开云体育

Hi everyone,
Just curious, is there a way to calibrate the output amplitude so that when measuring say a filter I can get it to read the same as a calibrated spectrum analyzer??
Thanks,
Ed

On Fri, Jan 10, 2020 at 08:54 AM, ejfelix4976 wrote:

Just curious, is there a way to calibrate the output amplitude so that when measuring say a filter I can get it to read the same as a calibrated spectrum analyzer??
===============================================

Ed,
When measuring a filter with a spectrum analyzer the usual procedure is to do so in concert with a tracking generator. The S21 (thru measurement) on the NanoVNA accomplishes the same thing within the limits of its dynamic range.

- Herb

You can measure a filter without a tracking generator and do want you desire... of course WHY. In any case, sure.

Set your generator to the same level as your vna. Stop the vna sweep so you can do this. Then run the generator direct to the SA. That is your zero reference and then insert filter under test. The measure is now IL and same as magS21.

Thanks for the replies!
Actually My spec an has a tracking gen which I normalize.
I noticed after doing a cal on the nanovna and nanoVNA F that there is a 2 to 3 dB difference in the measurement.
Used the same cables and the frequency was 6.5M to 14.5M on both
RBW was 3 KHz
I know the spec an is good.
I was curious if there is a way to adjust the Nanovna calibration to reflect the same value.

Is there a difference in S21 after thru calibration using a direct cable between CH0 and CH1?
That should not be as the thru calibration should normalize S21

--
NanoVNA Wiki: /g/nanovna-users/wiki/home
NanoVNA Files: /g/nanovna-users/files
Erik, PD0EK

There is a 2 to 3 dB difference between what the spec an says and the nanoVNA.
I know the spec an is good as it is calibrated.

Just to be clear, you are saying that the vna gives an insertion loss that is different than what the SA TG provides and the difference is as much as 2- 3 dB?

Yes, that’s exactly what I see

As a check try the same measurement with just a pad. Try a 3 or 6 db whatever pad you may have on hand. The suspect area I believe is either source or load termination either from the vna or the sa. One of these devices is NOT providing a proper 50 ohm termination. Under the assumption that the filter is designed to be source and load terminated into 50 ohms, if these terminations are not proper, than the loss will be off between the measurement devices. The experiment with the pad may provide a clue as to what is producing this error.

Under the assumption that one of the devices, vna or sa is Bad, pad the sa or the vna with say 10 db of pad. This will force a decent match, to the tune of a 20db return loss. Then do your filter loss measurements, you better get same numbers.

Filter response is very dependent on terminations. Higher load Z than filter designed load will increase coupling between filter elements, widening bandwidth and usually increases loss. Lower load Z will decrease coupling between elements and narrows filter bandwidth. Loss and ripple may decrease if the filter design was in Chebyshev region or loss could get greater if original design was Butterworth or Gaussian. The S21 input port Z is not very good so attenuator pad will help reduce the load variation. Spectrum analyzer has similar issue on its input. Any reactance in the load will detune the filter elements.

You did not mention the filter design frequency.

I have a nanoVNA, nanoVNA-F, and N1201SA units. The N1201SA is much better performance then nano's for S11 measurement. N1201SA at 1 GHz has about 50 dB dynamic range where the nano's are lucky to get 30 dB. Number on Z readout on N1201SA is much more stable and doesn't jump around like nano's.

The main difference in the N1201SA with its ADF4351 synthesizer is providing fundamental Tx signal where the nano's using the Si5351B must rely on harmonics for Tx above 300 MHz. Both use harmonic LO downmixer with the SA612 mixer.

I looked up specs for the synthesizer chips and the jitter for the Si5351B is very poor. The ADF4351(or ADF4350) is very good. The Si5351 is between 70 and 100 ps of jitter where the ADF4351 is less then 0.5 ps. That is a dramatic difference. The Tx jitter and Rx jitter are mostly independent since they are different dividers so their jitter stacks up to make things worse.

This impacts the dynamic range. The jitter at harmonics on the nano's is likely so bad that it is 'fuzzing up' the 5 KHz downconversion output to the point that it impacts the ability to measure the signal accurately.

There may be other factors such as if and how any smoothing averaging may being done or even layout/power supply noise differences. I am going to look at the Tx signals on spectrum analyzer on both units.

Good morning Alan,
I did as you suggested,
I know where the issue is now.
I neglected to take into account the dynamic range of the instrument.
The filter is a 40 meter filter down -78 dB so I was in the noise floor.
I did a cal and tested the nano with precision attenuators from mini circuits.
-10, -20, -40, and putting them together -70.
Down to - 70 dB it reads within .25 dB.
It still would be good to know if the amplitude could be adjusted in the software.

Classic case of pilot error!

Thanks for your help
Ed

Ok, great. If you are trying to look at the pass band insertion loss response while at the same time the rejection response, dynamic range may very well get in your way. If that sort of measurement is required than one method is separate out each measurement challenge. Go for IL first and then try to eek out as low a noise floor as possible while driving the filter with as much signal as the analyzer will tolerate. Unlike other vna's this has no BW adjust or video filter tweak. Although averaging helps quite a bit.

Perhaps, the displY can wait till the actual measurements were made and
logged for display.
This might help a noise reduction due to display drive signals.