Beware dodgy Vector Network Analysers

posted in: Ham radio | 40

Just a quick post about a bad experience with buying a Vector Network Analyser from AliExpress.

 

Update 10th July.

Thanks to everyone who has commented. I’ve updated the information in the post and added some more information to the end.

Please see the end of the post for updated information.

 

Original post starts here…..

 

The “NanoVNA” VNA arrived this morning, from this vendor on Aliexpress :MAXGEEK Tech Store

 

But when I opened the package I immediately noticed that the back PCB was scratched, and the screen also had a load of light scratches on it, as if it had been used and then perhaps returned and repaired (I’ve no idea) – See the photos at the bottom of the post.

I immediately raised a dispute with the vendor, and their response, was that the VNA I took the photos of, was not the one that they had sent to me, and I was maliciously claiming to get a refund.

I don’t normally video the opening of these sorts of things, but I guess I should do so in future, but they could of course claim I’ve open the box and resealed it etc

Seeing, the chances of any refund from this supplier look remote, I thought I may as well take the back off the analyser and look inside, I found some other things which could be done better number of serious problems with this device.

 

Firstly the shielding is missing from the 3 mixer sections, its obvious from the PCB that there should be shielding.

 

And I noticed that the are NXP SA602 rather than the SA612, as specified in the original design.

The SA602 is only rated to 45Mhz, so despite the analyser being sold as a 900Mhz device, it would really only function on HF.

Even if it was fitted with SA612’s, they are only rated to 500Mhz, which is half the claimed performance of the device

 

Following comments from lots of knowledgeable readers. I see that I was incorrect in my diagnosis, that the mixer chips would not be usable above 45Mhz.

The datasheets for the SA602 can be downloaded from here   https://www.nxp.com/docs/en/data-sheet/SA602A.pdf  and the SA612 from here https://www.nxp.com/docs/en/data-sheet/SA612A.pdf

 

Both the SA602 and SA612 seem to have similar specifications, but the data sheet for the SA612 introduction specifically mentions 500Mhz, but the SA602 didn’t.

I presume there was a reason that the original design used the SA612 rather than the SA602, but this could purely be something to do with availability.
I used to assume that devices with higher numbers in their model name, have a better specification, but I have since learnt from experience, that this isn’t always the case.

 

I had initially intended to replace the mixers with the SA612, and I have ordered them, however I may not end up replacing them if the analyzer seems to be function reasonable well using the SA602’s

I’m still going to make some shielding for the mixers, from some thin copper sheet, and tack that in place with a few points of solder

 

 

Updates 10th July

 

Doing some more research, and the original design is on github https://github.com/ttrftech/NanoVNA

 

https://github.com/ttrftech/NanoVNA/raw/master/doc/nanovna-sch.pdf

 

With links to the original manufacturer, who unfortunately seems to no longer make them,

 

https://ttrf.tk/posts/2016-11-12-tiny-vector-network-analyzer-nanovna-get-work-with-lcd/

 

I also found this post, by the original designer

 

https://groups.io/g/nanovna-users/topic/32309232?p=,,,20,0,0,0::recentpostdate%2Fsticky,,,20,2,0,32309232

 

Which shows the original design and the need for shielding

 

It also shows what they consider to be the “Bad clone” – which is the one I have.

 

I’ve no idea what makes the other clone worse than mine, so it would be interesting to see some photos of the PCB from that type.

 

 

Thinking about the problem with the display. My best guess is that they are using recycled / recovered screens, taken from other equipment.

I looked at a few videos on youtube, and I’m not sure if the analyser in this video has a plastic protector on it,  https://www.youtube.com/watch?v=fGO74yqDRe4

But it looks like it’s not in good condition.

Mine didn’t come with the plastic peel off, screen protector strip, but even if it did, I’ve had instances where I’ve received other LCD screens (e.g Nokia 5110 screens), which were supposed to be new, but which were obviously used, and had burn in of a logo or the date etc in the center of the screen, and yet still had a screen protector on it.

