The following are the classes of lines I often deal with (excluding
1) Short point to point, one 3-10 pf load, 1-3" unterminated and
I can use weak drivers for this say around 4 ma.
2) Long point to point, one 3-10 pf load, 3-25", usually tune for
speed. Ergo, I perform trade off analysis as to what termination scheme
is required. Usually series termination or no termination wins.
3) Short point to point, 3 to 16 3-10 pf loads(i.e. rebuffering), 1-3"
4) Long point to point, 3 to 16 3-10 pf loads(i.e. rebuffering),
3-25", usually tune for speed. Ergo, trade off analysis for termination
schemes is required.
5) Driving memories like devices. Up to 36 loads, 5-14" long. Usually
get I get stuck with unterminated lines but if I have a choice, I go for
a terminated one.
6) Spaghetti runs, If I don't care about speed, I go for as weak as I
7) Spaghetti runs. If I do care about speed I've got my work cut out. I
guess its called engineering :-)
8) Lines that need to be wired or'ed. As strong a drive as I can muster
if I'm looking for < 10 ns release times.
Usually we don't have many of these.
So what do I want in driver?
For source terminated transmission, ( I guess you can call it reflected
wave) I want to match the transmission line. It doesn't have to be an
exact match as long as I have margin from reflections. The gotcha is
usually the weak silicon case. I'll usually lower the driver impedance
to where I have a little margin but and still maximize speed. So TI
LVT162244 type of drivers do well for #2, #3, #4. i.e. some where around
40-65 ohms seen by the reflected wave. Symmetric drive is more desirable
Terminated drivers are another story. They offer the most in speed for
loaded lines. So for these I look for as strong and fast of a driver I
can have with out utterly being killed by SSO and power dissipation.
Oh, also we can't forget a fast buffer delay and low pin to pin skews.
... Rich Mellitz
>From: D. C. Sessions[SMTP:email@example.com]
>Sent: Tuesday, November 18, 1997 11:04 AM
>Subject: [SI-LIST] : Preferred PWB impedances
>I had an interesting meeting late last week. VLSI is
>drafting the specs for our newest I/O library and rather
>than wet-finger it, we're asking for a little help in
>making them meet real-world needs.
>Here's the question for the floor:
>If you could have only four 3.3v CMOS driver types, what
>would you have. (Aside from HSTL, SSTL, etc.) One obvious
>candidate is a reflected-wave driver for 50-ohm nominal
>lines; that comes to about 10mA at 350mV from the rails
>under worst case conditions. A 65-ohm reflected-wave
>driver would run about 25% lighter, or about 8mA.
>Incident-wave is another matter entirely.
>SO! the floor is open for nominations. Keep in mind that
>excessively strong drivers are both inherently slower and
>more vulnerable to SSO degradation.
>D. C. Sessions