1

How to produce UHF/PL259/SOT239 calibration standards

for use with the WNWA

About this document:

The aim is to provide the interested VNWA lover with the tool to create a calibration kit for the UHF/PL259

connector, which everyone can make. The document is very large but the first pages from 2 to 9 are dealing with the

background for a female calibration kit made from flange adaptors readily available. On page 6 and 7 are show the

calibration settings, which is based on pure mechanical measurement of the prepared adaptor with a sliding caliper

and the delay found for the Short on page 5 is proven to be the key for measurement after calibration, as being the

delay for Ext Port 1 delay permanently enabled, when measuring to compensate for the impact from the flange

adaptors not being 50 ohm. A simulation for the fringe C of the open adaptor is also required as explained on page 6.

The report is also demonstration how I made a reference calibration kit needed for proving the concept for the

flange adaptor calibration kit and solely of interest for the calibration nerds.

Besides use the List of Content to find the pages for further detail:

List of Content:

Preface: page 2

The initial mechanical measurements and preliminary calibration setting to get started page 2

The use of a simi rigid cable with UHF/PL259 adaptors for checking the calibration page 3

Finding the Load fringe C called C|| in the calibration setting page 4

Finding the Short apparent delay and it’s derived further activity page 5

Simulation of the fringe C of the open flange adaptor, and corrections to the calibration settings page 6

The revised short delay = 44.2ps page 8

Measurement of the flange load calibration standard and summary for the calibration kit page 9

CHAPTER FOR THE NERDS ONLY page10

FEMM analyzes of fringe capacitance for a PL259 female adaptor page11

Fabrication of the male reference calibration kit and initial comments page12

Calibration setting for the reference calibration settings page13

The initial semi rigid cable test for the reference calibration kit before finding load fringe C page14

Some Smith Chart educational comments page15

Real time recalibration explanation and how used page15

Finding the Fringe C for the reference calibration kit load page16

Fabrication of the female reference calibration kit page17

The reference calibration kit female short page18

The reference calibration kit female open page19

Some extra calculations page20

Calibration setting for the reference calibration kit prior to testing page21

Finding the fringe C for the reference calibration kit female load page22

Examining the calibration obtained so far page23

Load fitted with delay and 0ps female short fabricated and used for verification purpose page25

Checking the open and short reference calibration standard after corrected calibration settings page27

The thru adaptors male and female page28

Other experiments page29

Where there is a problem there is a solution hhow to secure the inserted PTFE isolation page31

Modelling the flange adaptor calibration kit for arbitrary calibration kit without Ext. port1 delay page31

Conclusion about arbitrary calibration modelling and obtained results page43

Considerations about a male calibration kit using a male male adaptor page44

What about using an empty female flange adaptor as male open ? page45

The female female adaptor page45

A T-Check page47

Final summary page48

2

How to produce UHF/PL259/SOT239 calibration standards

for use with the WNWA

Preface:

This connector type appear to be quite difficult to produce calibration standards for, as all the commercial available

adaptors from any supplier, even the better one such a Huber Suhner, Radial and so forth, are far from being 50 ohm

Z0 impedance. Most of them are around 35 ohm and I have even measured 24 ohm. It all depends on the mechanical

dimensions and the isolation material, which has relative dielectric constants varying from 2.05 for PTFE to 5.4 for

bakelite. The center conductor for male and female connectors is not identical in diameter which also adds to

difficulties.

A general used internal diameter for all adaptors is 12mm and in some cases 10.8mm is used, and the male

centerpin has 4mm diameter whereas the female centerpin has diameters ranging from 5 to 6mm.

However do not panic it is quite simple to generate a good set of calibration standards for PL259 by a very simple

trick.

Consider using a set of flange UHF adaptors and prepare them as shown on the picture below so they have exactly

the same delay – meaning distance from the front, being the calibration plane, to the end of the adaptors. For the

below adaptors this distance is 18.7mm. If the isolation material is PTFE which it should preferably be, then the delay

is 18.7/0.3/0.695= 89.69 ps . The 0.695 is the velocity factor for PTFE and dividing by 0.3 is for free space speed of

light.

