NanoVNA V2+4 SMA Cal load

[mawyatt]

I didn't start the thread asking the question and suggesting to know the reason why.   No need to get upset because you don't have data or that I question your thoughts on the matter.  I acknowledged I didn't know when I wrote " My guess is they found ..."    I made no claims otherwise.   Why would I contact anyone.  I've stated,  I don't have a need to measure to such accuracy.   

If you are not sure what the designer have explained to you in your private conversation, then why ask in the public forums.  Just ask them for further explanation.

I asked here about the 51 ohms load before I contacted the OEM. Why, because I thought this was an out of spec load and when I contacted the OEM I conveyed such. I got a reply and since you are reluctant to contact the OEM to answer your own questions and misbeliefs, here's a snip, "the cal load is specified at 51 ohms,..... this is to tune for a better RL at higher frequencies..."

Your measurement and Doug's confirms the 51 ohms is specified as stated by the OEM.

Later I asked the OEM about whether I should only use the 51 ohms load for cal and got a response "you can use 50 ohm loads". Much later I've asked other questions and awaiting a response.

I am not sure what your picture is showing.  Is this what was supplied with your V2+ for a short??  If so, that not even close to what they supplied with mine.

No this was in response to your home built SMA short, I built this SMA short as part of my 1st home built SMA cal kit before I received proper ones. Note this is built with pins on both the male and female SMA ends, the normal female end was filled with solder past and heated which surrounds the pin creates a good short to the SMA case. This SMA is the "R" type and also has hex nut flange to help hold the body in place and prevent rotation. Make sure the dielectric is teflon or other high temp insulator. Don't know if it's better or worse than the shorts that come with the kits, but sufficed until I got the supplied SMA cal kits.

Later these various cal kits will be taken to the university where I was an adjunct and one of my former grad students will help with the transfer. So I should have some pretty good transfer reference cal kits to work from.

Attached looking at SWR for three different resistors.   If you think that yellow is the best part, you would be wrong.  Red is actually the best of the three.  Because SI=SO.

Notes above in purple.

Best,

[joeqsmith]

snip, "the cal load is specified at 51 ohms,..... this is to tune for a better RL at higher frequencies...
 "you can use 50 ohm loads".  "

No doubt the parts are 51 ohms,  they offer a better return loss at GHz + compared with the 50 ohm standard supplied with my Nano or that you can use 50 ohms.  I don't think we learned much from any of that.   My guess is they used the 51 ohm part because the part had acceptable return loss and met their cost goal,  nothing to do with the V2+ hardware.   Maybe they told you otherwise but I don't know.

No this was in response to your home built SMA short,

Just to be clear,  that's the short supplied with my V2+, no home made.  I just machined up that nut from some hex stock and soldered it to the back.   


Here's a relic.   47 ohm load connected to my old SWR meter and home made amp.  The needle is hardly off 1:1.   I don't remember my antennas ever being tuned this good.      

[radiolistener]

Does anybody have any experience with these parts or know where I can find more info or a datasheet?

I bought two used RF resistors on aliexpress. They are used, but looks good. They have the following marking:
- 32A1213F FLORIDA RF C0C, measured with multimeter as 49.79 Ohm
- G150N 50W4B, measured with multimeter as 50.27 Ohm, VSWR=1.03 @ 200 MHz, VSWR=1.13 at 500 MHz

The price is just 1 USD for resistor. Seller wrote that they are 150W. But I'm skeptic about that, I suspect they cannot survive at 150W.

Both have pretty good VSWR. One with FLORIDA RF marking has a little better VSWR, but not much (lost my measurement, so cannot provide exact values at the moment, something like VSWR=1.01 @ 200 MHz).

[mawyatt]

Any ohmic (DMM) measurements of Precision Loads?

in VNA world i guess this is mute. you can measure precision in DC (not everybody have your precision kelvin setup) if you want and have the standard swings like aloha dance in GHz range.

