NanoVNA Calibration - When, Why, and How to cal a VNA

A somewhat detailed video on when, why, and how to calibrate a Vector Network Analyzer. Concepts and procedures are illustrated with a NanoVNA, but in general apply to other VNA instruments, with some changes in menus/etc.

[@marios2620]

Thank you very much for your fully detailed understandable!! tutorial!! , I really enjoy it while I am watching your tutorials!!

[@MegawattKS]

You are very welcome. Thanks for leaving a comment. It is encouraging.

[@microflite]

Thank you! There are too many nanoVNA videos made by people who don't understand its proper use. I for one, was doing my antenna measurements incorrectly. Now, I have the understanding to do it right.

[@MegawattKS]

Thanks! Glad it was helpful. The NanoVNA is an amazing instrument!

[@goofypettiger]

Well done! I just got my NanoVNA in the mail today and watching lots of videos like yours to get the most out of it.

[@MrSabram07]

I just learned you can do continuity tests with this thing

[@MegawattKS]

Yes - it goes to a low enough frequency that should be possible. It also has the ability to find the location of shorts (or opens) in a line partway to the load if that is suspected (using a frequency version of a reflectometer under Display>Transform). It's an amazing instrument comparable to the big expensive ones ! (Though the learning curve is sometimes a bit steep )

[@LtNelli]

excellent.. very clear presentation. Thanks for the video.

[@FAB1150]

That's EXACTLY what I've been looking for for while! Thanks!

[@MegawattKS]

Awesome. Thanks for the feedback. Glad it was helpful.

[@rfdave3980]

And as always THANK YOU for making these videos. I love it. Dam ... bed time. Thank you again. 73

[@thedmutz]

Excellent video thank you

[@AbuSous2000PR]

Excellent explanation; many thx

[@MegawattKS]

Thanks for the comment. Glad to hear it is helpful !

[@pilot7200able]

Thank you, very helpful information

[@MegawattKS]

Glad it was helpful!

[@RobB_VK6ES]

One aspect you did not cover is the sweep range and interpolation with the limited number of sampling points. If simply calibrated and used with the full sweep range it is likely details you are testing for will be missed. For best resolution it is advantageous to narrow the sweep range for the task at hand and calibrate with the reduced sweep range. Otherwise a very good explanation.

[@MegawattKS]

Thanks. Yes - agreed. I have had to narrow the sweep sometimes to look at things like an 80m 50 kHz bandwidth antenna. Interestingly, for the uses I've made of this unit, the interpolation has been sufficient and in that case I didn't have to re-cal (probably because I was only looking at the magnitude of S11). However, as you point out, that will not always be the case. If the coax is long for example, then the phase changes too rapidly with frequency for interpolation to work for full complex S11 and it would need to be re-cal'ed at the new, more narrow span. Other versions of the firmware or other VNAs may also have issues with the problem you point out. Indeed, the HP8753's in our lab didn't have the processing power to even do interpolation. I always had to recal those whenever changing frequency range or span. Thanks for pointing this out.

[@ramjet4025]

Its a pity that the menu did not show up on this video. Thank you for the explanation below.

[@MegawattKS]

Don't know if this will help, but here's a menu map from the nanovna site: nanovna.com/?page_id=87 . Sorry they're washed out in the video. The camera did an average exposure and the white became too white in the menu area. If other folks' videos are also washed out, that's probably why. FWIW, I have tried to be more attentive to this in subsequent videos and force the camera to expose on the menu section during the initial focus/exposure acquisition. Don't know if it's fixable in the video editing software. I will have to look to see if it can do that...

[@junior_engineer3852]

Thanks for the video! It was very useful to know not just HOW, but WHEN and WHY to calibrate NanoVNA. I think it's very important and interesting! Tell me please, what about the correct calibration of the S21? What should I do before clicking "ISOLN"? 73!

[@MegawattKS]

Thanks. That's a very good question, and I'd be happy for others to comment on as well. In our lab with Agilent/Keysight equipment, we generally just selected the omit option since the "S-parameter testsets" in the instruments had excellent internal isolation and we didn't need more dynamic range in S21 measurements we did. But for the NanoVNA, it's sometimes important to correct for isolation, especially at say 500 MHz and above if you need to see very low S21 values like -50 dB or less. My personal feeling (and what I've done sometimes) is put loads (50 Ohm terminations) on both ports before clicking ISOLN. The question is where to put the 50 Ohm terms - at the instrument connectors or at the end of any coax runs being used. Arguments can be made for either. If the coax cables are included, then there is some compensation for leakage from the cables in addition to calibrating out the NanoVNA's internal coupling errors. But a) coax coupling is usually small compared to the internal NanoVNA coupling between PC board circuit traces on the two ports, b) if the cables are included, they should probably be in the positions they'll be in for the measurement. But ... I'm not sure what happens when the device under test is not a good 50 Ohm itself on each port. So - IMO, probably just put a load (50 Ohm termination) on each port during the isolation cal step, and recognize that S21 values below say -50 dB are possibly not completely accurate...