In the calibration setting for open and short we enter twice the delay as we are dealing with forward, backwards

reflections.

After calibration, we sweep a 45cm long semi rigid cable over a frequency range of 500MHz

For the Open and load standard the

center pin flush with the end of adaptor

and for the short the center pin kept long

for soldering to the shorting disk. The

drilled hole in the shorting disk for the

center pin kept slightly larger than the

center pin diameter to allow solder to

flow all the way down to the PTFE

isolation

3

We observe the blue trace along the circumference of the Smith Chart and see it is not perfect In particular see the

trace S11dB with 0.2dB/Div resolution which demonstrates the impedance is not pure/optimum reactive as it should

be.

By enabling the real time recalibration feature in the VNWA calibration settings we are able to trim the calibration

setting and we have not yet compensated for the fringe C of the Load and for the open.

4

If we just compensate for the Load C || alone, by scrolling the mouse wheel, we can get a perfect trace for -280fF

which is found by observing the S11 dB trace (red trace) until maximum flatness.

The markers are brought to the strategic “compass” points where marker 2 and 6 are in “south” position where pure

negative (capacitive) impedance of 50 ohm, marker 3 and 7 in “west” position where we have pure short, marker 4

and 8 in “north” position where we have pure positive (inductive) 50 ohm, and finally marker (1) 5 and 9 in “east”

position where we have pure open.

Is that really all ?????

No ladies and gentlemen you are fooling yourself

The perfect result is only apparent. We have already observed we did not compensate for the fringe capacitance of

the open, which is quite simple to find by doing a FEMM simulation (more about that later), which would lead to a

different value for load C||, but before doing that lets have a look if we put a pure 0ps short to the calibration plane

and run a sweep. We see on the screen dump next page it is not a 0ps short because we have not taken into account

the measured delay in a 50 ohm system of an transmission line impedance deviating from 50 ohm has an apparent

delay. The real delay is always the same independent of Z0 for the physical transmission line, remember that !!! This

is the ugly face of mismatch !!! and impedance transformation caused by the adaptors not being 50 ohm

5

If we enable the Ext. port 1 delay we find that by moving the measurement plane forward to 41.4 ps, the 0ps

apparent physical short positon is found (where IMagZ is 0 for all frequencies) which has the implication that the

apparent calibration plane is 8.65 mm behind the physical calibration plane, but we know that the physical position

is 18.7mm measured with the sliding caliper. UPS what a difference. !!!

Does that also apply for the open so we only need to use an Ext port 1 delay of 41.2ps permanently when measuring

??.

Before we test that we need to find the fringe C for the flange open using a FEMM simulation.

http://www.femm.info/wiki/download

6

Above charge difference equals to the fringe C.

1.07624-0.944337=131.9fF as 20fF=1ps we get 6.59ps which multiplied by 2 gives additional 13.19ps in VNWA

calibration setting for the open.

Note: In the FEMM software you draw in an editor the profile of the adaptor from the centerline to one side, as the

simulation is an axial simulation and has rotational symmetry. Then you declare what is center conductor and what is

ground. In addition, the isolation material defined as PTFE (Teflon) and a dome with air content defined.

The revised calibration settings with added fringe C for the open. The C|| set to 0 fF again

Then new sweep of the 45cm semi rigid cable calls for adjustment of the load C|| by observing the red S11 trace

By enabling the real time recalibration and finding the C|| value for flattest S11dB trace we get -145fF

7

8

The 0 ps short is measured again and the Ext. port 1 delay is now 44.2ps (was before 41.4ps)

By measuring a homemade well defined 44,66ps Open standard which has a FEMM simulated fringe C of 0.15226fF =

7.61pS delay, we can make the following calculation: 0ps short provided that our measurement port is 44.2ps behind

the physical calibration plane. Adding the open standard of 44.66ps plus its fringe C of 7.61ps we reach an 8.07ps in

front of the physical calibration plane. We also have to add the fringe C delay of the open flange adaptor used for

the calibration of 13.19ps, in total 21,26ps. Below we measure that we must use 23.8ps to pull the measurement

back to the calibration plane. A difference of 2.54ps correspond to a physical displacement of 0.5mm and must be

stated here as acceptable

9

Measuring the load flange standard using the same Ext. port 1 delay setting (the resistor just soldered on top of an

open flange standard) shows incredible results.

