[TowerTalk] Antenna analyzer

[Ian White GM3SEK]

Jim Lux wrote:
>On 7/18/12 1:49 AM, Ian White GM3SEK wrote:
>> A few points from this and earlier postings....
>>
>> 3. True Vector Network Analyzers that do offer - and indeed, require -
>> OSL calibration at the exact location where the measurement is being
>> made. The instrument then compensates for its own inaccuracies and
>> presents the user with a much truer result. This is a defining feature
>> of the true VNA that kicks it into a totally different league of
>> accuracy.
>>
>> (For example, there is almost no difference in accuracy between the
>> home-built N2PK VNA and a high-end laboratory VNA. The only significant
>> differences are in extremely difficult measurements where it is highly
>> likely that both analyzers are wrong.)
>>


>I would suggest that there is a significant difference between two
>subflavors of the latter, depending on whether they used a tuned RF
>detector or not. The TAPR VNA, for instance, has a broadband detector
>scheme, so while it's great at measuring something like S21 of a filter,
>if there's out of band interfering signals (e.g. measuring S21 between
>two antennas), then there's a potential problem.
>
>I think, also, that the lab VNA will probably have somewhat better
>dynamic range than the less expensive kilobuck ones.  For the vast
>majority of uses (as you point out), whether you can accurately measure
>the gain and phase of an S21 of -60dB is probably not a big deal.

Thanks for the thoughtful comments, Jim. That last point is entirely 
fair: we *should* be able to connect an "antenna analyzer" to an 
antenna, without it falling over due to strong received signals.

The TAPR VNA is no longer available and seems to have been left behind 
by other, more advanced designs.

The dynamic range of the N2PK VNA is much better because it is 
essentially a robust direct-conversion receiver with a very narrow 
detection bandwidth. A dynamic range of 80-90dB is routinely achievable 
on the bench; and when connected to an antenna, out-of-band signals need 
to reach the blocking threshold in order to upset impedance measurements 
significantly. The main spurious responses of the N2PK involve LO 
harmonics, but they are still at a low level and they only appear in 
sweeps covering more than a 2:1 frequency ratio. Next along, the DG8SAQ 
VNWA is essentially an SDR using audio detection in a sound card, so 
once again the dynamic range can be quite good.

The big step up in cost and complexity is from those simple 
receiver/detector architectures to a full superhet design. Superhet 
architectures are the main reason why laboratory VNAs have a better 
dynamic range across their entire frequency coverage - but also the main 
reason why they cost tens to hundreds of kilobucks.


>Most of what you get with the lab box is better user interfaces, built 
>in computational features, and better remote control/export of data. I 
>don't think there's any LabView or Matlab drivers for the kilobuck 
>class VNAs. The latter sort of thing is what you get when you spend the 
>10-20k for the latest Agilent, R+S, or Anritsu portable boxes, which 
>performance wise are comparable to any of the tuned detector kilobuck boxes.
>
Well, not entirely. PC-driven test equipment has the potential to be far 
more flexible than an all-hardware box, because you can choose between 
several different kinds of control software to meet your needs of the 
moment. For example, N2PK wrote his original control software for DOS, 
and it was very much "bare-metal" to meet his needs as the original 
hardware developer. Then G4PMK and I wrote something completely 
different: VNA 4win was a "virtual front panel" that behaved much more 
like the hardware VNAs that we already knew. But the big difference with 
the "virtual front panel" was that whole areas could be completely 
reconfigured to give an optimized control interface for each specific 
function. That is something a hardware panel just can't do. Next up was 
G8KBB with MyVNA, which gave the same N2PK box a completely different 
UI, more of a "Windows" look that has been followed by the AIM software 
and others.

Only PC-controlled test equipment can achieve such total changes in 
personality... and not forgetting the potential for remote control using 
your existing wireless network.

As for the LabView and MatLab drivers... I am not aware of any, perhaps 
because the amateur approach is more commonly to hack together a 
special-purpose program to meet one specific need. But there's no doubt 
that more powerful all-purpose drivers could be written if needed. On 
the output side, there are no problems as the VNA control software can 
already dump all of its data out to a file.

>Oh yeah, you can also get an E-Cal widget for your Agilent, which is
>pretty convenient.  One could build such a thing pretty easily, though.

 From an amateur point of view, there isn't much need for a gadget to 
speed up the calibration cycle because calibration doesn't use up a lot 
of your time anyway. And before you could contemplate building an E-Cal 
unit, you'd need a precision-calibrated VNA to bootstrap yourself into 
the project.



-- 

73 from Ian GM3SEK
http://www.ifwtech.co.uk/g3sek