FreeLists / si-list / [SI-LIST] Differential Pair Characteristic Impedance Tradeoffs

[melvin bowman <mbowman@xxxxxxxxxxxxx>]

I find, in attempting to determine trace and space widths for multiple 
differential pairs with different impedance requirements on a single PCB 
layer, it can be impossible to find a perfect solution while staying 
within acceptable manufacturing constraints.
I.e. Given the following.

Impedance Requirements
Single-ended (default) trace is desired to have a Zo a of 60-ohms.
Diff. Pair #1 characteristic impedance = 60-ohms single-ended and 
100-ohms differential.
Diff. Pair #2 characteristic impedance = 60-ohms single-ended and 
90-ohms differential.
All impedances specified with a tolerance of +/-10%.

Manufacturing Constraints
Single-ended (default) trace with = 5 mils.
Minimum trace width = 5 mils.
Minimum space width = 5 mils.
Trace thickness (t) = 0.7 mils or 1.4 mils (1/2 oz. or 1 oz. copper)
Dielectric thickness (h) = 4 - 5 mils (microstrip).
Effective dielectric constant (Er)  = attainable values with FR4 (3.4 - 
4.8 ???) What is the reasonable range?

 From the constraints, above, we can see that both diff. pairs #1 and #2 
will have a trace width of 5 mils, as they have the same single-ended 
characteristic impedance as that of the default single-ended 5 mil trace.

Given this information we can determine a dielectric thickness (t) and 
effective dielectric constant (Er) to meet the requirements for diff. 
pair #1 while staying within the bounds of our manufacturing 
constraints.  However, at this point we have the diff. pair #2 
requirements to contend with.  As diff. pairs # 1 and #2 are both on the 
same layer and have the same single-ended characteristic impedance they 
must both have the same trace width (5 mils).  This leaves us with only 
one parameter to adjust to get a 90-ohm differential impedance for diff. 
pair #2 while maintaining a 100-ohm differential impedance for diff. 
pair #1.  When we attempt to set a differential pair spacing to attain a 
90-ohm differential impedance for diff. pair #2 we find that we can't 
get them close enough together to accomplish this without seriously 
violating our minimum space width constraint of 5 mils.

Our only solution appears to be to compromise on the characteristic 
impedances of our various traces, both single-ended and differential.  
When we do this we begin to eat away at the +/-10% tolerance that was 
intended to be a manufacturing tolerance, not a calculation tolerance.

If we make this compromise we come up with numbers such as the following 
scenario.

Single-ended (default) trace = 5 mils and 60-ohms.
Diff. Pair #1 = 5 mils and 60/101-ohms with a 5 mil spacing..
Diff. Pair #2 = 5.7 mils and 56/94-ohms with a 5 mil spacing.
This is accomplished with a trace thickness (t) of 1 mil, a dielectric 
thickness (h) of  4.3 mils and an Er of 4.2.
(Note that I used a 1 mil thickness rather than 0.7 or 1.4. This was 
just to split the difference for ease of analysis and has a small impact 
on the overall result.  Likewise, different impedance calculators may 
give different numbers but the trend should provide similar results)

Assuming that no one can tell me how to get the calculations to work out 
exactly to the requirements without violating the manufacturing 
constraints, this leads to my actual question(s).

1) How important is the single-ended impedance of the traces of a 
differential pair and what is the impact of a deviation of 5%  - 10%?

2) Is the best compromise to target the desired single-ended impedance 
for the benefit of the non-differential default traces and adjust the 
width and spacing of the differential pairs to give the desired 
differential impedances while violating their single-ended impedance 
specifications?

3) Is it better to compromise all impedances, default single-ended as 
well as diff. pair single-ended and differential impedances and just try 
to find a tradeoff that keeps them all relatively close to their desired 
values, if possible?

4) If the answer to question #1 hasn't already answered this question, 
what is the importance of the single-ended impedance of a differential 
pair, as long as it's balanced (both traces have the same single-ended 
impedance).




------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field

List FAQ wiki page is located at:
                http://si-list.org/wiki/wiki.pl?Si-List_FAQ

List technical documents are available at:
                http://www.si-list.org

List archives are viewable at:     
                http://www.freelists.org/archives/si-list
or at our remote archives:
                http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
                http://www.qsl.net/wb6tpu