On Tue, 28 Feb 2006 14:25:32 -0600, Bob Bob wrote:
The question is this. If one hangs a 1/2 wave dipole in free space I
assume it receives such that current maximums are at the centre and
voltage maxiums at the ends. Is this the case whether a feedline is
connected or not?

Yes.

If I then take a stripline cct terminated at one end with a 50r resistor
and a detect diode at the other and is a half wave long, what is the
current/voltage distribution in this configuration.

The point of your need at work is to insure that stripline presents a
50 Ohm characteristic as an untuned line, not a resonant line. Of
course, you can perform the same using a tuned line, but this seems
unlikely barring new details offered by you.

What I wonder is if it is a halfwave there may be no voltage at the
detect diode input. (It could be 3/4 wavelength when one factors in pcb
dielectric and end loading)

And this would be reason why to NOT have it be a resonant line. I
presume there's a ground in the vicinity for both this sniffer section
and the main feed (which also needs to exhibit a characteristic Z).
Such devices are generally useful over an octave range.

Thoughts?

This, of course, does require attention to factors such as pcb
dielectric as you have already taken care of end loading with the
resistor and the diode.

73's
Richard Clark, KB7QHC

Richard, Cecil and Tom

I should explain the "problem" somewhat.

The transmitter designs I believe are not very good. Perhaps they were
designed to work on a different or narrower frequency and were hastily
adapted by beancounters instead of designers. grin What is happening
is that at a certain critical band of freqs (around 11.2GHz) the output
from the detect diode is very close to zero and hence the power control
(via ALC) and calibration table dont work too well. Thinking it was a
design issue I actually moved the ref diode to the limit of its pads and
the problem resolved itself. The effect was of course to make the sensor
length electrically shorter by doing this. I was trying to establish
whether I had a 1/2 or 3/4 wavelength sensor section to maybe take it up
with engineering. (No doubt for another frequency I would move it in the
opposite direction)

I suspect there is also another issue whereby the resonant sensor
section is actually absorbing more power than it should and dissipating
it in the resistor. The effect of this being a loss of available output
power at the SMA connector.

Keep in mind that I am a production line tech rather than a designer. I
am not suppose to modify things, only test and align! (Am actually an IT
professional but had trouble finding that line of work!)

Many thanks for your input. As usual with you gents I am going to have
to work through it all slowly!

To answer you specifically Richard, the PCB is a multilayer fibreglass
thing with a largish ground being under both the output and sensor
striplines. I was surprised that the sensor line was so long and so
close when I first saw it thinking some major overcoupling might be
occurring. The tracks are maybe 3mm wide and about 1.5mm (edges) apart.
I am thinking also that the 50 ohm output is not being preserved as it
goes past the sensor stripline. I think the freq coverage for this model
is about 10.5 to 11.5GHz. Keep in mind that this really is the frist
time I have seen microwave TX's up close so my gut feelings about track
sizes/spacing may be way off.

Cheers Bob VK2YQA

On Tue, 28 Feb 2006 20:35:01 -0600, Bob Bob wrote:
I suspect there is also another issue whereby the resonant sensor
section is actually absorbing more power than it should and dissipating
it in the resistor.

Hi Bob,

That is why it is there, among other reasons.

The effect of this being a loss of available output
power at the SMA connector.

This is a symptom, and should not be a cause.

To answer you specifically Richard, the PCB is a multilayer fibreglass
thing with a largish ground being under both the output and sensor
striplines.

That is as it should be.

I was surprised that the sensor line was so long and so

On the order of 1.5 cM?

close when I first saw it thinking some major overcoupling might be
occurring. The tracks are maybe 3mm wide and about 1.5mm (edges) apart.

Sounds like a boilerplate design - which is to say right out of some
book or App. Note.

I am thinking also that the 50 ohm output is not being preserved as it
goes past the sensor stripline.

What does that mean?

I think the freq coverage for this model
is about 10.5 to 11.5GHz. Keep in mind that this really is the frist
time I have seen microwave TX's up close so my gut feelings about track
sizes/spacing may be way off.

