NanoVNA V2 / Forum

Note: this page is a mirror of https://groups.io/g/NanoVNAV2.
Click here to join and see most recent posts.

V2plus4 Problems in Low-Frequency Narrowband Measurement


Tenko46T 2020/11/28 04:22

Hello, everyone

Yesterday I received NanoVNA V2plus4 from DHL which I was looking forward to. Unfortunately, when I measured the 10.7 MHz monolithic crystal filter S21, there was an abnormality in the waveform, so I would like to report it. Attach the same filter results as measured with V2_2 and V2plus4. The waveform of V2plus4 is jagged and not smooth. This does not happen with V2_2, so I think it is the problem of the V2plus4 which I purchased. This phenomenon appears to be frequency dependent and did not occur with the 20 MHz crystal S21. I tried making and replacing the latest version of the git -20201124 firmware, but it was still the same. Waveform errors are the same for stand-alone operation or controlled by NanoVNA-App v 1.1.160 via USB.
I hope this problem will be resolved.
--
TENKO

OwO 2020/11/28 22:31

It's a known issue: https://nanorfe.com/nanovna-v2-faq.html

"the V2 is not suited to very narrowband measurements (coherent
bandwidth <1kHz) because of the fast switching of the transmit and
receive paths. In practice this only becomes a problem when measuring
crystals (not SAW filters)."

amirb 2020/11/28 06:51

his measurement span is about 80KHz (or was it 100KHz?) so it is not <1KHz

however, if by "Coherent BW" you mean Resolution BW (IF filter), then the usefulness of this device is very limited if it cannot properly work for RBW <1KHz

i'd rather have a slower sweep than not being able to measure in a 100KHz span

amirb 2020/11/28 06:51

[Edited Message Follows]

his measurement span was 100KHz so it is not <1KHz

however, if by "Coherent BW" you mean Resolution BW (IF filter), then the usefulness of this device is very limited if it cannot properly work for RBW <1KHz

i'd rather have a slower sweep than not being able to measure in a 100KHz span

Stephen Laurence 2020/11/28 07:35

Well, an opportunity to clear some confusion in my mind, please.

The resolution bandwidth is not controllable on the official v2 series firmwares, but the excellent version from Japan had an option of resolution bandwidth down to 100hz as far as I can remember.  When I compared the two firmwares, the effect of small rbw on noise floor and trace jitter was similar to increasing the averaging (which is also controllable on the Obijan firmware).

Does the Obijan firmware jitter or sweep the actual baseband frequency about (using fft I presume) or is it not real? Can you measure crystals with the Obijan software, because if it is possible, I will leave one of my v2 devices unmodified and on the Japanese firmware.

It would be a real pain to have to get out my noise generator, homemade bridge and fire up a Boatanchor spectrum anayser if I have to tweak narrow band crystal filters ever again, but at least I would be forwarned.

Steve L. G7PSZ

Tenko46T 2020/11/28 15:07

Thank you for your reply. ​I am very happy to hear from the developers directly.

​It was a known problem in V2. ​I'm sorry to bother you.

​However, I would like to tell you that the V2_2 I purchased in August did not have this problem.

​I'm sorry to be asking but this may be a problem specific to my V2 Plus 4. ​For example, DDS may not be in time for frequency hopping, or the jitter of the reference oscillation may be very bad.

​If it is not a specific problem, is it because of the circuit constant change made in the sweep speed and the AGC acceleration which are one of the differences between V2 Plus4 and V2_2? ​For example, when the gain is switched, charge is injected from the gate of MOS transistor, and the waveform of IF may change.
--
TENKO

Tenko46T 2020/11/28 15:24

​Thank you for your reply.

​I know that the Ojisan firmware uses CIC filters to implement RBW variably.
​I understand the importance of narrowing the RBW for narrowband measurements like crystal filters, but the Ojisan firmware does not seem to support V2plus 4 yet, so I am sorry but I cannot implement and verify it.

​I am also hesitant about the high-speed sweep modification of V2_2. ​Until this problem is resolved, it is likely that Ⅰ will be using two different V2's depending on the purpose.
--
TENKO

Tenko46T 2020/11/28 17:28

In response to your request, I installed Ojisan firmware 20201004 on my V2_2.
​Attached is the result of X'tal filter with F0 = 10.6935 MHz BW = 2.2 kHz measured with V2_2 set to RBW = 100 Hz, SPAN = 10kHz.

