Supply noise at higher harmonics #internals
|
When using the nanoVNA at frequencies above 300MHz it switches to harmonics mode
First attached picture shows the harmonics till 1600MHz of the SI5351 output when set at 100MHz. The 9th harmonic is at 900MHz only 34dB below the fundamental at 100MHz. To validate the performance of the SA612 I soldered two wires to output of the REF mixer and observed the output using a balanced USB audio input Second picture shows the 8k audio FFT when the nanoVNA is set to 1MHz. About 100dB SNR at this small audio BW. The nanoVNA uses less points (48) in its calculations and has therefore substantial less SNR The third picture shows the output when set to 2GHz. The SNR is reduced to something between 20 and 30dB This 80dB signal reduction is much larger as the reduction of power in the harmonics but can probably be attributed to the mixing loss in the SA612 as it it operated way out of spec (max 500MHz) The fourth picture shows the roughly 50dB SNR you get when you operate the nanoVNA at 2GHz from a clean 5V supply. A very welcome improvement making S21 measurements possible. A quick calculation based on the schematic suggests the REF mixer has the highest signal levels while the TRAN mixer is 10dB lower and REFL is 20dB lower If I have enough courage I may change the attenuators for the REFL and TRAN mixer to get more balanced power levels. |
|
Using the console command "sample" it is possible to measure the pure reference, refl and thru signal strength.
First picture shows the change over frequency of the reference signals (only look at the changes in level, the absolute levels differ due to some TAPR VNA internals) According to the internal gain setting table in the nanoVNA there is additional gain added as the frequency moves up. These are the vertical jumps, each about 5dB at 300, 600,900, 1200, 1500 and 1800MHz. In total about 45dB gain is added in the audio path to compensate for signal loss and with the 10dB delta between lowest and highest frequencies the delta is 55dB, similar to the 55dB measured with the direct connection to the SA612 so the nanoVNA adc and dsp algorithm do maintain the dynamic range. Next picture shows what happens when you use a clean 5V supply. Much less noise in the reference signal. The weird thing with the reference signal is the increase in strength above 1.5GHz. This could be caused by capacitive coupling in the attenuator before the reference mixer Third picture is the change of the amplitude signal (red line) of reflection mixer without anything connected to the SMA CH0 connector, a fairly good open. The change of amplitude over frequency does not show any weird things. The green line shows the leakage from the output signal into the forward mixer. The pattern suggest the attached cable acted like an antenna The fourth picture is the change of the amplitude of the forward mixer. (green line) with a through from CH0 to CH1. The bridge (red line) stops working above 1.5GHz as the characteristic pattern of the cable disappears, possibly because of insufficient measurement SNR Till now I have not been able to get valid phase signal from the bridge above 1.5GHz, only amplitude with at least 20% noise. Comparing the absolute levels of the signals suggests there is about 10dB headroom for increasing the thru signal and about 20dB for increasing the reflection signal. In particular the latter might have a big impact in the performance of the bridge at higher frequencies. I hope this information is not too detailed but writing it down helps me to understand it much better. |
|
Using some generous amplification after the 2GHz filter the nanoVNA performance for S21 measurements is getting to a usable level.
Attached S21 of a 2GHz cavity filter. It proves the SA612 has a lot of headroom when you eliminate the fundamental Next step is to test a high pass filter to get rid of the fundamentals so it will also become possible to measure S21 of low pass filters |
|
When @edy555 designed it and in Japan it was possible for others to build it in kit form, then in 2017 a description of the structure was made. At the end of this description is a list of known but unresolved issues.
The first was to not use 3.7V LiPo, but the product was redesigned at this point and made in large quantities. see: /g/nanovna-users/attachment/484/0/NanoVNA%20alpha1%20kit%20assembly%20manual%20_v1.2%20translated_.pdf 73, Gyula |
Oristo
Gyula Molnar