EMRFD Message Archive 252
Message Date From Subject 252 2006-12-22 08:28:44 w4zcb77 Homebrew boxes and shielding OK folks, I'm ready for a little advise. I'd even settle for a few
well thought out honest opinions, and some measured experiences
would be frosting on the cake.
For years, I've made little boxes and cabinets and sub assemblies
out of PC Board stock. Most of the time, I've used double sided
stock, sometimes bonding both sides with stained glass store copper
tape, and sometimes just wire vias and sometimes only dabbing
corners with a hot iron. The question for the day is, should I be
using double sided board or single sided board for shielding, in
consideration of the fact that there's appreciable capacity between
the two copper layers? (At one time, I even performed an experiment
of this concept but was either half hearted or inconclusive).
Looking at some of my MMIC construction, where I've got stable gain
to GHz and lots of it, where I've very carefully placed wire vias
every 1/4 inch between ground planes to emulate a 50 Ohm microstrip
makes me wonder if I've been doing it wrong. The consideration of
the use of single sided board for VFO construction, presumeably for
similar reasons (Lack of perfect bonding between the two conductors)
furthers my wonder.
Before I go spend a day or even a couple making a RF source in a box
and trying to measure both methods, anyone out there that's already
been here, done that, and has a spare T shirt?
Regards
W4ZCB253 2006-12-22 19:43:03 Rick Re: Homebrew boxes and shielding Hi-nice to see you on this forum. I've made some measurements and adopted some
practices that seem to work--but I do need to study this some more. First, the
measurements:
The rule of thumb we've used is about a nanohenry per mm or wire. I did some simple
bench measurements of resonance and that's about right. So all your wire vias are 1.5 mm
long, a bit more than 1 nanoHenry. Assume it's 2 nH. That's a scary inductance in a 50
ohm system with garden variety silicon MMICs, and way too much to be trusted if you are
using GaAs or SiGe MMICs with gain above 10 GHz. I don't use wire vias very often
anymore. If I need a real ground in the center of a PC board, I file a hole big enough to
use copper tape wrapped top and bottom. Little jewelers files from the Hobby shop make
it easy to do square and rectangular holes.
Now, on to the homebrew copperclad epoxy board enclosures. G-10/FR-4 is such a nice
structural material that this is a good way to build enclosures, but for shielding, I want a
continuous metal enclosure. I usually build MMIC and VFO enclosures with a combination
of EZ-Solder tinned sheet (hardware store/hobby shop) and double sided PC board. "Tin"
cans from the kitchen are even better, as they are a little thicker. I just ignore the outer
layer of copper on the double-sided board, and make sure the inside is a continuous six-
sided box with all seams soldered. Soldered up boxes are surprisingly rigid, and you can
make them more so with strategically placed bends and curves in the walls.
The circuit inside the box is often an ugly constructed assembly on it's own scrap of PC
board, with it's own ground. So the metal inside surface of the box is an electromagnetic
shield, not a circuit ground. I often fasten ugly boards or little microstrip boards down
inside the shield box with double-sided tape. Then I run the RF connections in coax all
the way from the connectors on the shield box to the appropriate spot on the little ugly
board inside, and don't worry about any other ground connection. That way I don't get
signal currents flowing in the shield box.
Electric shielding is effective just by having a lot of metal around a sensitive circuit for the
E-field lines to terminate on, but magnetic field lines don't terminate. For magnetic
shielding, you need closed paths. The eddy current losses in conductors are more
effective magnetic shielding than steel enclosures at frequencies above audio, but you do
need closed paths. A gap in the shield can turn a magnetic shield into a magnetic antenna
(you probably remember that from the DF loops in the old mobile handbooks.) So I solder
everything shut, all the way around.
I like to do fancy metal work, milled aluminum enclosures etc. Diecast boxes are
acceptable for some circuits. But for critical circuits: filters, VFOs, microwave amplifiers
(including HF amplifiers with MMICs), I always get it working with connectors and power
feedthroughs on the appropriate walls, and then solder the whole thing closed.