Since this graph does not convert the reference impedance from 50 ohms, the waveform is not correct. However, if the reference impedance is converted to 500 ohms, the insertion loss is reduced to about 4 dB, and the passband is slightly flatter.

​Then, changing RBW = 1 kHz and AVG = x10 instead produced a similar waveform.
​I hope this helps you.
--
TENKO

OwO 2020/11/29 11:34

Measuring crystals isn't supported on any V2, even if the curve looks
smooth it's still wrong. I recommend using a V1 NanoVNA to measure
crystals. I will try to fix this in a later firmware update that allows
you to select the wait time and integration time (and thus VBW).

Tenko46T 2020/11/28 20:59

​Thank you for a positive answer. ​I am looking forward to the firmware update. ​
​As you said, I understand that even if it looks smooth, it cannot be measured correctly.

​It is surprising that V2 implements VNA with only one AD converter and the interrupt processing on GD32 can sample R, A and B at the correct timing.
​Moreover, if the V2plus4 is changed IF frequency higher than V2_2, I imagine that the phase change for timing error is large and severe.

​Gabu-chan, Thank you very much this time.
--
TENKO

Stephen Laurence 2020/11/29 00:35

And thank you both for partially answering my question(s). I am still slightly muddled, but I will proceed to upgrade one of my v2 machines (V2N, my favourite) to plus, as it is reversible, with a soldering iron and firmware restoring.

I also have an original v1-H which I can turn out if needed. At present, it is the loan machine, but I will promote it to “first backup”.

Steve L

John Gord 2020/11/29 16:41

I have obtained good results on my -V2 (consistent with my 8753D) on crystals using the osijan firmware.  For a 1kHz wide sweep at 26MHz I used either RBW 10 Hz, AVG 1 or RBW 1kHz, AVG 50.  Either setting gives a slow sweep, but that is what is needed for a narrow resonator.

--John Gord

Tenko46T 2020/11/29 17:31

​Wow, really!
​The V2 Ojisan firmware works with the same RBW as the 8753 D. ​Your V2 is great. ​Is my V2 okay?
--
TENKO

John Gord 2020/11/29 21:18

TENKO,

As far as I can tell, the Ojisan firmware works on the 2.8" and 4" display versions of the NanoVNA-V2, but probably not the V2plus or V2plus4.  There are different files for the 2.8" and 4" versions, of course.
He also has a version for the original NanoVNA-H, but because of limited memory it does not have all the features.

--John Gord

On Sun, Nov 29, 2020 at 05:31 PM, Tenko46T wrote:

Tenko46T 2020/11/30 00:48

Thank you. John,

I have both V2_2(2.8") and V2plus4, and at the moment V2plus4 does not support narrow band measurement, but if I put Ojisan firmware in the V2_2 and set RBW to small, I may get the same measurement result as 8753 D.

​This is a good news. ​In fact, I used to measure crystal filters using HP8640B's FM external modulation and HP8754, but they were both old and heavy, and I disposed of them because I thought my workbench would be in danger in case of a big earthquake.
--
TENKO

Stephen Laurence 2020/11/30 01:06

All the above is well worth knowing.

I will keep at least one of my (soon to become antique) v2 devices on Ojisan software. I can get passable results up to 4ghz on two of my v2 and can increase the dynamic range somewhat by attaching little ferrite plates around the circuit board. They are various sizes 1-2cm and 2mm thick but I can buy 40 of them on Ebay for £8 - much less than a single Ferrite emi sheet.

I have used the through stripline on the block of four attenuators (£5 Ebay) to assess the attenuation of various materials connecting it between port 1 and 2 with a 0-4gkz sweep and logmag on 2db per division. Attenuation over a stripline is a measurement some of the emi manufacturers use.

When placed over the 2cm of stripline, a ferrite plate attenuates the signal zero at 1mhz increasing to 7db at 4ghz. This seems to be better attenuation than the emi manufacturers’ results of their most expensive emi absorbers which cost more than a V4 for one sheet.

Steve L

Thorsten DL9SEC 2020/11/30 03:14

Hi,

could you please clarify, what the "Ojisan" firmware is?

Tnx.

73, Thorste DL9SEC

amirb 2020/11/30 04:31

same question here

does it work on SAA-2N from deepelec?

Tenko46T 2020/11/30 06:01

Click the link below to download Ojisan firmware to your PC immediately.
​If you have trouble downloading, delete the "? dl = 1" at the end of the link.

