开云体育

All good questions and point. Thanks.

I shall post my measurement plots. I did both data capture as well as photo shot of the VNA as it looks a bit better on the VNA screen as opposed to the data capture.

I did a calibration span from 453-457 kHz. So with 101 points that should be ~ 40 Hz step and should resolve the filter BW fairly well. And in fact with 2 markers on, marker 2 will read DELTA Frequency in relation to marker 1. Nice feature. And you can discern a filter BW of ~ 400 Hz.

However, there is most definite what I will call phase modulation as there are distinct sidebands along the lower and upper skirts of the filter. Is the LO phase noise sufficiently low that with (assuming) 40 Hz steps this phase modulation of the skirts should not be present? Need to go look at their LO spec. Or is the interpolation challenged? Any form of averaging would certainly help.

Alan

________________________________
From: [email protected] <[email protected]> on behalf of amirb <amir.borji@...>
Sent: Wednesday, July 10, 2019 7:28 PM
To: [email protected]
Subject: Re: [nanovna-users] Narrow band filters and LO phase noise/jitter?

I may be wrong, but:
* is nanoVNA not derived from the EU1KY antenna analyzer?

No, EU1KY is quite a bit different

* in which case, mixers basically mix down to DC (no IF)
- bandwidth limited by RC networks into ADCs with 5.5 MHz gain bandwidth (at
100 kHz)

in any case, you have a limited resolution bandwidth even if it is not apparent as a physical filter circuit.
although I think there must be a lowpass filter after the mixer (zero IF)
The question is how wide/narrow is it?

On Wed, Jul 10, 2019 at 2:57 PM amirb <amir.borji@...> wrote:

certainly they are not implementing a narrow FFT bandwidth filter in the DSP

"DSP math libraries provided by ARM, with FFT code highly optimized for speed"

I didn't get where this quote is coming from but it does not mean the NanoVNA designer has
implemented a narrow band low pass filter (resolution BW) in DSP for small spans. Maybe he can explain it...

I may be wrong, but:
* is nanoVNA not derived from the EU1KY antenna analyzer?

No, EU1KY is quite a bit different

* in which case, mixers basically mix down to DC (no IF)
- bandwidth limited by RC networks into ADCs with 5.5 MHz gain bandwidth (at
100 kHz)

in any case, you have a limited resolution bandwidth even if it is not apparent as a physical filter circuit.
although I think there must be a lowpass filter after the mixer (zero IF)
The question is how wide/narrow is it?

On Wed, Jul 10, 2019 at 2:57 PM amirb <amir.borji@...> wrote:

certainly they are not implementing a narrow FFT bandwidth filter in the DSP

"DSP math libraries provided by ARM, with FFT code highly optimized for speed"

I didn't get where this quote is coming from but it does not mean the NanoVNA designer has
implemented a narrow band low pass filter (resolution BW) in DSP for small spans. Maybe he can explain it...

He is implementing a narrow bandwidth filter in the DSP to remove artifacts of the square wave.

It is the same procedure used by the FA-VA5 and EU-1KY antenna analyzers. Otherwise the square wave would be a big problem.
But using SA612 mixers could be a problem if they are driven into nonlinearity by all the harmonics involved.
For this reason, the FA-VA5 eschewed the SA612 in favor of a more linear mixer.

Michael Knitter has an English-language presentation where he describes, in detail, how the harmonics are handled internally by the FA-VA5.

I may be wrong, but:
* is nanoVNA not derived from the EU1KY antenna analyzer?
* in which case, mixers basically mix down to DC (no IF)
- bandwidth limited by RC networks into ADCs with 5.5 MHz gain bandwidth (at 100 kHz)

On Wed, Jul 10, 2019 at 2:57 PM amirb <amir.borji@...> wrote:

certainly they are not implementing a narrow FFT bandwidth filter in the DSP

"DSP math libraries provided by ARM, with FFT code highly optimized for speed"

in this instrument with a very simple schematics (although I don't have it in front of me and vaguely remember the general picture of it) I don't remember
seeing any type of variable filter or filter bank in RF stage. It must be just a simple fixed BW filter for the whole sweep range

certainly they are not implementing a narrow FFT bandwidth filter in the DSP (as some modern spectrum analyzers do for <10Hz RBW)
therefore, I am sure there is a fixed RBW filter in IF stage. if you have the schematics you should be able to figure it out.