If you follow the thread you will find I wanted some DCR measurements to question the V2+ supplied 51 ohm load, I was concerned this was an out-of-spec load and wanted to convey this to the OEM. The OEM response was this is specified as 51 ohms to improve the RL at higher frequencies. I know that a DCR reading isn't going to hold up at 1GHz, or 100MHz, but likely to have an influence below 10MHz. At 1MHz 1 ohm of inductive reactance is ~160nH, that's a lot of unaccounted for inductance, 2.4K shunt capacitive reactance is ~66pF, that's a lot of unaccounted for capacitance, these are the values to skew the 50 ohms load away from 50 by ~ +-1 ohm. This was the point to start out at near 50 ohms so the LF measurements are more in line and referenced to 50 ohms, not 51 ohms.

However, the OEM explanation was that 51 ohms load reference is specified and gives a better HF RL. You can still use the standard 50 ohms loads but maybe with a slightly inferior HF RL.

Now that I know this I can make the appropriate load selection when doing a cal based upon what the intended measurements are. So for GP work it doesn't matter, for HF work I'll use the 51 ohm supplied load, for precision LF work I'll use a 50 ohm load with a good DCR (read not +- 1ohm!) that behaves well in the LF area.

Anyway, this was the reason for the initial question about the load DCR readings....wasn't sure if the V2+ supplied load was in spec, evidently according to the OEM it was almost spot on, as were the others measured.

Best,   

[Mechatrommer]

At 1MHz 1 ohm of inductive reactance is ~160nH, that's a lot of unaccounted for inductance, 2.4K shunt capacitive reactance is ~66pF, that's a lot of unaccounted for capacitance

i'm not sure where you get the 160nH and 66pF figure. you are implying 2wL = 1ohm reactance @ 1MHz. thats not what i figured when 1 ohm deviation Load is simulated, assuming perfect zero stray capacitance (1st picture). 1 ohm reactance only achieved past 10GHz assuming on a perfect 50 ohm transmission line medium and about 1mm distance from measurement plane. increasing offset length will make it aloha jump around that 1ohm.j the worst and VSWR is invisibly changed from 1 (flat at maybe 1.05 from dc to daylight). we'll need larger deviation from 50 ohm to make reactance worsen. otoh 66pF shunt stray capacitance however will make it pretty useless as Load standard past few KHz even if perfect 50 ohm resistance and 0ps offset, which is not what we figured out from our provided CAL kit from NanoVNA, even the crappiest Load that i can make.

[mawyatt]

At 1MHz 1 ohm of inductive reactance is ~160nH, that's a lot of unaccounted for inductance, 2.4K shunt capacitive reactance is ~66pF, that's a lot of unaccounted for capacitance

i'm not sure where you get the 160nH and 66pF figure. you are implying 2wL = 1ohm reactance @ 1MHz. thats not what i figured when 1 ohm deviation

This is for a measured load, a 50 ohm resistance and a 160nH series inductance should produce a measurement of 50 +j1 ohm result @ 1MHz. A 50 ohm resistance and parallel 66pF capacitance should produce a 50 -j1 ohm result at 1MHz. My point is if I had calibrated to 51 ohms instead of 50 ohms, these measurements would be more inaccurate at this low frequency. 160nH is a large residual inductance and possible with a leaded resistor (~130mm long), 66pF is a large residual capacitance and likely more than stray capacitance of a leaded resistor or test fixture.

Anyway I got the answer I wanted about the 51 ohm load (thought this was an out-of-spec part, but now know it is the OEM specified load with the V2+). The V2+ SMA load has a rather large rear can (other SMA load is a typical short flat stubby can), which could indicate other effects (large physical R size?) helping the HF RL, don't know but do know it's spot on 51 ohms from the start.

Best,   

[virtualparticles]

I measured the supplied cal kit using a real VNA in the lab and found the load to have a return loss of about 22 dB. Pretty much unusable. You'll have 3.3 dB of return loss error at 12 dB.

