开云体育

Thanks Jim.
Still learning this neat tool.
So just the phase trace is all thats needed on s21 to measure isolation?

Are results frequency dependent?
Bryan, n0luf

Don't do "double calibration", it just forces the device to do
interpolation between points.
The note in documentation remains there for misbehaving devices
that would give out or range values when the device is used in an
uncalibrated state.

Thanks for all of the help. The plots are very enlightening. it seems like this explains this. In response to the on device calibration, is disabling it simply unchecking the apply button? And also the documentation suggests calibrating the device to full range then calibrating in the NanoVna-Saver software. Which one do you suggest?

On Thu, Jun 15, 2023 at 04:49 PM, Dragan Milivojevic wrote:

In case you were worried about "bricking" your device, don't be. It is
practically impossible to do.

If you do not have the version 0.8 as *.bin (or *.dfu) file I would propose to save it from device before you flash the new version. Dunno how to do it on windows, on linux I use "dfu-util -d 0483:df11 -a 0 -s 0x08000000 -U old_version.bin" to save (upload) the old version to the PC (128 KByte). The tool dfu-util works also for Windows, it is available e.g. from Erik's website together with the manual

Saa2n has a custom display pcb... So you must buy a "normal" 4 inch display and desolder the flatband cable (if display screen is broken)... Or you must ask the seller to send you a display (if it is not the screen but the display chip)
...
So if you screen is broken (mechanical) then buy a normal 4 inch screen and solder it to your existing pcb....
Yes its not easy.. But doable
(been there.. Did that)
73 sigi dg9bfc

Am 15.06.2023 15:12 schrieb K8HTB <k8htb@...>:

0.8 is an extremely old version.
This should work on your device:

In case you were worried about "bricking" your device, don't be. It is
practically impossible to do.

On Thu, 15 Jun 2023 at 12:04, Don Vosper via groups.io <donald.vosper=
[email protected]> wrote:

Roger,
Thanks for the link.
It doesn't look like my original Nanovna needs updating.
I assume that the updates for the later versions would not be suitable.
Don m5aky

I have an SAA 2 4 in nano VNA and I need a new screen any place available to buy a 4-in screen

De K8HTB Joe

On 6/15/23 4:38 AM, Bryan Curl wrote:

Hi Jim,
How do you do that?
[
"But it's good to check (and hey, you could use the NanoVNA to measure it)."
]
Bryan, n0luf

Measuring switch isolation.
Let's say your switch has a common, and 4 switched outputs A,B,C,D

Hook the VNA up to Ch0 to A and Ch1 to B, put a load on Common, set the switch to A, and measure the S21. Or select neither, or C, and measure S21 (which should be lower)

You could also hook the VNA Ch0 to Common, Ch1 to output A, then select A (S21 will be the loss/phase shift through the switch) then select one of the others (S21 is the common to unswitched port leakage).

If you want to be complete on this kind of thing (or small leakages perturb the system) then you may want to put termination on unused ports and see if it makes a difference.

I've done similar with antenna analyzers in the past and currently with a VNA. Switches I've used provided sufficient isolation...so far.

Hi Jim,
How do you do that?
[
"But it's good to check (and hey, you could use the NanoVNA to measure it)."
]
Bryan, n0luf

Roger,
Thanks for the link.
It doesn't look like my original Nanovna needs updating.
I assume that the updates for the later versions would not be suitable.
Don m5aky

Well guys n gals I found my issue at last!
The coax i was useing is (ok) if I'm using a tuner but horrible if running straight through. I was getting super high swr.
Today I remembered I had a string of RG8U I bought off a guy put ends on it and a ferrite bead near the antenna and a 6 wind choke back by the nano and BOOM working like a champ!
The reflected was so high I couldn't even get a decent reading! ? so at least I know what I need to do now!
Thanks EVERYONE that gave advice. I had a feeling after enuf trial n error something would give!

