开云体育

The group has been focused on SOLT or so called SOLM (match) for calibration. However, for various reasons the TRL; Thru, Reflect, Line calibration algorithm would be advantageous. Fabrication of such a kit usually requires at least 3 lines to be built plus one fixture piece and microstrip is not difficult using a straight edge and an Exacto blade knife! The TRL lends itself to NON INSERTION components. In another words, non coaxial elements. Just a thought and the copy below hints to the benefits.

From Keysight:

==============================================
A vector network analyzer requires that a measurement calibration be performed before error-corrected measurements can be made. For two-port measurements, the calibration algorithm used will determine the appropriate calibration kit, known either as SOLT or TRL/TRL*.

Traditional full two calibration calibration typically utilizes three impedance and one transmission standards to define the calibrated reference plane. These standards, typically a Short, Open, Load, and Thru, make up the SOLT calibration kit.

Another two port calibration type utilizes a minimum of three standards to define the calibrated reference plane. The measured parameters of the Thru, Reflect, and Line standards in a TRL/TRL* calibration kit provides the same information as a SOLT calibration via a different algorithm.


Depending on the availability of calibration standards and the functionality of the network analyzer, one may use either calibration kit.

In many non-coaxial , SOLT calibration standards are difficult, if not impossible, to build. The non-availability of calibration standards in proprietary or unique coaxial connectors also benefit from the TRL/TRL* calibration method.

============================================

On Tue, 15 Oct 2019 at 19:11, alan victor <avictor73@...> wrote:

The group has been focused on SOLT or so called SOLM (match) for
calibration. However, for various reasons the TRL; Thru, Reflect, Line
calibration algorithm would be advantageous. Fabrication of such a kit
usually requires at least 3 lines to be built plus one fixture piece and
microstrip is not difficult using a straight edge and an Exacto blade
knife! The TRL lends itself to NON INSERTION components. In another words,
non coaxial elements. Just a thought and the copy below hints to the
benefits.

From Keysight:

TRL calibration requires a VNA with 4 receivers. The HP 8753 series, which
have 3 receivers, can perform something HP refer to as TRL*. This requires
a good match at the test ports - typically 6 to 10 dB on each test port. It
is not as good, but it is not possible to do true TRL calibration on a
simple VNA.

I don’t think anything other than SOLT is likely to be practical on the
NanoVNA.

Actually, waveguide calibration is possible too, but waveguide is not very
useful under 1.5 GHz.

Dave
--
Dr. David Kirkby,
Kirkby Microwave Ltd,
drkirkby@...

Telephone 01621-680100./ +44 1621 680100

Registered in England & Wales, company number 08914892.
Registered office:
Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United
Kingdom

On Tue, 15 Oct 2019 at 20:28, Dr. David Kirkby <
drkirkby@...> wrote:

TRL calibration requires a VNA with 4 receivers. The HP 8753 series, which
have 3 receivers, can perform something HP refer to as TRL*. This requires
a good match at the test ports - typically 6 to 10 dB on each test port. It
is not as good, but it is not possible to do true TRL calibration on a
simple VNA.

What I meant to say is that when using TRL* calibration on an HP 8753, 6
to 10 dB attenuators need to be added to the test ports of 8753 VNAs.

The 8720D instrument I have, was available with an optional 4th receiver to
allow full TRL calibration. It was option 400 and quite expensive. My
instrument doesn’t have that capability.

Also a full S-parameter test set is needed.

I don’t believe that TRL, or any similar calibration procedure, will be
possible with a simple VNA.

Dave
--
Dr. David Kirkby,
Kirkby Microwave Ltd,
drkirkby@...

Telephone 01621-680100./ +44 1621 680100

Registered in England & Wales, company number 08914892.
Registered office:
Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United
Kingdom

True, however, I was thinking forward as I understand a 3 channel nano receiver is under development. Further, the technique for obtaining true 2 port s data currently requires unbolting the test device and turning it end for end. Not sure the calibration in that process would hold up. Thoughts? Unless a simple after market S parameter test set is developed.

