Re: Some questions

From: Andy Ingraham <ingraham@wrksys.ENET.dec.com>
Date: Mon Jul 28 1997 - 05:37:28 PDT

Hany Gad asks:

> 1- In the simulator you have two curves one for the Pull-up, Pull-down
> and the other for the Clamp diodes, these two curves are on the same
> pin I mean you can not get each of them individually so how can you
> check these curves with the actual Lab measurements. ?

The IBIS spec provides these four sets of curves for full flexibility,
but there are situations where not all four are either measurable, or
necessary. In that case, one or more of the curves are zero.

For an I/O (bidirectional or tri-stateable) pin, you can measure the
clamp curves when the pin is in the OFF state. Then measure the
characteristics in the HIGH and LOW states, and subtract the OFF state
curves to arrive at the Pull-up and Pull-down curves.

> 2- When You made these measurements in the LAB ,if you have the IC pin
> always high when you apply the VCC to that pin, how can you vary the
> voltage to measure the current variations?

Why would you apply VCC to that pin? Is this a power pin, or a signal
pin? I/V curves are needed for signal pins, not power pins.

If your signal pin always drives high (to VCC), then just connect a
variable voltage source to it and measure the current vs. voltage.

If this is a signal input pin that is "always" pulled high when the IC
is used, then you can either skip it (on the assumption that it "never"
needs to be simulated), or treat it just like any other input pin
(i.e., sweep its voltage and measure the current). Remember that IBIS
models show the analog behavior of signal pins without regard to their
purpose or whether the chip works when a pin is brought low or high.

> 3- Are you sure that the voltage change between -Vcc to 2Vcc will not
> harm the IC pin??

It may. But that's not the point (of IBIS modeling). The point is
to provide expected I/V curves for what the current would be if the
voltage on a pin momentarily reached these voltages.

If you can, quickly sweep the voltage and gather data without killing
the chip. Sweeping the voltage quickly reduces the chances of damaging
the chip. It also better mimics what happens when an overshoot or
undershoot signal hits the IC pin, which usually don't last more than
a few nanoseconds. DC measurements generally are not a good idea.

But not every point in the IBIS curves need to be measured. If you
can only measure from -2V to Vcc+2V before something strange starts
happening in your IC, then take just the good data and extrapolate it
to the full -Vcc to 2Vcc range.

Regards,
Andy
 
Received on Mon Jul 28 05:45:37 1997

This archive was generated by hypermail 2.1.8 : Fri Jun 03 2011 - 09:52:29 PDT