Sometimes I include little holes for access to tuning points. Then that enclosure goes
inside a metal box. Most ham gear isn't built that well, but then commercial ham gear
isn't even in the same league as the stuff you build at W4ZCB.
OK, what happens when I need access to the inside of the box I've soldered shut? First I
use solder wick to remove as much solder from the seam as I can. Then I put the box in a
vise, a soldering iron in my right hand and an exacto knife in my left, and run the
soldering iron down the seam followed by the exacto knife. I've had to do that a couple
times, but not often. I find that the circuits I solder shut are the most reliable modules I've
ever built--by perhaps an order of magnitude. There are lots of reasons for that...
Back to your original question: what about the other side of the circuit board? Well, it just
doesn't matter--it's outside the box. I usually just leave it floating, and then try to
remember that it's not a good place to solder the negative wire from the power supply.
I hope that helps. That's the way I do it from audio to mm waves... build and test the
circuit...build the shield box...put the circuit inside the shield box...test it one last time
and then solder it shut. The receiver on the cover of October QST works ok with the
plexiglass cover, but that's only for show. If I actually integrate that receiver into a serious
radio station, I'll solder on the shield covers.
Merry Christmas and Happy New Year,
Rick kk7b254 2006-12-22 19:57:20 Wes Hayward Re: Homebrew boxes and shielding Hi Harold, et al,
The question of using single Vs double sided PC board material for
box construction is an interesting one. I have used both, but have
never done serious measurements to determine what the difference
might be. Let me outline the procedure I have used. Then I'll
suggest what I would do to evaluate it. I suspect you have
instrumentation to pull this off. I don't, at least at the higher
frequencies.
The goal is to get really good shielding. This means having your
circuitry completely surrounded by a continuous metal wall. So I
start with a bottom of either single or double wall construction.
I've never done anything there to tie the two surfaces together.
Then I cut and put walls up as needed. If I'm using single sided
material, I put the copper on the inside. I then solder all inside
seams. The outside, if present, is soldered to the bottom base
sheet.
Now comes the subtle part. If I had used double sided material,
the inclination would be to put a lid of some sort on the top and
then solder the outside of the 4 walls to the lid. But here's the
rub--if you do this, there is nothing in the way of solder to
guarantee that there will be a connection between the lid and
the inside of the four walls. This would be needed to get that
complete electrostatic shield. So what I do is to make frequent
cuts in the edge of the four walls. A hack saw works fine, but you
can also do it with the corner of a file. Then I run wires through
the slits and solder them on both sides. Drilling will also work.
This then provides communications between the inside and outside of
the walls, making it come closer to a sheet of solid metal. Once
this is done, I can plop the lid down and solder it to the outside of
the walls.
Here's a subtle step that I often add to the process: I take a file
and remove the copper edge from the top of the outside of the four
walls. This means that when you lay the top lid in place, there
will be no copper to copper interface. But then I place a piece of
buss wire along the edge and solder that wire to both copper
surfaces. What this does is to provide something that can be
removed when (not if) you want to get back to the inside.
If I made walls from single sided material, I would have put the
small cuts in the walls and added wire. I then soldered the wires
to the inside copper surface, taking care to leave a small chunk of
wire on the outside. Then when you put the lid in place, you can
solder from the lid to the wire.
These methods seem to have worked OK. However, I have done no
measurements to evaluate the seal.
My favorite homebrew box is a hybrid using PC board for the top and
bottom. However, make the walls from hobby store brass strips.
Now for measurements: This is a scheme that I saw one of the guys
in the Tektronix Spectrum Analyzer group do. I think it was Bob Alm
who was our mm Wave guy. We were trying to evaluate the quality of
some shielding material that was at the edge of an aluminum box. A
lid would be attached and would be mechanically forced down against
the edge of the 4 walls that had the gasket material. Bob started
with a flexible cable with SMA connectors that was attached to the
input of a spectrum analyzer. The far end attached to an SMA
connector that had a small single turn loop of wire between the
center pin and a ground pin. The loop was about 1 cm in diameter as
I recall.