[virtualparticles]

This video is preposterous  . You CANNOT get good results using a bad calibration kit. Notice how the narrator never actually chose a calibration kit from the VNA menu. Basically the mathematics will force anything you apply to the spigot to match the chosen calibration kit definitions. The uncertainty of the measurements, particularly reflection measurements are ENTIRELY a function of the accuracy of the calibration kit. See the video below for more information.

https://coppermountaintech.com/reflection-vs-transmission-accuracy-in-vector-network-analyzer-measurement/

Also this first webinar in the "Master's Series"

https://coppermountaintech.com/webinar-series/

[Mechatrommer]

I measured the supplied cal kit using a real VNA in the lab and found the load to have a return loss of about 22 dB. Pretty much unusable. You'll have 3.3 dB of return loss error at 12 dB.

i remember chatting with Mr Kirkby that measurement uncertainty is quite high when try measuring Load accurately due to very little reflection signal coming back to VNA. i dont entirely understand what he meant though, or maybe i just explained it wrongly. but i guess... (continue below)

...You CANNOT get good results using a bad calibration kit. Notice how the narrator never actually chose a calibration kit from the VNA menu. Basically the mathematics will force anything you apply to the spigot to match the chosen calibration kit definitions. The uncertainty of the measurements, particularly reflection measurements are ENTIRELY a function of the accuracy of the calibration kit. See the video below for more information.

what i understand so far, real care must be taken on characterizing Open and Short standard (especially Open), thats why (older) HP "real" VNA have to support fringing capacitances effect of the Open. where Inductances effect (polynomial terms) is neglected for the Short and esp the Load, there is "none" model for it, not even the actual DC resistance nor offset delay. Kirkby CAL kit also doesnt provide Inductance terms (coefficients) for the Short, only its offset delay/length probably i guess the VNA Kirkby is using (20GHz calibrated HP VNA) doesnt support it. from this i concluded, the accuracy of the Load is lesser critical compared to Short and esp the Open. today's VNA may support Inductance effect for the Short, but i'm yet to see a common practice (in literatures at least) try to model the Load with some fancy stray/virtual elements model. Probably now its called "Arbitrary" load. the most complex model for the Load i can see so far is in Anritsu literature. and reflected in the above snapshots i attached. even that its just a simple shunt capacitance and series inductance, i have to make my own "hybrid" model by combining model for both Open and Short into the Load since some Load, ie diy and cheap version are not ideal 50 ohm at HF, so i need to model them as well even though some VNA (esp cheaper one) may not support it. i guess this is how modern VNA model the "Arbitrary" standard (if without "data-based" measured data) but i'm just imagining, i have no established literature backing me on this.

Luckily for Kirkby kit, he provided the S11/S21 plot (data-based) for all of his kit (SOLT) including the Load (not ideal as well past few GHz wrt HP CAL Kit he's using) so i can later reconstruct their coefficients when necessary when a VNA can support the model. having said this, there are countless calibration methods more out there, one of fanciest name is "unknown through" and less fancy is "tee" method, but lets not get into that i havent read about those methods. all of them have math and matrices behind them as a "backing proof". 

[joeqsmith]

Kirkby CAL kit also doesnt provide Inductance terms (coefficients) for the Short, only its offset delay/length probably i guess the VNA Kirkby is using (20GHz calibrated HP VNA) doesnt support it.

Good to know.  I support all the HPAK terms except offset loss and delay for both the thru and load with the V2+.  He should work with our local friend who helped out with their MatLab scripts to characterize them.   No big deal and I think the have added some documentation after I went through it. 

This video is preposterous  . You CANNOT get good results using a bad calibration kit....

  I'm pretty sure that was the whole point.  You may think you're getting useful data but yeah, no.    Shit in equals shit out.   This pretty much sums up my early attempts of playing at higher frequencies.  I was up and down converting the VNA, using a home made cal kit with an ideal model.   I was impressed with myself, until I learned some basics and how foolish I was. 

[Mechatrommer]

I support all the HPAK terms except offset loss and delay for both the thru and load with the V2+

you (and VNA) will need to support offset loss term if we want to improve model matching to measured data at high frequency, imho from playing with my tuning tool.