Check the resistor.

73's
Richard Clark, KB7QHC

Hi Richard

I was surprised that the sensor line was so long and so

On the order of 1.5 cM?

It is a "U shape". The section parallel to the output stripline is maybe
1cm long (going from memory). The leg length may then put the total size
at more like 2cm.

---

I am thinking also that the 50 ohm output is not being preserved as it
goes past the sensor stripline.

What does that mean?

I am suggesting that given the proximity of the sensor section it
presents a significant Z bump. Like I said I dont have a feeling for the
track dimensions for stripline etc at microwave. Not that it wouldnt be
difficult to look it up mind you!

---

I think the freq coverage for this model
is about 10.5 to 11.5GHz. Keep in mind that this really is the frist
time I have seen microwave TX's up close so my gut feelings about track
sizes/spacing may be way off.


Check the resistor.

One of the first things done. At DC of course. Dont think I ever
replaced it. Was considering playing with it last time but since moving
the frequency also bought it back to normal operating conditon I couldnt
see it as a reason.

Cheers Bob VK2YQA

On Wed, 01 Mar 2006 09:19:37 -0600, Bob Bob wrote:

Check the resistor.

One of the first things done. At DC of course. Dont think I ever
replaced it. Was considering playing with it last time but since moving
the frequency also bought it back to normal operating conditon I couldnt
see it as a reason.

Hi Bob,

DC is so remote from the application as to be only an approximation.
There's also the prospects of reactance to consider too. When you say
you can see a frequency dependency, you are almost guaranteeing that
"tuning" has been injected into an otherwise wideband design.

From your position in the company you have two paths:
1. Announce the design is FUBAR, or
2. Find an ad-hoc solution and forget theory because you are in no
position to re-engineer the design.

#2 is a dangerous path to take for the sake of the company's
perspective, although it may be more politic if the design department
is populated with prima-donnas.

#1 will accomplish one of two things, the design will be corrected, or
you will be educated - possibly both. Most designers appreciate
hearing what your experience has revealed. Most of my techs enjoyed
pounding my designs to find the weak seams. One fellow had a small
transistor radio that he would put on top of the microprocessor to
listen to the software running. He could always tell when one of my
patches went south.

73's
Richard Clark, KB7QHC

How thick is the board material, and what material is it? The
propagation velocity depends on the permittivity (dielectric constant)
of the material. The impedance depends on the material and the
spacings and trace widths. Presumably there's a ground plane behind
the microstrip lines (else they aren't microstrip). There are many web
sites that will let you play with microstrip designs, and some that
will give you the response of a coupler like you're describing. Do a
web search for things like "directional coupler" and "90 degree
[microstrip] hybrid". But if you plot the coupler's coupling versus
frequency, you'll find it's zero at DC, increasing to a fairly broad
maximum when the freq makes it 1/4 wave long (accounting for the
velocity factor), and falling again to zero at twice that frequency
where the line is 1/2 wave long. That pattern repeats. If you account
for the response, the coupler is useful over a broad range of
frequencies, as the directivity stays good even as the coupling
decreases (if it's accurately made). You can extend the frequency
range (make the peak even broader) by "tapering" the coupling. (Easier
to see in a picture than trying to explain in words...basically a
cascade of sections, with the center one coupled most closely.)

A point to note: if you make the coupled line say 5/4 wave long at
10GHz, it will couple nicely at 10GHz, but you only have to move by
2GHz in either direction to hit a null at 4/4 and 6/4 wave long for the
same physical line length. But if you make the coupled line 1/4 wave
long, then you don't see a null till 20GHz, and the coupler should be
quite useable between 8 and 12GHz. You can make the coupled section
short by leading the ends to the 50 ohm load and the diode detector
away from the coupled section, at right angles to it, so you don't have
to worry about the length of the resistor and the diode adding in some
difficult-to-calculate way to the overall length. Perhaps they already
are done that way, but from your description that's not clear to me.

Cheers,
Tom