On 10/15/2019 5:03 PM, alan victor wrote:

True, however, I was thinking forward as I understand a 3 channel nano receiver is under development. Further, the technique for obtaining true 2 port s data currently requires unbolting the test device and turning it end for end. Not sure the calibration in that process would hold up. Thoughts? Unless a simple after market S parameter test set is developed.

I use a 4 port (SMA) transfer relay and calibrate through that then when I want to switch ports for a 4 port measurement I simply activate the relays and it reverses the device under test as seen from the nano.I installed the relay into a box with rg405 cables and 2 input / 2 output jacks. It is always the same constant and doesn't change. My nano is fixed connected to it using semi rigid cables. If I ever need a portable one I just buy one, they are cheap enough to have several.

That is a nice solution. Who is the manufacturer for the relay? Quality SMA packaged relays are not cheap!

Thanks,

On Tue, 15 Oct 2019 at 22:03, alan victor <avictor73@...> wrote:

True, however, I was thinking forward as I understand a 3 channel nano
receiver is under development. Further, the technique for obtaining true 2
port s data currently requires unbolting the test device and turning it end
for end. Not sure the calibration in that process would hold up. Thoughts?
Unless a simple after market S parameter test set is developed.

*There are several reasons I believe TRL calibration is not too practical
on a low-cost VNA.*

1) The length of the line section needs to be between 20 & 160 degrees
electrical length - ideally 90 degrees.



So to work at 1 MHz, where the wavelength is approximately 300 m, you need
a line section which is a *minimum* of 300*20/360=16.7 m long, and ideally
300/4=75 m long. This really makes TRL impractical below a few hundred
MHz.

2) As far as I an aware, TRL calibration requires a VNA with an S-parameter
set, and *4 independent receivers.* These are the a1, b1, a2 and b2
receivers. So you need one local oscillator split 4 ways to go to 4
different mixers The output from each mixer has to go to a different
receiver. You can’t switch one mixer and one receiver to measure different
things.

I haven’t looked at the block diagram of the NanoVNA, but I suspect it uses
just one mixer and one receiver, which is what all simple VNAs do to keep
the cost down.

I would not be surprised if the cheapest VNAs from Keysight don’t have 4
receivers.

The real big advantage of a 4 receiver VNA is the ability to do unknown
thru calibration.



That would be the biggest advantage of a 4-receiver VNA to me.

I attach a block diagram of a 3-receiver VNA (top) and a 4 receiver VNA
(bottom). I believe that you will find it considerably more complicated and
so expensive than a NanoVNA.




--
Dr. David Kirkby,
Kirkby Microwave Ltd,
drkirkby@...

Telephone 01621-680100./ +44 1621 680100

Registered in England & Wales, company number 08914892.
Registered office:
Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United
Kingdom

Dow-Key Microwave 411C-220832 12V relays I got a handful from equipment that we tossed into recycling at worksites. They never seen a signal larger than -10dbm and are perfect for circuits like this. None of them are older than 5 years and if they have switched more than 10 or 15 times in that time frame I would be utterly surprised.

Bilbo

On Tue, Oct 15, 2019 at 03:59 PM, Dr. David Kirkby from Kirkby Microwave Ltd wrote:

I suspect it uses
just one mixer and one receiver, which is what all simple VNAs do to keep
the cost down

The nanoVNA is a full three mixer three receiver VNA without the switch to reverse the signal path

Very nice.

You may want to post your system to the message group that is discussing test fixtures.
Part of the discussion raises the query on testing devices so full 2 port data is required.

I raised the TRL piece as I have used a combination of TRL and SOL or SOM on another vna.
David is correct that at lower frequency the cal is problematic. However, my fixtures lend themselves to TRL since there is no easy way to equip the DUT to coax.

So going forward with the nanovna will require a pair of decent bias tees or least ones that can be made part of the calibration process with minimum error and a switching system to turn the device end for end.

Thanks,

On Wed, 16 Oct 2019 at 08:30, <erik@...> wrote:

The nanoVNA is a full three mixer three receiver VNA without the switch to
reverse the signal path

Em,
in that case it would suggest a 4-receiver VNA may be economic if the
switch was external and purchased by the user if they want it.

That’s what the VNWA 3E - it provides the signal, but not the actual switch.

