Episode 520 WARNING: do not input more than 0dBm (1mW) power into the NANOVNA Can the NANOVNA be used as a spectrum analyzer? The answer is complicated but probably not. Be a Patron: / imsaiguy
Interesting video! I tried the same test on the nanoVNA SAA-2N and has similar results. The two big limitations using these devices as spectrum analysers are the fixed narrow bandwidth "I.F.", with generally much larger discrete frequency steps, so you only see energy at narrow discrete frequencies and miss everything else between those spots. Throw the square-wave local oscillator and you'll get all sorts of anomalies! Interestingly when you start doing sweeps over a few octaves, you see significant images of the input signal at higher and LOWER frequencies. They aren't always visible, only visible when the harmonics of the sweeping oscillator happen to line up exactly with the input signals. If everything is just right, I see images of a 50MHz input at 1/3, 1/7, 1/9, etc frequencies.
Precisely...therefore we should look for nano spectrum analyser right away !!!!! I bet it won't work properly either. Low budget investments never pay back !
Well, we've determined that it's not really a good spectrum analyzer, but it was an interesting video. It actually showed me what the bottom port was for. Thanks for posting this video. Barry
I was wondering it myself, but unlike you I don’t have the tools to test it. The problem with your test is that you are not considering the limitations of the nanoVNA hardware. Let me explain: In your experiment, the nanoVNA collects only 100 samples spanning the 4 MHz bandwidth (148-144). Meaning, that the nanoVNA steps are 4 MHz / 100 = 0.04 MHz. Your generator is tuning in 0.1 MHz steps. Meaning that the generator generates signal at: X.10, X.20, X.30, X.40, X.50 MHz, etc. frequencies. The nanoVNA is ‘Listening’ to are X.00, X.04, X.08, X.12, X.16, X.20, X.24, X.28, X.32, X.36, X.40 MHz, etc. Notice that only frequencies X.00, X.20, X.40, X.60 and X.80 MHz are common to both TX & RX. Any frequency (TX or RX) outside of common frequency will read lower power because your generator is transmitting away from the frequency nanoVNA is listening to. Please make a note in your video, that when you tune to one of the common frequencies the power reading is what you expect. The moment you step away from frequency nanoVNA is sampling, you only get smaller magnitude reading for that frequency. A test that would demonstrate the right setup would be to sweep say 145 MHz to 146MHz and repeat your test with 0.1 MHz tuning steps. That way, your generator is transmitting on the same frequencies the nanoVNA is receiving. The signal magnitude should not change throughout the whole test.
You may have a point. I tried your test and yes the power stayed the same. but I went with 0.01 steps and it all stayed the same. At the 145-146 range the nano IF filter is wide enough to capture all steps. I also did a 145.000 to 145.100 sweep and it was garbage. I got the nasty double peak response and wide side bands that are 50kHz wide. so if you assume the IF is 50kHz fixed. then your input needs to match the 100 data points within the 50kHz filter. So wide sweeps only, steps must equal nano. no modulation. I've got the tinySA on order so will be fun to try that one.
@@IMSAIGuy What happens if you redo your calibration after choosing the narrower frequency range? Would the calibration points now line up with the sample points?
It acts like you are getting aliasing due to discretion effects. As a VNA, they are controlling the stimulus frequency. They could step the frequency to match ADC sampling rates and vise versa. It would be interesting to look at the code to see if thay re doing this.
The double jump could be from the unfiltered image response in the mixer. The audio IF results for RF input either above or below the LO frequency by the IF amount. RF and LO harmonics mixing and their image terms could cause the additional humps. The IF (audio) filter could be just fine.
Hello! So I can use it for example for measuring frequency of ring in a flyback DC/DC converter - it can be tens of Mega Herz and not each oscilloscope can display actual frequency.
Yes you are right. that would be too much. the .3W display is due to the way the HP8921 is constructed and measures. I'm not using that function and it is an artifact. I'm using the RF generator and if you look at the bottom left you will see the amplitude set to -10dbm. that is 0.1mW or 100uW.