​Firmware for 4-inch LCD (ST 7796) NanoVNA-V2.
https://www.dropbox.com/s/vwomqt986wkel2b/binary_st7796_20201004.bin?dl=1

​Firmware for 2.8 inch LCD (ILI 9341) NanoVNA-V2.
https://www.dropbox.com/s/g648ut5fqkajq4q/binary20201004.bin?dl=1

​I remember that Ojisan firmware was mentioned in another Groups.io before. ​Please refer to this.
https://groups.io/g/NanoVNAV2/topic/firmware_from_ojisankoubou_in/76854966?p=,,,20,0,0,0::recentpostdate%2Fsticky,,,20,2,0,76854966

--
TENKO

amirb 2020/11/30 06:20

does this work on SAA-2N?

thanks

Stephen Laurence 2020/11/30 06:24

Yes. Very well. However I have removed it as I will be converting to v2plus soon and it does not work on v2plus unfortunately.

I still have it on several of my others, however. It is easy to load and change firmwares, anyway, especially if you keep the various firmwares archived on your computer.

Steve L

hwalker 2020/11/30 06:49

On Mon, Nov 30, 2020 at 06:20 AM, amirb wrote:

>
> does this work on SAA-2N?
>
>

If your SAA-2N has a floating point unit then it will work.  Some early production units may not.  Check Config->Version and "Has FPU:" must be marked "yes".  There may also be a couple of bugs in the firmware that you may need to work around.  S21 measurements may have an ~1.5 dB offset due to the BBGAIN 3 value in the SAA-2N being -18.5 instead of ~-20.0.  Long marker drags using a stylus may also cause the SAA-2N to lock-up and require re-booting.

The author has also said that he no longer intends to support the firmware so no future improvement to the firmware maybe forthcoming unless he changes his mind.  One feature that will soon become available on other firmware is the ability to take screen shots, which is great for documenting measurements instead of having to take photos.

- Herb

amirb 2020/11/30 06:57

On Mon, Nov 30, 2020 at 09:49 AM, hwalker wrote:

>
> On Mon, Nov 30, 2020 at 06:20 AM, amirb wrote:
>
>> does this work on SAA-2N?
>>
>>
>
> If your SAA-2N has a floating point unit then it will work.  Some early
> production units may not.  Check Config->Version and "Has FPU:" must be
> marked "yes".  There may also be a couple of bugs in the firmware that you
> may need to work around.  S21 measurements may have an ~1.5 dB offset due
> to the BBGAIN 3 value in the SAA-2N being -18.5 instead of ~-20.0.  Long
> marker drags using a stylus may also cause the SAA-2N to lock-up and
> require re-booting.
>
> The author has also said that he no longer intends to support the firmware
> so no future improvement to the firmware maybe forthcoming unless he
> changes his mind.  One feature that will soon become available on other
> firmware is the ability to take screen shots, which is great for
> documenting measurements instead of having to take photos.
>
> - Herb
>

thanks for the valuable info
mine does have FPU but those bugs dont sound good

hwalker 2020/11/30 07:14

On Mon, Nov 30, 2020 at 06:57 AM, amirb wrote:

>
> thanks for the valuable info
> mine does have FPU but those bugs dont sound good

Not all SAA-V2N's experience the bugs.  As Stephen said, it is easy to change firmware, so if you don't notice the bugs you are good to go.  If you do, then it is easy enough to change back to your previous firmware if you decide the bugs are too annoying and outweigh the other nice firmware features.

If you do load the firmware, check Config->DMSEG and if BBGAIN 3 value is ~-18.5dB then you will most like have the 1.5dB S21 offset error when measuring attenuation values greater than 10dB.  The long marker drag lockup bug can be worked around by using the buttons instead of the stylus to move the markers.

- Herb

- Herb

Siegfried Jackstien 2020/11/30 17:44

set adf4350 power down to 0 or 1 ... save in slot 0 .. switch off and on
again ... now recheck bb3 gain ... should be very close to 20 db

thats what i found out here ... a bit lower clock power and the
calibration at startup works better

on high power i have seen 19.75 or 19.85 ... but with lower power i see
more or less exact 20 db (19.96 to 19.98)

dg9bfc sigi

Am 30.11.2020 um 15:14 schrieb hwalker:

DiSlord 2020/11/30 10:21

Most simple way for solve BBGAIN3 issue (if BBGAIN3 ~ -18.5, it come from bbgain0 overflow due to strong output power) is set STIMULUS->GFG SWEEP->ADF 4350 TX POWER set to 0 and save to 0 slot

After device powerup it should load this setting and correct calibrate BBGAIN.