Excellent point. Is there a tag between the analyzer sweep range and its resolution BW? I should do a cal over a reasonable narrow frequency range and see if this helps the situation. Otherwise it may very well be RBW limited.

________________________________
From: [email protected] <[email protected]> on behalf of amirb <amir.borji@...>
Sent: Wednesday, July 10, 2019 4:15 PM
To: [email protected]
Subject: Re: [nanovna-users] Narrow band filters and LO phase noise/jitter?

it's not that surprising, I think, considering that the RBW (or IF BW) of this instrument is definitely not
narrow enough to give you the resolution required for like a 1KHz or perhaps even 5KHz span. of course number of sample points is another bottleneck
but not as critical as the RBW.
Does anybody know what the RBW actually is?

it's not that surprising, I think, considering that the RBW (or IF BW) of this instrument is definitely not
narrow enough to give you the resolution required for like a 1KHz or perhaps even 5KHz span. of course number of sample points is another bottleneck
but not as critical as the RBW.
Does anybody know what the RBW actually is?

So ... I have played a little more ... Thanks to Larry for listing the serial commands ... I can now command my nanoVNA from ubnuntu using them .. and after a play and putting putty on the same virutal drive in PlayOnLinux have been able to confimr that simply linking the comport through works ..

ln -s /dev/ttyACM0 ~/PlayOnLinux\'s\ virtual\ drives/'NanoVNA'/dosdevices/com*

where * is whatever com port you wish to use and NanoVNA is the name of your Virtual Drive.

Putty will connect and work :) however NanoVNA Sharp still refuses to connect :(

More fun ... if anyone whishes me to try other software yell ... also getting POL to run the nanovnasharp program was a little fun ... when setting up the virtual drive you need to remove mono completely and then add the dotnet40 or above component .. if the mono compoent is still installed they clash and crash :(

Cheers

Tim M0THM

I started looking at some narrow band mechanical filters.
The measurement was challenged to look at a 500 Hz wide 455 kHz filter.
I will double check if the limits are 101 data points/interpolation or simply LO jitter.
It was very difficult to discern the transition skirt band.

However, other somewhat wider mechanical filters worked out well as shown here.
A 88 kHz sideband filter and a classic 455 kHz filter from a Collins. These were both
headed to the land fill. Such a shame. Here old meets new! Alan

I use an old blackberry 800mA charger.
Grab the datasheet for the IP5303 charger chip that's used for more info.
(Chinese)
1.2A Charging
1.0A Discharge

Thanks, I was wondering about the charger . It will at least charge a 800ma
Thanks

Frank - do a search though the forum and you'll find all kinds of useful info that others have already supplied...

Have a look at my posts & photos - I stuck an old Motorola 800mAH cellphone batt on mine - no probs.

...Larry

Does anyone know how big a battery in mAh (charging wise) the unit can handle?
Frank

Is anyone interested in an extended back cover for the NanoVNA? I designed and printed a sample it has a 6mm deep "pocket" that is 71mm X 40mm. Should be big enough to double sided tape a bigger battery in it

1

2

3

Frank
ka2fwc@...

I just ordered 1 from Banggood preorder, I hope this is the correct one to make sure Huegen gets paid for his work

Several discussions with my ham colleague Gary, N3GO and wrestling in a quite area,
I believe I have ONE approach that works well and is consistent with proper results.
The main items to note: Only TWO traces are present during the cal, LESS CONFUSION!
The frequency selection I chose, start and stop, is just an example.
The use of the RESET soft touch is crucial. And I should update message #83. I believe that RESET
eliminates the BOGUS firmware process that takes you to the DONE soft key premature!!
===================================================================
Calibration for 4 track nanoVNA (VNA)

Some notes: If you can have on hand another 50 ohm termination
and another female-female SMA adapter, it is helpful.

Supplied and used are the TWO- matched cables, the LOADS and one SMA adapter.
Each of the NUMBERED items below represent your PRESS (SOFT KEY TOUCH).