Thanks for the experiments. I just tried it on my NanoVNA-F (Hardware version 2.3, Firmware BH5HNU 0.1.5) and not too much better luck. While it responded resonably well if the center freq was right on the signal (and the signal was stable), I get nothing for most settings (e.g. 50k to 20MHz sweep with 12 MHz input signal shows nothing until I force it to hit the 12.000000 MHz signal exactly in the sweep). It seems to be that the step size is too big relative to the kHz wide filtering it is using. I wonder if someone could modify the firmware? But it would be too slow I'm afraid. Sure will be nice when the amazing developers of the NanoVNA come up with a cheap spectrum analyzer too !
It’s me again, thank you in advance for your patience. Considering your results, I think there should be no problem to build a reasonable audio filter. Maybe the LO of the VNA has some more phase noise (from fast reacting fractional n division). The LO shall react fast and phase noise is not so relevant as it is compensated in its original application where ch0 and ch1 are mixed with the same signal. If there is some external signal measured without the identical phase noise then it is not compensated. This could be checked by measuring the ch0 with a quality spectrum analyzer (in particular with a slow sweep at the VNA to cover different fractional n situations with their phase noise). I will make own measurements soon, at least I am almost sure, or I hope so.
@@Manf-ft6zk oh sorry. as you know many youtube providers have videos in queue to be published automatically in the future. Guess you'll have to wait ...
I don't know that you will ever see this..... I know it is an older video.. but I JUST GOT my Nano VNA H4 and this video caused a couple of questions. 1- DID you calibrate the NanoVNA and if so did you calibrate it for a "through" EVEN THOUGH you are only using the input side??? 2-I do notice that your generator is set to put out -10 dBm but it LOOKS LIKE the NanoVNA is showing something in the neighborhood of -20dBm on the grid. Am I reading/understanding this wrong? The reason I am so interested is that I am trying to see if I can use my NanoVNA to "more accurately set" my Boonton 102F and B&K 2040. Both of these can be set using the rotary attenuator and adjusting to the 0 dBm reading on the scale to output the desired dBm level... I am just wanting to determine whether or not I can use/trust the NanoVNA reading as an accurate setting! Thank you so much for what you do here!!!!! Bob AI5RR
@@IMSAIGuy - I've already put some thought in that direction! I have a REALLY old Hameg 5006... but it lacks a lot of features..... like the ability to read the frequency and level on the screen with a measure of certainty!!!! :) Thank you so much for the response.
hi, i just bought nanoVNA but now im afraid to measure my radio because I cannot find any example of people using nanoVNA to measure a radio. my radio is 900Mhz, and I want to know if my radio is really outputting 30dBm (I have 40db Attenuator). would i destroy my radio if I remove my radio antenna, and attach my radio to the Attenuator and to the nanoVNA(because it may act as if there is no antenna, or does the nanoVNA has an internal load that will compensate this)?
the nanovna will not measure the power level . it is not calibrated for that. you will need a spectrum analyzer (like the tinysa). you will need enough attenuation to keep the input below 0dBm. The 40db attenuator you have will be the primary load on the radio. the 30dBm will be reduced to -10dBm (0.1milliwatt) and the 1watt of power will be heat in the attenuator. if nothing is attached to the attenuator that 0.1mW will be reflected back and have to go through the 40db a second time so nothing will come back to the radio.
@@IMSAIGuy oh cool you replied so fast and detail! thanks a lot! btw can I at least use it to see if at least my transmitter is not broken? doesnt have to be percise power but at least if i can the graph hitting -20dbm (after the attenuation)
I could not tell at first if you were in uncalibrated mode or not, it looks like you were. A better spectrum analyzer would be the Airspy or other SDR. The up and down power response and weirdness is probably due to the Si5351 not able to generate power or small frequency steps accurately. The next NanoVNA video should be on measuring amplifiers S21.
Why weird filter? NanoVNA maybe HD, so it displays whatever is out there even zoomed-in. But it is mind-opening that people like you find so many different uses, cable length, inductor/capacitor measurement, oscillator, now spectrum analyser.
Thank you very much. This came right in time. I just had the question how well it might work. Does it get any better or does it change with more points on the sweep?
