开云体育

Unless I'm doing something stupid (which is entirely possible) or have been very lucky it may not make very much difference which clone you buy (as long as it has a battery and the soldering is ok!).

If my recollection is correct, according to Hugen (who we believe is the developer of this variant of the original 300 MHz NanoVNA) the worst clones not only have no screening but also the manufacturer tried to make up for this by using PCB for the rear panel. Apparently the PCB reflects signals back into the VNA, making things worse rather than better.

I have two NanoVNAs. One came from UK seller sqcase, has 900 MHz 2-trace firmware and no screening. The rear panel is plastic so I called it a semi-bad clone. The second was ordered before Hugen made his available on Alibaba. It came from AliExpress seller Guangyi0016, has 4-trace 900 MHz firmware, does have screening and a plastic rear panel. Both VNAs came with batteries.

In theory the lack of screening of the semi-bad clone should mean more crosstalk between TX and RX channels. I did the following to check:

On both VNAs I terminated ch0 and ch1 with 50 ohm loads and set the display to show ch1 logmag, scanning from 50 kHz to 900 MHz. As the two channels are not connected I believe that any signal appearing on ch1 should consist purely of crosstalk and noise.

Below 300 MHz the signal levels are off the screen on both VNAs (so lower than -70 dB). Surprisingly the unscreened clone has roughly 3 dB less xtalk/noise between 300 and 400 MHz, although it does have spikes at 300 and 600 MHz. (The spikes seem to vary between VNAs. The ICs are being pushed beyond specification so spikes seem to be a matter of luck rather than build quality.) What I'm seeing above 400 MHz is very similar on both VNAs, with displays showing noise that bounces around a lot but at 900 MHz is rarely above -50dB.

I also tried placing a metal sheet behind and then in front of the input/output areas of the unscreened VNA, to see if I could increase the amount of crosstalk. No effect was apparent.

These simple tests seem to suggest that screening may not be as critical as we all thought it was. Could the calibration isolation measurement & subsequent correction be reducing the effect of crosstalk to below the noise floor?

Mike

I think I have all three. A good Black one with shielding that has 80 dB dynamic range on the low end and 60 dB on the high end. A White one without shielding that has 70 dB on the low end and 50 dB on the high end. The worst one is a black one without shielding that I just got off AliExpress (was only $39). It has only 50 dB on the low end and only 20 dB on the high end. The noise is only on the receive port so I will only use the bad one for TX port measurements.

As someone else mentioned, the spike at 300 MHz seems to be hit or miss and is correctable in software.

Jim K.

Hi Jim

How are you measuring dynamic range? Can you publish a pic of the 80 dB?

WA8TOD

Would it be possible to tell us - in layman's terms - how the spike at 300 MHZ can be corrected through software?

Hi John

The 900 MHz NanoVNA switches from using the oscillator fundamental to using
a harmonic at 300MHz. The trouble is that the oscillator chip isn't
specified to work up to 300MHz. Although most work ok, some cause a glitch
(particularly at high temperatures).

The software fix is to install the 800 MHz version of the firmware, which
switches to using a harmonic at a lower frequency (presumably around 267
MHz).

See the group wiki & files for advice on installing firmware.

Mike

I actually question whether or not the “spike” is “real”. And I further question whether it actually affects operation.

There is a discontinuity at 300 MHz in the NanoVNA. The Stimulus is operating at the fundamental frequency of the Si5351 up to and including 300.000 MHz. Beginning at 300.001 MHz it operates on the third harmonic. So, at 300 MHz the Si5351 is producing 300 MHz….. at 300.001 it drops back to 100.000333 for the fundamental and the output is boosted by about 6 dB.

When ranges are selected that include passing through 300 MHz the detected level at the transition changes by about 2 dB from the fundamental to the harmonic. I suspect that the spike comes and goes depending upon where the bin boundaries hit.

I do not see the spike on my “worse” unit but I am wondering what effect others are seeing on measurements made with the spike visible.

WA8TOD

Ah, so you did see it.
You just didn't believe it is all.

Thank you, Sir.

Hi Warren,

Logmag S21

Jim K. K8SLC