开云体育

On 2/10/21 5:15 PM, Mel Farrer via groups.io wrote:

YOU missed the fact that the difference in diameter from wire to coax shield will make a difference in the inductance value........................
Mel, K6KBE

bifilar windings are essentially twisted pair. The two conductors are separated by the insulation thickness, which is "small".? So the mutual coupling is "large"

On Wednesday, February 10, 2021, 04:20:20 PM PST, Jim Lux <jim@...> wrote:
On 2/10/21 9:05 AM, David Eckhardt wrote:

? Thank you for the great summary and information about measure CMC CM
attenuation. I have a couple of follow up questions for you and/or the
group, if I may. I notice the process is for bifilar chokes. And, it looks
like you only connect one wire of the bifilar turns to the VNA. You also
said you can connect both wires if you short both ends. I am still trying
to make sense of shorting both wires and why their wouldn't be a difference
in CM impedance compared to only connecting one wire. To me, it would seem
that connecting both ends together would create a parallel path and would
not be a true measurement of impedance (i.e. due to wires in parallel,
similar to two resistors in parallel). But, I know we are dealing RF and
ferrite toroids ... my intuition can't make that jump, yet.


*? I have actually tried one vs. both wires in parallel with the VNA in
measuring CM attenuation..? There is a very minor difference.? The largest
practical effect is to reduce resistance (not so much the ±jX portion). *

*? ? ? ? ? ? Considering? each wire of the bifilar pair contributes equal
inductance by itself, so the total connected in parallel will be half that
of each wire alone.? This ignores mutual coupling.? A 3 dB difference in
30 dB of total *

the mutual L of two parallel conductors is very close to the inductance
of a single conductor. There will be essentially no reduction in the L
from paralleling.

(this is why flat strap has lower AC resistance, but very close to the
same inductance, as a round wire with the same cross section)

It would have been much more fun if you let him find that out for himself ;)

YOU missed the fact that the difference in diameter from wire to coax shield will make a difference in the inductance value........................
Mel, K6KBE

On 2/10/21 9:20 AM, Reinier Gerritsen wrote:

Op 10-2-2021 om 17:15 schreef Jim Lux:

On 2/10/21 7:36 AM, mender5@... wrote:

I made some small changes of my Nanovna-H.
Is there a way to test s11 and s21 dynamic range?

S21 - start putting attenuators in and seeing where it bottoms out (or where the displayed attenuation change doesn't match the actual attenuation change)

S11 - calibrated mismatches - attenuator into an open or short. If you put a 20 dB pad on the Tx port, can you see the difference between short, open, and load?? (20dB pad is -40dB S11, 20 dB going out, 20dB coming back from the reflection).


A couple decent step attenutors (one in 10dB steps, the other in 1 dB steps) makes this easy.? But a handful of fixed attenuators can also do it, just more time consuming as you swap them around.


Watch out for leakage from cables etc as you get to higher isolations. 80 dB is hard. >120dB is really hard.

I would do it in a different way. Calibrate the analyzer. Terminate port 2 in 50 Ohm. The noise floor is the lowest level you can measure. To get the dynamic range you would need to know the maximum signal that can be applied. Therefore you would need an RF amplifier (input to port 1). Do not overload port 2, but to find the maximum usuable signal, you must find the level where it is no longer linear. You can find this level bij adding a know attenuator in series with the amplifier output. S21 should follow the attenuation.
S11 dynamic range: calibrate the VNA and do the 50 Ohm load as the last one. Leave it connected. Apply the calibration. Make sure you tell the analyzer (or the PC software) that the load is a perfect load. Now the displayed value of s11 is the dynamic range since it sees a perfect load, perfect in the sense that it is exactly the same as the reference.
For noise floor, take the peak values and add a few dB for safety margin. For S11 a dynamic range of 30 to 40 dB is enough for all practical applications.
All values are frequency dependent.

Note that a 20 dB attenuator does *not* mean a 40 dB return loss. Most likely it is less in practice if the input is not exactly 50 Ohm (or better, exactly the same as your reference)

True, it depends on the two impedances and the design of the attenuator. But into a short or open, if the VNA is 50 ohms, and the attenuator is 50 ohms, the difference will be "small"

On 2/10/21 9:05 AM, David Eckhardt wrote:

Thank you for the great summary and information about measure CMC CM
attenuation. I have a couple of follow up questions for you and/or the
group, if I may. I notice the process is for bifilar chokes. And, it looks
like you only connect one wire of the bifilar turns to the VNA. You also
said you can connect both wires if you short both ends. I am still trying
to make sense of shorting both wires and why their wouldn't be a difference
in CM impedance compared to only connecting one wire. To me, it would seem
that connecting both ends together would create a parallel path and would
not be a true measurement of impedance (i.e. due to wires in parallel,
similar to two resistors in parallel). But, I know we are dealing RF and
ferrite toroids ... my intuition can't make that jump, yet.


