The SAA2 VNA uses a different DDS to the NanoVNA. The NanoVNA uses a SI5351 which can only generate frequencies up to 300MHz and uses harmonics to get above it. This reduced dynamic range significantly. The SAA2 uses an ADF4351 which can generate RF up to 4.4GHz without resorting to using harmonics, resulting in a better dynamic range across the board. Newer SAA2 revisions use the MAX2871 which is very similar to the ADF4351 but can sweep up to 6GHz. As you say, the shielding will also help with the SAA2. I also own a NanoVNA, one of the original SAA2 VNAs and a Siglent VNA :-)
Interesting and informative review! I don't do RF work but find this equipment interesting. I studied books in the 70's to get a first class FCC license but haven't ever used that in work. This review and that on the TinySA show that this technology can be obtained for a reasonable cost for schools and hobbyists. I got the TinySA just for kicks but not sure what I would do with the VNA. Thanks for two well done videos. I know it takes lots of time and effort.
Good review and very important points! The problem is you never know what you get when you buy cheap of of china/ebay/aliexpress... Even when one version was tested good, they build new fakes and it gets worse - but you don't know unless you have the capability to really test your device properly. Thanks a lot!
Impressed with the SAA2. Very close to the Siglent. I've just bought the NanoVNA-F V2 enclosed in metal case. Also has 4 metal cans on the circuit board, so hopefully well screened.
I have the same NanoVNA-F v2 and we did a similar test against our Megiq VNA0460, comparing both the bode plot and the smith charts, and they perform equally well. We use it a lot to perform impedance matching and the smith chart on both analyzers results in the same series LC (within some %) so I am very happy with the F v2
the NanoVNA-F is not the same as NanoVNA SAA2. Different companies using different designs. The F is a lower quality knockoff clone of the original design aimed at making them the most money. The SAA is newer design, redesigned from scratch with different goals in mind
@@sammiller6631 The NanoVNA-F v2 (not to be confused with the older NanoVNA-F) is a product developed by Sysjoint based on the original NanoVNA and the SAA v2. I have one and we use it a lot to match our RF circuits (430 up to 2450 MHz). We also have a Megiq VNA0460 and an HP spectrum analyzer with tracking generator and an external directional coupler. Measurement results from the NanoVNA-F v2 I have are as good as from the other equipment. So I would not say this is a lower quality knockoff. The metal case, metal RF shielding cans and the internal construction of the device look great.
Nice to see the comparison. I have nanoVNA-H4 from a reputable seller in The Netherlands and i suppose it is not a clone. The enclosure is a 'snap-on' type, so not easy to open up and look inside and say something about the shielding. I also don't have another instrument at home to compare, but could ask my colleague at work to do a quick compare. I have done some basic measurement filter measurements and on my EFHW 10m long antenna and could fine tune it with this nanoVNA. It all seems right, but once again not much to compare yet. About the port 21 isolation graph, that could be calibrated away? Or not?
I wonder why people always insist to compare a basic toy like tool to a semi professional one. It is like comparing a Cozy Coupe to a mid size Chevy or Volkswagen.
The SAA2 performing on par with the Siglent, at least on a noise level perspective up to 1.5GHz, was unexpected. There are other valuable pieces of information shared, such as a methodology to graph data from multiple brands of VNAs. Thank you for watching and for your feedback
Most of the people will use the nanoVNA way below GHz frequency so it performs great!My '97 Golf still runs like a swiss watch and takes me to places while seeing many luxury cars biting the dust...:) It gets the job done...:)
@@solarflare4259 True. There are other version of Nano that are fine in the GHz range, LiteVNA for example (same manufacturer as for H4), and pretty affordable. Golf bit reminds me of the Balkans 😁
@@galileo_rs Indeed. I've got a H4 and it works great. I find myself using it up to 60 MHz, mostly for HAM shortwave...Currently waiting to receive the 4401 ... Just wanted a larger display. As for the Golf, I grabbed it back in '99 and I've taken good care of it.It comes really handy now with rising of fuel prices (2L) engine MT... I barely see some like these on the road; even hard to find parts at salvage yards...:)
NanoVNA-H is the old V1 version, Si5335 based, that works on harmonics above something like 300MHz, so no wonder it has much worse specs. Si5335 synthesizer chip official spec: fmax 200MHz NanoVNA V2 is new version based on 2x ADF4350 with official fmax of 4400Mhz, so no harmonics needed. NanoVNA-F V2 clone version with somewhat better/nicer mechanical design, aluminum case instead of plastics, for a bit more money
As Andy mentioned DDS are different in you VNAs. That is very important, because RF generator output power is much lower when using 2nd harmonic and even lower when using 3rd harmonic, and if we say RX sensitivity and noise stays the same, you get much worse S/N ratio (you get more noise compared to the signal) on higher frequencies. Shielding would help, but just to a point. When you calibrate VNA it offsets that very low (2nd or up to 5th harmonic) signal level with big gain to get 0dB trace at all frequencies, so it is mostly RF generator output level problem. RX sensitivity is also getting worse at higher frequencies, because of higher mixer conversion losses, which is another 'RF signal level' problem. Inconsistencies between measurements are also due to cable bending (since those cables are not phase stable), SMA tightening and some common mode current outside the coax. It would be interesting to take one measurement with Siglent, then disconnect cable, connect it again to Siglent and taking second measurement. I suspect you would get almost the same difference as you did with comparison to SAA2. That would show cable bending and SMA tightening contributions plus other measurement uncertainties.