He should work with our local friend who helped out with their MatLab scripts to characterize them

his coefficients for Open is quite acceptable (good) up to rated BW, and his measured S11 plot is good for twice the rated BW i paid him. similar to Short with only delay coefficient. and he also provided offset loss parm/value too. maybe he intentionally left the Short coefficients out due to unsupported by most brand name VNA. i'm not sure what tuning tool he is using, but his C coefficients values are satisfactory. he does support and produces cal files (text) format for major brand names like anritsu/HP/Copper Mountain/Tek/R&S but it can be easily created by the VNA owner from his master file. i dont see any one of the brand names supporting L parm in their text format file.

but playing with my new tuning tool, without L/C coefficients (leaving them to zeros) ie by adjusting offset delay alone can match measured data closely quite satisfactory up to certain GHz BW, better with inclusion of offset loss adjustment. thats why i guess SDR-kit/Rosenberger came up with tuning tool for offset delay alone. but i guess we'll deviate from true 1 way trip offset delay/length in order to compensate for zero L/C terms ideality (from non-ideality). next step is to see this "virtual"/"elongated" delay effect on VNA measurement accuracy. btw, modelling the Kirkby Load or any other "sloppy diy" standard is several magnitude more difficult than modelling a good (Kirkby's) Open or Short. and with no existence or care about the model of the Load by VNA manufaturers probably imply, building a good Load should be easier than characterizing an Open or Short, or small/slight deviation on the Load from ideal 0 ohm will have less impact on VNA measurement accuracy, or teh said difficulty/uncertainty at measuring accurate S11 of the Load, either one or both , or either i'm wrong at any of these

[mawyatt]

Thanks virtualparticles for the posts above!!!

My point about starting from a good known 50 ohms, even at DC, is better than starting a something other than 50 ohms if you intend to make lower frequency measurements as shown with these measurements.

Test setup:

NanoVNA SAA2N with SMA Cal kit (load DCR 50.931 ohms)

NanoVNA V2+4 with type N cal kit (load DCR 49.643 ohms)

Siglent SAA3021X Plus expanded to 3.2GHz and VNA enabled.

Another SMA Cal kit (load 49.687 ohms, short stubby type).
SMA test devices (SMA loads with long rear extensions) Green 50.882 ohms, Black 51.395 ohms.

All DCR measured with new KS34465A and fixture residual R nulled out.

SAA2N calibrated with supplied type N cal kit

V2+4 calibrated with supplied SMA cal kit except when 49.687 ohm SMA used for calibration

Siglent SSA VNA calibrated with SAA2N type N cal kit.

Measurements at 10MHz utilizing averaging.

Short stubby SMA test load (DCR 49.687 ohms):
V2+4      48.7 ohms with RL of -37.92dB delta of -0.987 ohms
SAA2N    50.0 ohms with RL of -60.24dB delta of 0.313 ohms
SSA VNA 50.06 ohms with RL of -60.23dB delta of 0.373 ohms

SMA Green test load (DCR 50.882 ohms):
V2+4     49.9 ohms with RL of -54.04dB delta of -0.982
SAA2N   51.2 ohms with RL of -38.14dB delta of 0.318
SSA VNA 51.27 ohms with RL of -37.85dB delta of 0.388
V2+4 51.2 ohms with RL of -38.27dB delta of 0.318 (cal with 49.687 ohm SMA)

SAM Black test load (DCR 51.395 ohms):
V2+4     50.4 ohms with RL of 46.51dB delta of -0.995
SAA2N   51.7 ohms with RL of -35.87dB delta of 0.305
SSA VNA 51.84 ohms with RL of -34.66dB delta of 0.455
V2+4 51.7 ohms with RL of -35.14dB delta of 0.305 (cal with 49.687 ohm SMA)


A test at 100MHz with SMA test load (DCR 49.687 ohms):
V2+4     48.7 ohms with RL of 38.21dB delta of -0.987
SAA2N   50.0 ohms with RL of -41.69dB delta of 0.313
SSA VNA 50.19 ohms with RL of -41.03dB delta of 0.503


Note how the SAA2N and SSA VNA readings agree with each other and are more inline with the DUT load DCR value, much better than the V2+4 when calibrated with the supplied 51 ohm load. However the V2+4 also agrees well with the SAA2N and SSA VNA when calibrated with a load that is nearer 50 ohms as shown in blue.