What's nice to know is the max input level on the second port to avoid damage, and that it's possible to at least hunt down a source of interference in your immediate location (routers, smart meters, switching supplies, leaking powerline insulators, etc...). The rest of the comment blah-blah is irrelevant. Who cares if it's lacking this or that? It's made in China, is cheap, and questionable at best. Hypothesize all day long - proves nothing other than you don't have any idea. Buy a little spectrum analyzer and use it as intended if more detailed analysis is needed.
The answer to this question is definitely NOT!!! Here is a list of points in my humble opinion that show why i say NO. 1. The local oscillator in the VNA : The local oscillator cannot be guaranteed to be accurate ( no way to know if it has been calibrated and no way to calibrate it if it was off frequency). so frequency cannot be relied upon to have any real accuracy. 2. Sample points : Any signal that is between sample points is calculated in mathematics to be a rough approximation of something in between sample points. Sometimes there is not enough information to even show there is a signal between the sample points. This leads to the power level of the signal not being accurate or useful. if the sample point was actually at the frequency of interest then it may become more useful but inaccurate in frequency ( see number 1 above). NOTE: when using an external signal generator you have to be careful to not overload the input of the VNA's mixer. doing the same experiment at different power levels would possibly show the VNA's mixer is being over loaded (if different results at different power levels are shown). The 'weirdness' in the video could be simply the mixer being overloaded and causing undesirable harmonics to show up on something that was NEVER designed to function in the way it was being used. The TOP 2 points above show that you would never be able to trust any frequency or power level shown on the device. The main purpose of a spectrum analyser is to measure the frequency with some accuracy and measure the power level to some know accuracy. As these two functions of a spectrum analyser cannot be done with enough accuracy on VNA then the answer should be a definite NO!! Best regards Simon
You need to establish with a true spectrum analyzer that the output of the HP unit is clean. Everything that you say is unsupported. It may be accurate, but there is no way for me to evaluate your data.
I guess it would work for adjusting the direction of a Yagi antenna. Obviously it's not perfect for the application, but if it's all you have on hand...
I just want to check on which frequency a chinese device transmits. Even the manual says 424-469 Mhz (low power) lol. Guess this could work. Thanks for the video.
In your demonstration the “double hump thing” it’s a harmonic because you are over driving the nano. There is nothing strange or weird about this. It’s an S parameter tool NOT a spectrum analyzer WHICH IT CAN NEVER BE!
Interesting video! I tried the same test on the nanoVNA SAA-2N and has similar results.
The two big limitations using these devices as spectrum analysers are the fixed narrow bandwidth "I.F.", with generally much larger discrete frequency steps, so you only see energy at narrow discrete frequencies and miss everything else between those spots.
Throw the square-wave local oscillator and you'll get all sorts of anomalies!
Interestingly when you start doing sweeps over a few octaves, you see significant images of the input signal at higher and LOWER frequencies. They aren't always visible, only visible when the harmonics of the sweeping oscillator happen to line up exactly with the input signals. If everything is just right, I see images of a 50MHz input at 1/3, 1/7, 1/9, etc frequencies.
Precisely...therefore we should look for nano spectrum analyser right away !!!!! I bet it won't work properly either. Low budget investments never pay back !
If I'm not mistaken, I was told to recalibrate with each change in frequency spectrum.
Looking forward to seeing you do some videos with the tinySA. Excellent videos as usual. Thanks for sharing!
Well, we've determined that it's not really a good spectrum analyzer, but it was an interesting video. It actually showed me what the bottom port was for.
Thanks for posting this video.
Barry
I was wondering it myself, but unlike you I don’t have the tools to test it.
The problem with your test is that you are not considering the limitations of the nanoVNA hardware.
Let me explain: In your experiment, the nanoVNA collects only 100 samples spanning the 4 MHz bandwidth (148-144). Meaning, that the nanoVNA steps are 4 MHz / 100 = 0.04 MHz.
Your generator is tuning in 0.1 MHz steps. Meaning that the generator generates signal at: X.10, X.20, X.30, X.40, X.50 MHz, etc. frequencies.
The nanoVNA is ‘Listening’ to are X.00, X.04, X.08, X.12, X.16, X.20, X.24, X.28, X.32, X.36, X.40 MHz, etc.
