Stupidly I have blown up the SMA load fro my NanoVNA. I put accidentally about 20W into it. It blew instantly.

Does any one have any spares I could purchase?  It is just the load I need.

Neil G3RIR

Hi Neil

I can offer one calibrated) against an Anritsu 26GHz standard (long ago calibrated HI)  Specced (35dB RL to 26.5GHz)

I know I have several that had better return loss that the original supplied. (I don't use my original HI, it isn't good enough :) )

I can't offer traceability HI

Mark GM4ISM

Stupidly I have blown up the SMA load fro my NanoVNA. I put accidentally about 20W into it. It blew instantly.

Does any one have any spares I could purchase?  It is just the load I need.

Neil G3RIR

-- 
Mark  GM4ISM
gm4ism@...
dc2light.co.uk

All sorted now thanks to G4JNT

Neil G3RIR

Hi,

I am curious about how you calibrate an SMA load respectably supposing
access to another VNA (and its cal kit with coefficients!)

My Anritsu came without the calibration kit (usual comments about
people looking for test gear bargains apply) but I also tend to
measure stuff with SMA connectors rather than N connectors so probably
the N-type cal kit would just have sat on the shelf and gathered dust.
If I'm honest most of the time I use Rosenberger SMA load/thru/open
but have no idea what accuracy I'm losing doing this. There's a bit
of a documentation shortage on how one provides cal coefficients to
the Anritsu - I imagine it wants a capacitance at the end of the open
and inductance for the short as a function of frequency, but less
clear what is done about the load - which isn't exactly 50 ohm - I can
measure that on a 4 terminal bridge at DC. The manual of course just
says it wants a disk of cal kit coefficients which is a bit
meaningless.

Traceability not required here!

Alan


On Mon, 20 Mar 2023 at 18:54, Mark GM4ISM via groups.io
<gm4ism@...> wrote:

Hi Neil

I can offer one calibrated) against an Anritsu 26GHz standard (long ago calibrated HI) Specced (35dB RL to 26.5GHz)

I know I have several that had better return loss that the original supplied. (I don't use my original HI, it isn't good enough :) )

I can't offer traceability HI

Mark GM4ISM

On 20/03/2023 18:37, Neil, G3RIR via groups.io wrote:

Stupidly I have blown up the SMA load fro my NanoVNA. I put accidentally about 20W into it. It blew instantly.

Does any one have any spares I could purchase? It is just the load I need.

Neil G3RIR

--
Mark GM4ISM
gm4ism@...
dc2light.co.uk

Hi Alan

There is  quite a bit involved with getting confidence in the measurements associated, thus this email is a little long winded...

I don't know the instrument or cal kit you are using but typically the Open Short Load sets are  50 Ohm characterized.  When you calibrate most instruments with one, it measures and calculates all the offsets and errors  etc at the test point (where you put the standard) and measures all subsequent WRT the 50 Ohm cal standard which the instrument  assumes is 50 j0 ohms at every frequency. 

Other cal kit coefficients that may be 'needed' are electrical length to the load /  open / short.  These help calibrate the phase components  that a VNA will display. Assuming the instrument can be calibrated to the load, the measurement directivity will  match that of the load RL irrespective of the cal coefficients not being loaded

The ability to store and recall cal data is usually more to enable a full test system (including cables, adapters etc)  to be set up quickly for a given application easily if nothing has changed.  Something I rarely did professionally, re-calibrating is a little tedious but brings about a sense of certainty.

Its easy enough to damage a 50 Ohm standard load with over power but it usually stops measuring 50 ohms at DC at that point.  In a 50 Ohm system, a 51 Ohm impedance is still 40dB RL  A load with 34dB RL could be 52 Ohms at DC. The chances are your system is  calibrated well enough. That said, the DC reading of the load should not be that far off 50 Ohms.

Formal calibration  is about traceability in professional standards (pun intended) as much as anything else.

The equipment I have is all well passed its sell by date and overdue calibration  in those circles.   Having administered aspects of test gear standards professionally for years, I can say that I have never met an instrument (of this sort) that failed calibration and did not obviously have a fault.

I have network analysers, a scalar analyser to 26GHz  and reference standards, one with  specified return losses >49dB (sub3 GHz).  Having used them over the years and stored cal data, they do not appear to have changed in the years since formal calibration.

The other standard (3.5mm),  is specced at >40dB to 26.5GHz.  Using one as a reference for another will quickly show up any issue. I have also had access for years to formally calibrated standards and measurement comparisons were fine.

 So while I  have no formal traceability, I have confidence that my reference standards are good.

With any measurement of this type you need to reference a 'standard' and have confidence in that standard.

 If you calibrate a VNA with a standard of 49dB return loss, it should read the calibration standard as having a return loss of  infinity...  Noise floors etc  limit that but with a real network analyzer,  at least 80dB.  At these levels, just breathe on the test gear and the null into the reference load can be seen to twitch:)

With a 'calibrated' network analyser  showing >80dB return loss you can be assured that the directivity of the measurement system has become very close to the return loss of the standard (in this case 49dB worst case)

Now substitute the  load you want to measure (DUT)

The DUT return loss indicated will have an accuracy dictated by the directivity.. a reasonable gauge of the effect is here

https://www.markimicrowave.com/assets/appnotes/directivity_and_vswr_measurements.pdf

Measuring a DUT with a poor return loss  with a good directivity  system is really quite accurate.

Measuring a good DUT with a mediocre directivity and you could get quite large numerical errors in the answer.

