Maker Pro
Maker Pro

Basic question about differential pairs

Hi group,
I have been doing measurements lately to see fast rise-time (50ps)
"digital" signals on diffpairs. The question occured to me: what is
the scope displaying?
I am using an Infiniium (13GHz BW and 40GSa/s) and a 1169A 12GHz
probe.
The probe has no local ground. I just mash the two bits of wire on
exposed signal traces. There is no handy local ground nearby on the
PCB anyways. Apparently the probe looks like a 50k resistor with a .
25pF cap across it.
So is the scope displaying some kind of average of each line's rise
time?
What if one side is rising at 20ps (these are 10-90% figures) and the
other falling at 30ps?
I can't easily make the single-ended measurements I am more
comfortable with.
Sure we can make tons of measurements relating to bit error rates and
eye openings, but none of these relate easily (for me) to a rise-time,
and more generally what a differential rise-time means.

I am still trying to debug the disastrous loss of speed on our board.
I've seen a FR-4 board run at 6Gbps with better specs than my Rogers
monster. There is a big problem somewhere and I want to make sure I
understand the basics first. Cuz I think I don't.

TIA gang.
 

neon

Oct 21, 2006
1,325
Joined
Oct 21, 2006
Messages
1,325
the probe picks up surface information. at those hi frequency 1 pico is a big capacitor.
 
x-no-archive:
I am using an Infiniium (13GHz BW and 40GSa/s) and a 1169A 12GHz
probe.
The probe has no local ground. I just mash the two bits of wire on
exposed signal traces.

Is that FET probe a DIFFERENTIAL probe or single ended probe. Are you
putting the scope probe ground lead on one side of the diff line?

You should be using a DIFFERENTIAL probe or use two single ended
probes into two channels of the scope set up to display the
differential voltage. i.e V1-V2. In this case it does not really
matter what you do with the 2 ground leads of the two probes as long
as you connect them to each other with a short connection.

Most so called differential PWB traces are not really operating
differentially but rather are operating as two single mode lines to
the ground plane that happen to carry opposite singals. Read up on
even and odd mode impedance.

http://www.microwaves101.com/encyclopedia/evenodd.cfm
http://www.signalintegrity.com/news/3_10.htm

Mark






Mark
 
x-no-archive:


Is that FET probe a DIFFERENTIAL probe or single ended probe. Are you
putting the scope probe ground lead on one side of the diff line?

Yes, it's a diff probe.
http://www.home.agilent.com/agilent/product.jspx?pn=1169A&NEWCCLC=USeng
There is no scope ground lead on this probe.
Most so called differential PWB traces are not really operating
differentially but rather are operating as two single mode lines to

Yes, but how do I know what the rise time the scope displays means?
The scope is single ended, so the probe is doing a conversion.
What does a 50ps rise time mean on a single ended display when you are
looking at a diff pair?
The signals grow apart in 50ps, but what does it mean for each single
ended signal?
the ground plane that happen to carry opposite singals. Read up on

I've called them "complementary pairs" before because of the weak
coupling but it never caught on.
Thanks.
 
Thanks but it turns out the probe is more complex than I first
thought, there are several heads that can be used and you must use the
correctly trimmed lead ins...
So before freaking out about rise times I'm gonna see if it was
measured properly. Then I can continue freaking out.
 
J

JosephKK

Jan 1, 1970
0
Yes, it's a diff probe.
http://www.home.agilent.com/agilent/product.jspx?pn=1169A&NEWCCLC=USeng
There is no scope ground lead on this probe.


Yes, but how do I know what the rise time the scope displays means?
The scope is single ended, so the probe is doing a conversion.
What does a 50ps rise time mean on a single ended display when you are
looking at a diff pair?
The signals grow apart in 50ps, but what does it mean for each single
ended signal?


I've called them "complementary pairs" before because of the weak
coupling but it never caught on.

Thanks.

Since it is high speed differential the rise / fall times are usually
well matched for each driver leg and between legs. It kind of has to
be this way. So to the extent you can measure either leg individually
they should look the same but with smaller voltage swings. Kind of
like the differential outputs on ECL. Also think of current steering
pairs.
 
Top