Barry,
          
          I would be careful with taking your discussion below to 
          heart.  Your equation for frequency is actually a Bandwidth 
          equation.  Granted, if your calculating the HIGHEST 
          frequency of the bandwidth, that's fine.  But, frequency 
          does play a significant role in the "impedance" that one 
          wants.  Ask anyone dealing with board constructions with FR4 
          above 800 MHz.
          
          Also, your assumption is that all the frequencies required 
          for the transition time are in sync with one another (from 
          Fourier).  Bad assumption.  Group delays and shifts in phase 
          based on frequency are only part of what will determine the 
          actual outcome.  At best, you'll get a smeared edge at fast 
          edge rates.
          
          When board construction for controlled impedance is most 
          important, there's a whole other side to controlled 
          impedance that can get nasty.
          
          doug


______________________________ Reply Separator _________________________________
Subject: Re: Controlled Impedance - Dielectric Constant- Comment
Author:  [log in to unmask] at internet-mail
Date:    5/21/96 10:39 PM


Hi Luis
          
The dielectric constant of soldermask is slightly 
different to that of FR4 however in practice the 
variation is neglibible.
          
   (-1.55*H'/H)
>From IPC-D-317 e=er[1-e     ]
          
I also noticed from the discussion that it has been 
mentioned that the frequency of operation will 
determine the impedance of the substrate. That is 
not exactly correct as it is in fact the rise time 
of the leading edge of the pulse that is the 
influencing factor.
          
 f=0.35/Tr
   where f is in GHz
         Tr is in nS
          
At the end of the day what I do is just specify in 
the PCB Spec that the trace widths and prepreq 
thickness may be varied to obtain the desired 
impedance. And really who cares what material is 
used in the board manufacture as long as the 
impedance is what we want!
          
Cheers,
          
Barry Olney, In-circuit Design, Australia
          
Luis Rivera wrote:
> 
> Bob,
> 
> You are correct, the dielectric constant of a material does not change 
> between microstrip, embedded microstrip or stripline configurations.
> 
> Avner,
> 
> Some general words of caution:
> 
> Dielectric constant does change depending on resin and reinforcement 
> type, resin content, frequency, temperature and humidity. Typically, 
> dielectric constant decreases as frequency increases.
> 
> On microstrips, the fact that there is air on one side of the
> transmission line lowers the effective dielectric constant. Solder 
> mask coatings counteract this effect to some degree.
> 
> Please call off-line if you would like to discuss further. 
> 
> Luis Rivera
> [log in to unmask]
> CompuRoute, Inc.
> Dallas, TX
> (214) 340-0543
> 
> >
> >
> >   Avner--
> >          If I am wrong, I hope someone out there will make me smarter, but 
> >    the dielectric "CONSTANT" is just that--constant for that type of 
material.
> >    IMOP, when calculating impedance of microstrip or stripline, the 
dielectric
> >    constant of a material doesn't change, less the material's unique 
tolerance
> >    range REF: IPC-D-275 Table 3-2. Therefore, unless proven wrong, the 
constant
> >    DOES NOT change, but the impedance value WILL vary depending on which
> >    formula you are using depending on you pwb construction. Hope that helps.
> >    If wrong, please correct me via E-mail,
> >
> >                              Bob Vanech 
> > E-Mail : [log in to unmask]
> >
> >
> >
          
-- 
___________________________________________________
          
Barry Olney    
Managing Director   
          
In-Circuit Design Pty Ltd    Ph: +61 3 9205 9595 
VeriBest Solutions Centre Fax:+61 3 9205 9410 
Suite 211, Princess Tower Mbl:+61 4 1117 0827
1 Princess St,    
Kew, VIC 3101, Australia    Email:[log in to unmask] 
___________________________________________________