Re: [oc] Output loading in digital circuit

[Allan Herriman <allan_herriman.hates.spam@agilent.com>]

Kenneth,

The 50pF spec comes from the need to have *some* capacitance (you can
never have zero) and 50pF isn't too far from typical values you get in
real circuits.
Some older devices were specified with a 15pF load.  This made the part
look better (faster) but it was difficult to make a test jig that had a
capacitance that low.  So 50pF has sort of been settled on as an
industry standard.

In the days of TTL the input of the next device had a resistive
component (with a few diodes, yada yada) and if you bother to look in an
old data book, you'll find the test load specified so that it looks like
the input of a TTL gate (or possibly 10 of them in parallel, etc.).  As
another poster mentioned, in modern CMOS circuits there is no DC load,
so it seems reasonable to test without a load resistor.  ... that is,
until you realise that you need some sort of analog probe to measure the
timing.  I don't design chip testers (a different division of Agilent
does those), but I guess the 500ohm spec allows you to have cheap 10:1
passive probes.  These are made with a 450ohm resistor between the pin
under test and some 50ohm cable.  The cable is terminated at the other
end in 50ohm.

500ohms is effectively open circuit when compared with the 50pF load at
any reasonable clock frequency.

My apologies for the off topic post.  I'm all in favour of list
moderation to stop threads like this one, BTW.

Regards,
Allan.

Kenneth Hung wrote:
> 
> Hi John
> 
> Do u know why old TTL logic has hundreds of ohms in R ?
> I think I really want to know what information decides the R value
> and C value.
> 
> Say in typical CMOS 3.3V spec, you can see R=500ohm in parallel
> with C=50pF as an example. So I really want to know how those
> values come from ? From what theory or information to decide such
> values ? It's not from God, rite ? haha..
> 
> Kenneth
> 
> ----- Original Message -----
> From: "John Sheahan" <jrsheahan@optushome.com.au>
> To: <cores@opencores.org>
> Sent: Tuesday, April 02, 2002 3:39 PM
> Subject: Re: [oc] Output loading in digital circuit
> 
> > On Tue, Apr 02, 2002 at 11:48:07AM +0800, Kenneth Hung wrote:
> > > Hi Steve
> > >
> > > Ur answer might be correct in some sense in pF.
> > > But do u know why there's an R=500 or R=1kohm in parallel
> > > with the C ? I can't see no relation why there's a R and why is such
> > > value ? For a typical transmission line, even the termination resistor
> > > is 50ohm.
> > >
> >
> > Rl in the hundreds of ohms used to be specified for old TTL logic,
> > back when it actually consumed static input current.
> >
> > You need to to check DC limits to ensure you reach Vil and Vih (plus
> > noise margin)
> >
> > I have not seen Rl spec'ed for a while.  Can you point
> > me at an example?
> >
> > john
> > --
> > To unsubscribe from cores mailing list please visit
> http://www.opencores.org/mailinglists.shtml
> >
> --
> To unsubscribe from cores mailing list please visit http://www.opencores.org/mailinglists.shtml
--
To unsubscribe from cores mailing list please visit http://www.opencores.org/mailinglists.shtml