Note how the delta between the DUT DCR and the Cal DCR readings track. So it seems that starting from a reference near 50 ohms, the nearer the better, does indeed yield a superior measurement result at lower frequencies

Best,

[virtualparticles]

Interesting measurements. I would say this though, any return loss measurements below 35 dB are not terribly meaningful. The uncertainty of reflection measurement is entirely dependent on the return loss of the load standard. A load with 45 dB of return loss would allow VNA measurements of 35 dB with +/- 3.3 dB of accuracy and +/- 1dB at 25 dB. The best load in the world is at NIST and it's 52 dB. To get a better calibration one would have to use TRL where it is possible to get the directivity uncertainty down to 60 dB with a pristine air line.

[joeqsmith]

I support all the HPAK terms except offset loss and delay for both the thru and load with the V2+

you (and VNA) will need to support offset loss term if we want to improve model matching to measured data at high frequency, imho from playing with my tuning tool.

The V2+ is basically in dumb mode when running headless.  It's up to the software to handle all of the corrections.   There's no need for the VNA to support it when using software to control it.   That's true with all my old VNAs. 

He should work with our local friend who helped out with their MatLab scripts to characterize them

his coefficients for Open is quite acceptable (good) up to rated BW, and his measured S11 plot is good for twice the rated BW i paid him. similar to Short with only delay coefficient. and he also provided offset loss parm/value too. maybe he intentionally left the Short coefficients out due to unsupported by most brand name VNA. i'm not sure what tuning tool he is using, but his C coefficients values are satisfactory. he does support and produces cal files (text) format for major brand names like anritsu/HP/Copper Mountain/Tek/R&S but it can be easily created by the VNA owner from his master file. i dont see any one of the brand names supporting L parm in their text format file.

but playing with my new tuning tool, without L/C coefficients (leaving them to zeros) ie by adjusting offset delay alone can match measured data closely quite satisfactory up to certain GHz BW, better with inclusion of offset loss adjustment. thats why i guess SDR-kit/Rosenberger came up with tuning tool for offset delay alone. but i guess we'll deviate from true 1 way trip offset delay/length in order to compensate for zero L/C terms ideality (from non-ideality). next step is to see this "virtual"/"elongated" delay effect on VNA measurement accuracy. btw, modelling the Kirkby Load or any other "sloppy diy" standard is several magnitude more difficult than modelling a good (Kirkby's) Open or Short. and with no existence or care about the model of the Load by VNA manufaturers probably imply, building a good Load should be easier than characterizing an Open or Short, or small/slight deviation on the Load from ideal 0 ohm will have less impact on VNA measurement accuracy, or teh said difficulty/uncertainty at measuring accurate S11 of the Load, either one or both , or either i'm wrong at any of these

Keysight has their calibration database software you can download for free and play with.  There you will find all of the coefficients they support for their various kits.  My vintage VNA doesn't support their latest format but the software has conversions built in.   The right thing to do for the V2+ would be to add support for all of the coefficients to my model.  Even the load delay.     I think if you look, they left that one zero for every standard. 

Yeah,  I won't pretend to be able to make a load that will work at 4GHz.   My early attempts at an open and short were so poor, Mario's scripts were having problems with them.   

[mawyatt]

Interesting measurements. I would say this though, any return loss measurements below 35 dB are not terribly meaningful. The uncertainty of reflection measurement is entirely dependent on the return loss of the load standard. A load with 45 dB of return loss would allow VNA measurements of 35 dB with +/- 3.3 dB of accuracy and +/- 1dB at 25 dB. The best load in the world is at NIST and it's 52 dB. To get a better calibration one would have to use TRL where it is possible to get the directivity uncertainty down to 60 dB with a pristine air line.

After the holidays hopefully I'll be able to take these devices and cal kits over to the university and work with a real VNA and cal kits. One of my former grad students has offered to help, so likely I'll take him up on that

It's fun playing around with these things and trying squeeze out more, amazing little devices considering the cost, so hat's off to the designers

Best,

[Mechatrommer]

The V2+ is basically in dumb mode when running headless.  It's up to the software to handle all of the corrections.   There's no need for the VNA to support it when using software to control it.

yes this is a good thing.