The white Nano is a clone and the results you are getting are probably due to that fact. I have a modified clone H and an original H4 and those do not exhibit that kind of behaviour. Try NanoVNA Saver software and do not calibrate on device. Saver will offer much higher number of calibration points and a correction for the trough measurement. Makes sure that you get the latest firmware. DisLord firmware is recommended.
Nice video, thanks ! Yes, a review of the Siglent would be great. By the way, did you get to the SVA1032X from the SSA3021X Plus with just the s/w license upgrades Siglent offers or did you have to have h/w changes made at the factory ?
When you take measurements from duplexers, put a dummy load on the unused port. Leaving the unused port floating ussually results in phantom troughs all over the place.
I had to recheck the video and you're right, I totally forgot to do it. I believe the impact shouldn't be very significant due to the attenuation between the channels, but you're absolutely correct, this is an oversight on my part during this test. Thanks for pointing it out!
@@the_bootloader For what you were doing, I guess you could choose to deliberately ignore wierd responses outside of the band of interest. The only reason I noticed was because I have to tune commercial duplexers at work, and the phantom troughs can cause serious issues with figuring out what is going on.
@@digitalradiohacker yes you're absolutely right. As you point out, for my purpose it didn't really matter but generally it's good practice and the way to do it. Thanks again for participating and sharing your knowledge, it's great that we can all learn together from our respective experiences
@@the_bootloader Absolutely. I bought one of these as a potential alternative to the 10KG Aeroflex 3500 that I have in my van. I know that a new instrument of this class costs the same as a house, so they're not the same, but there is also 20 years of development between them as well. In reality, I'm finding the noise floor on mine to be quite pathetic (around 50dB?). This is no good for tuning commercial duplexers which are capable of -80dB rejection. Mine is marked as NanoVNA-H The search for firmware that won't make a brick is ongoing.....
The best is...., a HP8753E, ya it cost money, then again, if you want to go down the 1/4 mile track in 7 seconds, your not going to do it with a 4 banger, Datsun! LOL!@
Many thanks. This was a very interesting set of tests. Based on my own needs, my take away is that all three VNAs are very close to each other up to 400 or 500MHz and then they diverge, the "H" clone sharply so. Everything I do, though, is below 148MHz, and within that larger range usually test in much smaller spans of 10KHz or so and once in a while 200 or 300KHz. All below 148MHz. Any of these--the two nanoVNAs or the Siglent--would be adequate for my purposes. The 10 and 20dB offsets produced by the H-clone is weird. I wonder if the very-wide bandwidth scanned in these tests has anything to do with that
stay under 300MHz, its a 30 buck clone product, tripping on harmonics in the GHz range, and expecting reference values. use it in its range to a frew hundred MHz and here it would have been relevant to see how it performed vs its price.
When evaluating a product, it is always useful to evaluate it beyond its capabilities, to know up to where they may extend. There are many products that can do better than they should - and as a matter of fact, it is still a useful product beyond 300MHz. We just have to manage expectations. I hope that this video helped do that. Thank you for watching and commenting!
The SAA2 VNA uses a different DDS to the NanoVNA. The NanoVNA uses a SI5351 which can only generate frequencies up to 300MHz and uses harmonics to get above it. This reduced dynamic range significantly.
The SAA2 uses an ADF4351 which can generate RF up to 4.4GHz without resorting to using harmonics, resulting in a better dynamic range across the board. Newer SAA2 revisions use the MAX2871 which is very similar to the ADF4351 but can sweep up to 6GHz.
