Hi,

I'm in the process of designing a simple scalar network analyser. The
synthesizer will use 2 PLLs. One tuning from 1.6 to 2.6GHz and the
second on locked to 1.6GHz. I hope to create a DC to 1GHz output. I'll
use Minicircuits or similar for the broadband VCO. But it seems a
little wastefull to use the same thing for the fixed frequency PLL. So
here's the question, what would you experienced home brew gurus
recommend?

The oscillator phase noise does not have to be perfect, but it should
not be much worse that the Minicircuits VCOs. The cumbersome
old-fasioned multiplied crystal is not really needed. So what other
options should I look at?

Thanx for ideas,
74,
VK3FLP

3flp wrote:

Hi,

I'm in the process of designing a simple scalar network analyser. The
synthesizer will use 2 PLLs. One tuning from 1.6 to 2.6GHz and the
second on locked to 1.6GHz. I hope to create a DC to 1GHz output. I'll
use Minicircuits or similar for the broadband VCO. But it seems a
little wastefull to use the same thing for the fixed frequency PLL. So
here's the question, what would you experienced home brew gurus
recommend?

The oscillator phase noise does not have to be perfect, but it should
not be much worse that the Minicircuits VCOs. The cumbersome
old-fasioned multiplied crystal is not really needed. So what other
options should I look at?

Thanx for ideas,
74,
VK3FLP

Hi,
That sounds like a very interesting project. Can I try to persuade you to
make it a vector network analyser? After considering making such a thing,
I reckon that at least a third of the work is in making the signal source,
and so once you've gone to that trouble, you might as well make a VNA which
then allows proper calibration, so your return loss bridge etc. doesn't
actually have to have good directivity, just be stable over time.

Regarding your question, I think two full PLLs is probably the best
solution. If you don't need such wide tuning range for one (or both) of
them, you could consider one of the Analog Devices PLL chips with
integrated VCO.

Usually in a mixer, one of the inputs is driven hard and the other one is
driven weakly. If you drive both inputs of your mixer hard then I guess
that in addition to the wanted output you'll see the third harmonic of one
oscillator mixing with the third harmonic of the other, giving you a spur
at three times the wanted output frequency. What are your plans in order
to get a clean output?

Chris

3flp wrote:

Hi Chris,

I am planning to make it a VNA in a later stage. For now, I just need
something to check antennas, filters, etc. So a scalar analyzer with a
good return loss bridge will do OK.

After some thought & googling around, I'll just use the original
suggestion for now. Two identical VCOs + PLLs. So in effect, I am
optimising the design for speed & easiness :-). Not for cost or
performance...

I am aware that the mixer will produce various combinations of LO & IF
frequencies. I am hoping to optimise the power levels to get the 2nd
and 3rd harmonic down to -50dBc. That should be enough for return loss
measurement. For filter response measurements, this is probably not
good enough, so I am thinking about adding a switchable bank of low
pass filters at the output. But this is slightly messy, so I'll do it
later, and only if I convince myself that it's needed.

73
Fil
VK3FLP

Hi Fil,
I think that if you use a mixer or sampler as a detector then it may well be
possible to make the receiver relatively insensitive to harmonics in the
source signal but if you use untuned detectors then it would be more of a
pain.

If you were able to do proper VNA type calibrations then you would not need
such a good return loss bridge, cables, connectors etc. I wonder whether
in this age of cheap computing power, the VNA might actually be easier to
build, for a given level of directivity after calibration.

Good luck and let us know how you get on.
Chris

3flp wrote:

Hi,

I'm in the process of designing a simple scalar network analyser. The
synthesizer will use 2 PLLs. One tuning from 1.6 to 2.6GHz and the
second on locked to 1.6GHz. I hope to create a DC to 1GHz output. I'll
use Minicircuits or similar for the broadband VCO. But it seems a
little wastefull to use the same thing for the fixed frequency PLL. So
here's the question, what would you experienced home brew gurus
recommend?

The oscillator phase noise does not have to be perfect, but it should
not be much worse that the Minicircuits VCOs. The cumbersome
old-fasioned multiplied crystal is not really needed. So what other
options should I look at?

Thanx for ideas,
74,
VK3FLP


Hiya,

I'm using an AD9951 based 0-180MHz DDS as signal generator fed into four
( or if anyone want more ) AD8343's mixers - connected as doublers. The
sweep I get is supposedly 0-2.8G but I can only verify up to 2.5G.

the harmonics suppression in the AD8343's are roughly around 50 to 60 dB,
absolutely good enough. You do need a couple of directional couplers ( or
bridges ) as considerations need to be taken depending on the frequency.
I use tweaked broadband RF transformers from Coilcratft as directional
couplers. good for 1200MHz +. If you write the measuring SW yourself, you
can add the needed calibration additions as well..

There are sourcecode for this available on the net with the AD8302 as a
detector. You will have an absolute limit of 60dB dynamic range and abt
2.7GHz max... Good enough for me tho'...

Quick and dirty? Yes, but it's cheap, easy to build and do the job.

Cheers

Dan / M0DFI