I tried to use the SA621 mixers in combination with the ADF4351 (35MHz-4.4GHz) and its clear the bandwith limitation of the SA612 will anyway block going much above 1.5GHz.
As you can see in my measurements the NanoVNA can be calibrated till 1.5GHz but noise becomes an increasing problem above 1.2Ghz and between .9GHz and 1.2GHz the board design and the unbalance in the bridge due to stray capacitance starts to have an impact.
But it is possible to build a VNA using exactly the same principles as the nanoVNA but using IAM81008 mixers and 2 ADF4351 generators. ADC and dsp where done using the line input of a PC and SW running on a PC.
Due to the single ended input of the IAM81008 mixers you have to use a very wideband balun but there are many designs available. There is even a readymade 3GHz bridge on ebay for about 10$, thats the bridge I used and works till around 2.8Ghz, above there is no directivity. S21 works nicely to 4.4GHz and I did not even test going to harmonics mode.....
So 2.4GHz is out of the question with the SA612 but I was able to tune a 2.6GHz cavity filter using the ADF4351's and IAM81008 setup. No PCB, just dead bug style build
More info can be found here:
The bridge does not have to be all 50ohm. Any setup will work but you have to design for port 1 to be 50 ohm otherwise the resistor measurements will be off. The NanoVNA bridge is very well designed. The only concern I have are the two 390/50 ohm resistors towards the SA612 input. The SA612 input is differential 3kohm so why not use that high resistance to not load the bridge? I did not test the low impedance driving of the SA612 so I can not comment.
Noise from the CPU is not important. I am using an arduino zero clone for setting up the ADF4351's and never experience any noise from the zero.
Attached a picture of all the HW for the GHz VNA (except 12 volt power supply). Its not nice but it works
|
Re: NanoVNA compared to VNWA
Clearly, the NanoVNA is crying for more data points.
One is going to have to zero in on the desired detail frequency range of the DUT with the NanoVNA
in order to see desired details. I can clearly see the series resonate point on the left plot, not so clear on the right. I have seen the same story on other devices; loop antenna, crystals, etc... More data points or just COMPRESS the span range and do the cal over that COMPRESSED range.
Regards, Alan
|
Re: New FreeRTOS-based nanoVNA 4.3" 800x480 LCD on Aliexpress
Does anyone know how to order from taobao, to ship to outside China?
Cheers
Luis, CT2FZI
toggle quoted message
Show quoted text
On Sun, Sep 8, 2019 at 04:44 AM, SM Ling wrote:
This is the taobao for the demo kit:
¡¾RF Demo Kit NanoVNA ÉäƵ²âÊÔ°å ʸÍø²âÊÔ Â˲¨Æ÷
Ë¥¼õÆ÷¡¿
|
Erik,
Interesting notes, thanks for sharing them.
A few random thoughts, not necessarily expecting answers:
We do lose a lot of dynamic range by going to harmonics
when at 300mhz and above. The Si5340 I mentioned previously
would be a significant win at those frequencies,
good for clocks up beyond a GHz on the fundamental.
Of course, the next question would be, can we use harmonics
of the Si5340 to get this thing operate up into the 2400mhz ISM band?
Will the SA612's have any meaningful response at all up there?
Is there a better mixer for those frequencies that's moderately cheap
and keeps the very nice differential input of the SA612?
What, other than the use of harmonics, is limiting performance?
Is it board layout?
Firmware?
Noise from the uC?
Perhaps we'd be better off with an all 50 ohm bridge as in fig 2 of this article:
This code might be a good starting point for programming the Si5340:
Jerry, KE7ER
toggle quoted message
Show quoted text
On Sat, Sep 7, 2019 at 11:25 PM, <erik@...> wrote: On Sat, Sep 7, 2019 at 09:02 PM, Jerry Gaffke wrote:
This could be reduced by driving CLK0 and CLK2
into the bridge and mixers, shutting down CLK1 makes isolation better.
I did that in my own build VNA that uses the same bridge with SA612 approach
and the difference is small (couple of dB).