The spectrum analyzer also had an attached tracking generator that
went up to 1.8 GHz. A cable went from the TG to an SMA connector on
the box. The inside of the connector had a piece of wire on it that
stuck into the box by perhaps an inch or so. This would build up
fields within the box that could then be detected with the loop
probe. We would place the probe right at the edge of the box and
adjust levels and analyzer bandwidth for a robust response. Then
we would put a lid on the box. There were several different box
types that we evaluated.
I do remember a few qualitative results. One was that if we
actually soldered a brass lid to a brass box, we could see nothing,
even at 1.8 GHz; the shielding was perfect. If we tried to use a
lid that was casually screwed down to the box, we could see leakage,
especially if there was a sizable gap between screws. As I
recall, we concluded that we wanted a screw every inch or so. Some
of the mesh-like material that we forced down into a grove in the top
of a wall worked well so long as the lid was tight. The material we
were testing was a rubber that was impregnated with metal. The
rubber was in the physical shape of a U shaped extrusion that fit
over the edge of a metal box. The material was quite effective so
long as there was an even pressure on it. We ended up using that
shielding scheme in several instruments255 2006-12-23 07:05:19 w4zcb77 Re: Homebrew boxes and shielding Well, I knew that I had come to the right place, have no idea why I
haven't used this resource to more advantage earlier. My thanks to
both Wes and Rick for those well thought out and considered
responses, experience like that needs all the documentation we can
give it.
Rick, I own a couple hundred INA 02186 MMIC's, wonderful little
pills and the right combination for me, new enough to have the
feedback reduced enough to realize 30 plus dB of gain and old enough
NOT to have gain at 10 GHz! The data sheet for them comments on
using 0.032 board material for reduction of inductance in wire vias,
and I've on occasion used some 0.010 board material I have scrounged
for that purpose. I have however, cascaded those little miracle
pills and on 0.063 board by placing FIVE wire vias (#68 carbide
drill size) immediately adjacent to each ground tab and drilling a
little pocket to seat the device leads flush with the surface.
(Along with trying my best to make the striplines look like
striplines with more vias.) The nice thing about the 02186's is that
you don't need a GHz SA to find the oscillation. If it's oscillating
up there, the gain falls about 10-15 dB from advertised. When you
can get 30-32 dB gain from each pill, it's stable everywhere. But
then, I'd get a nosebleed if I operated at the frequencies you do!
Wes, thanks ever so much for sharing your experiences. I've used the
vias and the copper tape over the edges and the single sided board
tricks for years, even some squashed silver braid from some RG-316
between joints, (At times 3 out of four of the above) and have
generally been satisfied with the results. I also have a few
projects that look pretty ratty due to having DEsoldered a few lids
after going thru the process. The cause of my request for info was
the completion yesterday of a removeable lid for a solid state
bandswitched 10 band 5 pole BPF. I had a terrible time locating some
1/4 inch square Al bar stock despite having bought it before at the
local hardware store. I must have been their only customer, they
don't carry it any longer. I did find it stocked at "Small Parts
Inc" in Miami, with the usual disclaimer, the shipping and handling
roughly equals the cost of the material. (That's customarily my
reason for junque box accumulation, trying to dollar average the
S&H)!
I managed to drill and 4-40 tap 21 staggered holes in two faces of
four short pieces of the stuff to allow bolting one face to the
edges of the box and the other face for the lid. (Also made of PC
board stock although I may use that as a template for making a
matching one out of sheet tin can after Ricks suggestion.) Pleased
to report that my mechanical skills seem to have improved thru the
years, Christians 21 Lions 0 so far in that there were no broken 4-
40 taps consumed in the process, and the holes all lined up without
having to do more "hole drifting" than simply using a #31 clearance
drill in lieu of the appropriate #33. I was pleased with the ability
to remove the cover to get to the goodies and was wondering whether
I had shot myself in the foot in the purpose of the cover in the
first place.
