On 7/8/23 3:57 PM, Ma?l H?rz wrote:
Hello,
I have read various sources on the Internet and it seem that the classic
NanoVNA is best suited for testing filters (below 300 MHz) and measuring
crystals.
It also seems that the DiLord firmware for NanoVNA-H (not H4) can be
used on the original NanoVNA. Does that mean NanoVNA and NanoVNA-H have
similar performance.
What about NanoVNA-H4? I read it does have worse performance for this
application (<=300MHz) and measuring filters/crystals. But I would
prefer its larger screen.
I thought about additionally getting the LiteVNA for its larger
bandwidth, but apparently the noise floor for lower frequencies is worse
than the original NanoVNA (what about NanoVNA H-4)?
I could not find any info about the RBW of NanoVNA (original),
NanoVNA-H, NanoVNA-H2, or LiteVNA.
My main interest is really testing lower frequencies, as mentioned above.
Any clarification would greatly help. What I find on the Internet is
kind of confusing between various firmware versions and upgraded version
of the various models.
Best regards,
Ma?l H?rz
Without getting into the details of the various firmware versions, if you're measuring high Q devices (like crystals), then using one of the NanoVNAs with a computer running software that lets you have a lot more points is quite useful.
The problem with high Q is that the NanoVNA only samples for a millisecond - that is, it switches the stimulus and receiver frequencies to the next step, then delays a *short* time to let the synthesizers lock, then measures for a millisecond.
If the thing you're measuring has a time delay, or takes a while to settle, then the measurement won't be as accurate. The "time constant" for a resonator (i.e. the time for the energy to decay to 37%, 1/e) is Q radians. Since one cycle is 2pi radians, the number of radians is 2*pi greater. So Q/(2*pi) is the number of cycles, each of which is 1/f, so after Q/(2*pi*f) seconds, the energy will be 37%. (down 4.3 dB)
If your resonator has a Q of 5000, and resonates at 10 MHz, we can run some numbers: 5000/(2*pi*10E6) = about 0.08 milliseconds - so probably not a big deal.
I've not thought about what phase error one would get from the delay, though.
