Conversion of VNA cal kit data between formats

[radar_macgyver]

I have a cal kit with data in Keysight's calibration model format, so the losses are specified as offset delay. I wish to use this with a Rohde & Schwarz VNA, whose calibration format is different. Most of the other calibration coefficients are easy enough, it's just a unit conversion. How would one do the same with the loss terms? R&S specifies loss in dB/sqrtz(GHz).

[TheSteve]

This is the info I've used in the past:

https://www.rohde-schwarz.com/us/faq/calibration-kit-conversion-hp-to-r-s-faq_78704-30488.html

Seems they expect you to ignore loss - "Loss is neglected."

[radar_macgyver]

I found that exact page too. I entered in the coefficients without the loss terms into my ZVC and got believable results, of course when measuring a thru the S21 showed a tiny bit of gain at higher frequencies (6-8 GHz). For reference, the cal kit used here is the SMA kit from Applied EM Innovations, just came in today. I will compare against a Rosenberger 3.5mm kit I have at work when I get a chance.

[cdev]

Ive used resistors to make calibration standards. I have a heat sinked resistor intended for RF that works well. It's exactly 50 ohms. Calibration standards are by design very simple and so, not exactly rocket science to make. Just keep it simple. Shorts and opens are particularly straightforward to make. Just keep shorts short. And opens just end. They are the easiest.

Ive used "gnumeric" to convert data for use in different calibration (csv) data formats. It works well for that.  Excel can be used for this too.

[radar_macgyver]

Thanks @cdev for the response, though what I'm asking about is not file format conversions, but about the formula to convert the loss terms from the Keysight format (Offset loss in G /s and Offset Z_o in ) to the R&S loss format (dB/sqrt(GHz)). The rest of the conversions for the fringing capacitance and parasitic inductance model, and the delays are easy enough since they're just a unit conversion. This is outlined in the link posted earlier by @TheSteve.

The cal kit also comes with the s1p files for the open, short and load standards (but not the s2p files for the thrus). I found the thread describing the Octave scripts used to generate the fringing capacitance and inductance polynomials, but they also output loss terms in the HP format. Applied for a Metas VNA Tools download, waiting for that to come through.

[TheSteve]

Ive used resistors to make calibration standards. I have a heat sinked resistor intended for RF that works well. It's exactly 50 ohms. Calibration standards are by design very simple and so, not exactly rocket science to make. Just keep it simple. Shorts and opens are particularly straightforward to make. Just keep shorts short. And opens just end. They are the easiest.

Ive used "gnumeric" to convert data for use in different calibration (csv) data formats. It works well for that.  Excel can be used for this too.

An DIY OSL kit isn't too bad for lower frequencies. Try a DIY kit for above 6 GHz and it gets much tougher. Even above 3 GHz is pretty tough if you want really good performance and reasonably easy modeling.
Also most high end cal kit(18 GHz +) rarely have loads that are exactly 50 ohms. They care much more about the other properties.

[cdev]

I never done anything with such high frequencies, not having access to anything that works above 7 GHz. Am hoping to try some for moonbounce purposes, however.

I could see how machining standards precisely to offer perfect load might be difficult. Simple short and open might be easier, I don't know where oe would start looking for discontinuities. I'd like to see a youtube video of the basic attempt to make calibration standards and analysis f how well they worked, or didnt work..

My VNA only goes up to 4.4 Ghz so not even the high wifi band...

Ive used resistors to make calibration standards. I have a heat sinked resistor intended for RF that works well. It's exactly 50 ohms. Calibration standards are by design very simple and so, not exactly rocket science to make. Just keep it simple. Shorts and opens are particularly straightforward to make. Just keep shorts short. And opens just end. They are the easiest.

Ive used "gnumeric" to convert data for use in different calibration (csv) data formats. It works well for that.  Excel can be used for this too.

An DIY OSL kit isn't too bad for lower frequencies. Try a DIY kit for above 6 GHz and it gets much tougher. Even above 3 GHz is pretty tough if you want really good performance and reasonably easy modeling.
Also most high end cal kit(18 GHz +) rarely have loads that are exactly 50 ohms. They care much more about the other properties.

[cdev]

Thanks @cdev for the response, though what I'm asking about is not file format conversions, but about the formula to convert the loss terms from the Keysight format (Offset loss in G /s and Offset Z_o in ) to the R&S loss format (dB/sqrt(GHz)). The rest of the conversions for the fringing capacitance and parasitic inductance model, and the delays are easy enough since they're just a unit conversion. This is outlined in the link posted earlier by @TheSteve.

The cal kit also comes with the s1p files for the open, short and load standards (but not the s2p files for the thrus). I found the thread describing the Octave scripts used to generate the fringing capacitance and inductance polynomials, but they also output loss terms in the HP format. Applied for a Metas VNA Tools download, waiting for that to come through.

What do you think about the use of spreadsheets to straightforwardly convert these file formats..I hink it is likely to work.. although iits likely not plug and play..


It is likely to support the needed math functions, though.. And its easy to understand..

[radar_macgyver]

Sure you can use a spreadsheet for this, and that's what I did for the fringing capacitance, inductance and electrical length/delay. See attached.

[radar_macgyver]

Quick follow-up:

• I did finally measure the Applied EM kit against a more trustworthy calibration kit, results are here
• I also learned that the reason the loss terms can be neglected is that most VNAs can do a SOLR (aka reciprocal thru) calibration, so the loss terms fall out.

[cncjerry]

I have recently found a bunch of sma 50ohm terminators from minicircuits on ebay for cheap, less than $10 per.  The biggest POS's in sma are sma-bnc connectors.  The cheap ones fail constantly. i only use them when i absolutely must as generally even decent ones can be wiggled for an open.

[switchabl]

The biggest POS's in sma are sma-bnc connectors.  The cheap ones fail constantly. i only use them when i absolutely must as generally even decent ones can be wiggled for an open.

I have found these to be very reliable: https://www.amphenolrf.com/901-165.html

In particular, with high quality mating connectors, contact resistance was repeatable down to a few milli-ohms. The bad news is they are ~60$ each. Unfortunately, in my experience, with RF connectors and adapters you tend to get what you pay for. Machining tolerances and surface quality matter big time for precision measurements.