Hi Alexei!

> If you can, could you, please post me more on that modification,

I simply cut a pattern on normal double-sided fiberglass PCB material,
using a knife, with a layout such that the bridge resistors connect to
the SMA connectors with almost no trace length. The path widths were
calculated for 50 Ohm, (roughly 2mm wide). The detector parts were
arranged around the bridge, using thin (higher impedance) paths.
 
As long as you make it as small as possible, really even at 2.4 Ghz
the exact layout doesn't matter very much!

Cheers,
         Manfred.

--------------------------
Visit my hobby homepage!
http://www.qsl.net/xq2fod
August 12th: Improved scan quality of 52 gallery photos
--------------------------
_______________________________________________________________________

> why did you use ouble-sided PCB? is there any particular reason?

Yes, there is a good reason:

Every conductor, including resistors and capacitors, has a small series
parasitic inductance. If you place such a conductor or component close
to a ground plane, it will have some parallel parasitic capacitance,
which will tend to counteract the effect of the inductance. Bu chosing a
specific dimension and material, you can control this capacitance so
that it exactly offsets the inductance, for a given signal impedance.
For example, on normal glass-epoxy PCB, a trace of about 2mm width will
compensate its inductnace with the through-board capacitance, for 50 Ohm
signals. For this to work, the other side of the PCB needs a continuous
ground plane. 

>can it be single sided?

It can, but then you must sandwich it against the metal box, so that the
box wall serves as ground plane. If you use a single sided board in free
air, inevitably the components will look a little inductive, and this
will limit the frequency response. Still, by using very small
components, with very short connections, it should be possible to obtain
good results even at 2.4 GHz without using a groundplane. But using the
groundplane will in any case help.

>Sorry about bothering you with silly singth, but I have little expirience with RF,
>spesifically with very high ferq's...

No problem. These things are the ones that make RF work interesting, and
different from low frequency work!

Cheers,
        Manfred.

----------------------------------------------------
Visit my hobby homepage!
http://www.qsl.net/xq2fod
August 12th: Improved scan quality of 52 gallery photos
----------------------------------------------------


________________________________________________________________________
Hi Alexei!

> Just build a new version of SWR meter, but .....
> in all possible combinations it gives the same readings for bouth
> sides....

You mean that forward and reflected voltages are the same? Over what
frequency range did you test? If it happens only at very high
frequencies, it must be a layout problem. If it happens at lower
frequencies too, then you have connected something in the wrong way!

> but i've used copper clad board 0.5 mm sick, can that be the reason?

For such thin board a 50 Ohm trace is much thinner than 2mm. As long as
you use those extra small surface mount parts, and narrow traces, it
should work.

> I've use surface mounted components and positioned them really next to 
> each other, whith less than 2 mm traces...

Which size of SMD parts? The "standard" ones are best used with 1.6mm
board. The extra small ones can be used with thin board.

>  may I send you the foto of the assembly?

Yes. But try to make it smaller than 100kB, because I'm on a slow phone
connection!
 
> btw, the readings off the straight feer are 120 mV...
>  which is very reasonable, as power of the trans mitter is only 30 mW....

30mW over 50 Ohm would be 1.2V RMS. So, the forward voltage detector
should see 0.6V RMS, or about 0.8V peak. Since the diodes drop some
voltage, those 120mV look reasonable, but you should be aware that the
diodes are working in a very nonlinear range, and so the SWR is not the
real one. At this low power level you need to use some DC bias to get
correct values.
 
> can that be the reason?

I don't think so. The reflected voltage should be zero, when using a 50
Ohm load. And when leaving the load side open, of course the reflected
voltage will be the same than the direct one. So, in both cases the
output voltage levels would be correct. But at intermediate values of
SWR, the reflected voltage will look smaller than what it should be,
falsing the reading.

If you get both output voltages at the same level, when using a 50 Ohm
load, then something is wrong. It's not a problem of power level.
 
Bye,
     Manfred.

----------------------------------------------------
Visit my hobby homepage!
http://www.qsl.net/xq2fod
----------------------------------------------------