The resistance is 50 ohm, the shunt C (otherwise measured to -150fF)is in the calibration setting tuned to -145ps.

Remember we have by Ext port 1 delay of -23.8ps moved our measurement plane to the end of Flange adaptor or

said otherwise pulled the end of the adaptor back to the calibration plane, which is two sides of the same story.

Conclusion:

This very simple method described above to generate a flange UHF adaptor calibration kit with open, short and

load, where we use a permanent Ext. Port 1 delay of +44.2ps during all measurements, provides correct

measurement in our 50 ohm environment and this is a very successful result.

If all the adaptors used are prepared for exactly the same depth from front to where the short and 2x100 ohm 0.1%

resistors are fitted (the open just ends in the same distance) and the length is 18.7mm then you can use the

calibration setting as described above and the said Ext. port 1 delay of +44.2ps. A semi rigid cable test will verify if

you are successful. Later in the report there are more details about how to make these adaptors. I have secured the

PTFE isolation by 3pc. small hex screws ø2.5x3mm for each of the standards.

10

CHAPTOR FOR THE NERD’S ONLY

The two mentioned 0ps shorts and 44.66ps open standards, needed for verification of the concept, are described in

the following part of this document, where we go all the way how to generate a “Reference” calibration kit. This

section are for all the calibration nerd’s and VNWA lovers only

Removing the isolation material e.g. PTFE, the Z0 impedance can be combined by various combinations from

disassembled different adaptor be close to 50 ohm. For a short calibration standard this is not so critical as for an

open calibration standard. For a male open standard prefer to use nothing but the unterminated mating female

adaptor sitting on the end of a test cable test cable and thus simulate the fringe capacitance converted to delay.

Any attempt to use a commercial adaptor will fail, as the isolation and the characteristic impedance is far away from

50 ohm, and will cause delays much higher than expected and based on a simple length measurement as well an

anticipation of velocity factor VF which e.g. for PTFE is 0.695. Such an anticipation leads to a delay in ps by the length

in mm divided by 0.3 divided and by VF. This anticipation is only valid for a Z0 of 50 ohm. What even makes it more

difficult is that the adaptor has two or three sections having different Z0.

As well, a homemade 50 ohm load calibration standard is very sensitive to Z0 deviations away from 50 ohm in the

adaptor, and fringe capacitances is giving great deviation from the ideal situation, which will be shown. A method is

described how to find this fringe Capacitance, for the load standards, provided a well-defined and known open and

short standard is at hand.

In below description a method is shown how to create calibration standards which is accurate enough so other and

more practical creations for short, open and load can be evaluated, as already described page 1 to 8. In addition

these calibration standards can be used to study the “horrible” impedance transformation gender changer adaptor

create, e.g. from UHF to SMA, N to UHF and the like, and also to study and evaluate UHF male and female thru

adaptors used for S21 calibrations.

As mentioned the male open calibration standard is determined by simulation in the program FEMM which is more

accurate than the alternative Quickfield (http://www.quickfield.com/free_soft.htm), which besides has some other

limitation in the student version, not present in the free FEMM application. The same applies for the female

calibration standard so a FEMM simulation is preferred.

In general it is of great importance to have male and female short, open and load calibration standards accurate

enough for measuring on any other adaptor experiments, as else we are “blind folded” and not knowing what we are

doing.

This female female adaptor has 5

different Z0’s and not at all Z0=50

ohm

11

The Simulation is step 1.

FEMM analyzes of fringe capacitance for a PL259 female adaptor.