Measuring a good DUT with a better directivity reduces the uncertainly. Typically you need a directivity  15dB better than the DUT return loss to  have a reasonable  levels of uncertainty.

If you achieve an indicated >80dB RL, you have actually achieved a RL close to that of the directivity of the instrument (in this case 49dB)

Thus with  49dB directivity you can measure a DUT to about 34dB return loss and have a reasonable accuracy (within a couple of dB)  You are unlikely to get a flat line indicated RL >80dB! 

Having calibrated  my test gear to the 49dB RL standard, I measure the professional 3.5mm Marconi standard I have.

It is specced at 40dB RL (up to 26.5GHz) and looks way better than that up to 3GHz. It reads in the 60s dB RL up to 3GHz.  I can say is that it is very likely in spec ie >40dB

This comparison  gives a degree of confidence that the standard loads are in good condition and can be trusted.

 I substitute other SMA loads (including one from my little nanoVNA) and pick the ones that closest match the professional standard.

I was pleasantly surprised that the supplied standard for the nanoVNA was actually quite good, not quite as good as the Marconi standard but close. I measured a return loss of > 35dB +- 2dB

I tested sundry SMA loads.  Most were quite frankly not fit to be used as a return loss standard, with measured RL at 25dB or worse at higher frequencies.  Not surprising, look at the specs for even quality 50Ohm loads.. The spec required for a calibration standard is way higher than a random good load

If you used a 25dB RL load to calibrate a VNA, the VNA would have a directivity of 25dB.. Nearly useless in my books, as an indicated 20dB return loss could be anywhere from approx 16dB to 27dB .

If you adjust  the DUT for a 'perfect' match into such a measurement system and achieve a very high measured RL, you will actually have achieved a RL for the DUT that is equal to the directivity.    That may be good enough in many cases.

Of course a 20dB ret loss is fine as a dummy load.

I hand picked from my extensive stock of SMA loads and ended up with 3 or 4 (about 10%) that were good enough (RL in the 35dB region) up to 3GHz

 Only one  passed muster to >20GHz compared to the  Marconi standard.

Above 3GHz, I am just comparing the Marconi standard to the DUT.  Any load that reads similar (>60dB) after the 26GHz analyser has been 'calibrated'  to the Marconi standard, will have a similar return loss, ie about40dB

Above 3GHz I don't have another good reference  so there is a degree of faith here on the instrumentation and that Marconi standard load.

That faith is re-enforced by first principal measurements, measuring standing waves with a swept frequency measurement  using a non-directional coupler using a mostly forgotten technique called wobbulation (which can also indicate distance to fault so long as you can do inverse FFTs in your head). Some on this list will have also come across quaterwaving  as a technique for removing  offsets and improving accuracy.  I've got a waveguide with a probe in a long slot for match measurements in WG16. Again a 1st principal technique that doesn't rely on instrument directivity as such.

All these measurement techniques have intrinsic uncertainly but the more you get the same answer by different techniques, the higher the confidence is.

In short,  I have reasonable confidence that up to 3GHz, I can characterise a load to about 35dB RL with reasonable accuracy, which is adequate for my purposes and I suspect adequate for most amateurs as a standard against which to make measurements.  A bog standard SMA load  may not cut it and introduce significant uncertainly.

 That is what I consider 'calibration'.    Without  reference to what  accuracy one  expects to achieve or claims, the term calibration is open to interpretation.

Its a far cry from what a professional calibration house can access but the same principals are employed.

I hope that helps

Measuring return loss, frequency and power  is all  a bit of a minefield in the amateur world.  I am fortunate in having instruments that let me have a fair go at all these.  That said, you will not loose many QSOs and are unlikely to blow up your PA if the match of your antennas is only 17dB (so long as that isn't because it has water in it) .  Being 10 Hz of frequency is not often an issue  either....  Just don't believe your bird wattmeter  blindly ....

Regards

Mark GM4ISM

Hi,

I am curious about how you calibrate an SMA load respectably supposing
access to another VNA (and its cal kit with coefficients!)

My Anritsu came without the calibration kit (usual comments about
people looking for test gear bargains apply) but I also tend to
measure stuff with SMA connectors rather than N connectors so probably
the N-type cal kit would just have sat on the shelf and gathered dust.
If I'm honest most of the time I use Rosenberger SMA  load/thru/open
but have no idea what accuracy I'm losing doing this.  There's a bit
of a documentation shortage on how one provides cal coefficients to
the Anritsu - I imagine it wants a capacitance at the end of the open
and inductance for the short as a function of frequency, but less
clear what is done about the load - which isn't exactly 50 ohm - I can
measure that on a 4 terminal bridge at DC. The manual of course just
says it wants a disk of cal kit coefficients which is a bit
meaningless.

Traceability not required here!

Alan


On Mon, 20 Mar 2023 at 18:54, Mark GM4ISM via groups.io
<gm4ism@...> wrote:

Hi Neil

I can offer one calibrated) against an Anritsu 26GHz standard (long ago calibrated HI)  Specced (35dB RL to 26.5GHz)

I know I have several that had better return loss that the original supplied. (I don't use my original HI, it isn't good enough :) )

I can't offer traceability HI

Mark GM4ISM

On 20/03/2023 18:37, Neil, G3RIR via groups.io wrote:

Stupidly I have blown up the SMA load fro my NanoVNA. I put accidentally about 20W into it. It blew instantly.

Does any one have any spares I could purchase?  It is just the load I need.

Neil G3RIR

--
Mark  GM4ISM
gm4ism@...
dc2light.co.uk

-- 
Mark  GM4ISM
gm4ism@...
dc2light.co.uk