[time-nuts] DIY VNA design

[Attila Kinali]

On Sun, 21 Aug 2016 04:46:10 +0000 (UTC)
Bob Albert via time-nuts <time-nuts at febo.com> wrote:

> I was interested in this, but my needs are mostly below 100 MHz.
> I wonder what could be done similarly for this lower range...

As Orin mentioned, there are some designs for that range out there,
best known are probably the two Orin listed (N2PK and the VNAW by DG8SAQ).
Although these are good designs, they are not as good as the one by
Henrik Forstén. Henrik addresses many issues that the other leave out
for simplicity. 

What I would do instead is use Henrik's design and do some adaptions.
There are four parts that limit the frequency at the lower end:
the signal sources, the filters for the sources, the mixer and
the directional couplers.

For the signal source there are two choices: DDS and down-mixing.
The DDS is probably the obvious choice and delivers good results,
but limits the maximum frequency if you have price limit. 
The down-mixing approach uses one of the PLL's with VCO as the
original design uses, but only within a limited range, eg around
200MHz. This signal can then be down-mixed using a crystal oscillator
(or another PLL+VCO) and a suitable mixer (eg LTC5512 or a DIY diode mixer).
Advantage of this is, that the spurs of the PLL+VCO can be surpressed
to a large extend, as the frequency range is quite narrow relative to
the output frequency of the PLL+VCO.

For the directional couplers, the approach used with Henriks design
will not work for low frequencies, as this type of coupler needs a length
of approximately lambda/4 to work optimally. I.e. they would become
unweildingly large. The two choices I am aware of for the lower frequency
ranges are transformer based directional couplers or resistive bridges.
Transformer based couplers have the disadvantage of a non-flat frequency
response and an upper and lower frequency limit, given by the characteristics
of the transformer (number of windings/inductance and the used ferrite). 
Their advantage is that they have very little loss. Resistive bridges on
the other hand have a loss of 3db (respectively a -6dB signal at each output),
but are totally flat down to DC and up to several hundred MHz or even GHz if
RF resistors are used.


Most of the above mentioned methods have a lower frequency limit somewhere in
the range of 20kHz and ~100kHz. If you want to go below that limit, you will
need to adapt the circuit further:
For the signal source the DDS approach is the only one that will result
in a good SNR at a reasonable price. Easiest way to go is to use a 16bit
DAC at >1MHz and an uC or FPGA to feed it (but use some low jitter oscillator
as clock source for the DAC). The other components in the signal path
that are limiting are the baluns and mixers. I would get rid of those two
all-together and digitize the signal from the directional couplers directly
using an ADC with >1Msps and 16-18bit. If you limit yourself to the range
of 10Hz-20kHz, you can do all this using audio ADC/DACs and get a very
high performing system.

				Attila Kinali

-- 
Malek's Law:
        Any simple idea will be worded in the most complicated way.