The working method in FEMM is to create an axial symmetric mechanical model (from center line to one side) and

calculate the charge where frame (ground) is at 0V and Center conductor is at 1V.

The charge in Coulombs is identical to capacitance in Farad for an applied voltage = 1V and for PTFE the entire

simulated adaptor is shown below for the rotation symmetrical model and C is 2.28864pF

Removing the “Air” outside the calibration plane allows for simulation of the capacitance behind the calibration

plane shown below and simulated to 2.18356pF. This value we need in general to calculate the Z0 for adaptors as

such, but in this particular case we need only to determine the fringe capacitance outside the reference plane, but

will for the understanding just do a calculation of the Z0 despite we do not need it yet.

Difference is 2.28864pF – 2.18356pF = 105.08 fF. For Z0=50 ohm 20fF equals 1ps delay so we can calculate the delay

as 105.08/20=5.254ps . Despite the adaptor is not exactly 50 ohm we define it to be 50 ohm at the adaptor

calibration reference plane.

For the calibration settings we use 2x5.254= 10.508ps or just 10.51ps

The capacitance behind the calibration plane is 2.18356pF and the length of the adaptor model is (random) 13.1mm

giving an airline delay of 13,1/0.3 ps = 43.666..ps. However the isolation exist and as the delay 𝑡 = 𝐶 ∙ 𝑍0 then

𝑍0 =𝑡

𝐶 . 43.666/2.18356 = 19.99 ohm !!! This is the ugly face of the UHF connectors and adaptors. If we use a

coaxial conductor impedance calculator, then based the inner and center conductor diameters we get 44.78 ohm

without any isolation, so the isolation being not PTFE but probably some sort of PE with relative dielectric constant

of 2.28 is responsible for the drop from 44.78ohm to 19.99 ohm. The exact value is 44.78/19.99=2.24 so that looks

right.

12

For more information about such mathematical calculations read the document at

http://www.hamcom.dk/VNWA/Some basic knowledge about a transmission line.pdf

As a short we use a brass disk as shown below with 0ps delay due to the brass disk pressed into level with the

calibration plane.

The male and female load calibration standards are made of a small circular PCB (with 70u Cu) and with two SMD

603 100 ohm 1% resistors soldered between ground and center pin. The male PCB is held in place by the empty UHF

male adaptor housing.

The female load is held in place on top of a 14 mm adaptor piece from a female adaptor, from which the center pin

is taken as well. The female load is first mounted on the male adaptor and then the empty male adaptor house is

used to establish connection between female PCB ground ring and the 14mm adaptor. Below Pictures demonstrates

the configurations.

The female short is made in a similar way where the female center pin is soldered into a brass disk soldered to the

back of a 17mm segment from a female female adaptor. The female center pin also half of the center pin from the

female female adaptor. See later about how the delay is calculated.

The other half of the female female centerpin is used for the female open calibration standard and cut in length so it

Is flush with the end of the 14mm adaptor (used for the female load) when inserted and pressed against the

isolation in the mating male adaptor. See later how the delay is calculated

The male short is using a housing from a short

standard male adaptor for RG58, where the

isolation is removed and the centerpin

removed from the isolation. In this case melted

away by a soldering iron as material was not

made from PTFE.

An identical empty adaptor is used for fixing the

below described load calibration standards.

13

At this point we have now established the required hardware and details / informations for calibration settings and

below shown for the male calibration kit. The open delay from the FEMM simulation and the short with 0ps delay as

such designed. In my case the load measured with a four point measuring equipment but using straight 50 ohm is

fine.

After calibration, where C || is left at 0 fF, a 45 cm long semi rigid open cable is measured and as seen, the trace is

outside the Smith Chart and indicates calibration errors.

Simple

SOL

Arbitray

OPEN

Arbitrary

SHORT

Arbitrary

LOAD

14

The trace is running outside/inside the Smith Chart indication calibration errors. It should follow the circumference

of the Smith Chart

The nine markers are placed in 4 “strategic compass” positions on the trace in relationship to the Smith Chart.

