开云体育

On Thu, Oct 20, 2022 at 04:46 PM, N2MS wrote:

Isn't DiSlord's latest version 1.2.16?

DiSlord posted version 1.2.16 in message # 4004 on September 30, 2022.

Mike N2MS

At least in his GitHub repo the version is still at 1.2.15 (I get the version info by grepping for this line):

#define VERSION "1.2.15"

Martin

Using STM32CubeProgrammer I can get both NanoVNA-H-SI_v1.2.08.00 hex and NanoVNA-H-SI_v1.2.14 hex to download to the NanoVNA-H4 but only NanoVNA-H-SI_v1.2.08.00 hex will run. DfuSe Demo (v3.0.6) will down load NanoVNA-H-SI_v1.2.14 dfu and the NanoVNA-H4 will run. I prefer the cube programmer, so can anyone resolve this?

Tom W0IVJ

In case it may be of interest, I also measured one of the triode sections in a 12AU7 dual triode tube recently in grounded grid. I biased it at 0Vgk and 90V HT at 11mA anode current and extracted an S2p model using the VNA. I used bias tees and made direct connections to the pins of the tube (no tube base was used). Therefore, the model should be quite realistic for the raw tube before it is fitted to a base. The base will add capacitance between the pins and some inductance, so I made the measurements without the base.

The result for GMAX and K is given below. This shows that the tube can go unstable and oscillate up at about 500MHz as K is still below 1 here. This was quite a surprise for this basic little triode tube. I think the reason K fluctuates a lot up at UHF is because of the transit time in the tube. Presumably this causes reflections and this is probably why the fluctuations in GMAX and K are present up at UHF.

I've also included a classic Ft plot for the MMBTH10 BJT in common emitter at 10Vce and 5mA Ic. Ft is a fairly consistent 900MHz at each marker point. At unity current gain the Ft is 854MHz. In all cases, I always make sure the VNA source power is at a low level to prevent non-linear effects and I use an N4431B-60006 4 port Ecal and I exploit the fixture simulator option in the VNA to calibrate the (18GHz) test fixture such that the reference plane is right at the pins of the device under test as in the image below. All biasing and DC blocking is done outside the calibration plane.

To Donald S Brant Jr and Jim Lux, I want to match an old transmitting transistor. The coil nearest the base is about 6 ohms, that is about 1-3cm wire. Calibration and measuring those is hard.
It is too easy with fets.

Isn't DiSlord's latest version 1.2.16?

DiSlord posted version 1.2.16 in message # 4004 on September 30, 2022.

Mike N2MS

At the bottom of your e-mail you read:

Learn about the NanoVNA in the following forum areas:

Documentation & Update Files:/g/nanovna-users/files

If you visit that link you will find:

1. //[email protected] </g/nanovna-users>
2. //Files </g/nanovna-users/files>
3. Absolute Beginner Guide to The NanoVNA

and then you will find:

Absolute_Beginner_Guide_NanoVNA_v1_6.pdf </g/nanovna-users/files/Absolute%20Beginner%20Guide%20to%20The%20NanoVNA/Absolute_Beginner_Guide_NanoVNA_v1_6.pdf>

-PE0CWK-



Op 20-10-2022 om 14:42 schreef Frank:

Hi All. I am a newbie here and I need a manual for my new NanoVNA-H4 and also step by step instructions on how to calibrate and save the cal data.

Frank

Nice results.

I did the same experiment with a Hugen 2.3 and had to move to S21 measurements for accurate results on 1-30 MHz.

Maybe, if I get around to it, I? will try these low/high resistance S11 measurements on an H4.2.

Thanks.

Arie PA3A

Op 20-10-2022 om 01:33 schreef jmr via groups.io:

It's also possible to post process the VNA data to do other things. Here's a chart of current gain vs frequency for the same 2N3904 at 10Vce and 10mA Ic. This chart is derived from the same dataset that produced the GMAX and K plots. You can see the VNA shows unity current gain at about 300MHz at 10Vce and 10mA Ic as this is the frequency where the current gain falls though unity on the chart.

If Ft is calculated at 20.19MHz (marker 4) the Ft is predicted to be 20.19 * 16.73MHz = 338MHz.

I often use a lab VNA to measure and model semiconductors up to many GHz. These 2 port models work well in practice and a lot can be learned about a transistor from these measurements.

In the plots below I measured a 2N3904 in common emitter at 10Vce and 10mA Ic. I also measured a MMBTH10 in common base at 10Vce and 3mA Ic. The plots show GMAX and K in each case. These models are taken using bias tees and a decent test fixture where the reference plane is set right at the pins of the SMD BJT.