The right thing to do for the V2+ would be to add support for all of the coefficients to my model.

or even better if they support "data-based" method, like VNA View SW, we dont have to enter coefficients one by one, just provide s1p files. its wonderfull xoxoxo the original programmer provided it in the first place, its really eye opening to the "next level" since i asked how to use it properly in another thread. this is why Mr Kirkby will not recommend any VNA that doesnt support cal kit coefficients built-in, this means the NanoVNA, and "was" the Deepace KC901V, now Deepace responded to that by providing L and C coefficients input in their latest FW. for NanoVNA, the good thing is it spits raw/unprocessed S11 back to VNA View for post processing, by itself its just a "toy VNA" or another fancy VSWR meter that can do more, if without "near ideal" cal kit.

[switchabl]

The uncertainty of reflection measurement is entirely dependent on the return loss of the load standard.

It is maybe useful to add that this is usually true for good-quality, characterized calibration kits, as well as generally for low-reflection DUTs. For some of the improvised/home-made calibration kits discussed in relation with the NanoVNA, the residual source match error (relevant with more reflective DUTs) is likely to be set by the phase error on the open/short, as well the return loss of the load.

But it is true that, while you can get away with surprisingly much for non-critical measurements, a bad load standard is a no-go. The return loss on a calibration standard really needs to be a lot better than the one you are trying to measure. And consequently, if you want to evaluate a calibration-grade load properly, you do need a TRL or equivalent calibration or compare to a precision air line directly.

Incidentally, this is in my eyes also the main problem with the cheap Kirkby calibration kits. The load standard is only specified to >32dB RL and even if they managed to source better ones cheap, they might not be able to verify that with their measurement setup.

[Mechatrommer]

Incidentally, this is in my eyes also the main problem with the cheap Kirkby calibration kits. The load standard is only specified to >32dB RL and even if they managed to source better ones cheap, they might not be able to verify that with their measurement setup.

fwiw, this is RL derived from my Kirky's Load's profile. for 3GHz NanoV2 usage, it will be better than 40dB. will be glad to see other/Nano's Load's profile. of course this is measured wrt a much precise Cal Standard Kit of his (HP)...

[mawyatt]

The return loss on a calibration standard really needs to be a lot better than the one you are trying to measure.

Absolutely agree. While attending meeting at Keysight's Santa Rosa headquarters just before the terrible fires, we had a discussion with the CTO and the main designer behind the key semiconductor components in the latest high performance SA and VNAs, although we were discussing a new >100dB SFDR >20GSPS DAC and what to "expect" in the spectral results. They said that you need an instrument at least 10dB better than what you are measuring, which requires semiconductor components at least 10dB better than the instrument. They couldn't find chips good enough to meet this criteria, and thus had to design thier own in many cases, and why many of the critical ICs in these advanced instruments are custom designed and fabricated at Keysight. They even have their own special semiconductor fabs for Indium Phosphide (I've design in this >600GHz process), GaAs and other special processes in Santa Rosa. Even had to design special connectors and cables because they couldn't achieve the isolation required (>150dB I recall). All this makes sense as the measurements usually follow a Root-Sum-Square type and you are pushing the SOTA in dynamic range and frequency, so the ICs are also pushing SOTA.

BTW earlier version of the mentioned DAC had a measured phase noise better than -168dBm/Hz (output at 0dBm) at 1KHz offset at >1GHz, I know I saw it in 2009!! And this was dual DACs with Dual DDS on a single chip!! There is some really sophisticated stuff in these advanced instruments that most never see or even know about. The clever use of the AD and TI chips in these NanoVNAs is also impressive IMO, especially considered to cost.

Best,   

[switchabl]

fwiw, this is RL derived from my Kirky's Load's profile. for 3GHz NanoV2 usage, it will be better than 40dB. will be glad to see other/Nano's Load's profile. of course this is measured wrt a much precise Cal Standard Kit of his (HP)...