Marker 1, 5 and 9 in “east position” at the open position (and where the Real Z are very high impedance), marker 2

and 6 in “south position” where the impedance is 50 ohm pure negative imaginary (capacitive) , the marker 3 and 7

in “west position” at the Short position and finally the marker 4 and 8 in “North position”, where the impedance is

50 ohm pure positive imaginary (inductive)

15

By disabling the Smith Chart you can better study the makers and see where they are positioned on the phase, Imag

Z and Real Z traces. Quite a good idea to use this screen shot as an training in understanding the Smith Chart and an

open coaxial 50 ohm transmission lines behavior.

The only unknown is the load fringe capacitance set to 0 initially, and we now enable real time recalibration and with

the mouse scroll wheel we trim this C|| value.

16

The C|| is trimmed for smoothest S11 dB trace (observe 0.2dB resolution) Now the trace is within the Smith Chart

and run smoothly inwards just as it must be for a perfect calibration, with continuous increasing distance to the

circumference as frequency rises. This due to the cable loses in the semi rigid cable. An airline would not show this

effect.

C|| found to be 600fF and the calibration setting which now can be saved under a descriptive name.

17

The screenshoot shown again after the C || trimming and the marker position readjusted but only with small

changes

That concludes the verification, that we have a successfully created a precise set of male UHF/PL259 calibration

standards.

The Female calibration standard:

The next step is at first defining the female short standard as shown below. It consist as described of parts of a

female thru adaptor without isolation with a brass disk soldered at the end for the short adaptor. Internal depth

from reference plane to shorting disk is 17mm and for any Z0 it has a delay of 17 divided by 0.3 in ps = 56.666ps,

which must be double in the calibration setting = 113.33ps being entered, PROVIDED Z0 IS EXACTLY 50 ohm .

But unfortunately the Z0 is not 50 ohm so we have to do some simulation and calculation to compensate for this

effect when used as short in a 50 ohm environment. For the open standard to the right the depth is 13.4mm

18

The short:

Measuring with a sliding caliper the center conductor diameter d = 5.5mm and the internal diameter D = 10.8mm.

To find the nominal Z0 precisely we ca do a FEMM simulation and compare it with the result from a Coaxial

conductor calculator showing both Z0 (40.49 ohm), C and L

We can also do a calculation based on the formulas derived from the definitions of a coaxial conductor.

As 𝒁𝟎 = √𝑳

𝑪 then we have 𝒁𝟎 =

𝟓𝟗.𝟗𝟓𝟗𝟏𝟐

√𝜺𝒓∙ 𝐥𝐧 (

𝑫

𝒅) or 𝒁𝟎 =

𝟏𝟑𝟖.𝟎𝟔𝟎𝟗

√𝜺𝒓∙ 𝐥𝐨𝐠 (

𝑫

𝒅)

𝒗𝒍𝒊𝒈𝒉𝒕 = 𝒄 = 𝟏/√𝜺𝟎 ∙ 𝝁𝟎 and we get 𝒄 = 𝟐𝟗𝟗. 𝟕𝟗𝟓𝟔meters/second

The velocity factor 𝑽𝒇 = 𝟏/√𝜺𝒓 and for air the 𝜺𝒓 is 1 and 𝑽𝒇 = 𝟏

For the delay the following is valid 𝒕 = 𝒁𝟎 ∙ 𝑪 and 𝒕 =𝑳

𝒁𝟎

Another convenient expression to have at hand is the 𝐿

𝐶 ratio equal to𝑍0

2 as when C is found the L can be calculated

and vice versa. 𝑳 = 𝑪 ∙ 𝒁𝟎𝟐 and 𝑪 =

𝑳

𝒁𝟎𝟐

Let us just calculate Z0 C and L based on above formulas: Z0=40.46ohm C=82.51553pF/m and L=134.9596nH/m