After you can increase ADF 4350 TX POWER to 3

PS i suggest autoset ADF 4350 TX POWER to 0 on bbgain calibration, but at this moment this fix not added....

hwalker 2020/11/30 12:01

On Mon, Nov 30, 2020 at 07:14 AM, hwalker wrote:

...  If you do load the "Ojisan" firmware, check Config->DMSEG and if BBGAIN 3 value is ~-18.5dB then you will most like have the 1.5dB S21 offset error when measuring attenuation values greater than 10dB.  The long marker drag lockup bug can be worked around by using the buttons instead of the stylus to move the markers.
========================================

It's been a while since I tried the "Ojisan" firmware and couldn't remember if it allowed setting the the ADF4350 power level.  If it does, and your BBGAIN 3 value after upgrading is ~-18.5dB then you should be able to correct the S21 offset error by following the suggestions of Sigi and DiSlord.

- Herb

Stephen Laurence 2020/11/30 12:56

On Mon, Nov 30, 2020 at 12:31 PM, amirb wrote:

Stephen Laurence 2020/11/30 12:57

Yes, as long as you chose the one for the 4” screen, otherwise the screen js white. If that happens, just load the other one.

Steve L

Tenko46T 2020/12/06 00:24

Hi Steve,

Thank you for your advice on wave absorbers.

In the V2Plus4, the ferrite sheet as OwO said was used.
https://groups.io/g/NanoVNAV2/message/294?p=,,,20,0,0,0::relevance,,material,20,2,20,76323797
I found the ferrite sheet in the same picture is sold at the following site. It seems that I can buy it at a relatively reasonable price.
https://www.chinahao.com/product/4033059171/

Nicolson-Ross-Weir method is well known for its complicated permeability, but it is difficult because it requires the attachment of a coaxial line.
I think it is practical to use the microstrip line as you say.

However, I was surprised that OwO put a ferrite sheet in a microwave oven and examined its performance.
I felt the scales come off my eyes.

--
TENKO

Siegfried Jackstien 2020/12/06 11:49

put it in microwave ... if it heats up in a second or two its lossy at
13cm (put a cup of water also in the oven that it has some load)

and you need a material that is lossy in that area

the microwave oven maybe is not the best test ... the stripline method
is not much better

best maybe would be a cavity test (two probes in a cavity and measure
with and without sheets on the walls) cause we wanna surpress
reflections from the walls

greetz sigi dg9bfc

Am 06.12.2020 um 08:24 schrieb Tenko46T:

Tenko46T 2020/12/06 04:58

Cavity tests include reflection effects and require careful analysis to obtain correct results. It is important to convert the energy of electromagnetic waves into heat, not reflection.

The complex term of permeability contributes to the loss. I thought the microwave oven method was wonderful because it was full of amateur spirit to select such materials quickly.

By the way, water is not a magnetic material but a dielectric material. However, the complex term of the dielectric constant reaches its maximum at about 20 GHz, and even at 2.4 GHz, there is a loss, so it is suitable for cooking because the electromagnetic wave penetrates inside moderately and generates heat.
--
TENKO

Siegfried Jackstien 2020/12/06 13:18

Am 06.12.2020 um 12:58 schrieb Tenko46T:
> Cavity tests include reflection effects and require careful analysis
> to obtain correct results. It is important to convert the energy of
> electromagnetic waves into heat, not reflection.
yepp ... if signal goes down the wave is absorbed in the cavity ... to
compare different meterials .. set 0 db without sheet and then add them
inside ... lowest signal is best material to use
>
> The complex term of permeability contributes to the loss. I thought
> the microwave oven method was wonderful because it was full of amateur
> spirit to select such materials quickly.
cooking sheets is a fast test yes ... to compare ... heat all sheets at
once in the oven ... hottest sheet after a few seconds of cooking is
best :-)
> By the way, water is not a magnetic material but a dielectric
> material. However, the complex term of the dielectric constant reaches
> its maximum at about 20 GHz, and even at 2.4 GHz, there is a loss, so
> it is suitable for cooking because the electromagnetic wave penetrates
> inside moderately and generates heat.
and that is why you should not dry your wet pet (dog, cat, hamster) in
your microwave ... it penetrates to deep ... grin

To reply to this topic, join https://groups.io/g/NanoVNAV2

View this thread on groups.io