开云体育

I tried to repeat Roger's test of a 'Hi Q' 470pF cap with my nanoVNA and managed to see an indicated Q that was very noisy (as expected) below 10MHz. At 10MHz it was probably averaging a indicated Q of 2000 with the nanovna. By 20MHz the indicated Q dipped below 800 and the trace was still very noisy but it was easier to spot the average.

Manfred's plots of the 2000pF 'Hi Q' capacitor (on the EFHW transformer thread) obviously had a problem somewhere with the test setup or maybe his 2000pF capacitor was faulty? I'd expect the nanoVNA to perform better than that across the HF bands if given a fighting chance with a decent 2000pF test capacitor. Obviously, the nanoVNA can't measure the Q of the 2000pF cap down at lower frequencies but it should perform reasonably well across the HF bands for example.

In Roger's case, a decent 470pF cap is probably best tested across 5MHz to 20MHz. Below this the nanoVNA will increasingly limit the measurement and it becomes meaningless to even try measuring the Q of the cap much below 5MHz unless it is not a high Q cap.

Thanks for the update. I appreciate the effort.

Looking at the notes you have a fix in the H4 version "-For H4 now allow 7 save slots H - fix tupo on average (now noise floor more better)" Do you recommend upgrading my H4 version from 1.0.45?

Mike N2MS

If the transmission line leading to the load and the load are good 50Ohms the length of the line is irrelevant.

It would be interesting to see what type of 470pF cap Roger was testing. The plot shows a Q that collapses to 200 by 1.3MHz. I'd expect the nanoVNA to be able to indicate a much higher Q than 200 at 1.3MHz for a decent 470pF cap. So I'm wondering what is limiting the Q to just 200 at 1.3MHz. Is it the 470pF cap or something else? Up across the HF bands I think the nanoVNA can probably measure the Q of some types of 470pF cap but probably not a fairly decent ceramic cap.

On Sun, Mar 21, 2021 at 02:12 PM, Carl-Mikael Zetterling wrote:

Very useful discussions around this demo kit.

Hello Carl,
I have also some documentation around the RF Demo Kit board:
/g/nanovna-users/wiki/16592

73, Rudi DL5FA

Look for the Boonton 34A resonant line coupler. A reasonable approach of course is build your own. A target unloaded Q that would be of value is 500 to a 1000 at some target freq... say 400 MHz. Fun!

Quite some time ago I was qualifying ceramic caps for high power PA designs. The vna measurement just can't quite make it. The approach that did work and correlated well with practice in design was to build a cavity resonator and embed the cap under test in series with the Tline resonator. Then measure the BW and center frequency and hence the unloaded Q. Eventually that became one of the standard techniques used in the industry and I believe Boonton provided such a resonator for cap Q testing. If I can find the design note I will post.

I think it's a bit optimistic to hope that a VNA could measure that component with a basic s11 measurement. I did a few quick sums and at 100kHz, a 485pF capacitor with ESR 0.58R, -3282R reactance, (Q of 5655) will have a reflection coefficient of something like mag 0.9999946. I did some tests on my nanoVNA a couple of days ago with a test load consisting of an ATC 100B 39pF porcelain cap in series with a 1R 1% 0805 SMD resistor.

The aim was to see how well it could measure the Q of this network across the 2-30MHz HF band. This cap will have a similar reactance magnitude as a typical handwound toroid using either Micrometals 2(red) or 6(yellow) material and this is why I chose these component values to put in series. Things generally get difficult for a modern VNA when the magnitude of the reflection coefficient creeps above about 0.998.

The plot below shows the theoretical Q curve in green for 1.02R in series with 39pF. The blue trace is for the Q measured with the nanoVNA and the Q was also measured with a lab VNA and this is the brown trace. The nanoVNA does OK until the Q goes above about 200 and then it gets very noisy. This is a plot taken with no averaging on the nanoVNA. I've improved my cal kit definitions since I took that plot so I'll try and find the RC network and measure it again. It isn't a perfect yardstick because the ESR of the 1R resistor will tend to creep up slightly with increasing frequency and the ATC cap probably has an ESR of 0.02R across the HF bands. However, I think it is a useful test.

In another topic the limitations of the NanoVNA when measuring the Q of inductors and capacitors was discussed by Manfred Mornhinweg.

/g/nanovna-users/message/21404

This has been my experience as well. For this test I measured a capacitor from my parts bin. It was labelled as Hi-Q 470 pF. and was measured on a DE-5000 LCR meter to determine its characteristics at 100 kHz. It was 485 pF with a Q of 3281, ESR of 0.58 ohms and reactance of 3282 ohms
.
Next a small SMA test jig was used with a NanoVNA-H4 and the OneOfEleven NanoVNA app. The NanoVNA app has calibration averaging and you can also average successive sweeps to reduce "noise" in the measurements. From the attached screenshot you can see that the NanoVNA capacitance measurement was close at 490 pF which is quite good considering the measured reactance was 3227 ohms. However the ESR was way off, just like Manfred experienced in his measurements.



So the pitfall here is to not expect high Q measurements when using the NanoVNA.

Roger

On Sun, Mar 21, 2021 at 10:13 AM, <reuterr@...> wrote:

Roger Need has made a comparion of two different RF Demo kit PCB's.
See: /g/nanovna-users/message/20702
and the attached photo: RD-Demo-Kit_PCB_comparison.png

Thank you for pointing out the other thread. I haven't seen it yet. I agree that the board on the left has a better layout - all three standards are in the same position.