[@junior_engineer3852]

Thanks for the detailed answer. This is good advice! Will definitely use this! 👍 If I'm not mistaken, in the instructions for NanoVNA, it was written: «Connect SMA Load to PORT1, connect second Load to PORT2, wait more than 3 seconds (wait for the screen to stabilize) and press ISOLN. Note: if you don't have second Load, just leave PORT2 open». But it baffled me. That's why I was looking for answers to my questions. Thank you!

[@rdfrq222]

Thank you for your detailed video. So before taking our NanoVNA readings of X, R, and SWR of antennas for 2200M (136 kHz), 630M (472 kHz), 160M (1.8 MHz) bands, if we are wanting to use a 5 foot extension section of RG213/8U coax from Port 1/S11, should we calibrate the NanoVNA with that section of coax connected to the Nano, and if so also use the open, short, load connectors on the end of that coax during the calibration, or is just calibrating at the Nano's port without the coax inline good enough? Finding mixed answers on this- some say calibration with the coax is not needed since it's in such low bands, others say better to do it anyway. Thank you for your help.

[@MegawattKS]

Sounds like you can probably get away without calibrating with the 5 foot section in place. 5 feet is on the order of 2 meters - a little longer if we adjust for velocity of propagation. So that's about 1/50th to 1/100th a wavelength for the worst-case 160m band. That should result in reasonably negligible rotation on the Smith Chart. 1/100 would be 7.2 degrees of rotation. The only place I can think of where this might be an issue is if the impedance being measured is significantly away from the line impedance - say trying to measure a 600 Ohm Zant using a 5 foot 50 Ohm cable feed. Then the Smith Chart rotation could change the Z readout somewhat significantly (it would show a capacitive X even if its resonant). But if it's a nominal 50'ish Ohm antenna with a 50 Ohm line, it should be fine - even if the antenna has a 2:1 VSWR... (Opinions may vary :-) )

[@hubercats]

Great video! Thank you. One question follows: Around the 19 minute mark you save the S11 cal data in the 4th slot. Why does the display still show a lower case letter “c” which I believe means the device is not in a calibrated state? - Cheers - Jim

[@MegawattKS]

Good question. According to the book "A Guide to the NanoVNA", the lower-case means its using interpolated data - i.e. the frequency range is not the same as it was during the cal. But in reviewing the video, I don't see that I changed it (50 kHz to 500 MHz was used during both cal and during verification). My guess is that it still has old T and X (transmission/isolation) S21 cal data together with the new S11 (SOL) data. So the S21 part of the cal set is still interpolated. Honestly I've never really looked at those letters. I just do the verification step to be sure its OK. Now that we have this puzzle, I'm even more inclined to not trust those letter readouts. After all, they don't tell us if the cable type/length was changed. But in their defense, they are reminding me that I didn't cal S21 this time I guess.

[@clems6989]

Calibration IS required to establish the measurement plane. And it is required if you change the freq span, the connection method, anything that alters the measurement plane. Just added fyi.....

[@MegawattKS]

I've been pleasantly surprised with the NanoVNA's ability to interpolate calibration data to new frequencies within a wider range. Of course I usually check to be sure the cal is still providing good data, using the SOL standards. And as you said, you can't change the interconnect distance more than about a few percent of a wavelength, or the phase data (Smith chart rotation) will be off for sure.

[@JulieanGalak]

Great video, just got a Nanovna, still trying to figure things out with it. Question about calibration: I tend to use BNC connectors on everything, so first thing I did was put some SMA-BNC adapters onto the device. This immediately threw the S11 trace out of whack (haven't played with anything else yet). So I need to recalibrate for the new plane, but I don't have a BNC "short" standard. Can I just solder a BNC connector closed? Or should I go and buy a professionally manufactured one? (Open is easy enough, and I have a quality 50ohm BNC load that should work for the load standard) I am mostly interested in measurements up to the 70cm ham band, so being accurate into the GHz range isn't a concern.