On 6/14/23 5:08 PM, Jim Lux wrote:

On 6/14/23 2:17 PM, Jim Lux wrote:

On 6/14/23 12:20 PM, DiSlord wrote:

I hope you disable internal NanoVNA calibration and use only NanoVNA Saver calibration
Linear Interpolation (used in internal calibration if points count or frequency range not equal) can give errors

This is a good point..

In addition, you might use a sim program to simulate an ideal line with a small discontinuity, and see what kind of wiggles you get.

Think of this - you're doing an S21 measurement - a small reflection at either end will constructively or destructively interfere with the main line signal.

On a group delay plot like this, you're looking at fractions of a degree in phase -
A reflection (or spurious signal) that is 90 degrees out of phase, and down 40dB would give you approx atan(.01) = .6 degrees.

just to illustrate how a small error can show up in group delay, I simulated what you'd get with a 2 meter cable, and adding a "triple transit" echo (i.e. a reflection at the Port 2 end, then another reflection at the Port 1 end).? So what winds up inside the Port 2 (Ch1) receiver is the sum of the direct signal and an attenuated delayed signal.
60 dB (essentially a -30dB mismatch at each end) gives you about .03 ns of ripple on 9 ns. 40dB is tenths of a ns, etc.
I ran it for 3 cases, 20dB, 40dB, and 60dB down.

and the source code for the plots:

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""

phase effects of small time delay

Created on Wed Jun 14 14:40:45 2023

@author: jimlux
"""


import numpy as np
import matplotlib.pyplot as plt

f = np.linspace(1,701,num=1220)

dbdown = 40
cablelength = 2
d1ns = cablelength * 3 / .66 # 2 meters @ 3ns/meter, 0.66 VF
d2ns = 3*d1ns # triple transit

periods = 1000./f #ns

ph1 = d1ns/periods*2*np.pi #radian

ph2 = d2ns/periods*2*np.pi

""" now convert to I/Q """

sig1 = np.cos(ph1) + 1j* np.sin(ph1)
sig2 = np.cos(ph2)+ 1j * np.sin(ph2)

""" sum them
"direct path" is down by half dbdown
"triple reflection" is down by full dbddown """

output = (1-10**(-dbdown/40)) * sig1 + 10**(-dbdown/20)* sig2 #40dB down
""" amplitude and phase """
outamp = np.abs(output)

outph = np.unwrap(np.angle(output) )
outph = outph * 180/np.pi
""" plot them """

plt.figure()
plt.plot(f,20*np.log10(outamp))
plt.xlabel('MHz')
plt.ylabel('magnitude (dB)')
plt.title(" %5.1f meters, %5.1f dB down "%(cablelength,dbdown))

""" compute group delay & plot """

difph = np.diff(outph)
diffrq = np.diff(f)
gd = difph/(360 * diffrq) *1000. #ns
gd = np.around(gd,decimals=3)
plt.figure()
plt.plot(f[:-1],gd)
plt.ylabel('ns')
plt.xlabel('MHz')
plt.title(" %5.1f meters, %5.1f dB down "%(cablelength,dbdown))

On 6/14/23 2:17 PM, Jim Lux wrote:

On 6/14/23 12:20 PM, DiSlord wrote:

I hope you disable internal NanoVNA calibration and use only NanoVNA Saver calibration
Linear Interpolation (used in internal calibration if points count or frequency range not equal) can give errors

This is a good point..
In addition, you might use a sim program to simulate an ideal line with a small discontinuity, and see what kind of wiggles you get.
Think of this - you're doing an S21 measurement - a small reflection at either end will constructively or destructively interfere with the main line signal.
On a group delay plot like this, you're looking at fractions of a degree in phase -
A reflection (or spurious signal) that is 90 degrees out of phase, and down 40dB would give you approx atan(.01) = .6 degrees.

just to illustrate how a small error can show up in group delay, I simulated what you'd get with a 2 meter cable, and adding a "triple transit" echo (i.e. a reflection at the Port 2 end, then another reflection at the Port 1 end). So what winds up inside the Port 2 (Ch1) receiver is the sum of the direct signal and an attenuated delayed signal.

