There is no embeded installer, you must simply unpack the zip archive and run the executable.
Like previous releases you need .NET Framework 4.0 to run the program. Minimal supported operating system is Windows XP SP2, but it work fine on all recents versions of Windows (32 and 64 bits). You can also run the program under Linux system (or Apple Mac OSX ) using Mono Framework
Under Mac OSX you need to force the 32 bits version of Mono ( .NET Winform is not fully supported on 64 bits architecture)
Command line : mono — arch=32 SNASharp.exe
Changelist:
Hide label for non visible curve
zoom frame now disapear when the mouse leave the spectrum windows
Remove level display when the mouse leave the spectrum windows.
Add analyzer model on bottom graph display
And the major improvement : display layer management was rewritent to be compatible with recent buggy mono version. You can run this version on recent mono release with fast and nice display.
As some people know, I buy all the analyzers made in China that are compatible with DL4JAL software (because they will also work with SNASharp). This way I can verify that it works and also write a SNASharp configuration file for that model.
I bought this analyzer available for a few months. It is indicated as being usable from 10kHz to 40MHz.
At the reception I made this high resolution photography , it is easier for me to zoom on an image to determine the schematic.
I quickly measured a quartz filter, and the results were catastrophic, I deduced that the analyzer did not implement correct impedance matching on his ports.
I then observed the output circuit of the tracking generator, and realized that there was a big design error. The output consists of an OPA695 operational amplifier (with feedback loop) followed by a 3dB attenuator. This configuration can only offer an output impedance around 18 Ohms, far from the expected 50 Ohms .
Output of tracking generator
I then observed the input circuit and there too there is a big mistake. We have a 3dB attenuator followed by the input of the logarithmic amplifier AD8307. The AD8307 offers an input impedance of 1100 Ohms, the consequence is that the input port offers an impedance of about 240 Ohms, we are also very far from the expected 50 Ohms.
Analyzer input
Fixing the issues
For the output it was quite simple, I removed the two 300R attenuator resistors and replaced the 18 Ohms series resistor by a 47 Ohms resistor, so we have an output that offers a ~50 Ohms broadband impedance.
Tracking generator output impedance fixing
For the input I had to do a little math. I wanted to change the minimum of components, I just replaced the first 300R resistor of the input attenuator by a 62 Ohms resistor. This offers almost 50 Ohms in broadband (49.4 Ohms measured)
Upgraded input, The resistor was directly soldered on the SMA socket because I damaged the PCB when unsoldering the 300R resistor.
Test and conclusions
Port matching is ok and the analyzer keeps 82dB of dynamic range on the low frequencies and 74dB on the high frequencies, which is very correct.
S21 with SNASharp Raw mode ( uncalibrated)
The linearity is also very good, here up to 60dB.
Linearity checkMesure of a 4MHz SSB filterMesure of an 7.5MHz LPF
Once modified, this 20€ analyzer is perfectly usable for HF bands. My tests indicated that it could make measurements from 1kHz to 40MHz. The generator provides a fairly clean signal, the harmonics are at -48dBc (H2) and -50dBc (H3).
A year ago I bought an LTDZ analyzer, without finding the needed time to test it. It’s done now. It is an analyzer that can be found for less than 40 € (with a casing) or around 30 € without a casing. The schematic is identical to the original D6, so with the same design errors, and the only differences concern the firmware and the size of the device. The LTDZ analyzer is the smallest one.
D6 and LTDZ analyzers
Like the D6 this analyzer suffers from a significant overload when used it in SNA mode, the TG output delivers a signal too strong for the mixer located on the input port
Original LTDZ. made with SNASharp in Raw mode, for different attenuators values.
For this reason, if you hope to made relevant measures in tracking mode, you will add (at least) 12dB of attenuators (6dB on each port), as example with these small SMA attenuators.
LTDZ with two 6dB additionals attenuators
As the D6 the input impedance is not matched, the analyzer presents about 30 Ohms instead of the expected 50 Ohms, the author having added an unnecessary 50 Ohms resistor on the input while the mixer is already matched to 50 Ohms.
Return loss of poor matched LTDZ entry
But, as expected, the matching is improved when we add 6dB attenuator.
Return of LTDZ entry with an additional 6dB attenuator
You can also improve the input matching by removing the 50 Ohm resistor..
If you have removed the 50R resistor, the entry become more sensitive and you must add 3dB more attenuation (15dB total) in tracking mode, otherwise the analyzer could saturate.
LTDZ with 50R removed. made with SNASharp in Raw mode, for different attenuators values.