I still intend to test with two SI5351 as they cost nothing.
The jitter etc of the SI5351 has no impact. Also noise of the SA612 has no
impact.
|
Re: NanoVNA compared to VNWA
Hi Jeff,
the VNWA had 301 points.
73, Norbert, DG1KPN
|
Re: What about fittng N conectors like this?
On Sat, Sep 7, 2019 at 06:43 PM, Dr. David Kirkby from Kirkby Microwave Ltd wrote:
.......you will never get a good calibration of N connectors directly, but you will with the addition of an adapter to female
SMA.
Could you clarify this? where would the female SMA adapter be connected? to the N-connector or to the NanoVna? Don
|
Re: NanoVNA compared to VNWA
Thanks, Norbert.
How many points are in a VNWA sweep? (I know the NanoVNA sweep is 101 points).
73,
- Jeff, k6jca
|
Hi all,
I am new to this group and like many others I could not resist in getting the NanoVNA.
Ten years ago I had built the VNWA from a SDR-kit and I am still very happy with Toms outstanding machine.
Anyway I don't need the NanoVNA but I want to have it, because I was curious to see what I would get for 44€.
A few days ago I received this tiny VNA and yesterday I did some tests in comparing the NanoVNA with the VNWA.
Due to the fact, that the VNWA can import s1p and s2p files the comparison of the two VNAs was easy.
A good candidate as a DUT is a homebrew H-field probe. Please find attached two VNWA plots, where one plot shows the s1p data from the NanoVNA.
Interpretation of the plots I will leave up to you!
73, Norbert, DG1KPN
|
Re: New FreeRTOS-based nanoVNA 4.3" 800x480 LCD on Aliexpress
Well it just landedme in a chinese page peddling fancy shoes and makeup
kits...
Den s?n 8 sep. 2019 13:44SM Ling <sm.ling11@...> skrev:
toggle quoted message
Show quoted text
This is the taobao for the demo kit:
¡¾RF Demo Kit NanoVNA ÉäƵ²âÊÔ°å ʸÍø²âÊÔ Â˲¨Æ÷ Ë¥¼õÆ÷¡¿
|
Re: Early app for the NanoVNA
Just to clear up what's happening in the app: I've changed the S21 Smith
Chart into a Polar Plot, and I'm keeping it as a default for now, because
*I* personally like that way of presenting phase and magnitude in the same
chart.
At some point, I'll make the display layout saveable, so you don't have to
get rid of it every single time you start the software. ;-)
Currently, I'm working on implementing imperfect cal standards. I've
implemented the calculations, but I don't know how to verify that I'm doing
it correctly. If any of you have any pointers, for example what kind of
results I should see when sweeping an "imperfect" open or load, please let
me know.
--
Rune / 5Q5R
toggle quoted message
Show quoted text
On Sun, 8 Sep 2019 at 13:26, Jeff Anderson <jca1955@...> wrote: Good morning, Alan,
I think we are arguing about two different things, and that we are
actually in agreement regarding both.
My point is that a Smith Chart overlay on a captured S21 sweep makes no
sense because there is no way (that I know of) to mathematically relate a
Smith Chart (and its lines of constant R, G, B, and X) to the definition of
S21. (Unlike S11, in which there is a very clear mathematical relationship
between the two).
(To Tom's point -- perhaps there is a mathematical relationship underlying
S21 that, when plotted, generates curves of the same *form* as a Smith
Chart -- but these curves won't be in terms of R, G, B, X).
Alan, I believe you are saying (please correct me) that you can place S21
data on a Smith Chart when S21 is being plotted as a function of load or
source impedance.
I completely agree.
The difference between my point and yours is the difference between a
single S21 sweep, sweeping over multiple frequencies, for a *single* load
and source impedance, versus data (for example) at a single frequency,
captured over multiple S21 sweeps, each sweep made with a different
impedance.
If this is *not* what you mean, would it be possible for you to send a
link to a site containing more material? For example, earlier you posted a
plot for an MRF914. It would be great to get more detail on exactly what
is being plotted, and why they chose to represent it on a Smith Chart.