Notice that only frequencies X.00, X.20, X.40, X.60 and X.80 MHz are common to both TX & RX.
Any frequency (TX or RX) outside of common frequency will read lower power because your generator is transmitting away from the frequency nanoVNA is listening to.
Please make a note in your video, that when you tune to one of the common frequencies the power reading is what you expect. The moment you step away from frequency nanoVNA is sampling, you only get smaller magnitude reading for that frequency.
A test that would demonstrate the right setup would be to sweep say 145 MHz to 146MHz and repeat your test with 0.1 MHz tuning steps. That way, your generator is transmitting on the same frequencies the nanoVNA is receiving. The signal magnitude should not change throughout the whole test.
You may have a point. I tried your test and yes the power stayed the same. but I went with 0.01 steps and it all stayed the same. At the 145-146 range the nano IF filter is wide enough to capture all steps. I also did a 145.000 to 145.100 sweep and it was garbage. I got the nasty double peak response and wide side bands that are 50kHz wide. so if you assume the IF is 50kHz fixed. then your input needs to match the 100 data points within the 50kHz filter. So wide sweeps only, steps must equal nano. no modulation. I've got the tinySA on order so will be fun to try that one.
@@IMSAIGuy What happens if you redo your calibration after choosing the narrower frequency range? Would the calibration points now line up with the sample points?
Waiting for you to give the same love to tinySA as you did to the NanoVNA over the many videos since the beginning of this year
Have you used the TinySA yet? It looks awesome.
@@hebrewhammer1000 Yes, received mine 1 week ago. Its pretty awesome, the PC software, the TinySA analog of nanoVNA-saver is coming along too
It acts like you are getting aliasing due to discretion effects. As a VNA, they are controlling the stimulus frequency. They could step the frequency to match ADC sampling rates and vise versa. It would be interesting to look at the code to see if thay re doing this.
Thank you for answering the question so well.
The double jump could be from the unfiltered image response in the mixer. The audio IF results for RF input either above or below the LO frequency by the IF amount. RF and LO harmonics mixing and their image terms could cause the additional humps. The IF (audio) filter could be just fine.
good point, hadn't thought of that one
tried it. works great. love my 8921
Hi, does it detect the exact frequencies without being directly connected to the device like you have done so in this video?
same way a radio tunes a station
Hello! So I can use it for example for measuring frequency of ring in a flyback DC/DC converter - it can be tens of Mega Herz and not each oscilloscope can display actual frequency.
Very interesting test! For that price tag you can't expect too much and it was never designed as a (precise) spectrum analyzer 🙂
What would be considered safe power to send to the RX port of the unit ? Isn't 0.3 watt a lot for such a small unit ? I got my nanonva-f today.
Yes you are right. that would be too much. the .3W display is due to the way the HP8921 is constructed and measures. I'm not using that function and it is an artifact. I'm using the RF generator and if you look at the bottom left you will see the amplitude set to -10dbm. that is 0.1mW or 100uW.
Thanks for the experiments. I just tried it on my NanoVNA-F (Hardware version 2.3, Firmware BH5HNU 0.1.5) and not too much better luck. While it responded resonably well if the center freq was right on the signal (and the signal was stable), I get nothing for most settings (e.g. 50k to 20MHz sweep with 12 MHz input signal shows nothing until I force it to hit the 12.000000 MHz signal exactly in the sweep). It seems to be that the step size is too big relative to the kHz wide filtering it is using. I wonder if someone could modify the firmware? But it would be too slow I'm afraid. Sure will be nice when the amazing developers of the NanoVNA come up with a cheap spectrum analyzer too !
You need the TInySA
ua-cam.com/video/n6WEM3--Npc/v-deo.html
@@IMSAIGuy Awesome. Checking it out now in your video :-)
@@MegawattKS I have several videos. It takes a while to learn to drive it well and I'm learning. It is quite good if used below -30dbm
It’s me again, thank you in advance for your patience.
Considering your results, I think there should be no problem to build a reasonable audio filter.