Well, that is the point isn't it. On the website they mention they are using an Agilent 85052B cal kit. The included broadband load is >44dB @3GHz. Now that is quite respectable, but not nearly enough to know how good your load really is. Just going by the plot, yours might be 45dB up to 3GHz, quite good isn't it? But what is the uncertainty here? Pretty large actually, the real value might be as low as 38.5dB!

Not to say this load will not usually be good enough @3GHz. If you are going to use it @6GHz though, you definitely need to keep the limitations in mind.

[joeqsmith]

The right thing to do for the V2+ would be to add support for all of the coefficients to my model.

or even better if they support "data-based" method, like VNA View SW, we dont have to enter coefficients one by one, just provide s1p files. its wonderfull xoxoxo the original programmer provided it in the first place, its really eye opening to the "next level" since i asked how to use it properly in another thread. this is why Mr Kirkby will not recommend any VNA that doesnt support cal kit coefficients built-in, this means the NanoVNA, and "was" the Deepace KC901V, now Deepace responded to that by providing L and C coefficients input in their latest FW. for NanoVNA, the good thing is it spits raw/unprocessed S11 back to VNA View for post processing, by itself its just a "toy VNA" or another fancy VSWR meter that can do more, if without "near ideal" cal kit.

Certainly doable on the PC side but I'm not sure how you would get a Touchstone file loaded into a stand alone V2+.  I think before I do that, I would want to add interpolation for the calibration as it's something I would actually use.       

Using Google to search for VNA View software  and it comes back with a LabView program.  Not thinking this is what you are referencing.   
http://wb9jps.com/Gary_Johnson/VNAView.html

Post a link and I'll have a look.     

[Mechatrommer]

fwiw, this is RL derived from my Kirky's Load's profile. for 3GHz NanoV2 usage, it will be better than 40dB. will be glad to see other/Nano's Load's profile. of course this is measured wrt a much precise Cal Standard Kit of his (HP)...

Well, that is the point isn't it. On the website they mention they are using an Agilent 85052B cal kit. The included broadband load is >44dB @3GHz. Now that is quite respectable, but not nearly enough to know how good your load really is. Just going by the plot, yours might be 45dB up to 3GHz, quite good isn't it? But what is the uncertainty here? Pretty large actually, the real value might be as low as 38.5dB!

if Agilent's Load is only 38-44dB, what do you expect from $4 kit? or even a $500 Kirkby's Kit? 60dB? (dividing equally from pieces i got, the female Load the plot i attached is about $50 price incl characterizing service)

Not to say this load will not usually be good enough @3GHz. If you are going to use it @6GHz though, you definitely need to keep the limitations in mind.

if the VNA supports data-based Load/Arbitrary standard, and if the math implementation is right and leakage and error terms are taken cared of, measurement should be good, theoritically up to 12GHz (the Kirkby's s1p plot) plus minus whatever the uncertainty is. otherwise if the VNA doesnt support any kind of reasonable Load model, i believe my measurement is as good as any others budget/Deepace/NanoVNA + the provided cal kit. i will compare Kirkby's vs NanoVNA's vs my diy best/worst Load's profile later, for now i have some other tasks slipping in. and i will be outstationed tomorrow.

The right thing to do for the V2+ would be to add support for all of the coefficients to my model.

or even better if they support "data-based" method, like VNA View SW, we dont have to enter coefficients one by one, just provide s1p files. its wonderfull xoxoxo the original programmer provided it in the first place, its really eye opening to the "next level" since i asked how to use it properly in another thread. this is why Mr Kirkby will not recommend any VNA that doesnt support cal kit coefficients built-in, this means the NanoVNA, and "was" the Deepace KC901V, now Deepace responded to that by providing L and C coefficients input in their latest FW. for NanoVNA, the good thing is it spits raw/unprocessed S11 back to VNA View for post processing, by itself its just a "toy VNA" or another fancy VSWR meter that can do more, if without "near ideal" cal kit.

Certainly doable on the PC side but I'm not sure how you would get a Touchstone file loaded into a stand alone V2+.  I think before I do that, I would want to add interpolation for the calibration as it's something I would actually use.       

