Hi

A question.

Would the in-circuit losses for RF equipment operating around 1 - 3
GHz be greater if built on 1.6mm or on 0.8mm FR4 laminate?

In particular, on which laminate thickness would a VCO in this range
be easier to get running?

Any explanation for the answers would be helpful.

Thanks

MikeN

Loss will be greater on the thicker board, because a greater fraction of
the fields surrounding traces and components will be in the board
material as opposed to being in the air. Having less field in the board
material will also make your circuit less sensitive to the temperature
and humidity dependency of the board material. The actual amount of loss
or temperature/humidity sensitivity depends on a whole host of factors,
so it's hard to say how much difference it'll really make in practice.

Roy Lewallen, W7EL

MikeN wrote:
Hi

A question.

Would the in-circuit losses for RF equipment operating around 1 - 3
GHz be greater if built on 1.6mm or on 0.8mm FR4 laminate?

In particular, on which laminate thickness would a VCO in this range
be easier to get running?

Any explanation for the answers would be helpful.

Thanks

MikeN

"Roy Lewallen" wrote in message
...
Loss will be greater on the thicker board, because a greater fraction of
the fields surrounding traces and components will be in the board
material as opposed to being in the air. Having less field in the board

This is correct as far as dielectric loss is concerned, but copper resistive
loss decreases as the board gets thicker. Which one will dominate depends
on frequency and characteristic impedance. At low frequencies, dielectric
loss will probably dominate. At high enough frequencies,
copper loss will dominate. For low enough impedance traces, the
field will be mostly in the air for either board thickness. Complex
problem to analyze.

Rick N6RK

What's the mechanism for the copper loss decreasing as the board gets
thicker? Of course, if you assume that the traces consist only of
microstrip or stripline transmission lines with some fixed impedance,
then the line width will be greater on the thicker material, resulting
in lower loss. Is that the rationale, or is there some other phenomenon
at work?

Roy Lewallen, W7EL

Rick Karlquist N6RK wrote:
"Roy Lewallen" wrote in message
...

Loss will be greater on the thicker board, because a greater fraction of
the fields surrounding traces and components will be in the board
material as opposed to being in the air. Having less field in the board


This is correct as far as dielectric loss is concerned, but copper resistive
loss decreases as the board gets thicker. Which one will dominate depends
on frequency and characteristic impedance. At low frequencies, dielectric
loss will probably dominate. At high enough frequencies,
copper loss will dominate. For low enough impedance traces, the
field will be mostly in the air for either board thickness. Complex
problem to analyze.

Rick N6RK

In article , "Ian White, G3SEK"
writes:

Rick Karlquist N6RK wrote:
"Roy Lewallen" wrote in message
...
Loss will be greater on the thicker board, because a greater fraction of
the fields surrounding traces and components will be in the board
material as opposed to being in the air. Having less field in the board

This is correct as far as dielectric loss is concerned, but copper resistive
loss decreases as the board gets thicker. Which one will dominate depends
on frequency and characteristic impedance. At low frequencies, dielectric
loss will probably dominate. At high enough frequencies,
copper loss will dominate. For low enough impedance traces, the
field will be mostly in the air for either board thickness. Complex
problem to analyze.

Yes, all of the above... and 0.8mm ("1/32in") FR4 also has many
practical advantages for home constructors.

"Losses" due to substrate material don't begin showing up until
one gets into S Band (above 2 GHz or so). All of that microwave
stuff is rather old hat to the short wavelength folks, in texts for
over 3 decades...but expensive to collect due to textbook costs.
:-)

I've "designed" directional couplers, hybrids, Wilkinson dividers,
etc., (at 1 GHz) by simply copying textbook instructions while at
Teledyne Electronics, a place where lots of folks were busy with
distributed constant structures on stripline. The watchword ought
to be Dimensional Stability rather than "losses." In that case,
Duroid (Teflon-fiberglass) is a better substrate. That isn't needed
at 400 MHz and below since the "losses" in any circuit down
there are in the components, not the substrate.