60 dB (essentially a -30dB mismatch at each end) gives you about .03 ns of ripple on 9 ns. 40dB is tenths of a ns, etc.



I ran it for 3 cases, 20dB, 40dB, and 60dB down.

On 6/14/23 12:20 PM, DiSlord wrote:

I hope you disable internal NanoVNA calibration and use only NanoVNA Saver calibration
Linear Interpolation (used in internal calibration if points count or frequency range not equal) can give errors

This is a good point..

In addition, you might use a sim program to simulate an ideal line with a small discontinuity, and see what kind of wiggles you get.

Think of this - you're doing an S21 measurement - a small reflection at either end will constructively or destructively interfere with the main line signal.

On a group delay plot like this, you're looking at fractions of a degree in phase -
A reflection (or spurious signal) that is 90 degrees out of phase, and down 40dB would give you approx atan(.01) = .6 degrees.

I hope you disable internal NanoVNA calibration and use only NanoVNA Saver calibration
Linear Interpolation (used in internal calibration if points count or frequency range not equal) can give errors

On 6/14/23 9:55 AM, astech119 wrote:

Can anyone explain these wiggles. I am just measuring the group delay of a 2m cable with NanoVNA Saver using a 1MHz to 701MHz 1212pts 3average sweep. I used both a NanoVNA H4 and NanoVNA V2, but the wiggles are more pronounced in the H4. Any explanations?

Two easy explanations (assuming a perfect cable) are:
1) The 2 port cal doesn't necessarily take into account that the Ch1 (port 2) isn't perfect.
2) Your S/O/L loads for cal aren't perfect (or more properly that they have a small physical offset), so when the S21 measurement is made, there's small variations in phase from idealized expected.

One thing you might do is try an S11 measurement and process to TDR of the cable with a short or load.

That takes port 2 issues out of the picture.

Delay is delta phase (in cycles)/delta F, and for your measurement, you're measuring 700 MHz in 1200 points, so about 580 kHz/point. A delay of 9 nanoseconds would be about 0.005 cycles or 2 degrees. You're seeing maybe 0.3 ns wiggle - 1/15th of a degree. That's not huge.

Can anyone explain these wiggles. I am just measuring the group delay of a 2m cable with NanoVNA Saver using a 1MHz to 701MHz 1212pts 3average sweep. I used both a NanoVNA H4 and NanoVNA V2, but the wiggles are more pronounced in the H4. Any explanations?

On 6/14/23 7:44 AM, Dallas wrote:

I have my antenna going thru my MFJ tuner then to the common terminal of a 6 position coax switch. I leave the VNA connected to position 6. Coax 1,2,3&4 each go to different radios I have.
I can quickly switch the VNA to the tuner and adjust for a match and then switch radio to properly tuned antenna.
I’m getting ready to add a second coax switch between the tuner and antenna so I can switch between 3 antennas I have.
Having the VNA a click away is very handy.

Since you've been doing this a while, you don't have an issue, but in general, one would want to make sure the isolation between switch ports is high enough to make sure that when you run 100w to the antenna, the leakage to your VNA is <10 mW or so (my guestimate of "no damage max power" for the VNA. That's 40 dB of isolation, which most relay based switches can do.

But it's good to check (and hey, you could use the NanoVNA to measure it)..

I have a sort of similar setup using RCS-8Vs
Coax to switch and tuner at the antenna. The switch is cabled up so I can connect open, short, load to the end of the coax, as well. At the shack end, more switches to connect rig and VNA to various feedlines.

I have my antenna going thru my MFJ tuner then to the common terminal of a 6 position coax switch. I leave the VNA connected to position 6. Coax 1,2,3&4 each go to different radios I have.

I can quickly switch the VNA to the tuner and adjust for a match and then switch radio to properly tuned antenna.

I’m getting ready to add a second coax switch between the tuner and antenna so I can switch between 3 antennas I have.

Having the VNA a click away is very handy.

Dallas
N5fee