Yeah, material is what I meant by construction really. But I think it does depend (to a second order) on the layup. Stripline and microstrip are a little different I believe.
Once I dug into some of the details of PCB construction affecting the electrical features and I found that there doesn't seem to be a good theoretical basis for the formula used. Most of what I found was empirical with equations plucked from air to fit observations.
One guy would make some measurements and they would be used as the basis of PCB design for 10-15 years, then someone else would take some measurements that didn't agree with the first guy and he would provide a different set of equations to use that weren't all that much like the first guy's stuff. In the end I got the impression that none of this is used rigorously and each board design has to be "tuned" to the vendor's process. Coupons are often used to verify that the vendor is holding his process and you are getting what you need.
I haven't done much with tightly controlled impedance so this is not familiar territory to me.
Try this, adapted from a Philips (Mullard) research publication of the 1970s. Ported by me from FortranIV to Fortran77. Dimensions changed from cm to mm. Stated to run in 2-3 seconds, on an IBM360, runs like lightning on a PC ;-) I'd be interested to see how it compares with what you use.
Any old F77 compiler will compile it.
Program for computing the dimensions
of microstrip couplers using Schwartzmann's
formulae
*
100 format(20X,'Give data in the following order:'// +' Relative dielectric constant'/ +' thickness of dielectric in mms'/ +' Thickness of copper layer in mms'/' coupling in - dB'/ +' Center frequency of operation in Hz'/ +' Coupler impedance in ohms'///)
200 format(///' Relative dielectric constant = ',F6.3/ +' Thickness of dielectric = ',F8.5,' mm'/ +' Thickness of copper layer = ',F8.5,' mm'/ +' Coupling coefficient = ',F8.5,' or',F8.3,' dB'/ +' Frequency of operation = ',E10.3,' Hz'/ +' Coupler impedance = ',F8.3,' ohms'///)
300 format(20X,'Number of allowed iterations exceeded'//)
400 format(35X,'Results :'//' Width of lines = ',F8.3, 'mm'/ +' Spacing between lines = ',F8.3,' mm'/ +' Length of coupling region = ',F8.3,' mm (odd mode)'/ +' Even mode impedance = ',F8.3,' ohms'/ +' Odd mode impedance = ',F8.3,' ohms'/)
500 format(1H ,'Width of',F8.3,' ohm connecting line = ', +F8.5,' mm'/) write(*,100) read(*,*)er,h,t,db,freq,z0 x=db/20.0 couple=10.0**x write(*,200)er,h,t,couple,db,freq,z0 b=376.6/sqrt(er) d=1.0/(3.0*sqrt(er)) f=1.35/alog(4.0*h/t) zoe=z0*sqrt(1.0+couple)/sqrt(1.0-couple) zoo=z0*sqrt(1.0-couple)/sqrt(1.0+couple) k=1 w=0.0 m=0 step=0.01
8 if(k.gt.1000)go to 1 go to 2
1 write(*,300) go to 3
2 w=w+step a=w/h g=(a+1.0)*(a+1.0) aa=d*(a-(1.0/g)) bb=d/g cc=a+aa+0.5*d*a-bb+1.5*f dd=0.5*d*a+0.5*f x=((b/zoe)-cc)/dd y=1.0/x xx=4.0*(y-1.0) s=w/xx if(s)13,13,14
13 teszoo=1.0e10 go to 15
14 zcppu=2.0*aa m=m+1 zcppud=4.0*d/((s/w)+1.0) zcdf=1.35/alog(4.0*s/(3.141592*t)) teszoo=b/(a+zcppu+f+zcppud+zcdf) if(m.eq.1)step=step/10.0
15 check=(teszoo-zoo)/zoo chabs=abs(check) if(chabs.le.0.01)go to 4 if(check)5,4,7 if(m.eq.1)step=step*10.0
5 w=w-step step=step/10.0 k=k+1 go to 8
7 k=k+1 go to 8
4 const=er/(2.998e8*376.6) cppu=2.0*const*aa cdppu=8.0*d*const/((s/w)+1.0) cpp=const*w/h cf=const*2.7/alog(4.0*h/t) cdf=const*2.7/alog(4.0*s/(3.141592*t)) coo=cpp+cppu+0.5*cf+0.5*cdppu+0.5*cdf aloo=1.0/(1.0+((cppu+cdppu)*(1.0-(1.0/sqrt(er)))/(2.0*coo))) akoo=1.00/sqrt(1.0+aloo*aloo*(er-1.0)) cleng=0.25*2.998e11*akoo/freq write(*,400)w,s,cleng,zoe,zoo l=1 w0=0.0 stepj=0.1
9 if (l.gt.1000)go to 1 w0=w0+stepj a0=w0/h g0=(a0+1.0)*(a0+1.0) aa0=d*(a0-(1.0/g0)) tz0=b/(a0+2.0*aa0+2.0*f) che=(tz0-z0)/z0 cheab=abs(che) if(cheab.le.0.01)go to10 if(cheab)11,10,12
11 w0=w0-stepj stepj=stepj/10.0 l=l+1 go to 9
12 l=l+1 go to 9
10 write(*,500)z0,w0
3 stop end
--
"For a successful technology, reality must take precedence
over public relations, for nature cannot be fooled."
(Richard Feynman)
I just use Appcad, and sometimes (like, 5 minutes ago) TXline. TX handles some cases that Appcad doesn't, specifically diff pairs.
Practically all single formulas break at some physical limits. For example, take the old Motorola ECL handbook microstrip impedance formula and keep increasing trace width; Zo will pass through zero and go negative. Appcad does a good job of handling a wide range of geometries, and warns you if it can't.
In extreme cases, we crank up ATLC and actually do the field solving. That's labor intensive.
formatting link
It's fun to run a trace geonetry through different impedance calculator programs and see how they do/don't agree. Some of the online stuff is horrible.
--
John Larkin Highland Technology, Inc
jlarkin at highlandtechnology dot com
http://www.highlandtechnology.com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser drivers and controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro acquisition and simulation
Yeah, most of the common equations are sort of accidental curve fits and get in trouble when the geometries get away from their sweet spots.
I often add a test trace and an SMA connector to a board layout, so I can TDR the traces. If I hit anywhere from 45 to 55 ohms on FR4, for an intended 50, I consider that to be pretty good. Board houses can tweak the Gerbers to hit a target, but we seldom ask them to do that.
--
John Larkin Highland Technology, Inc
jlarkin at highlandtechnology dot com
http://www.highlandtechnology.com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser drivers and controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro acquisition and simulation
Why don't you compile the Fortran and check the accuracy for us. It's not as if you're too busy designing electronics.
--
John Larkin Highland Technology, Inc
jlarkin at highlandtechnology dot com
http://www.highlandtechnology.com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser drivers and controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro acquisition and simulation
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.