Turn on the VNA and note register 0 is ON.
CONNECT THE EQUAL LENGTH MALE-to-MALE COAX CABLES TO CH0 AND CH1.

1. DISPLAY
2. TRACE 0
3. SINGLE
4. TRACE 1

Now we are displaying S11 and S22 ONLY both as LOGMAG.

5. STIMULUS
6. START 100 KHZ
7. STOP 500 MHZ
8. BACK
9. CAL
10.RESET
11.CALIBRATE

Note! None of the softkeys on the right side touch screen are highlighted.
Place an OPEN on end of CH0 cable and this will be followed by a SHORT.

12.OPEN
13.SHORT

Place a 50 ohm LOAD(s) on the end of cables for CH0 and CH1.

14.LOAD
15.ISOLN

Connect the 2-cables together using the female-female SMA adapter.

16.THRU
17.DONE

Finally save to a desired register.

That's it. Check and be sure that the open, short and 50 ohm are on the proper
side of the chart. Note, the cables are CONNECTED to CH0 and CH1 and the calibration is to the end of the cables. The reference plane for insertion loss measure magnitude is in place of the adapter. However, the
phase is subject to error and that needs addressed in another message.

Also, with a thru cable connection between CH0 and CH1, the
loss measure, S21 should be nearly 0 dB.

Alan W4AMV

You know, of course, that the signal out is a square wave.
According to the schematic, the Si5351 feeds the output directly through the resistor bridge..

I believe that's exactly how they have extended the range of this instrument to 900MHz: by using the harmonics.
otherwise Si5351 is only rated up to 200-300MHz. This is the same principle used in EU1KY antenna analyzer

I was interested in the performance of the output signal. So at first I measured the output power:
10 MHz -12,13 dBm
50 MHz -12,57 dBm
100 MHz -13,21 dBm
200 MHz -14,85 dBm
300 MHz -7,92 dBm
400 MHz -8,07 dBm
500 MHz -8,25 dBm
600 MHz -8,90 dBm
700 MHz -8,57 dBm
800 MHz -8,76 dBm

Frequency Offset:
Set to 100 MHz measurement was 99,999 908 MHz which is an offset of 92 Hz. Excellent !!!

Output Spectrum:
Spectrum is fine up to 270 MHz with harmonics at least 20dB down
However above this frequency there is a strong subharmonic or spurious 12dB higher than the wanted signal. See picture enclosed. I have no idea how this may influence our measurements. In the pictures some sent to the group I can not see negative results.
This probably explains why output power is higer at 300MHz and above.

Ernst

This is a very interesting group. I am not a radio expert or hobbyist. I am using vector network analysers to make biological and chemical sensors from piezoelectric resonators ranging from 200MHz surface acoustic wave devices, to 4GHz bulk acoustic wave devices. My devices of course are equivalent LCR circuit units, so I use the Butterworth van Dyke model to try to make sense of my resonators outputs.



Thank you for showing me where the software was to run the nanoVNA. I also have a Pocket VNA and a HP8753D. I am very pleased with the functionality of the nanoVNA.



Sent from Mail<> for Windows 10



________________________________
From: [email protected] <[email protected]> on behalf of alan victor via Groups.Io <avictor73@...>
Sent: Tuesday, July 9, 2019 7:59:47 PM
To: [email protected]
Subject: [nanovna-users] Higher frequency dynamic range... UHF Helical

This is the highest filter I had from the junk box without exceeding the limits of the instrument. The next filter up was centered at 902 MHz. So trying to get a sense of measurement DR would be a stretch.

This is a 6 pole UHF helical filter centered at ~ 490 MHz after re-tuning.
This noise floor of the response is HEAVILY dependent on the dressing of the cables the vendor supplies.
The cal routine I used included ISOLN as required, however, both CH0 and CH1 ports are
terminated with 50 ohm. So at UHF, if you average the noise in your head, about 70-75 dB DR.

Despite getting consistent cals with the 2 track VNA (the white gecko) or antenna analyzer, I am not certain the convoluted cal routine of the 4 track or black VNA is IDENTICAL. I would think it would be, however, some peculiar cal results do occur. Incidentally, the black unit has a firmware boot message, GEN11.TAOBAO.COM. Is this the latest firmware? I am going to guess maybe not.