At this point, the issues encountered were anticiped (because they were also present on the D6) , but the reality is worse. I realized that the analyzer was returning wrongs values, it was exaggerating the level variations by 60% (16dB of variation for 10dB efective). I think it’s an error in the firmware, maybe the author wanted to compensate for the lack of dynamic range of the device, by returning « amplified » results, sadly this made the measurements wrong . To compensate this issue, i need to change vertical sensibility in SNASharp device description. (This new setup will be available of the next SNASharp release, but LTDZ owner can edit this value on current version). With the right sensibility, the dynamic range of this analyzer down to less as 50dB.
SNASharp device editor VerticalResolutiondB updateD6 Vs LTDZ on dynamic range evaluation, note the poor TG isolation on LTDZ
For this reason it is better to buy a D6 than an LTDZ, especially since the manufacturer of the D6 has taken into account certain comments from this blog to improve it.
Some improvements can be made on this analyzer, the same as the D6 (untested on LTDZ)
There is no embeded installer, you must simply unpack the zip archive and run the executable.
Like previous releases you need .NET Framework 4.0 to run the program. Minimal supported operating system is Windows XP SP2, but it work fine on all recents versions of Windows (32 and 64 bits). You can also run the program under Linux system (or Apple Mac OSX ) using Mono Framework
Under Mac OSX you need to force the 32 bits version of Mono ( .NET Winform is not fully supported on 64 bits architecture)
CaptureDelay_µs : This allow to insert a « pause » between 2 captures. This is needed to support LTDZ analyzer.
DefaultPPMCorrection: This is a temporary workaround, in feature release i will add PPM correction on calibration sequence.
TrackingModeFrequencyShift : I added this frequency offset to compensate a firmware bug on Azeroth_Si5351_SpectrumAnalyzer, otherwise the analyzer will work with 200 000Hz frequency error.
Bugfix:
I added device unexpected port close detection, and automaticly reconnect the device if needed.
If device connection is lost in loop acquisition, you can now easily cancel with stop button.
Don’t hesitate to put a comment below this article if you have any question.
There is not embeded installer, you must simply unpack the zip archive and run the executable.
Like previous releases you need .NET Framework 4.0 to run the program. Minimal supported operating system is Windows XP SP2, but it work fine on all recents versions of windows (32 and 64 bits). You can also run the program under Linux system (or Apple Mac OSX ) using Mono Framework
Under Mac OSX you need to force the 32 bits version of Mono ( .NET Winform is not fully supported on 64 bits architecture)
Command line : mono — arch=32 SNASharp.exe
About this version..
I added the possibility to disable the calibration if needed. With this new capture mode you can evaluate the real bandwidth of your analyzer (without the calibration correction). This really help to evaluate an hardware improvment. To activate this mode you need to check « Raw capture« . The 0dB reference (for each analyzers) can be set on device editor panel on « RawMode_0dB_Reference » entry.
The displayed level is done by:
Displayed Level In dB = (Analyzer captured value – RawMode_0dB_Reference ) * VerticalResolutiondB .
Later don’t miss to re-enable the calibration to make accurate measurements (The raw mode capture option is automatically saved and restored when the program exit and start).
v2019_08_30_0 new options
D6 Analyzer capture in raw mode
Another new option is « Use Att cal » ( Checked by default). This option is dedicated for embeded attenuators analyzers ( like NTW70,NTW300,NWT500). If you keep this option checked, the behaviors stay the same as previous versions of SNASharp. If you unchek this option the analyzer will avoid to use specific attenuator levels calibration set ( In this case the 0dB calibration set serves as reference).
Why this option ?
It’s usefull to check if embeded attenuators of your analyzer works fine, and evaluate the linearity of your hardware. In addition, this can reassure some users who did not understand why the measurements remain on a reference of 0 dB while attenuators were engaged.
NWT70 capture with differents attenuator sets and « Use Att cal » unchecked
I added The Geekcreit® Spectrum Analyzer USB LTDZ_35-4400M on supported hardware list( not tested yet).
Geekcreit® Spectrum Analyzer USB LTDZ_35-4400M
Don’t hesitate to put a comment below this article if you have any question.
Manoes, a reader of this blog, have send to me a description of a set of modifications for the « D6 » analyzer (My words remains in italics).
Mod’s I made.
I did put a lot of C’s 100pf on existing and newly made places on the powerline traces. I also did try one on the input C it did not have effect that I expected but…… my measurement could be false, maybe someone can do it again. Upside
Upside of the board; maybe I forgot some C’s to mention but better one to much then one forgotten
Underside of the board a lot of them, the places are experimental found…..
Just try and error with a non-conductive plier!!
Also I made some new screening on the upside and underside of the board.
I did change the SMA connectors for a wider type witch has nuts on the outside.
Before I put the whole board inside the cabinet I glued some conductive foam on the board it did give a bit better result.
All the measurements are made with input and output closed with 50 Ohm the last picture is input and output connected with a short SMA cable.
