RE: [IBIS-Users] RE: [IBIS] I-V and V-t curve mismatch

From: Tom Dagostino <tom_at_.....>
Date: Tue Oct 23 2007 - 12:24:48 PDT
All

 

NC is not always the best choice for analog pins.  We often use a terminator
model for analog pins.  It allows simulators to at least check for crosstalk
into analog nets.  And some simulators will complain if a net is connected
in a system but has no model associated with it.  The terminator model can
be just the key word, the voltage range and no other elements.  Having a
terminator model present for true NC pins also allows the stub effect of a
pin to be simulated if a net goes through a NC and connects to it.  For
example some static protections device like the ones from CMD recommend
routing through the footprint and picking up NC pins in the path between the
active pin and the input to the next IC in the net.  If a NC model is used
then some simulators will either not allow that connection or not simulate
the stub effect of the package.

 

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 

From: owner-ibis@server.eda.org [mailto:owner-ibis@server.eda.org] On Behalf
Of Muranyi, Arpad
Sent: Tuesday, October 23, 2007 10:13 AM
To: ibis@server.eda.org; ibis-users@server.eda.org
Subject: RE: [IBIS-Users] RE: [IBIS] I-V and V-t curve mismatch

 

If you use legacy IBIS, then yes, you will have to

put NC for any analog buffers.  But if you are using

IBIS with *-AMS, you can also make analog models and

list them on the [Pin] list.

 

Arpad

=====================================================

 

  _____  

From: owner-ibis-users@server.eda.org
[mailto:owner-ibis-users@server.eda.org] On Behalf Of Sudarshan H N
Sent: Tuesday, October 23, 2007 10:06 AM
To: Todd Westerhoff
Cc: ibis@server.eda.org; ibis-users@server.eda.org
Subject: Re: [IBIS-Users] RE: [IBIS] I-V and V-t curve mismatch

Hi All,

Thanks Tom ,Todd Scott and Arpad for your valuable answers. Arpad, with your
clear explanation i am able to solve all the problems i had in my modeling. 

One general question i have to the forum is , in the IBIS website is all the
mails are stored in some central location. Because i was accessing IBIS
forum mails from my other email id 3 months back. Now i am accessing from
other email id. 

I think sometime back there was a question on analog pads IBIS model. Do i
need to put a NC for these pads in pin mapping section ?

Regards
Sudarshan

On 10/23/07, Todd Westerhoff <twesterh@sisoft.com> wrote: 

Oops,

 

You're right: -VDDQ to 2*VDDQ.

 

Todd.

 


Todd Westerhoff
VP, Software Products
SiSoft
6 Clock Tower Place, Suite 250
Maynard, MA 01754
(978) 461-0449 x24
twesterh@sisoft.com
www.sisoft.com

  _____  

From: owner-ibis@eda.org [mailto: owner-ibis@eda.org
<mailto:owner-ibis@eda.org> ] On Behalf Of Muranyi, Arpad
Sent: Monday, October 22, 2007 2:01 PM
To: ibis@eda.org; ibis-users@eda.org
Subject: RE: [IBIS-Users] RE: [IBIS] I-V and V-t curve mismatch

 

While both Tom and Todd are suggesting the right thing,

I think both need some correction, or clarification...

 

Todd wrote "-2*VDD to 2*VDD" in which the first -2*VDD

should have read -VDD.

 

Tom's comment about using typical VDD is correct for the

range calculation but you need to be careful an not

apply that suggestion to the VDD of the power you

apply to the device under test and the VDD-relative

calculations (in case you don't measure it VDD relative

directly).

 

So to sum it up, my numbers in my first reply were

calculated as follows:

 

For the pulldown the sweep range is always:

-VDD_typ to 2*VDD_typ  =  -3.3 to 6.6,

regardless of which case you are generating,

i.e. whether the supply voltage to the chip 

is typ=3.3, min=3.0, or max=3.6 volts (assuming

that the chip's pulldown is connected to GND

or 0 volts).

 

For the pullup, which is swept VDD relative,

the numbers for the sweep ranges in GND relative

orientation will work out this way:

 

-VDD_typ to 2*VDD_typ with respect to VDD_typ = 6.6 to -3.3

-VDD_typ to 2*VDD_typ with respect to VDD_min = 6.3 to -3.6

-VDD_typ to 2*VDD_typ with respect to VDD_max = 6.9 to -3.0

 

Note that this numbers are only this confusing because

you are looking at the sweep ranges in a GND relative

way, while you are really doing the same rule as above

for the pulldown with respect to VDD.