I went to school long enough ago that I can still learn a lot, a
significant portion of which I find in the writings of Roy, W7EL. It
hasn't been more than a couple of years since I learned that
magnetic shielding is the result of eddy currents in the shielding
material. Up till then, I had considered that if one copper wall
was "good", then two ought to be better!
I thank you both for documenting your experiences and for hosting
this URL. It's an international treasure.
Regards and Merry Christmas.
W4ZCB256 2006-12-23 07:30:31 Jim Miller Re: Homebrew boxes and shielding one solution to the tight fitting lid problem might be the use of the
waveguide below cutoff principle. if the lid can be made to fit to the box
so that the space between them has at least a right angle rather than
straight path i think that should work. if it has two such angles, e.g. a
bead dropping into a notch even better.
i haven't given enough thought to this to figure out how to homebrew it yet.
73
jim ab3cv274 2006-12-26 18:56:09 Jason Milldrum Re: Homebrew boxes and shielding Hi Wes,
I have a few comments and a question below:
Wes Hayward wrote:
> Bob startedWe still use a similar probe now on the RTSA line (we call it the sniffer
> with a flexible cable with SMA connectors that was attached to the input of
> a spectrum analyzer. The far end attached to an SMA connector that had a
> small single turn loop of wire between the center pin and a ground pin.
> The loop was about 1 cm in diameter as
> I recall.
probe). It can be useful for finding RF leaks much like the way in which
you described it.
> The material we were testing was a rubberThat conductive gasket material is still being used today in the RSA3408A
> that was impregnated with metal. The rubber was in the physical shape of
> a U shaped extrusion that fit over the edge of a metal box. The material
> was quite effective so long as there was an even pressure on it. We
> ended up using that shielding scheme in several instruments on some of the
> modules that I designed.
and RSA6100A instruments. I've seen quite a few problems with spurious
products caused by misplaced or missing gasket material. Its pretty
amazing how even a small amount of missing gasket can cause large,
widespread spurs thoroughout the frequency range of the instrument.
Anyway, on to the question. I have been working on building a transceiver
for the FDIM Potluck contest this year. I was looking at borrowing the
post-mixer IF amp from the general purpose front end design in Chapter 6
of EMRFD (the one in Fig. 6.69). I noticed that the parts list for the amp
on page 6.34 under Fig. 6.70 calls for an AC emitter degeneration resistor
(R10) with a value of 6.8 kohms. It seems to me that it would make more
sense to use a value of 6.8 ohms (or at least at that magnitude). I didn't
see any errata regarding this on your website, so I wanted to get your
feedback on this. I've never tried to make an amp with two paralleled
BJTs, so I don't want to make too many assumptions about the circuit. Any
insight you could give me on this would be very helpful.
Thanks very much,
Jason Milldrum, NT7S275 2006-12-26 19:01:32 Jason Milldrum Re: Homebrew boxes and shielding Sorry folks,
That message wasn't supposed to go the group, but since it accidentally
made it, maybe someone else could chime in about the question as well.
Thanks,
Jason Milldrum, NT7S
Jason Milldrum wrote:
> Hi Wes,
>
>
> I have a few comments and a question below:
>
>
> Wes Hayward wrote:
>
>
>> Bob started
>> with a flexible cable with SMA connectors that was attached to the input
>> of a spectrum analyzer. The far end attached to an SMA connector that
>> had a small single turn loop of wire between the center pin and a ground
>> pin. The loop was about 1 cm in diameter as
>> I recall.276 2006-12-26 20:51:38 Wes Hayward Re: Homebrew boxes and shielding Hi Jason, et al,
Yep, I was surprised when I finished updating the errata for EMRFD on
my web site and then thought I'd tale a quick look at the postings on
this group. I noticed Jason's note again and got confused. But
not a problem, for Jason's comments are quite interesting.
Evidently, he works in the present version of the old Tektronix
spectrum analyzer group. I think there are perhaps two or three of
the folks who where there when I was. The major engineer from the
group, Larry, W7JBY, is now SK, having died last June. He will be
missed.
But yes, it is 6.8 Ohms and not 6.8K. The resistor was correct in
the parts listing for the "Easy 90 Receiver" described