Using Google to search for VNA View software  and it comes back with a LabView program.  Not thinking this is what you are referencing.   
http://wb9jps.com/Gary_Johnson/VNAView.html

Post a link and I'll have a look.     

no! i thought i've seen you used it, and i believe its linked in NanoVNA website (from where i got it), here..
https://github.com/nanovna/NanoVNA-QT/releases
from https://nanorfe.com/nanovna-v2.html
opps i think my bad, not sure its official name from the website, i guess its the NanoVNA-QT? but the program's windows' caption is "VNA View"..

so may i ask you a favor in return... which link you are refering to the Agilent's cal data in below comment.. so i can have a look

Keysight has their calibration database software you can download for free and play with.  There you will find all of the coefficients they support for their various kits.

sometime my google-fu is failing esp with lack of sleep and time, and brain cells resources used up during past few weeks typing codes. 

[switchabl]

fwiw, this is RL derived from my Kirky's Load's profile. for 3GHz NanoV2 usage, it will be better than 40dB. will be glad to see other/Nano's Load's profile. of course this is measured wrt a much precise Cal Standard Kit of his (HP)...

Well, that is the point isn't it. On the website they mention they are using an Agilent 85052B cal kit. The included broadband load is >44dB @3GHz. Now that is quite respectable, but not nearly enough to know how good your load really is. Just going by the plot, yours might be 45dB up to 3GHz, quite good isn't it? But what is the uncertainty here? Pretty large actually, the real value might be as low as 38.5dB!

if Agilent's Load is only 38-44dB, what do you expect from $4 kit? or even a $500 Kirkby's Kit? 60dB? (dividing equally from pieces i got, the female Load the plot i attached is about $50 price incl characterizing service)

That is not what I am trying to say. I am trying to illustrate how hard it is to characterize a calibration standard. The Agilent load is >44dB, which is quite respectable. But the uncertainty in the measurement you posted is still so high it can only show that yours is >38.5dB (below ~3GHz).

For what it's worth, I think the Kirkby kit is pretty good value. It's just worth keeping in mind that the load standards are not real calibration-grade loads (that would increase the price _a lot_) and in particular for >3GHz that may introduce significant measurement uncertainties.

If you can do data-based calibration, you are in a better place because then you are depending on the stability of the VNA and the load instead, which will be quite a bit better. But unless you are using a modern lab-grade VNA, that likely means offline processing on a PC.

[joeqsmith]

Post a link and I'll have a look.     

no! i thought i've seen you used it, and i believe its linked in NanoVNA website (from where i got it), here..
https://github.com/nanovna/NanoVNA-QT/releases
from https://nanorfe.com/nanovna-v2.html
opps i think my bad, not sure its official name from the website, i guess its the NanoVNA-QT? but the program's windows' caption is "VNA View"..

so may i ask you a favor in return... which link you are refering to the Agilent's cal data in below comment.. so i can have a look

Keysight has their calibration database software you can download for free and play with.  There you will find all of the coefficients they support for their various kits.

sometime my google-fu is failing esp with lack of sleep and time, and brain cells resources used up during past few weeks typing codes. 

I missed you were talking about the OEMs software.   I thought they had a load and save cal was all.  It's a pain to run on my PC.  I'll fire up the old laptop and have a closer look at what this database is that you referred to.

You may find the Keysight software here:
http://na.support.keysight.com/pna/apps/applications.html

...if the VNA supports data-based Load/Arbitrary standard, and if the math implementation is right and leakage and error terms are taken cared of,
...

I learned that lesson with the V2+, you need to account for leakage (crosstalk) terms or it falls apart above 3GHz.

[virtualparticles]

The uncertainty of reflection measurement is entirely dependent on the return loss of the load standard.

It is maybe useful to add that this is usually true for good-quality, characterized calibration kits, as well as generally for low-reflection DUTs. For some of the improvised/home-made calibration kits discussed in relation with the NanoVNA, the residual source match error (relevant with more reflective DUTs) is likely to be set by the phase error on the open/short, as well the return loss of the load.

Excellent point