On Sun, 9 Nov 2003 16:00:21 -0500, "Tom Holden"
wrote:

Thanks to everyone who has responded. I think the conclusion is that at the
frequencies I'm interested in, and the available space inside the radio (a
little RS DX-394 table radio), it's impractical to absorb the 455 kHz
crosstalk energy from 2nd IF to frontend. Better to attempt to
compartmentalize the radio. That may prove to be impractical also as it
would appear very difficult to make small (removable) shields over the IF
section that would not have gaps. However, I'll examine the pcb layout more
closely to see if there are any viable paths for the sides of the shield
box.

73, Tom


You'll never know before you have tried, and remember that Racal
solved a similar problem with their famous RA-17 series receivers
using a hacksaw to make a little mark in the chassis to stop unwanted
radiation from one point to another. Believe I've seen the application
of those carbonized foam used in the lids of boxes which were
definitely not microwave equipment

Such things are impossible to predict

It is also some definite requirement for the thickness of the walls to
act as screen on certain frequencies, as an example could be mentioned
that pcb laminates are not thick enough for good screening on 80m in
an application with two oscillators which need good screening to avoid
coupling to be used for third order IP measurements

73
Jan-Martin, LA8AK
Amateur radio techniques http://home.online.no/~la8ak/c.htm
--
remove ,xnd to reply (Spam precaution!)

Most novice designers get burned somewhere along the line by failure to
realize what Ian is describing -- that all amplifiers are differential
amplifiers, and calling one of the inputs "ground" doesn't impart magic
properties. Once that lesson has been learned (alas, usually the hard
way), the probability of making circuits work as designed increases
dramatically.

Even so, though, a surprise sometimes comes along at the last minute. I
know of one expensive, high quality test instrument that had a ground
path broken at the last minute because of ground current that was
induced in the signal path by the field from the display CRT deflection
yoke. Forewarned, I discovered 100 mA of induced current in the shield
of a multi-turn delay line from the same source, in an instrument in
whose design I was participating. Simply soldering the turns of
semi-rigid coax together reduced the effective transformer secondary to
one turn from 20 or so, making it possible to reduce the coupling and
its effect to an acceptable level by other means.

But I remember that shortly after I went to work at Tektronix, I was in
the main lobby admiring the example of the first oscilloscope model
produced by that company, which was on display there. Its success and
the reputation it gathered as a quality test instrument was instrumental
in launching the company. It had two chassis, one above the other,
supported by four L - cross section aluminum legs between the chassis.
One of those four legs had very neatly been cut, leaving a small gap. A
sturdy piece of phenolic was mounted between the gapped ends, again very
neatly, with countersunk flathead screws. I remember that seeing this
gave me a good feeling that ground currents could be a problem for the
best of 'em.

The 'scope in the lobby was a production model, but somewhere, I'm sure,
there was a prototype much like the Racal device -- with a hacksaw cut
across the leg.

Roy Lewallen, W7EL

Ian White, G3SEK wrote:
J M Noeding wrote:


You'll never know before you have tried, and remember that Racal
solved a similar problem with their famous RA-17 series receivers
using a hacksaw to make a little mark in the chassis to stop unwanted
radiation from one point to another.

[...]

More than a "little mark" - a big cut, deliberately intended to
interrupt an unwanted current path through the chassis.


Such things are impossible to predict

If you stop to think about it, such things often *can* be predicted...
or at least explained with the benefit of hindsight.

When we connect a signal from point A to point B, we often forget about
the return path through 'ground'. But every signal path must also have a
return path - that's why we call these things "circuits". Even if we
shield the signal conductor to provide a preferred return path, there
are still other return paths available in parallel with that, through
the box.

Return currents will divide between these paths according to the RF
impedance that each path presents (Ohm's law), so you will always get
some RF currents in the body of the box. The box is just another RF
conductor, and when currents flow through it, voltages will appear.

The key thing is to remember that there's no magic that makes the box a
perfect zero potential. For all us Sixties children: "'Ground' is a
delusion, Grasshopper."

Therefore in a complex project you will always get some interaction
between different stages. What's hard to predict is how much effect this
will have on overall performance.

The problem that plagued Racal was spurious receiver responses. Racal
was a startup company with a revolutionary all-band receiver concept...
but it involved several oscillators in different stages, that were
getting into the wrong parts of the receiver and causing "birdies". They
couldn't sell a single unit until that problem was fixed.

They had already gone to the huge expense of a diecast chassis divided
into many shielding compartments... and it wasn't working. Very soon,
the money would run out and Racal could cease to exist. Hence the
desperate solution with the hacksaw - and it saved the company. Legend
has it that the prototype receiver (complete with saw cut) was on
display in Reception for many years... maybe they shoulda kept the
hacksaw too?