It’s often possible to pick up HP/Agilent 18 GHz SPDT mechanical relays
for a modest amount of money in small quantities. I have 10 here I picked
up very cheaply. I believe there’s a way of configuring just two, which
enables one to reverse the direction of the signal. There’s information
somewhere on the VNWA resources how to do this, without actually needing a
more specialised relay designed for that purpose.

I would see the major advantage of 4-receivers to be the ability to
perform *unknown
thru calibration*. That is incredibly useful, as you can perform a
calibration with two male cables, ignoring the properties of the thru. The
adapter can even be SMA to N, or even a waveguide to coaxial.

The NanoVNA is an amazing bit of kit for the money.

Dave.

--
Dr. David Kirkby,
Kirkby Microwave Ltd,
drkirkby@...

Telephone 01621-680100./ +44 1621 680100

Registered in England & Wales, company number 08914892.
Registered office:
Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United
Kingdom

New to VNAs, I picked up a NanoVNA after a fellow ham showed one around at a hamfest. (I think I got "lucky" and got a good one.) I've been following my nose, learning as I go, and studying various technical VNA web sites. I came to wonder why all the NanoVNAs that I see advertised come with at /most/ one short, one open, one load test fixture. It seems to me that we /need/ 2 of each to do a decent cal...or is what we are doing actually a valid, but "incomplete" cal...we have actually only calibrated Port 1 (Ch0), and I am not sure about the implications of that. There are no instructions with my device, so I have been reading docs and messages here.

One of the issues that I see as hams do various "tests" and "demos" is that those activities seem to be not well understood by those executing them or those reading|observing them. Because those activities are presented with a modicum of authority, they are repeated around the ham community, leading to confusion (I observe).

I intend to keep pointing folks to these pages...we need a central place where vetted information can be presented.

One of my questions is this: if I cal with only one set of short and load fixtures, then is the through valid? And does it really matter at frequencies below, say 600 MHz? Most of the stuff that I do as a ham does not seem to need super precision. OTOH, I want to use this device in the best way possible, just as I do all my gear, including test gear. I've ordered a 2nd NanoVNA from hugen (or so I hope), and it will come with a 2nd set of cal fixtures.

--
On the banks of the Piscataqua
Rich NE1EE

On Mon, 28 Oct 2019 at 15:37, Rich NE1EE <NE1EE@...> wrote:

New to VNAs, I picked up a NanoVNA after a fellow ham showed one around
at a hamfest. (I think I got "lucky" and got a good one.) I've been
following my nose, learning as I go, and studying various technical VNA web
sites. I came to wonder why all the NanoVNAs that I see advertised come
with at /most/ one short, one open, one load test fixture. It seems to me
that we /need/ 2 of each to do a decent cal...or is what we are doing
actually a valid, but "incomplete" cal...we have actually only calibrated
Port 1 (Ch0), and I am not sure about the implications of that. There are
no instructions with my device, so I have been reading docs and messages
here.

I am not following you 100%, but just because the VNA has two ports,
doesn’t mean you need two calibration kits. However, you do in general want
male and female standards.

In ideal wor you would have the following.

* A set of a set of male standards (open, short, load), as well as *one
extra male load* for an isolation measurement.
* A set of a set of female standards (open, short, load), as well as *one
extra female load* for an isolation measurement.
* Female-female adapter, so you can do a thru calibration with cables
having two male ends where you connect the device. *This is a common
occurrence. *
* Male-male adapter so you can perform a thru measurement is the cables
connecting to the DUT have two female ends, which is *very rare*.

What is supplied with NanoVNAs appears to be a subset of those, to keep
costs down, on what is an *amazing value for money*. For example, you can
combine the male load and a female-thru to get a female load. Exactly what
can be done, depends on exactly what parts you have been supplied.

The extra loads I mentioned do not have to be good loads. In fact, in
professional VNAs an isolation measurement is rarely desirable, and no
extra loads are supplied. The NanoVNA does however benefit from an
isolation measurement.

One of the issues that I see as hams do various "tests" and "demos" is
that those activities seem to be not well understood by those executing
them or those reading|observing them. Because those activities are
presented with a modicum of authority, they are repeated around the ham
community, leading to confusion (I observe).