* I have actually tried one vs. both wires in parallel with the VNA in
measuring CM attenuation.. There is a very minor difference. The largest
practical effect is to reduce resistance (not so much the ±jX portion). *

* Considering each wire of the bifilar pair contributes equal
inductance by itself, so the total connected in parallel will be half that
of each wire alone. This ignores mutual coupling. A 3 dB difference in
30 dB of total *

the mutual L of two parallel conductors is very close to the inductance of a single conductor. There will be essentially no reduction in the L from paralleling.

(this is why flat strap has lower AC resistance, but very close to the same inductance, as a round wire with the same cross section)

I did an S21 test on my NanoVNA-H4 using precision SMA attenuators. This is what I got for 70 dB, 80 dB and 86 dB. You can get 80 dB out to about 200 MHz.

Roger

There was a recent thread in the TinySA group in regards to "resistive vs inductive" RF taps, not sure if that's what you are referring to.
/g/tinysa/topic/79601195

On Wed, Feb 10, 2021 at 01:35 AM, Peter Ivanooff wrote:

Halloooo.... miro - Your answers are nagging. Only bla, bla, bla.

You are right! I just assumed that you are genuinely interested in this subjects, but as many others using nanoVNA not formally educated in this "branch of science", so I tried to word it in a way that even a beginner can understand.

Now I see that I have made mistake, you do have "knowledge", even have a diagram that shows correct terminology, but are just sloppy with terminology and plain rude dude.

Should have said "check your question/statement as they make no sense. Once you do it all the rest will get addressed on it's own".

Obviously my bad!

Won't even dignify your other comments that have no place in this group.

Joao CU3AA what is the duplexer model you want to adjust?

This video also helped me to make the separate adjustment of each of the TPRD-1554.

Gary, Sig and Kadir
Thanks a lot for your replies.
Already have enough information to tune up the cans, hopefully with better results than last try. Then I tuned the notch down but had no power out and never figure out the band pass measurement.
Again thanks and will let you know of the results.
73 de CU3AA, Joao

You wrote: "KEY Down on CW and 175 volts p/p"
Please behave like an adult.

Wind 5 turns on one side of the core and measure the inductance.
Wind additional 5 turns on the other side, parallel them and measure
again. Do you get half of the inductance?


* Considering each wire of the bifilar pair contributes equal

This video helped me a lot in understanding how to adjust a set of
cavities, I hope it will serve you too 73.

Greetings from cm3kma, I comment that last month I had to readjust the TPRD-1554 in the place where they are installed and the only tool I have to be able to do this readjustment was a modest VNA of the brand (Nanovna-F HW3.1 Deepelec ), which at the end is the same principle of the NANOVNA of this forum, in the attachment are the photos of the notches and the band passes corresponding to the frequency (TX: 145.210 and RX: 144.610) of this repeater that we have mounted on the hill de candela, güines, mayabeque, cuba.
If you look at the photos, the nanovna configures the range to be measured in such a way that the jumps were from 10 to 10khz, starting at 144,500 to 145,500Mhz since it is the range assigned for the repeaters in Cuba and also to coincide with the 101 points of the vna for jumps of 10 in 10KHZ.

Very important to have greater precision in measurements of this type, I recommend that you recalibrate your nanovna in the range that you are going to work with.

I hope my humble contribution will help you 73.CM3KMA

Op 10-2-2021 om 17:15 schreef Jim Lux:

On 2/10/21 7:36 AM, mender5@... wrote:

I made some small changes of my Nanovna-H.
Is there a way to test s11 and s21 dynamic range?

S21 - start putting attenuators in and seeing where it bottoms out (or where the displayed attenuation change doesn't match the actual attenuation change)

S11 - calibrated mismatches - attenuator into an open or short. If you put a 20 dB pad on the Tx port, can you see the difference between short, open, and load?? (20dB pad is -40dB S11, 20 dB going out, 20dB coming back from the reflection).


A couple decent step attenutors (one in 10dB steps, the other in 1 dB steps) makes this easy.? But a handful of fixed attenuators can also do it, just more time consuming as you swap them around.


Watch out for leakage from cables etc as you get to higher isolations. 80 dB is hard. >120dB is really hard.