As you say, the shielding will also help with the SAA2.
I also own a NanoVNA, one of the original SAA2 VNAs and a Siglent VNA :-)
Interesting and informative review! I don't do RF work but find this equipment interesting. I studied books in the 70's to get a first class FCC license but haven't ever used that in work.
This review and that on the TinySA show that this technology can be obtained for a reasonable cost for schools and hobbyists.
I got the TinySA just for kicks but not sure what I would do with the VNA.
Thanks for two well done videos. I know it takes lots of time and effort.
Very comprehensive and interesting. Thx for sharing!
Very interesting and clear! Thank you very much (for the utility gnuplot too !!! )
Good review and very important points! The problem is you never know what you get when you buy cheap of of china/ebay/aliexpress... Even when one version was tested good, they build new fakes and it gets worse - but you don't know unless you have the capability to really test your device properly. Thanks a lot!
Fascinating...cheers.
very helpful 🙂
Impressed with the SAA2. Very close to the Siglent. I've just bought the NanoVNA-F V2 enclosed in metal case. Also has 4 metal cans on the circuit board, so hopefully well screened.
I have the same NanoVNA-F v2 and we did a similar test against our Megiq VNA0460, comparing both the bode plot and the smith charts, and they perform equally well. We use it a lot to perform impedance matching and the smith chart on both analyzers results in the same series LC (within some %) so I am very happy with the F v2
the NanoVNA-F is not the same as NanoVNA SAA2. Different companies using different designs. The F is a lower quality knockoff clone of the original design aimed at making them the most money. The SAA is newer design, redesigned from scratch with different goals in mind
@@sammiller6631 The NanoVNA-F v2 (not to be confused with the older NanoVNA-F) is a product developed by Sysjoint based on the original NanoVNA and the SAA v2. I have one and we use it a lot to match our RF circuits (430 up to 2450 MHz). We also have a Megiq VNA0460 and an HP spectrum analyzer with tracking generator and an external directional coupler. Measurement results from the NanoVNA-F v2 I have are as good as from the other equipment. So I would not say this is a lower quality knockoff. The metal case, metal RF shielding cans and the internal construction of the device look great.
Nice to see the comparison. I have nanoVNA-H4 from a reputable seller in The Netherlands and i suppose it is not a clone. The enclosure is a 'snap-on' type, so not easy to open up and look inside and say something about the shielding. I also don't have another instrument at home to compare, but could ask my colleague at work to do a quick compare. I have done some basic measurement filter measurements and on my EFHW 10m long antenna and could fine tune it with this nanoVNA. It all seems right, but once again not much to compare yet. About the port 21 isolation graph, that could be calibrated away? Or not?
Edit: nonsense what I said about the enclosure 🙈 it has 4 screws.
Ótimo vídeo
Obrigado/a 😊
I wonder why people always insist to compare a basic toy like tool to a semi professional one. It is like comparing a Cozy Coupe to a mid size Chevy or Volkswagen.
The SAA2 performing on par with the Siglent, at least on a noise level perspective up to 1.5GHz, was unexpected. There are other valuable pieces of information shared, such as a methodology to graph data from multiple brands of VNAs. Thank you for watching and for your feedback
Nano is not (especialy the official versions) a toy up to 250MHz.
Most of the people will use the nanoVNA way below GHz frequency so it performs great!My '97 Golf still runs like a swiss watch and takes me to places while seeing many luxury cars biting the dust...:) It gets the job done...:)
@@solarflare4259 True. There are other version of Nano that are fine in the GHz range, LiteVNA for example (same manufacturer as for H4), and pretty affordable. Golf bit reminds me of the Balkans 😁
@@galileo_rs Indeed. I've got a H4 and it works great. I find myself using it up to 60 MHz, mostly for HAM shortwave...Currently waiting to receive the 4401 ... Just wanted a larger display.
As for the Golf, I grabbed it back in '99 and I've taken good care of it.It comes really handy now with rising of fuel prices (2L) engine MT... I barely see some like these on the road; even hard to find parts at salvage yards...:)
NanoVNA-H is the old V1 version, Si5335 based, that works on harmonics above something like 300MHz, so no wonder it has much worse specs.
Si5335 synthesizer chip official spec: fmax 200MHz
NanoVNA V2 is new version based on 2x ADF4350 with official fmax of 4400Mhz, so no harmonics needed.
NanoVNA-F V2 clone version with somewhat better/nicer mechanical design, aluminum case instead of plastics, for a bit more money
Yes, please do a video about the Siglent. Also, a full lab tour would be interesting! Thank you!
