开云体育

Still new to using a VNA.

I've calibrated my NanoVNA-H4 and double checked the calibration with the 3 tips once again as per hexandflex' guide, with the isolation and through, since I want to measure the coax loss.
So far so good.

Now when I plug in the cable to both S11 and S21, I get some really odd readings that don't seem to match up with what I'm seeing in guides and videos. My logmag is all over the place and nowhere near showing a proper reading.

What is it that I'm missing here? Did I do something wrong?

Here's with a reset calibration, I don't think S21 should be noisy like that, should it?

The value for S11 looks "about" right. Yes, S21 is noisy. Run the same
test with the coax you used for the through cal, of course, after another
complete cal.

Dave - W?LEV

--
Dave - W?LEV

This is when I calibrated and left the cable in through, used the thin coax that comes with it.

Hello,

Try setting the THRESHOLD setting to 280 and do another complete calibration.
Some clock chips are not stable at the default value of 300(MHz). These chips are severely overclocked in the NanoVNA design, their nominal Fmax is 200MHz.
The instability shows up as "noise" in many traces.
This will reduce your "maximum" frequency to 1400MHz using the 5th harmonic, but most manufacturers state a useful maximum up to about 1250MHz anyway.
Make sure you do a RESET and CLEAR ALL to clear all old Calibration data before you do the new cal.
HTH.

73...Bob VK2ZRE

Very cool tip, thanks for detailing it!

It removed the movement, but it's still not a line, still looks like noise with inaccurate values unfortunately.

Could the S21 port just be busted?

don't think it is busted. Your photo shows the port unterminated. Try terminating both ports in 50 Ohms and rerun the scan.
The Port2 input is pretty sensitive to internal crosstalk (noise), so I would not be too concerned.
Remember, this is a ~$100 instrument. It performs exceptionally well for that amount of money:-)

73...Bob VK2ZRE

An easier way to check coax loss is available on the H4 with most recent
firmware. You connect the coax to only the s11 port, and leave the far end
of the coax open. Then from the menu choose measure / coax. The
measurements will appear at the left of the screen, including a loss
measurement at whatever frequency the cursor is at. So you set the
frequency range to include the frequency of interest, and directly read the
loss at that freq.

The nano makes this measurement from the s11 reflected wave, and divides by
2 because the raw measurement is for both up and back along the cable.
It's a nice feature. Not also that the frequency range must include a low
enough frequency to be a quarter wavelength for the length of the cable, so
I often just calibrate for 50kHz - 30MHz when doing this type of measure so
I don't have to think of it. Even with this wide range it gives good
accuracy because the nano firmware does a good job of interpolation between
measurement points (thank you, Hugen!).

(Note that the length measurement shown depends entirely on the accuracy of
the velocity factor - so be careful with that value.)


On Thu, May 5, 2022, 5:48 PM Bob Ecclestone VK2ZRE <becclest@...>
wrote:

don't think it is busted. Your photo shows the port unterminated. Try
terminating both ports in 50 Ohms and rerun the scan.
The Port2 input is pretty sensitive to internal crosstalk (noise), so I
would not be too concerned.
Remember, this is a ~$100 instrument. It performs exceptionally well for
that amount of money:-)

73...Bob VK2ZRE


On 6/05/2022 10:36 am, kosmos wrote:

Very cool tip, thanks for detailing it!

It removed the movement, but it's still not a line, still looks like

noise with inaccurate values unfortunately.

Could the S21 port just be busted?

Oh that's an awesome feature I didn't even know about, thank you! Though I'm still going to troubleshoot to try and get the other port to work properly.
Is this calculated value as accurate as having it plugged in both though? I would assume it's a little more accurate when it is.

I don't think the 2-port thru measurement will be any more accurate for
simple cable measurements. The nano is very accurate with a good dynamic
range on reflected (s11) measurements. If the loss were very high, e.g.
greater than 30dB, then the thru measurement may be more accurate, because
the two-way loss measured by s11 would be >60dB, coming closer to the
dynamic range limits.

On Thu, May 5, 2022 at 08:01 PM, Stan Dye wrote:

An easier way to check coax loss is available on the H4 with most recent
firmware. You connect the coax to only the s11 port, and leave the far end
of the coax open. Then from the menu choose measure / coax. The
measurements will appear at the left of the screen, including a loss
measurement at whatever frequency the cursor is at. So you set the
frequency range to include the frequency of interest, and directly read the
loss at that freq.

The techniques used in the latest firmware to calculate the loss is probably the the "Return Loss Method" of calculating Matched line Loss for a transmission line using an open circuit (or short circuit) line. It is easy to do because you just measure the Return Loss (RL) and divide by two. However it gives an approximation that is poor with low loss lines. A better approximation is obtained if the RL open circuit and RL short circuit measurements are added together and the result divided by 4. An article by Witt explained the rationale behind this method and it is discussed here...


I have attached some measurements done on a 30.75 foot (9.37M) length of Belden RG58 A/U cable. One graph is a plot of Matched Line Loss calculated using RLoc/2, RLsc/2 and using S21 on a NanoVNA-H4. There are large ripples in the RL plots due to the cable impedance not being equal to the reference impedance of 50+j0 ohms. The second graph shows (RLoc+RLsc)/4 and S21 and the two plots are very close. For reference a plot of this cables attenuation from the manufacturer is also attached. The manufacturer data shows 3.53 dB of loss at 50 MHz. for 100 feet which is equal to 1.09 dB for 30.75 feet. This compares well with the 1.16 dB measured using the S21 method

Roger

On Thu, May 5, 2022 at 08:34 AM, kosmos wrote:

I've calibrated my NanoVNA-H4 and double checked the calibration with the 3
tips once again as per hexandflex' guide, with the isolation and through,
since I want to measure the coax loss.
So far so good.

Now when I plug in the cable to both S11 and S21, I get some really odd
readings that don't seem to match up with what I'm seeing in guides and
videos. My logmag is all over the place and nowhere near showing a proper
reading.

Here are some screenshots done with my nanoVNA-H4
- One show S21 being 0 dB with cable attached between ports (right after calibration)
- Next shows S2 noise floor with no cable attached between ports (about 70 db or so of range)
- Last one shows 30 feet of RG-58 connected between ports.

Hope these help you to know what to look for...

Roger