开云体育

I started this topic because the original thread got hi-jacked to antennas matching. :-(
H Walker posted in Msg #4702
/DMR,
/Thanks for your contribution. A previous DFU file was uploaded for making 75 ohm system measurements
/ ("50-75 Ohm_Use transformer-10.10.2019.ra").
/The uploader made special note that in order for measurements to be accurate 50-75 ohm matching pads were required on CH0 and CH1. Does your
/mod also require matching pads/transformers, or are you applying some type of mathematical correction factor?
/
/- Herb

** Herb, I can't find that file. Could you give us a message number? Have you opened it? Thanks Jim

Dr. David Kirkby posted in msg. #4540
/1) Minimum loss pads can be used to allow calibration with 75 ohm
/calibration kits. There are 75 ohm versions of both N and BNC connectors,
/and calibration kits for both available

**David, I understand the 7 db loss pads.. Would wide band transformers work also? Coilcraft makes several, one set is sold on Mr. Kurt Poulson's SDR-Kit site and on eBay, or maybe we could roll our own , at least for 3-30 Mhz or 6, 4 or 2 meter bands. One pad/transformer for each VNA port used?

* Anyone. Assuming I understand me using pads/transformers, and calibrating at the desired impedance (75 ohms) at the test fixture, then the NanoVNA Smith chart, while showing 50 ohms is actually 75 ohms reference, and the graphs will be ref. to 50 ohms, what math operation do I do to get a 75 ohm graph and (maybe using Excel) to get 75 ohm values? URLs or pointers to web sites are welcome.

thanks,
Jim, KA6TPR

On Mon, 14 Oct 2019 at 05:14, Starsekr via Groups.Io <Starsekr=
[email protected]> wrote:

I

Dr. David Kirkby posted in msg. #4540
/1) Minimum loss pads can be used to allow calibration with 75 ohm
/calibration kits. There are 75 ohm versions of both N and BNC connectors,
/and calibration kits for both available

Yes, and F connectors too.

**David, I understand the 7 db loss pads.. Would wide band transformers
work also? Coilcraft makes several, one set is sold on Mr. Kurt Poulson's
SDR-Kit site and on eBay, or maybe we could roll our own , at least for
3-30 Mhz or 6, 4 or 2 meter bands. One pad/transformer for each VNA port
used?

Yes, transformers will work, and given the fairy low dynamic range of the
NanoVNA, transformers would be better in *most* cases. However, by their
very nature they are not very wide-band and will present a short at DC.

Depending on what your requirements are, you may determine that not adding
any resistors or transformers is perfecly acceptable.

One thing to bear in mind is that the reflection coefficient at port 2
might not be low at 50 ohms. In fact from a previous note I see posted,
this would vary depending on the actual batch of chip used. I do actually
intend trying to improve the match at port 2 to 50 ohms by a series
resistor or shunt resistor. Before even considering adding a transformer on
port 2, you should determine what it’s impedance currently is.

Something else to consider is that given the cost of a NanoVNA, it might be
worthwhile to buy a second one and change the hardware to 75 ohms. I
suspect not only requires changing a few resistors, but I have not looked
at the circuit.

* Anyone. Assuming I understand me using pads/transformers, and
calibrating at the desired impedance (75 ohms) at the test fixture, then
the NanoVNA Smith chart, while showing 50 ohms is actually 75 ohms
reference, and the graphs will be ref. to 50 ohms, what math operation do I
do to get a 75 ohm graph and (maybe using Excel) to get 75 ohm values?
URLs or pointers to web sites are welcome.

If the VNA supposed 75 ohms properly, all that should be taken care of.

Whilst someone is going to the effort of supporting an impedance other than
50 ohms, it would make sense to me to make the impedance being able to set
to any value, but defaulting to 50 ohms.

thanks,
Jim, KA6TPR

Dave.

--

Dr. David Kirkby,
Kirkby Microwave Ltd,
drkirkby@...

Telephone 01621-680100./ +44 1621 680100

Registered in England & Wales, company number 08914892.
Registered office:
Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United
Kingdom

If you change just firmware for Z0=75 ohm it won't work for 75 ohm properly. It may show you proper resistance for zero reactance circuit (such as dummy loads and resistors). But if your circuit contains any inductor, capacitor or transmission line, you will get wrong measurement. Because it's Q-factor will be affected by 50 ohm hardware port, so you will get shifted resonance and different S11 and S21 reading.

I am chasing the same issue and have been lead to this

Maybe you have it already, but since your journey seems much farther along than mine - I don't even have a nano yet - let me know if this helps.
If so I will try to understand it.
thanks

I seem to recall from the distant past, reading instructions for one of the early hobby vna devices about this.