Many thanks,
- Jeff, k6jca
|
Re: Early app for the NanoVNA
Good morning, Alan,
I think we are arguing about two different things, and that we are actually in agreement regarding both.
My point is that a Smith Chart overlay on a captured S21 sweep makes no sense because there is no way (that I know of) to mathematically relate a Smith Chart (and its lines of constant R, G, B, and X) to the definition of S21. (Unlike S11, in which there is a very clear mathematical relationship between the two).
(To Tom's point -- perhaps there is a mathematical relationship underlying S21 that, when plotted, generates curves of the same *form* as a Smith Chart -- but these curves won't be in terms of R, G, B, X).
Alan, I believe you are saying (please correct me) that you can place S21 data on a Smith Chart when S21 is being plotted as a function of load or source impedance.
I completely agree.
The difference between my point and yours is the difference between a single S21 sweep, sweeping over multiple frequencies, for a *single* load and source impedance, versus data (for example) at a single frequency, captured over multiple S21 sweeps, each sweep made with a different impedance.
If this is *not* what you mean, would it be possible for you to send a link to a site containing more material? For example, earlier you posted a plot for an MRF914. It would be great to get more detail on exactly what is being plotted, and why they chose to represent it on a Smith Chart.
Many thanks,
- Jeff, k6jca
|
Re: What about fittng N conectors like this?
Hi Peter,
I also had this idea a while ago. Have a look at post 1212.
When I made the case, I discovered that everything is so small, I could not get a wrench in to tighten the nut on the adapters.
So now I'm looking at modifying hexandflex's housing design.
You can get the adapters cheaper on eBay.
Larry
|
Re: New FreeRTOS-based nanoVNA 4.3" 800x480 LCD on Aliexpress
This is the taobao for the demo kit:
¡¾RF Demo Kit NanoVNA ÉäƵ²âÊÔ°å ʸÍø²âÊÔ Â˲¨Æ÷ Ë¥¼õÆ÷¡¿
|
Re: nanoVNA Real Resistance Measurement Range
Tom,
Thanks for your extensive measurement.
I could not help noticing the seemingly systematic deviations from perfect real impedances for the actual resistances so I downloaded the touchstone files and noticed when using a 60ps delay they nicely rotate to almost real impedance independent of the actual resistance used.
The "open" and "short" measurement deviates more but for the open that seems to be caused by not connecting a open BNC as it rotates up
Now I am no expert but it keeps me wondering where the 60ps could come from????
|
Re: New FreeRTOS-based nanoVNA 4.3" 800x480 LCD on Aliexpress
I don't know that anyone who has a NanoVNA-F has released any details on
what the communications is like.
I do know that adapting for a different kind of device is trivial in
reasonably designed software.
Rune / 5Q5R
================================
Rune,
Fingers crossed! I note that Wireshark has a USB option....
73,
David GM8ARV
--
SatSignal Software - Quality software for you
Web:
Email: david-taylor@...
Twitter: @gm8arv
|
Re: Getting started with the nanoVNA guide
I have jsut updated my post to include some of the useful feedback received here.
Many thanks
|
Re: New FreeRTOS-based nanoVNA 4.3" 800x480 LCD on Aliexpress
George,
I asked flyoob if his NanoVNA-F supports shell commands, he said tha he is working on it. He will upload the code when it's ready. Since the commands are not exactly the same as the ChibiOS version, he will publish his own PC APP.
From Usr/appvna.c, you can see the list of commands he plans to offer.
grep "static void cmd_" Usr/appvna.c
static void cmd_pause(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_resume(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_reset(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_offset(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_freq(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_power(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_time(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_dac(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_saveconfig(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_clearconfig(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_data(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_dump(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_gamma(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_scan(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_sweep(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_list(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_fatfs(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_pwm(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_beep(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_lcd(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_task(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_cal(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_save(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_recall(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_trace(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_edelay(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_marker(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_touchcal(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_touchtest(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_frequencies(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_test(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_gain(BaseSequentialStream *chp, int argc, char *argv[])
static void cmd_port(BaseSequentialStream *chp, int argc, char *argv[])
|
Re: Early app for the NanoVNA
Hi Tom,
thanks for the corrections! I've pushed an update to the repository.
