What happens when you compare different calibration stubs—the 50 ohm load. The NanoVNA comes with 1 50 ohm calibration stub, but you need 2 (and 2 F-F barrel connectors if using the supplied cables) to do a proper ISOLATION calibration for S21 type tests.

So I bought for about $15 another stub kit with an identical looking 50 ohm load and F-F barrel connector (a bonus was an M-M barrel connector the original kit didn’t have). Then, also for about $15 I saw a pair of different looking 50 ohm load stubs. I bought those too, wondering what would happen if I compared 50 ohm stubs? What if I CALIBRATED with one type of load stub and then TESTED (S11) the other?

The three pics below tell the story. The two stub types match (50ohm, no reactance) only at the low end of my test frequency (1MHz–800MHz) range. They progressively diverge as frequency goes up, and as might be expected, the divergence is inverted. (A) becomes capacitative, and resistance goes down compared to (B), while (B) becomes inductive, and resistance goes up compared to (A)!

Is there a difference in real measurements? I tested a J pole antenna across my calibrated range (1-800 MHz). Between 750-800 MHz the frequencies reported for the lowest SWRs varied by 4+ MHz depending on which calibration was used. At 100-150 MHz the measured difference was only 50 kHz.

The device can only be as good as its calibration. Professional calibration kits cost THOUSANDS of dollars—I looked. Regarding absolute measurements the NanoVNA using the supplied or other inexpensive calibration stubs is “very close” at lower frequencies and progressively less reliable as the frequency goes up. Users who are doing measurements in the upper parts of the NanoVNA’s frequency range must keep this in mind.

—----------
Picture 1: Load stubs (A/Top) supplied with the NanoVNA. (B/Bottom) a second set of 50 ohm loads purchased (cheap) from Amazon.

Picture 2. 1MHz - 800MHz calibrate using the (B) stub from picture 1 but testing the (A) stub. Trace 1 is Resistance. Trace 2 the Smith chart. The A (DUT) stub is more capacitative and resistance goes down with increasing frequency.

Picture 3. 1MHz - 800MHz calibrate using the (A) stub from picture 1 but testing the (B) stub. Trace 1 is Resistance. Trace 2 the Smith chart. The B (DUT) stub is more inductive, and resistance increases with increasing frequency.

On Mon, Oct 7, 2024 at 02:36 PM, Matthew Rapaport wrote:

Picture 1: Load stubs (A/Top) supplied with the NanoVNA. (B/Bottom) a second
set of 50 ohm loads purchased (cheap) from Amazon.

Those gold ones have horrible performance. If you cut one open you will find a leaded resistor inside with one long lead that just brushes against the connector. You will get rising resistance with frequency due to skin effect on resistor and the long leads and resistor construction will have noticeable inductance.

Roger

Thanks Rodger... It had occurred to me that the original, and the one I got in the second kit (feels identical) are better built than the gold ones. Must settle for 1 set and be consistent in the future.

You can spend $14.27 single quantity at Minicircuits https://www.minicircuits.com/WebStore/dashboard.html?model=ANNE-50%2B
and know that it is of good quality, 30dB return loss (1.065:1) up to 4GHz, rated to 12GHz.
73, DonN2VGU

On Oct 7, 2024, at 11:36 PM, Matthew Rapaport <quineatal@...> wrote:

What happens when you compare different calibration stubs—the 50 ohm load. The NanoVNA comes with 1 50 ohm calibration stub, but you need 2 (and 2 F-F barrel connectors if using the supplied cables) to do a proper ISOLATION calibration for S21 type tests.

I’m new to this (and haven’t been an active ham for a couple of decades); but I figured I had a use for a dummy load anyway. So I got a large one to use at home, and a 15W BNC dummy load for direct testing of smaller radios. Then on the nanovna I need to use an SMA-BNC adapter, but that’s realistic since some of my cables have BNC connectors too. Maybe it’s better to just use dummy loads on both terminals, so as to test the same adapters that I usually need in practice, and also in case the little calibration stub is not so accurate?

