis-users@eda.org<mailto:ibis-users@eda.org> Subject: Re: [IBIS-Users] {Spam?} {Filename?} LVDS IBIS model - simulation issue Re-sending with updated attachment Hi, I am making IBIS models for LVDS (low voltage differential signal) IO buffer. To model an LVDS buffer I created the following test circuit, assigned the generated IBIS model to both buffers, set one buffer as inverted output. [cid:image001.png@01CEB462.5402D6F0] Problem: In the image below the YELLOW waveform is what I saw and the PURPLE waveform is what I expected to see. [cid:image002.png@01CEB462.5402D6F0] YELLOW = IBIS model PURPLE = Spice ( reference ) model Here is combined pullUp (yellow) and pullDown (green) IV curves (actually it is spice curves from which IV tables were derived): [cid:image003.png@01CEB462.5402D6F0] Lets assume that simulator tries to find operational point for LVDS buffer driving high. Then for the non-inverted buffer a point (Vp, Ip) will be taken from pullUp table, for the inverted buffer a point (Vn, In) from pullDown table, the following conditions should be satisfied: Ip = In, Vp - Vn = Ip*50. On the image above it is seen that there are 2 points satisfying the condition: Vp =~ 1.45V and Vp =~ 2.2V. Seems that something like that happened during simulation and simulator got to the second point. (BTW, there are 60 points for the region 0.9 - 1.6V and 10 points for the region 1.6-2.5V in the IBIS model IV tables). Question: What can be done to resolve this situation? Should IV tables or another IBIS structures be modified and how? I am making ibis ver. 4.2 models. Thanks, Andrey On 09/18/2013 06:33 PM, Andrey Babintsev wrote: Warning: This message has had one or more attachments removed (fail.min.png). Please read the "EDA-Attachment-Warning.txt" attachment(s) for more information. Hi, I am making IBIS models for LVDS (low voltage differential signal) IO buffer. To model an LVDS buffer I created the following test circuit, assigned the generated IBIS model to both buffers, set one buffer as inverted output. [cid:part1.07090501.06090707@freescale.com] Problem: In the image below the YELLOW waveform is what I saw and the PURPLE waveform is what I expected to see. [cid:part2.00080404.04020708@freescale.com] YELLOW = IBIS model PURPLE = Spice ( reference ) model Here is combined pullUp (yellow) and pullDown (green) IV curves (actually it is spice curves from which IV tables were derived): [cid:part3.05090108.02030307@freescale.com] Lets assume that simulator tries to find operational point for LVDS buffer driving high. Then for the non-inverted buffer a point (Vp, Ip) will be taken from pullUp table, for the inverted buffer a point (Vn, In) from pullDown table, the following conditions should be satisfied: Ip = In, Vp - Vn = Ip*50. On the image above it is seen that there are 2 points satisfying the condition: Vp =~ 1.45V and Vp =~ 2.2V. Seems that something like that happened during simulation and simulator got to the second point. (BTW, there are 60 points for the region 0.9 - 1.6V and 10 points for the region 1.6-2.5V in the IBIS model IV tables). Question: What can be done to resolve this situation? Should IV tables or another IBIS structures be modified and how? I am making ibis ver. 4.2 models. Thanks, Andrey -- This message has been scanned for viruses and dangerous content by MailScanner<http://www.mailscanner.info/>, and is believed to be clean. -- This message has been scanned for viruses and dangerous content by MailScanner<http://www.mailscanner.info/>, and is believed to be clean. -- This message has been scanned for viruses and dangerous content by MailScanner<http://www.mailscanner.info/>, and is believed to be clean. -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. -------------------------------------------------------------------- |For help or to subscribe/unsubscribe, e-mail mikelabonte@eda-stds.org |or ibis-request@eda-stds.org | |IBIS reflector archives exist under: | | http://www.eda-stds.org/ibis/email_archive/ Recent | http://www.eda-stds.org/ibis/users_archive/ Recent | http://www.eda-stds.org/ibis/email/ E-mail since 1993Received on Wed Sep 18 14:20:46 2013
This archive was generated by hypermail 2.1.8 : Wed Sep 18 2013 - 14:21:11 PDT