I haven't put in changing the label ... I really don't like the idea of
switching between displaying "capacitance" and "inductance" as a label, as
I think it would be confusing to the user. I wish there was a term that was
reasonable to use for both. Any ideas?
I understand that "reactance" isn't the right term, so it certainly needs
to be replaced. I'd just prefer that it be with one rather than two
different words :-)
Again, thanks for your help!
--
Rune / 5Q5R
toggle quoted message
Show quoted text
On Sun, 8 Sep 2019 at 03:42, Tom VA7TA <tma.7ta@...> wrote: Greetings Rune,
Thank you so much for NanoVNA Saver - it is running well for me under
Windows 10 Pro 64 bit - very user friendly app with lots of nice features!!!
I noticed a bug with regards to the capacitance value calculation.
Attached is a screenshot of a sweep of a 1000 Ohm load. The mouse pointer
is at 126.126 MHz and the Xc reactance value is shown as -j358.226 Ohms.
The value given for the capacitor is 457 nF when actually it should be
0.0035226 nF or more commonly specified as 3.5226 pF. For the MF/HF/VHF/UHF
spectrum (300 KHz to 900 MHz of the spectrum use here) capacitance values
are more appropriately specified in pF. The formula for C in pF should be:
Cpf = (10EE12) / (2piFXc) where:
pi = 3.14159
F = frequency in Hz
Xc = reactance in Ohms
In the case of +j reactances the label "Reactance:" should be replaced
with "Inductance:".
In the case of -j reactances the label "Reactance:" should be replaced
with "Capacitance:"
The Inductance value calculation appears to be OK.
I hope this feedback is helpful and thanks again for your fine work!
--
Best Regards,
Tom, VA7TA
|
Re: New FreeRTOS-based nanoVNA 4.3" 800x480 LCD on Aliexpress
I don't know that anyone who has a NanoVNA-F has released any details on
what the communications is like.
I do know that adapting for a different kind of device is trivial in
reasonably designed software.
--
Rune / 5Q5R
toggle quoted message
Show quoted text
On Sun, 8 Sep 2019 at 00:18, George <steber@...> wrote: Hi David,
I've been thinking or ordering one too. NanoVNA-F
Can you tell me if there is any PC software available?
I haven't been able to confirm if the old NanoSharp software and its
cousins will work on this unit.
Also I'm a bit worried about the new firmware download procedure using a
U-Disk.
While the boot loader on github seems okay, I'm concerned about the
proliferation of download methods.
Your thoughts greatly appreciated.
"Perhaps foolishly, I've ordered one of these units as I liked the idea of
the larger display, so I should be in a position to test any Windows-32/64
software offered. I've not yet done a firmware update on my present
2.8-inch unit (but I've read the procedure), but all being well I should be
able to test updates on my new unit.
73,
David GM8ARV
|
Re: New FreeRTOS-based nanoVNA 4.3" 800x480 LCD on Aliexpress
From: George
Hi David,
I've been thinking or ordering one too. NanoVNA-F
Can you tell me if there is any PC software available?
I haven't been able to confirm if the old NanoSharp software and its cousins will work on this unit.
Also I'm a bit worried about the new firmware download procedure using a U-Disk.
While the boot loader on github seems okay, I'm concerned about the proliferation of download methods.
Your thoughts greatly appreciated.
================================================
George,
I don't know about PC software, that's why I was offering to test and/or help for someone to adapt the existing software.
As I said, I've not yet even updated the firmware on the basic nanoVNA, so I hope someone else can help there. The block diagram (attached) does show "USB Communication".
Cheers,
David
--
SatSignal Software - Quality software for you
Web:
Email: david-taylor@...
Twitter: @gm8arv
|
CT2FZI
David J Taylor
HexAndFlex
