Subject: Re: [IBIS-Users] Clamp curve subtraction
From: Ingraham, Andrew (a.ingraham@ieee.org)
Date: Fri Jun 27 2003 - 08:36:04 PDT
If this is an output-only pin, then I would agree with you that overshoot is
probably not that significant (although light loads might lead to enough
ringing to activate the clamps). But in this case I believe you would omit
the clamp tables anyway; i.e., your model would have Pullup and Pulldown
tables which include the clamp diodes, and no separate Clamp tables; so
clamp subtraction would not be an issue.
If the pin can be tri-stated (whether or not it is an I/O), and the net to
which this device is attached can be driven by another device on the net,
then the clamps ARE significant and should be modeled relatively well.
But, if the non-monotonicity starts happening only when your outputs are
well into the clamping region, then it could be argued that the accuracy
"out there" (especially when driving) is less important, since the device
should spend very little, if any, time that far out along the clamp curves.
At -3V, or at Vcc+3V, accuracy is less important than just being in the
ballpark with a reasonable slope to pull the simulator back to the "normal"
operating region.
Also, if the non-monotonicity is slight and due to measurement noise,
another option to consider is to smooth out the measured data, perhaps by
reducing the number of points.
Personally, I really don't care very much if subtracting the clamps causes
the raw Pullup and Pulldown tables to become non-monotonic. Monotonicity is
not an IBIS requirement! When the IBIS model is used by a simulator, and
the simulator adds the Clamp tables back to the Pullup and Pulldown tables
before using them, the non-monotonicity goes away and the simulator engine
never sees it.
It's just that the IBIS syntax checker doesn't add the tables back together
(I assume in part because it would need to do some interpolation or spline
curve approximation), so it looks only at the raw curves, which could be
non-monotonic. It calls it out as a warning ... not an error. Don't get
too hung up on that.
Regards,
Andy
----- Original Message -----
From: "Derwin Mattos" <dhm@cypress.com>
To: <ibis-users@eda.org>
Sent: Thursday, June 26, 2003 02:08 PM
Subject: [IBIS-Users] Clamp curve subtraction
> IBIS developers/users,
>
> I'm attempting to work through the non-monotonic
> issues raised in the standard.
>
> Can someone tell me if I'm missing something?
>
> The IBIS standard recommends the subtraction of clamp curves from
> pull-up and pull-down curves. This can cause "non-monotonic"
> warning messages to pop up during a syntax check.
>
> When I want to replace my clamp curves with curves
> measured from silicon (to improve clamp accuracy), the
> combined curve has the potential of being non-monotonic
> because the pull-down/pull-up curve has the original clamp
> subtracted from it.
>
> I'm thinking that double-counting clamps at the outputs
> is not a significant issue (most signal overshoot occurs
> at receiving-ends, not driving-ends), so would like to NOT do the
> clamp subtraction. A simple note explaining why would
> be included in the model.
>
> My preference is for input clamp curve accuracy, not output driver
> clamp curve accuracy.
>
> Are there any write-ups on this particular issue?
>
> As an IBIS user, what is your preference? Do you require that
> the clamp curves be subtracted? Why? Is it a technical reason?
>
> Thanks so much for any feedback/guidance. It would be much
> appreciated.
>
> Regards,
> Derwin
>
|------------------------------------------------------------------
|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 1993
This archive was generated by hypermail 2b28 : Fri Jun 27 2003 - 09:20:51 PDT