I have the variant of the board shown on the right. Here "short" and "open" are directly at the u.fl connector while "load" is on the end of approx. 2mm of a microstrip line. This adds some additional phase shift to the load (around 30 deg. at 3 GHz), but I don't think this matters much for calibration. If there is approx. 0 reflection off load, a phase shifted 0 is still 0.

The offset of the "load" might matter if I could better characterize the load resistor (e.g. describe it with inductance + resistance, like in the Agilent white paper I cite in my blog post). However so far I haven't found a method to do that that would work within the limitations of my equipment.

Also, I should stress that my 0.6 pF estimate of the stray/fringing capacitance for "open" depends only on the measurements of "short" and "open". These two seem to have pretty well matched lengths on my board (although I can't say how the PCB looks like underneath the u.fl connector)

I might order the other kind of board as well. It would be interesting to see how it compares. However I would still expect the other board to have some stray capacitance on the "open". Even professional calkits have that - the difference is that they usually come with the capacitance value specified (the C0 parameter)

Best regards
Toma?

I think found problem (it exist on H version only)
Try last loaded firmware version v1.0.50

/g/nanovna-users/files/Dislord%27s%20Nanovna%20-H%20Firmware

On Sun, Mar 21, 2021 at 08:24 AM, n2msqrp wrote:

By any chance does it work with the NanoVna-F. When I start my NanoVna-F it
does detect the device version 0.2.1 but the seems to time out.

NanoVNA app version is 1-1-1-193.

No it does not work with the NanoVNA-F. The firmware in the -F does not have the "new " data transfer protocol.

Roger

Hi Manfred,

I'll add the GR 1606A to WB2UAQ's list -- Dick Benson, W1QG, who wrote the MATLAB app mentioned in my previous post, has used his to measure capacitance Q in the thousands range (e.g. 8000). If you want to do this over a frequency range, I think you would first need to rebalance the bridge at each new frequency before measuring the capacitor, so it could be time-consuming.

One can make Q measurements of capacitors using S11 over a frequency range, but you have to be very careful. The HP 4191A does this -- it is essentially an S11 measuring device, but the maximum Q it will display is 1000, and this is with an *ovenized* directional coupler. But it does have a fixture that lets you measure SMD caps.

Best regards,

- Jeff, k6jca

P.S. The GR-821 is a pretty cool device. I had one, but ultimately sold it because it was so unwieldy to use (and large, too). But I was very impressed when I removed the unit from its case and saw that the case itself was completely copper lined.

Only thing possibly missing is the Smith Chart. Looks good to me as well.
Have a look through the Wiki for starters.

Dave - W?LEV

On Sun, Mar 21, 2021 at 2:18 AM Tony Michel <tonyfromjupiter@...>
wrote:

looks OK to me

On Sun, Mar 21, 2021 at 12:52 AM <aleks07111971@...> wrote:

What is wrong with the device?

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

By any chance does it work with the NanoVna-F. When I start my NanoVna-F it does detect the device version 0.2.1 but the seems to time out.

NanoVNA app version is 1-1-1-193.

Mike N2MS

I had a play yesterday with the newer firmware versions and the 111 VNA App. Overall I wasn't that impressed and went back to my own PC tools and the earlier firmware. With my PC tools I can correct my SMA cal kit and I can also collect true s2p files from the DUT. With my cal kit corrections I'm now exploring above 50MHz as I'm getting really good results when compared to a decent Agilent VNA with a N4431-60006B Ecal.

See the image below. This compares a full 2 port model extraction of a VHF bandpass filter. The graphs compare between the nanoVNA and the Agilent VNA for things like insertion loss phase and complex impedance at port 1 and at port 2. The agreement is very good as the traces almost overlay perfectly. The nanovna saver software and the 111 software look nice with lots of features but I want to try and optimise the performance of the VNA and the cal kit so I've reverted back to my own PC tools.

Hi,

Very useful discussions around this demo kit. I was trying to do the measurements on it but I realized that the OPEN structure nr 14 was causing me large problems.
After calibration it did not measure correctly, so most of the other structures will of course show up incorrect. I see my board is the left version above.
Can it perhaps be fixed by cutting the open line closer to the surface mounted connector?

I purchased this board for potentially using it for teaching, but I was unaware of the U.FL female contact issues.
What could be really useful is if this board could be redesigned for SMA, but sold without the SMA connectors.
The the buyer can select what price range SMA they want to solder in place.

Thanks for the links to the SDR-Kits testboard and the VNA Sandbox.
Perhaps a joint effort could result in a useful PCB design for teaching, with the Gerber files for anyone to order and assemble?
Splitting it to two or more 10x10 cm PCBs would reduce costs.

On Sat, Mar 20, 2021 at 12:43 PM, <tomaz.solc@...> wrote:

I found out that I get better results if I take into the account that the
"open" standard has some stray capacitance.

Hello Toma?

Roger Need has made a comparion of two different RF Demo kit PCB's.
See: /g/nanovna-users/message/20702
and the attached photo: RD-Demo-Kit_PCB_comparison.png

How about adding the 0.6 pF with an isolated wire at the Open connector?
About 1 cm length should be sufficient, see:


73, Rudi DL5FA

I may be wrong, but for NanoVNA-H and H4 the requirement is Virtual COM Port Driver.


Martin 9A2JK

looks OK to me