[@MegawattKS]

Hi. Yes - as long as your soldered short is physically short (e.g. half the diameter of the connector), it should work OK up to maybe 1 or 1 1/2 GHz. But try to keep the shorting wire length to about 1/2 the diameter of the BNC shield if possible. I'm guessing you're envisioning a Male BNC and putting a wire from the center pin to the shield on the back of the connector. Don't loop it out and back to the outside. Maybe you can use bare copper wire soldered/crimped to the center pin,, tin the wire, and then insert a small diameter soldering iron and solder down into the shield from the backend of the connector? We used BNC shorting caps like this one at work and they worked fine. www.amazon.com/Amphenol-Rf-coaxial-Shorting-Cap/dp/B00M1I2U9I

[@JulieanGalak]

@@MegawattKS - I was actually thinking of a female BNC, but I'm just going to order the cap you linked - I didn't find it in my initial search, which is why I thought about doing a DIY one. Thanks! (I think my mistake was searching for "calibration set" rather than just "shorting cap"...)

[@romario21]

hello ,I don't see vertical grid lines ,like yours .I only have horizontal lines shown. Can you please help me set my graph to have both? thanks

[@MegawattKS]

Sorry, I haven't changed anything other than the format type (logmag vs Smith, etc), and am not sure you can with the one I have. Maybe the NanoVNA model or firmware differ? This is the "-F" version with stock code build from earlier this year when I bought it (I haven't done any updates). If I discover anything, I'll try to post

[@romario21]

@@MegawattKS Thanks a lot

[@geoff37s38]

Am I correct in thinking calibration is not really required if I just want SWR measurement for antennas in the 3MHz to 30MHz range?

[@MegawattKS]

Probably, but it depends on the state of the analyzer. When I got mine, it came calibrated. The easiest way I can think of to check (or at least inspire some confidence in using it as-is) would be to setup the sweep range you want and the trace/format (e.g. S11 trace only and SWR or Return Loss format). Then put known loads on port 1 and see what we get. Left open-circuited, the SWR should be very high (e.g. > 10), and when the "Load" (50 Ohm cap that came with the analyzer) is attached to port 1, it should read close to 1 (e.g. < 1.1). Note that for HF antennas, some are very narrow-band, so be sure to narrow the span as needed. Otherwise it is possible the sweep (which is only 101 steps by default), may miss the resonance point. Hope that helps. 73

[@geoff37s38]

@@MegawattKS Thanks for reply. I have not yet bought a device but am considering a 4.3in screen model due to my poor eyesight. So, if I set range 3-30MHz and do a calibrate then this calibration should be ok for all antennas in this range? I can narrow sweep to say 7-7.3MHz, leaving calibration alone? Is I set trace to SwR only and sweep to a specific band can this setting be saved, or will all settings reset again next power up?

[@MegawattKS]

@@geoff37s38 It is hard to say for sure, since different unit variations come with different 'firmware'. With the unit I have (4.3in NanoVNA-F purchased in 2020 with firmware 0.1.5), there are 5 memories: 0 through 4. The unit appears to power-up with whatever was last recalled from memory. For example, I saved my most recent call in "RECALL 4". It power up with that. But if I do a: RECALL > RECALL 0, it changes to that and then when I power down and back up, it is in that state. But if I change sweep for example and then power off and back on, it doesn't remember that. It's back to what was most recently recalled. So not quite the same as the typical "power-up from last state" behavior that HP/Agilent units might do. (As an aside, I keep the original cal it came with in memory 0 and use the others to mess around with. The original was valid from 50 kHz to 1500 MHz and displayed S21 in Trace 1 in logmag format together with S11 in Trace 0 in Smith format.)

[@MegawattKS]

And yes - the 2020 NanoVNA-F unit I have does do interpolation, so if I narrow up, it will interpolate cal points within the new range from the coarser date. I think it may actually extrapolate off the ends if you widen up from a more narrow call range - but that is fraught with some inaccuracies at best.

[@MegawattKS]

All that said - be careful. I have seen mention of some units that don't have good memory and interpolation. One was a model SAA2N - which is not technically a NanoVNA - but kinda looks like one. I saw stories of such units in the NanoVNA and TinySA Facebook group: facebook.com/groups/368777730463838

[@donepearce]

The reason I calibrate a VNA is because I am going to make a measurement.

[@EddieVfgi]

Outstanding, thank you. I have seen a lot of videos on calibration, yours is the tops! Eddie kj4fgi

[@MegawattKS]

You're welcome ! Glad it helps :-)