New screening under Changed screening upside and new SMA connectors Changed connectors Yellow before mod’s, green after mods without casing and no foam. Yellow before mod’s, green after mod’s board inside casing, new SMA and conductive foam. Trace in/ out connected with mod’s : -3 dB input and -6 dB output and changed 120 kHz. Filter
All the measurements are made with the above mods. Terrible enough my spectrumanalyser is broken during the measurements so I can not produce better pictures as above ones. I do hope to repair the old boy when I have the time to live and fun to do it. (High tension unit defect, transformer into smoke so I have to made a new one by hand . Software VMA of our friend Vitor Martins Augusto. Grtz, Manoes
For several months i don’t understand why the frequency response curve of this analyzer deteriorated when we exceeded 3GHz. By drastically reducing this problem Manoes give me an answer. I will add an option on SNASharp to be able to trace the frequency response of an analyzer by disabling the calibration, this can help to understand how a hardware modification work. David.
Dans ce post de blog Yvon F6GLE présente certaines mesures réalisées avec le NanoVNA et les compare avec celles obtenues avec un VNA de référence.
Il utilise un NanoVNA « noir », c’est à dire le clone Chinois de meilleure qualité. C’est un modèle doté d’une batterie et dont les mélangeurs d’entrée sont blindés. Il existe d’autres clones dont les mélangeurs ne sont pas blindés, dotés ou non dotés de batterie. Un d’entre eux est blanc avec un gecko dessiné sur la droite.
L’appareil utilisé en base de comparaison est un R&S ZNC3. C’est un analyseur qui n’est plus actuellement fabriqué, on le trouve cependant sur le marché de l’occasion à des prix se situant dans une fourchette de 12000$ à 20 000$
Mesure de dynamique :
Concernant la dynamique en transmission (S21), je me suis
mis en full span (50KHz-900MHz). A priori c’est idem avec d’autre span. Et j’ai
regardé sur quelques fréquences avec un atténuateur ajustable entre le port 0
et le port1 :
A 10MHz : de 0 à 60
dB : erreur de mesure sur marqueur < 0.5 dB. A 70dB, la mesure est
encore OK mais avec un bruit d’environ 1.5dB
A 100MHz : idem
A 290MHz : idem , mais avec
un bruit de l’ordre de 2dB a 70dB
A 500MHz : de 0 à
40dB : erreur <1dB. A 50dB : affiche 48.5 avec du bruit environ
2dB
A 700MHz : environ idem
A 900MHz : jusqu’à
30dB : erreur <1dB, après ça se gate avec pas mal de bruit
Concernant la phase,
ça semble à peu près bon, il y a du bruit sur la phase quand il y a du bruit
sur le niveau.
Pas en fréquence
Le pas s’ajuste en fonction du span que l’on met :
Exemple :
avec un span de 3 à 30MHz, le pas est de 280KHz, avec un span de 144 à 145MHz,
le pas est de 10KHz.
Le pas mini que l’on peut paramétrer à 100MHz et que l’on
voit sur analyseur de spetcre en P0 est d’environe 100Hz
Adaptation des ports :
J’ai aussi fait une mesure de RL sur les 2 ports pour voir
leurs adaptation 50ohms :
Sur P0 : sur
toute la bande : RL environ -30dB +/-2dB
Sur P1 : jusqu’à
150MHz : RL meilleur que -30dB, puis monte progressivement 400MHz :
-22dB et 900MHz -16dB
Mesure sur filtre passif 10MHz :
S21 mag (en rouge : sur analyseur R&S ZNC3, en bleu sur nanoVNA) :
S21 phase sur nanoVNA :
S21 mag sur R&S avec zoom sur 10MHz : (en rouge : sur analyseur R&S ZNC3, en bleu sur nanoVNA)
S11
Mesure S11 de : resistance 12 ohms en serie avec 47pF au bout de 9cm de coax. Span 2MHz à 100MHz (en rouge : sur analyseur R&S ZNC3, en bleu sur nanoVNA)
J’ai probablement mal réalisé la calibration….ce qui
expliquerai le décalage en haut de bande
Mesure S11 de : resistance 12 ohms en serie avec 47pF au bout de 9cm de coax. Span 300MHz à 400MHz : (en rouge : sur analyseur R&S ZNC3, en bleu sur nanoVNA)
J’ai probablement mal réalisé la calibration….ce qui
expliquerai le décalage…
Spectre mesuré sur le port 0 (avec Span 0). Mesure avec
analyseur R&S FSU
F à 10MHz :
A 100MHz :
A 250MHz :
A 500MHz :
A 750MHz :
A 900MHz :
Spectre mesuré en port 0 en maxhold nanoVNA en span de 50KHz à 900MHz :
Dans ce span, le pas en fréquence est plus fin entre 100MHz
et 300MHz.