The software with the vna (mandatory pc to use it) could automatically readjust the response. The demonstration was an ssb sideband filter with input and output inpedance of several 100ohms. The initial result seemed to be a poor filter with maximum skirt attenuation of only 20db but after the corrections were applied, was the expected 80+ db. This was at hf of course.

What was going on? Can this sort of correction be applied. How do I assess a filter with several hundred ohm in and out impedance, or do I resort to a spectrum analyser with tracking generator ( do not have one) or noise source, which is what I have done in the past. I still have that kit ( does 0-120mhz using a scope for display) for posterity.

Is this an opportunity for a firmware ( or saver) addition?

Steve L. G7PSZ

The software with the vna (mandatory pc to use it)
could automatically readjust the response.

I made naive calculation for software correction of (nanoVNA) bridges at other than balance:


Bottom line: math is doable,
but reduced sensitivity slope elsewhere and quantization errors work against precision;
for > 2:1 impedance ratios, add a transformer.

On 8/17/20 7:36 AM, Stephen Laurence wrote:

I seem to recall from the distant past, reading instructions for one of the early hobby vna devices about this.
The software with the vna (mandatory pc to use it) could automatically readjust the response. The demonstration was an ssb sideband filter with input and output inpedance of several 100ohms. The initial result seemed to be a poor filter with maximum skirt attenuation of only 20db but after the corrections were applied, was the expected 80+ db. This was at hf of course.
What was going on? Can this sort of correction be applied. How do I assess a filter with several hundred ohm in and out impedance, or do I resort to a spectrum analyser with tracking generator ( do not have one) or noise source, which is what I have done in the past. I still have that kit ( does 0-120mhz using a scope for display) for posterity.
Is this an opportunity for a firmware ( or saver) addition?

There's a couple ways to approach this:
1) make a test fixture and cal kit that has 100 ohm loads - run the cal, and it should automatically fix everything - I would be concerned that various displays wouldn't "know" about 100 ohms, so displays of R and X would be wrong, even if S11, S21 are correctly displayed. The "center" of the Smith chart would be 100 ohms.

2) calibrate with 50 ohms, and post process to change to 100 ohms - one could implement such a thing in a program like NanoVNA-Saver, and that's probably the cleanest way.


Ultimately, though, for both of these the measurement uncertainty will maybe be slightly worse - the "bridge" in the NanoVNA is designed for 50 ohms, so you'll always get reflected power and the transmitted power will always be less - How that would quantitatively affect your results, I can't say.

3) build a test fixture that has a 2:1 impedance transformation (non trivial, since it's a transformer with a 1.414:1 turns ratio).
Now you're back to sort of a #1 scenario.


I note that this is of some interest to people making measurements on twisted pair cables (i.e. Cat 5), since they're 100 ohm nominal impedance balanced. And a TDR is a great thing for finding places where the installer bent the cable too tightly or some similar problem.


On a fancy analyzer, they'll provide software inside to change the reference impedance, so you can do a cal with 50, measure 100, display with 100, or any other combination. Here, I think NanoVNA-Saver is the right kind of place. Keep the inexpensive box with limited firmware space simple. Put the fancy features in software that can run on any platform.

For a 2:1 impedance transformer, 14:10 turns ratio gives 1.96:1 impedance ratio which may be close enough, or 17:12 turns ratio gives 2.007:1 impedance ratio (<0.35% error). A few minutes playing around with a calculator will probably find even better approximations.

73
John
M0JBA

Hmmmm,

I have a question.

The test/ education board available for about ?15 has simple loads, capacitors attenuators, etc with leads which have those inferal minute connectors to select which circuit you wish to play with. Each little circuit has the expected screen display overlaid on the circuit board. In my arrogance (ignorance) I have resisted buying one.

One of those little circuits is a ceramic filter. I presume its input and output impedance is anything but 50 ohm, more like 200-500 ohm. What happens there? Or is the filter especially selected for 50 ohm termination?

Steve L. G7PSZ

On 8/17/20 10:33 AM, John Baines via groups.io wrote:

For a 2:1 impedance transformer, 14:10 turns ratio gives 1.96:1 impedance ratio which may be close enough, or 17:12 turns ratio gives 2.007:1 impedance ratio (<0.35% error). A few minutes playing around with a calculator will probably find even better approximations.

yeah, I'd just go hunting through the Minicircuits or Johansen catalogs and look for something that does the function - I have no idea how they make them internally (well, I know it's some windings on a core) - but wideband transformer design is partly an art, partly engineering.

That would also solve a "unbalanced test equipment to balanced DUT" issue.