Maybe the LO of the VNA has some more phase noise (from fast reacting fractional n division). The LO shall react fast and phase noise is not so relevant as it is compensated in its original application where ch0 and ch1 are mixed with the same signal. If there is some external signal measured without the identical phase noise then it is not compensated.
This could be checked by measuring the ch0 with a quality spectrum analyzer (in particular with a slow sweep at the VNA to cover different fractional n situations with their phase noise).
I will make own measurements soon, at least I am almost sure, or I hope so.
I did measure the RF generator once:
ua-cam.com/video/MWEVJAjMuhs/v-deo.html
@@IMSAIGuy The video is not available it is stated to be private.
@@Manf-ft6zk oh sorry. as you know many youtube providers have videos in queue to be published automatically in the future. Guess you'll have to wait ...
Was thinking maybe it could be connected to a 2m/440 antenna for EMI detection.
TinySA is better: ua-cam.com/video/nn14JNxNT1Y/v-deo.html
@@IMSAIGuy good video, thanks.
really helpful! i just got a Lite VNA, which seems to have the same interface. Not sure how it's anomalies compare yet though.
Hmm. Looks like the nano vna could have gating/prescaler/aliasing issues, somewhat like frequency counters may have.
I don't know that you will ever see this..... I know it is an older video.. but I JUST GOT my Nano VNA H4 and this video caused a couple of questions.
1- DID you calibrate the NanoVNA and if so did you calibrate it for a "through" EVEN THOUGH you are only using the input side???
2-I do notice that your generator is set to put out -10 dBm but it LOOKS LIKE the NanoVNA is showing something in the neighborhood of -20dBm on the grid. Am I reading/understanding this wrong?
The reason I am so interested is that I am trying to see if I can use my NanoVNA to "more accurately set" my Boonton 102F and B&K 2040. Both of these can be set using the rotary attenuator and adjusting to the 0 dBm reading on the scale to output the desired dBm level... I am just wanting to determine whether or not I can use/trust the NanoVNA reading as an accurate setting!
Thank you so much for what you do here!!!!! Bob AI5RR
no, I suggest getting a TinySA ULTRA
@@IMSAIGuy - I've already put some thought in that direction! I have a REALLY old Hameg 5006... but it lacks a lot of features..... like the ability to read the frequency and level on the screen with a measure of certainty!!!! :)
Thank you so much for the response.
@IMSAI Guy thanks .... ok nanoVNA work like a spectrum analyzer... but in home mode i set-up in the varoius MENU ? help me ? 🙂
ok nanoVNA work like a spectrum analyzer... but in home mode i set-up in the varoius MENU ? help me ? 🙂
hi, i just bought nanoVNA but now im afraid to measure my radio because I cannot find any example of people using nanoVNA to measure a radio. my radio is 900Mhz, and I want to know if my radio is really outputting 30dBm (I have 40db Attenuator). would i destroy my radio if I remove my radio antenna, and attach my radio to the Attenuator and to the nanoVNA(because it may act as if there is no antenna, or does the nanoVNA has an internal load that will compensate this)?
the nanovna will not measure the power level . it is not calibrated for that. you will need a spectrum analyzer (like the tinysa). you will need enough attenuation to keep the input below 0dBm. The 40db attenuator you have will be the primary load on the radio. the 30dBm will be reduced to -10dBm (0.1milliwatt) and the 1watt of power will be heat in the attenuator. if nothing is attached to the attenuator that 0.1mW will be reflected back and have to go through the 40db a second time so nothing will come back to the radio.
@@IMSAIGuy oh cool you replied so fast and detail! thanks a lot! btw can I at least use it to see if at least my transmitter is not broken? doesnt have to be percise power but at least if i can the graph hitting -20dbm (after the attenuation)
@@stonedDawg the nanovna does hot have any units of dbm you cannot use it.
I could not tell at first if you were in uncalibrated mode or not, it looks like you were. A better spectrum analyzer would be the Airspy or other SDR. The up and down power response and weirdness is probably due to the Si5351 not able to generate power or small frequency steps accurately. The next NanoVNA video should be on measuring amplifiers S21.
Does your HP 8921A spectrum analyzer have a noisy fan? I am thinking of buying one from ebay.