 

Of course the direction of the sweep is irrelevant, as

long as the device doesn't act up.  Some times devices

can do funny things depending on which direction you

do the sweep.  Also, to help the simulator to find an

operating point I suggest that you start the seep at

the same rail voltage the device is driving to (i.e.

VDD for logic high or GND for logic low) and then do

the sweep you need for the IBIS model.  There is a

better chance to get the simulator converge that way...

 

I hope this helps,

 

Arpad

===========================================================

  _____  

From: owner-ibis-users@server.eda.org [mailto:
<mailto:owner-ibis-users@server.eda.org>  owner-ibis-users@server.eda.org]
On Behalf Of Todd Westerhoff
Sent: Monday, October 22, 2007 10:09 AM
To: ibis@server.eda.org; ibis-users@server.eda.org
Cc: tom@teraspeed.com
Subject: [IBIS-Users] RE: [IBIS] I-V and V-t curve mismatch

I agree with Tom.

 

The choice of -2*VDD to 2*VDD is somewhat arbitrary . it's simply meant to
ensure that you have valid V/I data for any voltage that the part will be
exposed to.  Thus, you don't need to adjust the voltage ranges based on the
MIN and MAX cases.  For most technologies, if you have overshoot that's
equal to the supply voltage, you have bigger problems than model accuracy !

 

Todd.

 


Todd Westerhoff
VP, Software Products
SiSoft
6 Clock Tower Place, Suite 250
Maynard, MA 01754
(978) 461-0449 x24
twesterh@sisoft.com
www.sisoft.com

  _____  

From: owner-ibis@eda.org [mailto: owner-ibis@eda.org
<mailto:owner-ibis@eda.org> ] On Behalf Of Tom Dagostino
Sent: Monday, October 22, 2007 12:11 PM
To: 'Sudarshan H N'; 'Mirmak, Michael'
Cc: ibis@eda.org; ibis-users@eda.org
Subject: RE: [IBIS] I-V and V-t curve mismatch

 

Model all three corners from -Vdd to 2*Vdd where Vdd is the typical voltage.

 

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 

From: owner-ibis@server.eda.org [mailto: <mailto:owner-ibis@server.eda.org>
owner-ibis@server.eda.org] On Behalf Of Sudarshan H N
Sent: Monday, October 22, 2007 7:43 AM
To: Mirmak, Michael
Cc: ibis@server.eda.org; ibis-users@server.eda.org
Subject: Re: [IBIS] I-V and V-t curve mismatch

 

Hi Michael & All,

Can you people answer to my question regarding the voltage ranges in MIN and
MAX case as explained below. As Arpad mentioned in one of the earlier mails
the range for MIN and MAX case is

typ:  -3.3 to 6.6

min:  -3.6 to 6.3

max:  -3.0 to 6.9

How do we select this range for MIN and MAX case. Is there any document
which explains the selection criteria for these corners ? If there is no
logic , then how can we decide if the supply is 1.8 V supply or 2.5V supply?

Let me know your answers.

Regards
Sudarshan

 

On 10/19/07, Sudarshan H N < hn.sudarshan@gmail.com> wrote:

Hi Michael, 

Thanks for your detailed reponse. I made a small mistake in putting the
voltage ranges as you mentioned. As i was writing a new tool , at this point
of time i didnt give much  attention to printing correct voltage range , and
i was only worried about the correct extraction of V-t and I-V tables.

I have considered the formula you have mentioned for calculating
Vcc-relative Pullup and Powerclamp tables. I changed the voltage range to
(3.3, 3.0 and 3.6 which is the actual case) and now i am able to get rid of
all these errors.

But still i have one query regarding the values for I-V curves when it is in
the Min or MAX corner. I asked about this in my previous mail. Lynne and
Arpad gave some answers  but that didnt solve my problem.

The problem with MIN and MAX corner is , what is the voltage range we need
to consider for MIN and MAX case . For example if we consider 3.3 V supply,
the typical range would be -3.3 to 6.6 and this is the voltage range in the
IBIS file for all 3 corners. As per the cookbook we need to consider voltage
range -Vcc to +2Vcc and hence for the MIN case the range will be -3.0 to
+6.0. So in that case what current values we need to put for the voltage
range -3.3 to -3.0 and +6.0 to +6.6 whose values will be missing in MIN
corner simulations.

I hope i have explained the problem correctly and let me know your answers.