Great video. Can you make a video on upgrading the SSA3021x to SVA1032x ?
As Andy mentioned DDS are different in you VNAs. That is very important, because RF generator output power is much lower when using 2nd harmonic and even lower when using 3rd harmonic, and if we say RX sensitivity and noise stays the same, you get much worse S/N ratio (you get more noise compared to the signal) on higher frequencies. Shielding would help, but just to a point. When you calibrate VNA it offsets that very low (2nd or up to 5th harmonic) signal level with big gain to get 0dB trace at all frequencies, so it is mostly RF generator output level problem. RX sensitivity is also getting worse at higher frequencies, because of higher mixer conversion losses, which is another 'RF signal level' problem. Inconsistencies between measurements are also due to cable bending (since those cables are not phase stable), SMA tightening and some common mode current outside the coax.
It would be interesting to take one measurement with Siglent, then disconnect cable, connect it again to Siglent and taking second measurement. I suspect you would get almost the same difference as you did with comparison to SAA2. That would show cable bending and SMA tightening contributions plus other measurement uncertainties.
Why not try the LibreVNA ?
The white Nano is a clone and the results you are getting are probably due to that fact. I have a modified clone H and an original H4 and those do not exhibit that kind of behaviour. Try NanoVNA Saver software and do not calibrate on device. Saver will offer much higher number of calibration points and a correction for the trough measurement. Makes sure that you get the latest firmware. DisLord firmware is recommended.
Nice video, thanks ! Yes, a review of the Siglent would be great. By the way, did you get to the SVA1032X from the SSA3021X Plus with just the s/w license upgrades Siglent offers or did you have to have h/w changes made at the factory ?
Thanks a lot for zour effort and the great comparision!
When you take measurements from duplexers, put a dummy load on the unused port.
Leaving the unused port floating ussually results in phantom troughs all over the place.
I had to recheck the video and you're right, I totally forgot to do it. I believe the impact shouldn't be very significant due to the attenuation between the channels, but you're absolutely correct, this is an oversight on my part during this test. Thanks for pointing it out!
@@the_bootloader For what you were doing, I guess you could choose to deliberately ignore wierd responses outside of the band of interest.
The only reason I noticed was because I have to tune commercial duplexers at work, and the phantom troughs can cause serious issues with figuring out what is going on.
@@digitalradiohacker yes you're absolutely right. As you point out, for my purpose it didn't really matter but generally it's good practice and the way to do it. Thanks again for participating and sharing your knowledge, it's great that we can all learn together from our respective experiences
@@the_bootloader Absolutely. I bought one of these as a potential alternative to the 10KG Aeroflex 3500 that I have in my van. I know that a new instrument of this class costs the same as a house, so they're not the same, but there is also 20 years of development between them as well.
In reality, I'm finding the noise floor on mine to be quite pathetic (around 50dB?). This is no good for tuning commercial duplexers which are capable of -80dB rejection.
Mine is marked as NanoVNA-H
The search for firmware that won't make a brick is ongoing.....
The best is...., a HP8753E, ya it cost money, then again, if you want to go down the 1/4 mile track in 7 seconds, your not going to do it with a 4 banger, Datsun! LOL!@
Many thanks. This was a very interesting set of tests. Based on my own needs, my take away is that all three VNAs are very close to each other up to 400 or 500MHz and then they diverge, the "H" clone sharply so. Everything I do, though, is below 148MHz, and within that larger range usually test in much smaller spans of 10KHz or so and once in a while 200 or 300KHz. All below 148MHz. Any of these--the two nanoVNAs or the Siglent--would be adequate for my purposes. The 10 and 20dB offsets produced by the H-clone is weird. I wonder if the very-wide bandwidth scanned in these tests has anything to do with that
Merci pour ce comparatif, hyper intéressant ! ;)
stay under 300MHz, its a 30 buck clone product, tripping on harmonics in the GHz range, and expecting reference values.
use it in its range to a frew hundred MHz and here it would have been relevant to see how it performed vs its price.
When evaluating a product, it is always useful to evaluate it beyond its capabilities, to know up to where they may extend. There are many products that can do better than they should - and as a matter of fact, it is still a useful product beyond 300MHz. We just have to manage expectations. I hope that this video helped do that. Thank you for watching and commenting!
For the data processing I found exporting the s1p or s2p files and using Python with scikit-rf a nice tool too
Good suggestion, thanks for sharing! I may give it a try