Network Analyzer Connection/Calibration

[smithmat]

I'm new to working with vector network analyzers (Rohde & Schwarz ZNB4) and would like to ensure that the measurements I'm making are accurate.

The project I'm working on has two phases. In the first, I'd like to measure the impedance of an unknown load. The connection is made directly to the VNA using a 6" length of RG-58 which has been "doctored" such that it terminates at the load using flying leads. The first question, then, is should I use this connector in the calibration by simply removing the load and making a reflected osm by opening, shorting, and terminating the leads with a 50-ohm discrete resistor? Would it be better to remove this connector and use a cal. kit directly off the port?

After calibration using the latter method, I took a measurement and found that the load is far less than 50 ohms at its design frequency (around 12 +j10 ohms). For this reason it was advised that I check into the offset de-embedding function of the VNA to reconfigure the port to simulate a matched line. Is this something that should be done?

In the second phase, the load will be driven by the VNA after adding a 50 dB amplifier and about 6 ft. of coax. Will I need to recalibrate?

 

[The Electrician]

I'm new to working with vector network analyzers (Rohde & Schwarz ZNB4) and would like to ensure that the measurements I'm making are accurate.

The project I'm working on has two phases. In the first, I'd like to measure the impedance of an unknown load. The connection is made directly to the VNA using a 6" length of RG-58 which has been "doctored" such that it terminates at the load using flying leads. The first question, then, is should I use this connector in the calibration by simply removing the load and making a reflected osm by opening, shorting, and terminating the leads with a 50-ohm discrete resistor? Would it be better to remove this connector and use a cal. kit directly off the port?

After calibration using the latter method, I took a measurement and found that the load is far less than 50 ohms at its design frequency (around 12 +j10 ohms). For this reason it was advised that I check into the offset de-embedding function of the VNA to reconfigure the port to simulate a matched line. Is this something that should be done?

In the second phase, the load will be driven by the VNA after adding a 50 dB amplifier and about 6 ft. of coax. Will I need to recalibrate?

What is the design frequency?

What is the nature of your "50 ohm discrete resistor"? Is it a coaxial resistor?

 

[Mikebits]

What frequency? The cal kit will do no good once you put on the pig tail. So I would say find another solution to connect to the circuit under test, such as solder a SMA connector to the circuit under test. But if you must use the pigtail, then yes calibrate with the pigtail using your method above. But for the load use a small SMT 50Ω resistor. When shorting keep the short very short.
Instead of using rg58 you should use rg316.

 

[smithmat]

What is the design frequency?

What is the nature of your "50 ohm discrete resistor"? Is it a coaxial resistor?

In this case the frequency range is 3.5-6 MHz and my matching load is just a through hole carbon film resistor with minimal leads. I'll swap it out for an SMT.

I'm confused about trying to use an SMA connector. It makes sense in that it would preserve the integrity of the transmission line, but with the impedance mismatch under either circumstance I don't see how the configurations would differ unless the elimination of the 2" pigtails would be that significant. I should also note that the leads terminate at the load in very close proximity, so they are wound as a twisted pair all the way to the load.

 

[WTP Pepper]

Ideally you need to use a cal kit supplied by the manufacturer on the end of the transmission line.
SC/ OC and 50ohm cals can be made with SMA panel mount connectors. SC flood the back of the connector with leaded solder, OC cut the pin off at the surface and 50ohm use a carbon composite SM resistor from the pin to connector ground and cut the pin as far down as poss.
On the OC and 50ohm, dab a bit of epoxy on to prevent moisture effects and smudgy fingerprints that can affect the measurement.
Should be good upto a few GHz.
We do this at work as like any other connector, they get lost and are ridiculously expensive to replace. Especially from Rhode & Schwartz.

Use decent cables. RG58 like UR167 is dreadful. For calibration and reliable measurements then RG 316 is OK or solid.

 

[Mikebits]

Agreed the cal kit is ideal. I was gonna mention that we made our own cal kits same as Pepper described above. Not the most accurate method compared to the cal kit but a far cry better than using pig tails.

As a learning tool. Cal your analyzer using the cal kit, then set to S-11 and connect your pig tail with 50 load. Now just bend the coax around and watch the display. It will move all over the place.