The recyclers seem to just remove the screens, and stick a new plastic protector on them. They don’t even bother to clean the screen before applying the protector, probably because that costs time and money.

 

Another minor problem with my device is the push button , jog control seems to be partially defective. This could be because the unit is not new, or it could just be a manufacturing fault, its impossible to tell.

 

One interesting thing I noticed, is that the PCB seems to have an unpopulated set of PCB pads, which looks like its for a micro SD socket.

The original design, seems to show the SD card as used, so it may not be supported in the firmware.

 

 

Anyway. I’ll continue with my investigation into this device

 

 

 

 

 

 

40 Responses

  1. ken
    |

    thanks for the info

  2. Roger Clark
    |

    No worries.

    When I get chance I’ll write a more detailed blog on this e.g. put the links to the original open source design etc

    https://github.com/ttrftech/NanoVNA

    I will also probably attach a debugger to the STM32 chip and read out the firmware, if its not protected and see if it contains an exact copy of the open source firmware or perhaps an improved version.

  3. Anonymous
    |

    Removing and replacing the chips should be simple enough, liquid flux around the pins and hot air will remove them without damage. Then desolder wick to remove old solder, solder paste on the pads and place new chips use hot air to reflow solder.

  4. Roger Clark
    |

    Thanks.

    That’s my plan 😉

  5. James
    |

    I had actually considered bying one of these, not now thank you roger

  6. Roger Clark
    |

    No worries

    Apparently the one I bought is not the worst clone, just a bad clone.

    When I get time I will fix it, but finding the time is always difficult

  7. Anonymous
    |

    Thanks. I was about to buy.

  8. Roger Clark
    |

    Glad to help..

    So you don’t waste $60 USD, like I did.

  9. gerrykav
    |

    Nice write-up Roger. Thanks.

  10. Roger Clark
    |

    Its just a quick post, since I’m busy with a lot of other stuff..

    I just wanted to alert people to the problems with these devices

  11. dave c
    |

    write suppliers name large ALIEXPRESS pure chinese crap i use paypal to ensure that these bent chinese suppliers do not get away with robbery

  12. Roger Clark
    |

    I’ve not had an issue with AliExpress’s dispute resolution in the past, but this could be the first time.

  13. Don Miller
    |

    Tell us the vendor so we can steer clear of them. If you bought it on eBay you have recourse. If you used PayPal you also have recourse.

  14. Shaun
    |

    Sorry to hear the bad news regarding this device as we were talking about last week, Rob VK4HAT brought a simular unit and seems to work ok.

  15. Lasse Moell
    |

    Both the SA602 and 612 are specified up to 500 MHz as per NXP data sheet

  16. Peter-PA3FQH
    |

    Hi Roger,
    According to the NXP datasheets,there is very little difference between the NE602 and NE612. Both can handle input frequencies up to 500MHz, not 45 as you suspect.
    Not sure if other versions do have shielded compartments for the NE6x2, mine doesn’t (looks identical to yours, also with the NE602’s)

  17. Roger Clark
    |

    Hi Peter

    Thanks.

    I’ll update my post.

    I looked again at the data sheets and the SA612 explicitly says its for applications up to 500MHz, but the SA602 doesn’t contain that text.

    However, looking at the data tables further down in the document, I agree there does not seem to be much difference in these 2 devices, and I noticed that they seemed to be virtually the same price.
    I presume the original design chose the SA612 instead of the SA602 for a reason. Perhaps its lower noise.

    I had another look at the other IC’s on the board and it uses he same clock (frequency) generator, Si5351 which can generate square waves up to 200Mhz.

    These analysers are being sold as operating up to 900Mhz, but I find it hard to believe they can operate that well above 500Mhz, because of the mixers and the clock generator limitations.

  18. Roger Clark
    |

    Thanks. See my comment to Peter PA3FQH

  19. Roger Clark
    |

    Thanks.