In the OP's case, measuring a filter, there's sort of two cases:
1) "what the heck is this unlabeled thing" and for that the "software correction in NanoVNA Saver" strikes me as the easy way
2) "adjusting a filter or doing manufacturing checkout" - it's over a narrow frequency range, so you can probably find a transformer that would work.


Minicircuits ADT2-1T+ goes 400kHz to 450 Mhz and 2:1 as an SMT device, for $3.85. Stick it on a board, solder it to the SMA connector, however you fixture it.

FOr a bit more ($5.85) you can get the T2-1-X65 in a 6 pin DIP, goes from 50kHz to 600 MHz.
Bigger pins to solder to if that matters - the pins are 0.1" apart (just like the SMT device) and on a standard "solder in" SMA, the center post is about 0.14" from the corner posts.

Minicircuits DOES sell a board for things

(with the fascinating note that you have to use your PNA's "impedance conversion function" ) - but they cost >$100.

73
John
M0JBA

On 17 Aug 2020, at 18:19, Jim Lux <jimlux@...> wrote:

On 8/17/20 7:36 AM, Stephen Laurence wrote:

I seem to recall from the distant past, reading instructions for one of the early hobby vna devices about this.
The software with the vna (mandatory pc to use it) could automatically readjust the response. The demonstration was an ssb sideband filter with input and output inpedance of several 100ohms. The initial result seemed to be a poor filter with maximum skirt attenuation of only 20db but after the corrections were applied, was the expected 80+ db. This was at hf of course.
What was going on? Can this sort of correction be applied. How do I assess a filter with several hundred ohm in and out impedance, or do I resort to a spectrum analyser with tracking generator ( do not have one) or noise source, which is what I have done in the past. I still have that kit ( does 0-120mhz using a scope for display) for posterity.
Is this an opportunity for a firmware ( or saver) addition?

There's a couple ways to approach this:
1) make a test fixture and cal kit that has 100 ohm loads - run the cal, and it should automatically fix everything - I would be concerned that various displays wouldn't "know" about 100 ohms, so displays of R and X would be wrong, even if S11, S21 are correctly displayed. The "center" of the Smith chart would be 100 ohms.

2) calibrate with 50 ohms, and post process to change to 100 ohms - one could implement such a thing in a program like NanoVNA-Saver, and that's probably the cleanest way.


Ultimately, though, for both of these the measurement uncertainty will maybe be slightly worse - the "bridge" in the NanoVNA is designed for 50 ohms, so you'll always get reflected power and the transmitted power will always be less - How that would quantitatively affect your results, I can't say.

3) build a test fixture that has a 2:1 impedance transformation (non trivial, since it's a transformer with a 1.414:1 turns ratio).
Now you're back to sort of a #1 scenario.


I note that this is of some interest to people making measurements on twisted pair cables (i.e. Cat 5), since they're 100 ohm nominal impedance balanced. And a TDR is a great thing for finding places where the installer bent the cable too tightly or some similar problem.


On a fancy analyzer, they'll provide software inside to change the reference impedance, so you can do a cal with 50, measure 100, display with 100, or any other combination. Here, I think NanoVNA-Saver is the right kind of place. Keep the inexpensive box with limited firmware space simple. Put the fancy features in software that can run on any platform.

Here is an idea for those who only want to make VSWR and RL measurements at other than 50 ohms using their NanoVNA and are not overly concerned with accuracy.

The company RigExpert has a software package (AntScope2) that they recently modified to work with the 2.8" NanoVNA. It has an option in the settings to set the system impedance for the SWR and RL calculations to 75 ohms. I assume what they do is calculate the DUT complex impedance from the NanoVNA reflection coefficient and then compute a 75 ohm complex reflection coefficient using these complex impedance values. From there they compute the VSWR and RL in the usual manner. I ran a couple of tests using a 75 ohm system impedance. The first was with the 50 ohm cal load and the second with a 25 ohm load. The VSWR and RL results were pretty good and I put them in the attached file.

Roger

Those little connectors on the VNA educational boards apparently rip off their pads way too easily.
This looks better to me, a proto board with 6 sma's for $8:

That plus some axial leaded R's and C's and L's should be fine for learning how things work
at moderate frequencies of 30mhz or so. And might actually be useful as a project board.

Jerry, KE7ER

99:70 gets pretty close, but I wouldn't want to wind it.

73,

Gordon KF5JWL

For those of you who would like to use the resistive pad to match 75 ohms to the 50 ohm nanovna, the series resistor is 25 x sqrt of 3 (about 43.3 ohms) and the shunt resistor on the 50 ohm side is 50 x sqrt 3 (about 86.6 ohms) . With that you can use 1% and 1/2% resistors and get as close as you want! Or you can distribute the resistance around the connectors to minimize inductance and capacitance.