The FEMM model simulation shown below. C equals to 1.36249pF

The inductance can now to be calculated from the formula 𝑳 = 𝑪 ∙ 𝒁𝟎𝟐 (the FEMM C value) being 2.364253nH

However that is not completely correct as the center conductor consist of 2.1mm of 4 mm diameter and 14.9mm of

5mm diameter. We can calculate each sections induction and add them as total L. The formula to use is:

𝑳 = 𝟐𝟎𝟎 ∙ 𝐥 𝐧 (𝑫

𝒅) The 14.9mm section=2.010898nH and the 4mm section 0.4171657nH in total 2.428063nH

The value to use in the calibration settings model is: C=1.36249pF L=2.428063nH and calculated Zx=42.21466ohm

Regarding contact resistance, it is measured to 2mohm.

The coaxial conductor calculation show for

the dimension d= 5.5mm and D=10.8mm a

Z0=40.46029 ohm As 17mm long the figures

are C=1.402764pF L= 2.294313nH and

t=56.666ps as defined.

The FEMM C value is slightly smaller due to a

slimmer section with 4mm diameter from the

mating adaptor. We may correct Z0 based on

t=Z0*C as the delay t is fixed and frozen.

Real Z0 = 40.46029*1.402764/1.36249

=41.66ohm

𝐶 = 2𝜋 ∙ 𝜀0 ∙𝜀𝑟

ln(𝐷

𝑑) Farads/meter where 𝜀0 = 8.854 ∙ 10−12 Farads/meter

𝐿 =𝜇0

2𝜋∙ 𝜇𝑟 ∙ ln (

𝐷

𝑑) Henry/meter where 𝜇0 = 4𝜋 ∙ 10−7 Henry/meter

As 𝜇𝑟 𝑎𝑠 usual is = 1 for all dielectrics we can simplify the expressions to

𝑪 = 𝟓𝟓. 𝟔𝟑𝟏𝟑𝟐 ∙𝜺𝒓

𝐥𝐧(𝑫

𝒅) or 𝑪 = 𝟐𝟒. 𝟏𝟔𝟎𝟑𝟕 ∙

𝜺𝒓

𝐥𝐨𝐠(𝑫

𝒅) pF/meter

𝑳 = 𝟐𝟎𝟎 ∙ 𝐥 𝐧 (𝑫

𝒅) or 𝑳 = 𝟒𝟔𝟎. 𝟓𝟏𝟕 ∙ 𝐥𝐨𝐠 (

𝑫

𝒅) nH/meter

19

Next step is to calculate the female open parameters

To offset variation from this one male adaptor to another ( at the end of the test cable) we use the extender

bushing in combination with a center conductor extender of length 13.4mm, which ends up at the same plane as the

bushing (see the picture on previous page to the right).

The female open calibration standard C and L values are calculated based partly of the delay determined by the

distance from calibration plane to the end of the bushing/center conductor extender, in combination with the FEMM

simulation of the C value. We calculate the C value with and without the fringe capacitance.

For the Zx determination L being determined without the fringe capacitance. and C value used including the fringe

capacitance.

The FEMM simulation

The capacitance behind the calibration plane in the left picture (assumed to be PTFE isolation but not relevant) has a

capacitance of 0.919055pF and total capacitance in the right picture without fringe capacitance is 1.99861pF. The

difference 1.99861 – 0.919055=1.079555pF is the capacitance used for calculation of Zx in combination with the L

value. Delay t is 13.4mm divided 0.3 =44.6666ps

As the complete capacitance is 2.15087pF the difference 2.15087 – 1.99861 = 0.15226pF is the fringe capacitance,

and the “internal” C value 1.079555pf is for a center conductor with two diameters, first 10mm of 5.6mm diameter

and then 3.4mm of 6.42mm diameter. Based on the formula 𝑳 = 𝟐𝟎𝟎 ∙ 𝐥 𝐧 (𝑫

𝒅) nH/meters. we calculate these two

inductances and add them together. As D=12mm the 10mm section is 1.52428nH and the 3,4mm section is

0.4253322nH in total 1.949612nH.