Thanks Donald! Good reference! They look nice. Maybe for Christmas! 😁

When it comes to cal standards, stay clear of AliExpress and Amazon and a
few others from China. They are assured to be marginal in quality,
especially in the upper frequency ranges of the NanoVNAs. They may be just
a qualified "OK" at HF frequencies. As mentioned in another email in this
thread, stick with MiniCircuits. They aren't that expensive and you know
you have a quality standard.

Dave - WØLEV

On Mon, Oct 7, 2024 at 9:36 PM Matthew Rapaport via groups.io <quineatal=
gmail.com@groups.io> wrote:

What happens when you compare different calibration stubs—the 50 ohm load.
The NanoVNA comes with 1 50 ohm calibration stub, but you need 2 (and 2 F-F
barrel connectors if using the supplied cables) to do a proper ISOLATION
calibration for S21 type tests.

So I bought for about $15 another stub kit with an identical looking 50
ohm load and F-F barrel connector (a bonus was an M-M barrel connector the
original kit didn’t have). Then, also for about $15 I saw a pair of
different looking 50 ohm load stubs. I bought those too, wondering what
would happen if I compared 50 ohm stubs? What if I CALIBRATED with one type
of load stub and then TESTED (S11) the other?

The three pics below tell the story. The two stub types match (50ohm, no
reactance) only at the low end of my test frequency (1MHz–800MHz) range.
They progressively diverge as frequency goes up, and as might be expected,
the divergence is inverted. (A) becomes capacitative, and resistance goes
down compared to (B), while (B) becomes inductive, and resistance goes up
compared to (A)!

Is there a difference in real measurements? I tested a J pole antenna
across my calibrated range (1-800 MHz). Between 750-800 MHz the frequencies
reported for the lowest SWRs varied by 4+ MHz depending on which
calibration was used. At 100-150 MHz the measured difference was only 50
kHz.

The device can only be as good as its calibration. Professional
calibration kits cost THOUSANDS of dollars—I looked. Regarding absolute
measurements the NanoVNA using the supplied or other inexpensive
calibration stubs is “very close” at lower frequencies and progressively
less reliable as the frequency goes up. Users who are doing measurements in
the upper parts of the NanoVNA’s frequency range must keep this in mind.

—----------
Picture 1: Load stubs (A/Top) supplied with the NanoVNA. (B/Bottom) a
second set of 50 ohm loads purchased (cheap) from Amazon.

Picture 2. 1MHz - 800MHz calibrate using the (B) stub from picture 1 but
testing the (A) stub. Trace 1 is Resistance. Trace 2 the Smith chart. The A
(DUT) stub is more capacitative and resistance goes down with increasing
frequency.

Picture 3. 1MHz - 800MHz calibrate using the (A) stub from picture 1 but
testing the (B) stub. Trace 1 is Resistance. Trace 2 the Smith chart. The B
(DUT) stub is more inductive, and resistance increases with increasing
frequency.

--

*Dave - WØLEV*


--
Dave - WØLEV

My experience as an engineer at HP/Agilent:  you can make a decent female/socket SMA load by clipping the center pin of a socket short and soldering two 100 ohm 0603 or 0805 1% (or 0.1% if you want) resistors directly to the flange and center pin, diametrically opposite - or four 200 ohm resistors similarly symmetrically opposed.  I was once asked to design a 3-way 50 ohm splitter that would be reasonable to use to 6GHz and was able to do it using SMT resistors with zero lead length; it had a brass screw to adjust the high-end performance (adjusting the capacitance to the 4-way junction inside).  It was built in a machined aluminum block that at least attempted to maintain transmission line impedances around the resistors.  Up to low GHz, at least, it's not rocket science to get decent performance, but it requires attention to detail to use off-the-shelf inexpensive parts.  (The tech that had to assemble that splitter was not overjoyed with me, as there was very little space to work in to solder the resistors together.)

Cheers,
Tom
----------------
"Weird hou men maun aye be makin war insteid o
things they need." -- Tom Scott (1918-1995)