Thanks & Regards
Sudarshan

 

On 10/18/07, Mirmak, Michael <michael.mirmak@intel.com
<mailto:michael.mirmak@intel.com>  > wrote:

Sudarshan,

 

Thanks for the message.  The major issue appears to be problems in
generating the Vcc-relative information, but some general supply issues are
cropping up too.  Assuming this is just a regular I/O buffer without
internal terminations...

 

1) I would expect the Pullup I-V tables to pass through the zero V, zero I
axis intercept.  Yet they do not, with the max and min data shifted by ~0.3
V above and below 0 V, respectively.  This suggests that the math used to
generate the tables may be incorrect.  Remember that, if the data at Vout is
collected ground-relative, you can make the power clamp and pullup table
data in Vcc-relative by the formula (Vtable_corner = Vcc_corner -
Vout_corner), where xxx_corner refers to a value collected for typ, min or
max.  For example, if Vcc_min is 4.5 V, data collected ground-relative for
0.0 at the output would be entered in the table for the 4.5 V row.

 

The supply voltages and the Pullup I-V axis intercepts (when plotted
ground-relative) don't agree, which implies a problem with the supply listed
for the [Voltage Range] keyword not matching either the real circuit or the
values used to generate the Vcc-relative tables.

 

2) The voltage for the buffer is 5 V +/- 0.5, but the V-t fixture is 3.3 V
+/- 0.3 V.  This complicates the math (it gets very hard for me to calculate
the intercepts if they don't match, at least before my first cup of coffee),
but it also suggests that the buffer V-t and I-V data sets were collected
using different settings or otherwise using different conditions.  I would
recommend checking the actual fixtures used for the V-t extraction; matching
the fixture voltages to the voltage supply would make checking easier.

 

Calculating the right I-V intercepts vs. V-t levels will be easier once
these issues are addressed.

 

- Michael Mirmak

  Intel Corp.

  Chair, EIA IBIS Open Forum

 

  _____  

From: owner-ibis@server.eda.org [mailto:owner-ibis@server.eda.org] On Behalf
Of Sudarshan H N
Sent: Thursday, October 18, 2007 2:16 AM
To: ibis@server.eda.org; ibis-users@server.eda.org
Subject: [IBIS] I-V and V-t curve mismatch

Hello Experts,

I am finding problem with one of my IBIS model in I-V and V-t curve
mismatch. Please find the attached model for the refernence.

ERROR - Model dummy: The [Rising Waveform]
      with [R_fixture]=80 Ohms and [V_fixture]=0V
      has TYP column DC endpoints of  0.01V and  2.66v, but
      an equivalent load applied to the model's I-V tables yields
      different voltages (-0.00V and  3.90V),
      a difference of  0.25% and 31.85%, respectively.

I wrote a new tool to generate IBIS models and i am not able to make out
where it is going wrong.
I actually tried to manually calculate the currents at the points , that it
has reported in the above error.

As it is a rising waveform having a R_fixture to ground  the current at
2.66v (steady state) would be 2.66v/50 = 33.25mA.

I looked at the voltage corresponding to 33.25mA in pullup table and i found
it is coming around 0.66v. There is a mismatch in the way i am caluculating
also. But i observed a different voltage of 0.66v compared to what ibischk4
has reported i.e, 3.9v. 

Let me know is the way i am cross checking is correct or not ? If not let me
know the exact procedure.

Also let me know what might have been gone wrong in the curves. From the
shape of the curves i am not seeing any problem with the way i have
generated. 

Please let me know your answers as soon as possible.

Note : Please dont consider any other values apart from the curves like Ramp
, C_comp etc as i have dummy values for these place holders.

Thanks & Regards
Sudarshan

-- 
This message has been scanned for viruses and 
dangerous content by MailScanner , 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 <http://www.mailscanner.info/> , and is 
believed to be clean. 


-- 
This message has been scanned for viruses and 
dangerous content by  <http://www.mailscanner.info/> MailScanner, 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  <http://www.mailscanner.info/> MailScanner, 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 majordomo@eda-stds.org
|with 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 e-mail a request to ibis-request@eda-stds.org.
|
|IBIS reflector archives exist under:
|
|  http://www.eda-stds.org/pub/ibis/email_archive/ Recent
|  http://www.eda-stds.org/pub/ibis/users_archive/ Recent
|  http://www.eda-stds.org/pub/ibis/email/         E-mail since 1993
Received on Tue Oct 23 12:27:42 2007

This archive was generated by hypermail 2.1.8 : Tue Oct 23 2007 - 12:28:13 PDT