There’s a *lot* of rubbish written about VNA calibration kits.

My own company



designs, builds and sells calibration kits - we have sold the to many
places, including the US military . I can assure you that there’s a *lot*
of junk written about VNA calibration.

Someone mentioned on here the other day about a good YouTube video. I
looked at it and pointed out it was *seriously flawed*. The author is on
this list, and between me and someone else on the list, we managed to
convince him that the advice he was given was just plain wrong.

One of my questions is this: if I cal with only one set of short and load
fixtures, then is the through valid? And does it really matter at
frequencies below, say 600 MHz?

At 600 MHz you should be concerning yourself with these issues. There are
some limitations in the NanoVNA firmware which are stopping ideal
calibrations, but the I think these will be addressed.


Most of the stuff that I do as a ham does not seem to need super precision.

OTOH, I want to use this device in the best way possible, just as I do all
my gear, including test gear. I've ordered a 2nd NanoVNA from hugen (or so
I hope), and it will come with a 2nd set of cal fixtures.

All you really need is an extra load for the isolation measurement, which
doesn’t need to be a good load.

Please understand that you don’t need to have as many calibration standards
as ports on the VNA. I think you are under the impression that you need two
opens and two shorts because you have two ports. That is simply wrong. Lots
of professional VNAs have 4 ports, and several can be configured with more
- certainly 32 port VNAs have been produced. But only one calibration kit
is required.

--
On the banks of the Piscataqua
Rich NE1EE

Dave, G8WRB
(not too far from the banks of the river Crouch ????)


--
Dr. David Kirkby,
Kirkby Microwave Ltd,
drkirkby@...

Telephone 01621-680100./ +44 1621 680100

Registered in England & Wales, company number 08914892.
Registered office:
Stokes Hall Lodge, Burnham Rd, Althorne, Chelmsford, Essex, CM3 6DT, United
Kingdom

HI Rich NE1EE -

It seems to me that we /need/ 2 of each to do a decent cal

nanoVNA supports 1-way OSLIT, where Isolation is better with 2 loads,
but otherwise a single short, two pigtails and female connector barrel suffice.
My data for CH1 with load is ~ -72dB vs -55dB unterminated @ 45MHz
CH1 shorted is about the same as unterminated..

$ nt trace
0 LOGMAG CH0 15.000000000 7.000000000
1 SMITH CH0 1.000000000 0.000000000
3 LOGMAG CH1 12.000000000 8.000000000

On Mon, Oct 28, 2019 at 12:35 PM, Dr. David Kirkby from Kirkby Microwave Ltd wrote:

Someone mentioned on here the other day about a good YouTube video. I
looked at it and pointed out it was *seriously flawed*. The author is on
this list, and between me and someone else on the list, we managed to
convince him that the advice he was given was just plain wrong.

I think that you could, without offending anyone, recommend for and "not for" various YouTube vids. It is precisely because I see what seems to be conflicting advice, here and on YouTube, written and vids, that I wrote the post you replied to. I don't wish to burden you with the rank of arbiter, but I welcome intelligent commentary.
--
On the banks of the Piscataqua
Rich NE1EE

On Tue, Oct 15, 2019 at 03:28 PM, "Dr. David Kirkby from Kirkby Microwave Ltd" <drkirkby@...> wrote:

Actually, waveguide calibration is possible too, but waveguide is not very
useful under 1.5 GHz.

Dr. David Kirkby,

Hi Dave,

I assume you're talking primarily about LRRM calibration, when saying it's possible to do waveguide calibration on a nanoVNA? Has anyone worked out a method or created a program to allow this? It appears the onboard GUI will only accept SOLT standards.

Presently actually working with LiteVNA64's, but wouldn't mind adding a nanoVNA or two to the stable, as well. Anything that helps reduce the need to move my DUT to where I can access a 4-receiver VNA with TRL capability would be of high value, and I love the extreme portability of these cheap little VNA's. About half the waveguide work I do these days is well below 1.5 GHz, where the waveguide starts looking like air-conditioning ductwork, and takes a pickup truck to move it anywhere.

Thanks!
Thomas