I would do it in a different way. Calibrate the analyzer. Terminate port 2 in 50 Ohm. The noise floor is the lowest level you can measure. To get the dynamic range you would need to know the maximum signal that can be applied. Therefore you would need an RF amplifier (input to port 1). Do not overload port 2, but to find the maximum usuable signal, you must find the level where it is no longer linear. You can find this level bij adding a know attenuator in series with the amplifier output. S21 should follow the attenuation.
S11 dynamic range: calibrate the VNA and do the 50 Ohm load as the last one. Leave it connected. Apply the calibration. Make sure you tell the analyzer (or the PC software) that the load is a perfect load. Now the displayed value of s11 is the dynamic range since it sees a perfect load, perfect in the sense that it is exactly the same as the reference.
For noise floor, take the peak values and add a few dB for safety margin. For S11 a dynamic range of 30 to 40 dB is enough for all practical applications.
All values are frequency dependent.

Note that a 20 dB attenuator does *not* mean a 40 dB return loss. Most likely it is less in practice if the input is not exactly 50 Ohm (or better, exactly the same as your reference)

What you are measuring is DM loss through the balun. Ideally, this should
be low. The Smith Chart, S21, and S11 all look about right for that
measurement. If you wish to measure CM attenuation, connect just the braid
between the two ports. That is the path the CM energy 'sees'.

Dave - W?LEV

--
*Dave - W?LEV*
*Just Let Darwin Work*

Thank you for the great summary and information about measure CMC CM
attenuation. I have a couple of follow up questions for you and/or the
group, if I may. I notice the process is for bifilar chokes. And, it looks
like you only connect one wire of the bifilar turns to the VNA. You also
said you can connect both wires if you short both ends. I am still trying
to make sense of shorting both wires and why their wouldn't be a difference
in CM impedance compared to only connecting one wire. To me, it would seem
that connecting both ends together would create a parallel path and would
not be a true measurement of impedance (i.e. due to wires in parallel,
similar to two resistors in parallel). But, I know we are dealing RF and
ferrite toroids ... my intuition can't make that jump, yet.


* I have actually tried one vs. both wires in parallel with the VNA in
measuring CM attenuation.. There is a very minor difference. The largest
practical effect is to reduce resistance (not so much the ±jX portion). *

* Considering each wire of the bifilar pair contributes equal
inductance by itself, so the total connected in parallel will be half that
of each wire alone. This ignores mutual coupling. A 3 dB difference in
30 dB of total *
* attenuation, to me, is of little concern. The total
inductance of two inductors in parallel can be calculated from L(total) =
[L(1) X L(2)] / [L(1) + L(2)], just like resistors in parallel.*

Second question, what is the procedure for coax chokes? I *think* what I
have seen and understand is that you connect the *braid* only on CH0 and
CH1 (Port 1 and Port 2). This is the path that the CM would take. Is this
correct for coax chokes?


*You are correct. Connect the braid from one end to CH0 and the other end
to CH1. *

I have the NanoVNA and love it. Building and measuring CMCs is my next
adventure. Thanks to you and the group for a wonderful resource.

*Anything to encourage learning the NANOs and building your own
"whatevers". Experience is the best teacher!*

*Dave - W?LEV*

--
*Dave - W?LEV*
*Just Let Darwin Work*

Brian, no pix. I took it apart as I had another application for the
RG-142. In the future, I'll take pictures.

Dave - W?LEV

--
*Dave - W?LEV*
*Just Let Darwin Work*

Hi DiSlord,
I just tried to compile your H code with the updates from a few days ago and got the following errors in the ChibiOS area.
I installed ChibiOS from your current repo.

ChibiOS/os/hal/osal/rt/osal.h:241:34: note: in expansion of macro 'chDbgAssert'
#define osalDbgAssert(c, remark) chDbgAssert(c, remark)
^~~~~~~~~~~
ChibiOS/os/hal/src/hal_usb.c:932:5: note: in expansion of macro 'osalDbgAssert'
osalDbgAssert(false, "EP0 state machine error");
^~~~~~~~~~~~~
ChibiOS/os/hal/src/hal_usb.c:934:3: note: here
case USB_EP0_ERROR:
^~~~
Compiling nvic.c
Compiling hal_lld.c
ChibiOS/os/hal/ports/STM32/STM32F0xx/hal_lld.c:54:40: error: expected identifier or '(' before 'void'
static void hal_lld_backup_domain_init(void) {
^~~~
ChibiOS/os/hal/ports/STM32/STM32F0xx/hal_lld.c: In function 'hal_lld_init':
ChibiOS/os/hal/ports/STM32/STM32F0xx/hal_lld.c:235:30: error: expected expression before ')' token
hal_lld_backup_domain_init();
^
make: *** [ChibiOS/os/common/startup/ARMCMx/compilers/GCC/rules.mk:216: build/obj/hal_lld.o] Error 1

Any idea where to look?
The last code I compiled for the H was from Aug 2020 and it still compiles OK today, so did something change in ChibiOS?

Thanks,
Larry