    After reading the comment from Peter PA3FQH, I think the SA602 may be adequate.. Just probably not the optimum device, as specified in the original Open Source design.

    I will still order some SA612’s since they are not expensive, and I may replace them if I can think of a good way to compare the performance, before and after changing them

  20. Gerry Kavanagh
    |

    I guess if the device is calibrated up to 900HHz using SOL calibration loads, any non-linearities would be included in the calibration curve, so theoretically should work fine, albeit with a loss in resolution.
    / Gerry

  21. gerrykav
    |

    I have used that for tricky removals, like relays. For SMD I just use hot air and lots of flux.

  22. Roger Clark
    |

    Hi Shaun

    I think the problem is mainly the supplier sending me a used unit and claiming its new. Goodness knows where they got hold of a used unit.

    I’m wondering if possibly they are using recovered screens out of something else.

    I’ve had problems when buying Nokia 5110 LCD screens, for use with my projects ( https://duckduckgo.com/?q=nokia+5110+LCD&t=h_&ia=images&iax=images )

    Because I’d often find they were scratched and dirty, as if they’d been removed from an old phone and they don’t even bother to clean them with alcohol etc, before putting them into an anti-static bag and passing them off as new

  23. Roger Clark
    |

    I did mention the vendor in the post.

    I’m not sure its directly their fault, because most of these vendors don’t make these devices, they simply by them down the electronics markets in China and then package and sell them.
    Its possible that they didn’t know that it was in poor condition. But what is bad, was that their standard response is to simply claim that I already had one another one of these and was trying to rip them off by claiming a refund.
    Its a great argument for them to use, as unless something has a serial number etched into it etc, its impossible to prove whether the one you are showing is the one they sent you, and this can be applied to anything they sell.

  24. Roger Clark
    |

    Gerry,

    I’m not sure how it could be calibrated to 900Mhz, because the frequency generator only can produce 200Mhz square waves, and the mixers are only rated to 500Mhz.

    BTW. I know square waves are full of harmonics, but don’t know whether that, normally unwanted aspect, could be usefully used in this context

  25. Roger Clark
    |

    With devices like this, I normally try just to snip the legs where they join onto the main body of the IC, or attempt to break up the IC without putting too much strain on the legs and the PCB.

    On this board there are a lot of small SMD caps and resistors, which are very close to the IC, and I think it would be hard not to heat those, and potentially move them, while trying to get the IC hot enough

  26. Anonymous
    |

    The response they gave you was why I was interested in the vendor. I will only deal with honest vendors and their response was far from honest.

  27. Lasse Moell
    |

    The SA602/612 do offer a lot better performance than what you suggest. The highly acclaimed DG8SAQ VNWA uses the same mixer, and runs well over 1GHz, with a “low” frequency oscillator with plenty of harmonics! The performance do drops off above 500 MHz but that is not just the mixer but how the square wave of the oscillator is utilized. Calibration is the key here, and if there is no specific h/w errors to your VNA, it should work as specified, if your cal kit is OK. Try find the documetation on how the DG8SAQ VNWA works.

  28. Roger Clark
    |

    Thanks.

    I was just working from the spec’s and thought 900Mhz was too optimistic for a IC designed to operate at 500Mhz, and a oscillator with max freq of 200Mhz.
    But the 4th or 5th harmonic would be quite strong.

    I still have the problem of recycled parts being used, that I’m not happy with 🙁

  29. gerrykav
    |

    Hi Roger…
    Yes, apparently the higher order harmonics are at a high enough level that they can be utilised.
    EU1KY AA uses this ability quite successfully: https://bitbucket.org/kuchura/eu1ky_aa_v3/wiki/Home
    Firmware updates have gotten reasonable success at ~1.3GHz
    / Gerry

  30. Roger Clark
    |

    Thanks. I’ll take a look

  31. Steve
    |

    Hi Roger , the VNA performance above 500Mhz is called out on the data sheet and it is poor, so I don’t think anything is being hidden. You just don’t get great performance a 900Mhz for £50.