The sole exception to "losses" is the case of substrate problems
such as I encountered at Hughes Aircraft way back in 1957, new
to southern California aerospace. Hughes had subcontracted a lot
of terminal board assemblies for airborne radar and was suddenly
encountering way-too-low insulation resistance in some assemblies.
At the time that was important to the relatively high impedances of
vacuum tube circuits. After organizing the lower classes in several
shifts and tying up every MegOhmMeter in the corporation, a crash
round-the-clock testing discovered a single fabricator being the
culprit. Said fabricator also made rubber products and some of the
(carbon?) dust in the fabricator's air system was drifting through
their plant and settling on the substrate press where they finished
raw "circuit board" material. Fine for overtime (one of the last
places where I got any overtime in the career) but a pain in the
posterior and most boring and dull to do. Problem solved and
the radar sets got finished. [weird old anecdote]

One is that unlike the regular 1.6mm board, you can cut 0.8mm FR4 easily
with hand shears - or even strong scissors - without significantly
damaging either the board, the shears or yourself.

The thinner material is not as resistant to flexing as 1.6mm - it's
about as flexible as a credit card - but that's still strong enough for
most purposes. If it's an RF circuit, and you're going to add shielding
walls of hobby brass strip all around the outside, the resulting
assembly will be very rigid.

Whether shielding or not, a simple strip of copperclad soldered in
perpendicular to the circuit board adds enormous stiffness and
dimensional stability.

I just uncovered the motherboard-backplane to a Midwest Scientific
6800 microprocessor kit in a workshop storage box. It used strips
of copperclad soldered to the backplane +5 VDC and ground traces
(roughly 0.37" wide). It had been in the box better than 20 years and
at a diagonal, other parts putting weight on it as if to bend or bow it.
Sighting down the surface, it was still as flat as it was when I put it
together two decades earlier!

50-ohm microstrip on 0.8mm FR4 is about half its normal width on 1.6mm,
which makes it a much better match (in all senses) to the width of the
SMD components hanging off the ends. It's still wide enough to make the
impedance reasonably insensitive to HB fabrication tolerances.

Since 0.8mm FR4 is the basic material of multi-layer boards, increasing
numbers of PCB materials suppliers and board makers can now offer 0.8mm.

In the UK, Mega Electronics supply both SS and DS photo-resist-coated
0.8mm FR4 (and also many other specialist PCB-making materials):
http://www.megauk.com

One of my personal gripes is the cost of basic copperclad at dealers.
In bulk quantities the material is quite cheap in anyone's monetary
units, yet the dealers mark the prices up well above bulk cost. The
excuse is "cutting to manageable shipping size" and other mythology
to disguise getting as much for it as they can. :-)

I've found that true in the USA, especially around southern California,
once one of the aerospace centers and a big user of copperclad.
The few Canadian dealers doing mailorder have better prices for
copperclad than in the USA. I've gotten "surplus" copperclad (when
spotted with the bargain-hunter's eye) in sheets at less than a tenth
of the price of cut copperclad from dealers. A small fee paid to a
sheet metal fabricator and they will use a shear to cut such down to
manageable size. Overall copperclad cost that way is still less than
a tenth of dealers' cut copperclad prices.

The four-sided box type of circuit board shielding will do wonders for
physical stability once soldered in place, even with the thin 0.032"
thick single-sided copperclad.

Len Anderson
retired (from regular hours) electronic engineer person

If you keep the impedance constant, the loss goes down
for the simple reason you state that the width increases.
If you keep the width the same, the loss also goes down
because the characteristic impedance is higher. This is
because the copper resistance, relative the the characteristic
impedance, is lower. This is analogous to open wire line.
Suppose you make two OWL's with #14 wire. Suppose
one line has 1/2 inch spacing and the other has 1 inch
spacing. The 1 inch line will be lower loss owing to its
higher characterisitic impedance. OTOH, a #14 twisted
pair will be much higher loss due to its much lower impedance.

Rick N6RK

"Roy Lewallen" wrote in message
...
What's the mechanism for the copper loss decreasing as the board gets
thicker? Of course, if you assume that the traces consist only of
microstrip or stripline transmission lines with some fixed impedance,
then the line width will be greater on the thicker material, resulting
in lower loss. Is that the rationale, or is there some other phenomenon
at work?

Roy Lewallen, W7EL