The VNA predicts that (at 10Vce and 10mA Ic) the 2N3904 can produce about 15dB gain at 145MHz and about 8dB gain at 432MHz. It also predicts that the 2N3904 could go unstable and oscillate at just over 1GHz if configured as an oscillator as the K factor is still below 1 at 1GHz at this operating point. The MMBTH10 plot shows that the MMBTH10 BJT can oscillate at up to about 2.4GHz at 10Vce and 3mA Ic as K is still less than 1 here. This is the SMD version of the MPSH10 BJT.

You can't predict this behaviour from the datasheet. This shows how useful and powerful a VNA can be. I used an Agilent VNA for these plots rather than a nanovna.

On 10/19/22 3:26 PM, Leif M wrote:

"You can calibrate for any resistance"
Sure. But if those things can be made would it improve accuracy, when measuring low and high impedances.

Sure - and people do that - sometimes as part of a research project or dissertation.

Ultimately, what you do is look at what you need to measure, assess the uncertainty with your "50 ohm" equipment and go from there. Usually, it's a matter of SNR and accuracy of measurement of small changes. Looking at a 5 ohm circuit, the reflection coefficient is huge (0.8=45/55) and if you want 10% accuracy, you need to look at telling the difference between about 0.02 out of 0.8 (2%).

Z mag gamma
4.5 .834
5 .818
5.5 .801

The NanoVNA can easily make measurements to 2% or even 1%.

OTOH, if you need 1% accuracy, you're looking at needing a measurement to 0.2%.
The NanoVNA can still do that, but you'd better have really good cal standards, etc.

It's also going to be painful making accurate S21 measurements of lossy components, because you've got 15dB loss through the first reflection and the same at the output, so it's like having a 30dB pad in series with your UUT.



If you want to get really wild, look into "dembedding" - it's of vital importance in measuring circuits on a semiconductor die and they're rarely 50 ohms.

In addition, waveguide has a frequency dependent impedance that is also propagation mode dependent. And it calibrates out.

I've used near field antenna ranges which use a VNA as their core, with a open waveguide as the probe, and it's nothing even remotely close to 50 ohms. But it *can* be precisely calculated, and verified.

There's also VNAs that don't use the coherent transmitter/receiver like the NanoVNA. Look up 6-port analyzers - which use only power detectors (which can be made very accurate).

VNA errors get high when the mismatch is also high; we try to calibrate them out but it is an imperfect process. When we needed to measure the impedances of large L-band power bipolar transistors, with impedances in the single digits, we would use a low-impedance series microstrip line matching transformer to pre-match the impedance to something closer to 50Ω. We would use a break-apart fixture where we could measure the matching section separately and back its transformation out of the measurements. We'd measure two matching sections back to back to determine the losses. It was time-consuming and narrow-band.

For lower frequencies it might be worthwhile using a commercial (Mini-Circuits) 9:1 or 16:1 transformer backwards and do the same trick. Then the analyzer only needs to deal with measuring a 45Ω impedance (VSWR 1.11:1) rather than 500Ω (VSWR 10:1).

In the REAL Dark Ages We had an Alford 5Ω slotted line for measuring low impedances; SUPER tedious work. For high impedances, maybe an open-wire "Lecher line" could be used:
73, Don N2VGU

I have had success on my Baofeng using a female barrow between the Baofend RP and a rubber duck with a normal BNC.

Dave - WB6DHW

Across the 5MHz to 50MHz frequency range, it's still possible for a basic nanovna to make fairly accurate resistance measurements across a wide range of resistances using a basic s11 measurement. In the plot below I measured a suite of test resistors from 0.167? through to 330k? from 2MHz to 30MHz. You can see in the chart below that my little nanovnaH did quite well in these tests. It failed to measure the 330k? resistor but did OK across 0.167? through to about 100k?. At higher test frequencies the performance of the nanovna will degrade but I think the performance of the nanovnaH is quite impressive in the plot below. This was all done with a single s11 calibration.

I have had problems when I have accidentally used RP connectors. Two female connectors back to back do not work.

I can't edit my posts, so a second post. What I understand, NanoVna asn other VNAs are optimised for 50 ohms, they have for instance 50 ohms impedance bridge inside.

"You can calibrate for any resistance"
Sure. But if those things can be made would it improve accuracy, when measuring low and high impedances.

The test equipment standard in the RF/?wave industry is and has been for
some time 50-ohms. No, others are not made. The TV industry may have a
few dedicated instruments at 75 ohms, but that's about it. 50-Ohms is our
standard.

Dave - W?LEV

--
Dave - W?LEV

On 10/19/22 1:12 PM, Leif M wrote:

Do we need a VNA for 5 ohms and an other for 500ohms. Are they possible. Lets forget about 5 or 500 ohms attenuators, coaxials and such. But would it help if VNA is made for other impedances.

You can calibrate for any resistance

Do we need a VNA for 5 ohms and an other for 500ohms. Are they possible. Lets forget about 5 or 500 ohms attenuators, coaxials and such. But would it help if VNA is made for other impedances.