For the arbitray calibration we must in the model enter the physical existing value which is:

L=1.949612nH and C=1.2381pF calculation

The Z0 impedance for the “internal” open calibration standard is found using 𝒁𝟎 = √𝑳

𝑪 for C0=1.079555pF and

L=1.949612nH and we get 42.49635 ohm.

As delay 𝒕 = 𝒁𝟎 ∙ 𝑪 we find 𝒁𝟎 as

44.6666/1.07906 = 41.39405ohm

The L value based on 𝑳 = 𝑪 ∙ 𝒁𝟎𝟐 = 1.849859nH

L value to use could be 1.85nH

C value to use is (1.07955+0.15226)=1.2381pF

These values can be used to model the female

short in arbitrary calibration settings, but is not

the optimum data for the inductance , being of

two Zx sections and we can do better than this.

20

Some extra calculations.

The Fringe capacitance creates a delay which is independent of Z0 so to calculate the apparent delay in the actual “Zx

environment” of the open calibration standard we need to find the total delay and transform it to “Zx environment”

The total delay is adaptor delay 13.4mm/0.3 = 44.6666ps and fringe C delay 𝒕 = 𝒁𝟎 ∙ 𝑪 = 7.6135ps in total 52.2801ps

To resume for Female open:

L=1.949612nH C=1.2381pF to enter in arbitrary calibration settings as these are the real physical values

As tx*Zx=L=t50*Z50 the corrected and apparent delay with respect to Zx is 52.2801*(50/42.49635)=61.51ps

All required simulation and calculations made for the arbitrary calibration settings:

To resume for the female short:

L= L=2.428063nH C=1.36249pF Zx=42.21466ohm and t=56.6666ps

But before we take that step let us see why we are going to do these “advanced techniques” and while having the

VNWA calibrated with the male calibration kit, which we trust, a male male adaptor in series with the female short is

measured.

The combination runs perfectly along the circumference of the Smith Chart so why bother?

A male Male thru adaptor incl. female short with 56.66ps delay shown below where we were applying Ext Port 1

delay to try to find the delay of the combination which shows an apparent total delay 110.0ps. As can be seen the

Imag Z is not a flat response because our male male adaptor and female short is not exactly Z0 = 50 ohm .

Thus the male male adaptor seems to be 53.34ps but only for low frequencies. We will see later what the real delay

is. That is the way it is and by using the arbitrary calibration method and model for the female short and open we

can calibrate correctly.

21

We have now established all if needs for starting to create the arbitrary calibration setting and the female load

design has already been described, apart from the resistance measured to 49.92 and inserted as conductivity with a

shunt C initially set to 0 fF.

After the female calibration is performed a semi rigid cable of length 45cm with female adaptor is connected.

As seen before the trace is outside the Smith chart as with the male calibration and same procedure used with

tuning the load C|| after real time recalibration enabled.

The open

The short

The load with C|| initial set to 0fF

22

Trace running outside/too far inside relative the circumference of the Smith Chart

The C||trimmed ad determined to 1100fF

A perfect trace obtained as seen below

23

Examining the calibration obtained so far:

By running a sweep of the female short calibration standard, and setting the Ext Port 1 Delay to 56.666, we see the

sweep is as observed before and trimming the delay so it is running perfectly asymptotic from 1MHz upward, we

reach a delay of 47.84ps, exactly the ratio between short Z0 41.66 and 50 ohm lower (56.66666*42.2147/50=47.84).

As previously described the curve upwards, when frequency increases, is due to the short Z0 not being 50 ohm but

that is exactly what we have compensated for in the arbitrary calibration settings.

Just adding 1 ps and we are spot on

24

By running a sweep of the female open we observe the same phenomena and setting the ext pot 1 delay to the

calculated apparent delay 61.51ps 4 pages backward, we observe almost as expected running flat from the low

fequencies upward.