I once had a problem of this type with spurious responses from a
receiver that was constructed in three (apparently) shielded diecast
boxes with coax signal interconnections. The boxes were stacked side by
side inside the case of the receiver. Remembering the Racal experience,
something made me shove a piece of folded paper under one of the boxes,
to insulate it from the receiver case. The spurs disappeared!

Looking further into this, the real problem was with the power
interconnections between the boxes, which I'd been ignoring until then.
RF was getting into the power rails, and contact between the boxes
formed a ground loop. Unwanted signals were being transferred between
the two boxes via this 'back channel'. With better bypassing of the
power rails, it was possible to re-connect the boxes to the case of the
receiver.

Most novice designers get burned somewhere along the line by failure to
realize what Ian is describing -- that all amplifiers are differential
amplifiers, and calling one of the inputs "ground" doesn't impart magic
properties. Once that lesson has been learned (alas, usually the hard
way), the probability of making circuits work as designed increases
dramatically.

Even so, though, a surprise sometimes comes along at the last minute. I
know of one expensive, high quality test instrument that had a ground
path broken at the last minute because of ground current that was
induced in the signal path by the field from the display CRT deflection
yoke. Forewarned, I discovered 100 mA of induced current in the shield
of a multi-turn delay line from the same source, in an instrument in
whose design I was participating. Simply soldering the turns of
semi-rigid coax together reduced the effective transformer secondary to
one turn from 20 or so, making it possible to reduce the coupling and
its effect to an acceptable level by other means.

But I remember that shortly after I went to work at Tektronix, I was in
the main lobby admiring the example of the first oscilloscope model
produced by that company, which was on display there. Its success and
the reputation it gathered as a quality test instrument was instrumental
in launching the company. It had two chassis, one above the other,
supported by four L - cross section aluminum legs between the chassis.
One of those four legs had very neatly been cut, leaving a small gap. A
sturdy piece of phenolic was mounted between the gapped ends, again very
neatly, with countersunk flathead screws. I remember that seeing this
gave me a good feeling that ground currents could be a problem for the
best of 'em.

The 'scope in the lobby was a production model, but somewhere, I'm sure,
there was a prototype much like the Racal device -- with a hacksaw cut
across the leg.

Roy Lewallen, W7EL

Ian White, G3SEK wrote:
J M Noeding wrote:


You'll never know before you have tried, and remember that Racal
solved a similar problem with their famous RA-17 series receivers
using a hacksaw to make a little mark in the chassis to stop unwanted
radiation from one point to another.

[...]

More than a "little mark" - a big cut, deliberately intended to
interrupt an unwanted current path through the chassis.


Such things are impossible to predict

If you stop to think about it, such things often *can* be predicted...
or at least explained with the benefit of hindsight.

When we connect a signal from point A to point B, we often forget about
the return path through 'ground'. But every signal path must also have a
return path - that's why we call these things "circuits". Even if we
shield the signal conductor to provide a preferred return path, there
are still other return paths available in parallel with that, through
the box.

Return currents will divide between these paths according to the RF
impedance that each path presents (Ohm's law), so you will always get
some RF currents in the body of the box. The box is just another RF
conductor, and when currents flow through it, voltages will appear.

The key thing is to remember that there's no magic that makes the box a
perfect zero potential. For all us Sixties children: "'Ground' is a
delusion, Grasshopper."

Therefore in a complex project you will always get some interaction
between different stages. What's hard to predict is how much effect this
will have on overall performance.

The problem that plagued Racal was spurious receiver responses. Racal
was a startup company with a revolutionary all-band receiver concept...
but it involved several oscillators in different stages, that were
getting into the wrong parts of the receiver and causing "birdies". They
couldn't sell a single unit until that problem was fixed.

They had already gone to the huge expense of a diecast chassis divided
into many shielding compartments... and it wasn't working. Very soon,
the money would run out and Racal could cease to exist. Hence the
desperate solution with the hacksaw - and it saved the company. Legend
has it that the prototype receiver (complete with saw cut) was on
display in Reception for many years... maybe they shoulda kept the
hacksaw too?


I once had a problem of this type with spurious responses from a
receiver that was constructed in three (apparently) shielded diecast
boxes with coax signal interconnections. The boxes were stacked side by
side inside the case of the receiver. Remembering the Racal experience,
something made me shove a piece of folded paper under one of the boxes,
to insulate it from the receiver case. The spurs disappeared!

Looking further into this, the real problem was with the power
interconnections between the boxes, which I'd been ignoring until then.
RF was getting into the power rails, and contact between the boxes
formed a ground loop. Unwanted signals were being transferred between
the two boxes via this 'back channel'. With better bypassing of the
power rails, it was possible to re-connect the boxes to the case of the
receiver.