    I am interested in how the performance improves when you have all the correct shielding in place.

    Regards
    G0AIN

  32. Roger Clark
    |

    Hi Steve

    I totally agree with the data sheet. Theses are 500Mhz mixers, but the analysers are being advertised as 900Mhz in most of the listings

    I realised that I’d not get a 900Mhz analyser for £50, but the vendors should really be truthful and not list them as 900Mhz when really they are a 500Mhz analyser

    I’m very busy with my work on the OpenGD77 firmware at the moment, but I hope to be able to do some tests on the VNA in a week or two when things quieten down.

    I also know someone (Ron VK3AFW) who is an expert in these things and has some expensive commercial analysers he could compare it against, so I will have a chat to him when I see him at the local radio club meeting next Tuesday.

  33. Jens
    |

    Has anybody measured the accuracy of this simple device? I am more interested in those results and don’t care about used mixer devices and any scratches on particular shipments.

  34. Vk4hat
    |

    Just because a manufacturer specs a part at 500meg, at 501meg it does not magically stop working. Its just they have not characterized the part at higher frequencies. Most will work way beyond that, in this case the 602 is the better mixer.

    Same goes with the si5351a, read the code for the firmware and the data sheet and you can see exactly whats happening, the 25meg pll is multiplied by 36 and that gives 900meg, in the divider stages of the ic you divide by 1, this is the upper bounds of what the ic can output.

  35. Steve Withnell
    |

    I’ve done some comparisons against my miniVNAtiny and I’ve tested the dummy load kit on my Siglent to 1.5GHz. No issues so far. I only intend to use the Nano upto 30MHz and I’m delighted with it so far. The device is open source hardware, opensource software and there is transparency on the specification. It’s around 10% the price of my miniVNAtiny and is just as good upto 30MHz. The PC software is a bit clunky but fully featured.

    Free download https://www.banggood.com/NanoVNA-Vector-Network-Analyzer-50KHz-900MHz-Digital-Display-Touch-Screen-Shortwave-MF-HF-VHF-UHF-Antenna-Analyzer-Standing-Wave-p-1471576.html?akmClientCountry=GB&cur_warehouse=CN

  36. Roger Clark
    |

    @Jens
    Re: Has anybody measured the accuracy of this simple device?

    I have not seen an comparison with the results from this device, versus commercial equipment.

  37. Roger Clark
    |

    Steve

    These clones are not actually open source and the PC software in the link is closed source. (Windows Exe)
    The clones seem to have different PCB layouts, depending on which clone version you receive, which seems to vary, and is not related to how the device looks from the front panel.

    I’ve compared my clone with photos of other people’s clones and there may also be schematic and component differences between the clones.
    Also the clones use different versions of the mixer chip, and again there is no way of knowing what you are going to get.

    I friend of mine stuck lucky and got a clone which had the RF shielding over all the mixers, and I think possibly the better spec mixer chips, and more debugger connections on the PC etc, but I was not so lucky.

  38. EvO
    |

    Very informative.
    Also in this field you are the forerunner of all, really a great job.
    Beyond what has already been written, about NE602 and NE612 it might be useful to read this:

    http://soldersmoke.blogspot.com/2009/06/na5n-on-ne602.html

    Thank you very much Roger!

    EvO

  39. Tom Berger (K1TRB)
    |

    Roger,
    Steer your readers to
    nanovna-users@groups.io
    and
    http://www.cqham.ru/forum/showthread.php?40166-nanoVNA-%E8%E7-%CA%E8%F2%E0%FF
    They will find out all about clones and good sources.
    Also they will find out about the amazing developments on the nanovna.
    The design evolves from that of the VNWA. (Which followed after the N2PK VNA.)

    K1TRB

Leave a Reply