It has an audible fan.
Thank you for sharing your knowledge. 73 Bret/AC0AE
The question is can you use it as a signal generator?
ua-cam.com/video/RvYwAn32Dl4/v-deo.html
ua-cam.com/video/MWEVJAjMuhs/v-deo.html
Guess you can't turn off tx on channel 0 (does it even matter) ? When I turn off tx-sweep also channel 1 data is not updated anymore.
How many dbm input nanoVNA can stand?
Why would you use a NA as a SA ?
HackRF One used as a spectrum analyzer? 🤔
How would I be able to hook up a ham radio to the NanoVNA?
I have a video on the subject of attenuators:
ua-cam.com/video/Sdb_cs13njk/v-deo.html
ua-cam.com/video/ljJcYFQS9KU/v-deo.html
you can also do something like this:
ua-cam.com/video/4sxPSJKPOs4/v-deo.html
Why weird filter? NanoVNA maybe HD, so it displays whatever is out there even zoomed-in. But it is mind-opening that people like you find so many different uses, cable length, inductor/capacitor measurement, oscillator, now spectrum analyser.
Thank you very much. This came right in time. I just had the question how well it might work. Does it get any better or does it change with more points on the sweep?
seems to just change
@@IMSAIGuy Thank you for the answer I will try it out later. I am sure for my first trials it will be helpful.
You know you should be re-calibrating the NanoVNA every time you modify the Start. Stop, Center or Span.
I thought channel 0 was RF input, why not use it?
Channel 0 is for S11 measurements. It is an RF output and also a return loss bridge.
What's nice to know is the max input level on the second port to avoid damage, and that it's possible to at least hunt down a source of interference in your immediate location (routers, smart meters, switching supplies, leaking powerline insulators, etc...). The rest of the comment blah-blah is irrelevant. Who cares if it's lacking this or that? It's made in China, is cheap, and questionable at best. Hypothesize all day long - proves nothing other than you don't have any idea. Buy a little spectrum analyzer and use it as intended if more detailed analysis is needed.
The answer to this question is definitely NOT!!!
Here is a list of points in my humble opinion that show why i say NO.
1. The local oscillator in the VNA :
The local oscillator cannot be guaranteed to be accurate ( no way to know if it has been calibrated and no way to calibrate it if it was off frequency). so frequency cannot be relied upon to have any real accuracy.
2. Sample points :
Any signal that is between sample points is calculated in mathematics to be a rough approximation of something in between sample points. Sometimes there is not enough information to even show there is a signal between the sample points. This leads to the power level of the signal not being accurate or useful. if the sample point was actually at the frequency of interest then it may become more useful but inaccurate in frequency ( see number 1 above).
NOTE: when using an external signal generator you have to be careful to not overload the input of the VNA's mixer. doing the same experiment at different power levels would possibly show the VNA's mixer is being over loaded (if different results at different power levels are shown). The 'weirdness' in the video could be simply the mixer being overloaded and causing undesirable harmonics to show up on something that was NEVER designed to function in the way it was being used.
The TOP 2 points above show that you would never be able to trust any frequency or power level shown on the device. The main purpose of a spectrum analyser is to measure the frequency with some accuracy and measure the power level to some know accuracy. As these two functions of a spectrum analyser cannot be done with enough accuracy on VNA then the answer should be a definite NO!!
Best regards
Simon
You need to establish with a true spectrum analyzer that the output of the HP unit is clean. Everything that you say is unsupported. It may be accurate, but there is no way for me to evaluate your data.
I've looked at the output of the HP8921 with an HP8591E. It is just fine.
of course NOT,... not even a little bit
I guess it would work for adjusting the direction of a Yagi antenna. Obviously it's not perfect for the application, but if it's all you have on hand...
I just want to check on which frequency a chinese device transmits. Even the manual says 424-469 Mhz (low power) lol. Guess this could work. Thanks for the video.
In your demonstration the “double hump thing” it’s a harmonic because you are over driving the nano. There is nothing strange or weird about this. It’s an S parameter tool NOT a spectrum analyzer WHICH IT CAN NEVER BE!
u didn't calibrate it.
Vague