Spectre mesuré en port 0 en maxhold nanoVNA en span de 50KHz à 1MHz :
Spectre mesuré en port 0
en fréquence fixe à 100MHz
Sur une bande de 1MHz :
Sur une bande de 100KHz :
On voit l’OL 5 KHz
plus haut
Sur une bande de 1KHz :
Spectre mesuré en port 0
en fréquence fixe à 100MHz
Sur une bande de 1MHz :
Sur une bande de 100KHz :
Sur une bande de 1KHz :
Dans ces 1eres mesures, on peut voir que les performances sont
plutôt bonnes pour un appareil qui tient dans la main à 50€.
There is not embeded installer, you must simply unpack the zip archive and run the executable.
Like previous releases you need .NET Framework 4.0 to run the program. Minimal supported operating system is Windows XP SP2, but it work fine on all recents versions of windows (32 and 64 bits). You can also run the program under Linux system (or Apple Mac OSX ) using Mono Framework
Under Mac OSX you need to force the 32 bits version of Mono ( .NET Winform is not fully supported on 64 bits architecture)
Command line : mono — arch=32 SNASharp.exe
Bug fix:
A incompatibility bug ( only D6 analyzer with factory firmware) was introduced on previous version of SNASharp. Now fixed
Improvments/new features
The VFO frequency is automaticly saved and restored
F1 & F2 key shortcuts where swapped, this is now more natural ( F1 to decrease and F2 to increase zoom frequency range).
There is not embeded installer, you must simply unpack the zip archive and run the executable.
Like previous releases you need .NET Framework 4.0 to run the program. Minimal supported operating system is Windows XP SP2, but it work fine on all recents versions of windows (32 and 64 bits). You can also run the program under Linux system (or Apple Mac OSX ) using Mono Framework
Under Mac OSX you need to force the 32 bits version of Mono ( .NET Winform is not fully supported on 64 bits architecture)
Command line : mono — arch=32 SNASharp.exe
Bug fix:
A bad calibration file was used in rare case (the issue is related to a wrong choise of attenuation level setup), this occur only with analyseur models NWT70,NWT500,NWT300. Full fixed now.
The VFO stop command now work on D6 analyser.
Device editor « delete » option was removed ( not working)
Transformer combo box removed ( useless)
New « analyse only » feature
Improvments/new features
Minimal configuration : SNASharp can now run efficiently on 1024*768 display configuration. This can be usefull if you own an old laptop computer.
SNASharp start on full resolution available ( like the majority of PC software).
Some improvments on crystal analyzer function
A « Analyse only » button was added. Using this new function:
Automatic tracking is unactived, the user must zoom manualy to crystal resonances range ( like screen capture above).
The sampling resolution is not forced, they can be defined by the user.
Detector is not forced, user car select the best detector ( linear if we need only serie resonnance, both in all others case).
With this new option, is easy to fastly sort a big set of crystal ( by frequency and quality factor)
(Full auto mode is identical to previous crystal analysis function, the sotware automaticly fix capture resolution, detector, and scan wide frequency range to found the stronger crystal resonnance).
Tout à commencé par un message intercepté sur Twitter.
On trouve ensuite, assez rapidement des vendeurs sur Aliexpress et sur Ebay.
En fouillant un peu, on trouve même un dépot GitHub sur le projet, et c’est la que ça devient vraiment intéressant car on rentre dans le monde de l’open source et de l’open hardware.
Le GitHub nous indique que l’auteur du projet est un certain edy555 , Japonais, ce qui permet de trouver son twitter
Ses tweets étant rédigés en Japonais, on demande à la traduction automatique de tenter d’en saisir le contenu et on tombe sur quelque chose d’intéressant.
On apprend donc que le clone en vente sur Aliexpress et ebay a été fait « dans le dos » de l’auteur initial, ce qui est finalement assez courant. Plus intéressant, l’auteur a retrouvé la trace de l’auteur ( Chinois cette fois ci) du clone, sur un forum.
Voila un lien vers un PDF du contenu de ce forum traduit automatiquement.
On y apprend que l’auteur du clone est admiratif de la qualité du travail effectué par l’auteur initial, qu’il a programmé un outil (en C#) pour piloter l’analyseur depuis le PC, et rajouté quelques fonctionnalités pour gérer une batterie.
J’ai obtenu ( auprès d’un vendeur à qui je l’ai demandé) une archive contenant les schémas, le logiciel PC, et les firmwares alternatif pour le clone, elle est ici.
Ce qui semble particulièrement intéressant, c’est le comparatif posté par l’auteur du clone Chinois entre ce VNA et un Agilent E5062A.
Mis à part au niveau de la dynamique ( limitée à 70dB sur le NanoVNA) les mesures semblent comparables.
Sur l’analyseur lui même, voila le manuel d’utilisation, qui contient la synoptique et le schéma.