Just adding 1ps and we are correct

Another verification that we also have a quality UHF/PL259 female calibration kit. The addition of 1ps might just be

derived from the inaccuracy in the sliding caliper and are totally acceptable (1ps correspond to a displacement of

0.15mm of the reference plane or from the fact that we have not taking into account that the load is sitting behind a

13.4mm transmission line.

25

Load fitted with delay and 0ps female short fabricated and used:

Center conductor has a length of 12.6mm with diameter 5.5mm and 0.8mm length with 4 mm diameter total

13.4mm.

C= 1.09648pF L = 5.5mm section=1.966nH + 4mm section = 0.17578nH. Total = 2.142nH These values are used in

arbitrary calibration settings.

As seen before complete wrong calibration with the 45cm semi rigid cable fitted as C|| set to 0fF

Next we trim the C|| value

C|| 780fF found

26

The S11 dB trace is extremely flat and linear. This proves we have got a complete control of all the elements to

handle in the arbitrary calibration. For this exercise we have used a female short calibration standard of 0ps length.

The trick was to find an male adaptor with a chest on level with the calibration plane onto which a shorting disk

could rest. A special adaptor was made which could put pressure in the center of the shorting disk (made of a piece

of single sided PCB with 70u Cu) The shorting disk better made in e.g.0.5mm bronze to flex slightly

C||

27

Running a sweep of the open standard and entering the calculated apparent delay in ext. port 1 delay and bang on!

Running a sweep of the short standard and entering the calculated apparent delay in ext. port 1 delay and bang on!

This is indeed very satisfactory and the entire project successfully completed……….

28

The Thru adaptors:

While still having the VNWA calibrated with the female calibration kit it is time for determination of a female female

adaptor for S21 calibration.

Female Female adaptor in series with 0ps male short. The curve upwards after 100MHz due to the deviation from 50

ohm for the Female female adaptor. A touchstone file was saved for later analyzes and parameter determination.

Other experiments performed

A set of female short, open and load calibration

standards, based on shown flange adaptor, was made

after the centerpin grinded down the end of the

adaptor. The only one which “survived” was the short

as after a few mounting and removal the isolation

was pulled out. As shall be shown below the isolation

has a very negative effect on the open (and load also

but is not documented) as the Z0 is not 50 ohm and

serious impedance transformations take place.

The short is useable as below measurements

demonstrates, but the open is quite hopeless

29

Female Flange load as above shown

Female Flange Short has a delay of 48ps and curves upwards slightly at 400MHz (remark 0.1ohm/div)

30

Next follows the terrible flange open

If the Z0 had been 50 ohm the delay would be based on length 18.5 mm and PTFE 18.5/0.3/0.695 ps= 88.75ps +

endpoint radiation of about 4.5ps in total 93ps but measured to be 167.85ps and considerable increasing above

100MHz. It indicates a Z0 in the region of 50*93/167.85 = 27.7ohm

Using the application ZPlots by AC6LA it is possible to subtract a transmission line (a delay) for a VNWA

measurement. Here the length 16.7mm used (should have been 18.5) and VF for PTFE 0.695. The Nom. Z0 entered

as 24ohm to simulate flat phase up to more than 100 MHz as in the measurement above. Due to the shorter length

we arrive at 24ohm so if 18.5mm had been entered we would have arrived at about 28 ohm.

Conclusion: Such an open is not useable. It requires saving of a Touchstone file and analyzes by the VNWA Tool

optimizer to figure out the delay and C coefficient for arbitrary calibration method. However it requires to have the

precise short, open and load calibration standard to measure the S11 plot so some work ahead.

31

Where there is a problem there is a solution

I found a way to secure the PTFE insert by drilling 3 x 2mm holes in the rim (when the PTFE inserts removed) and cut

2.5mm threads and use 3 pc 3x2.5mm hex screws to prevent the PRFE insert to turn or move.

With the VNWA calibrated with the Reference calibration kit these adaptor was measured and result saved as

Touchstone files. With the VNWA tool “Optimizer” a model created for each of them and the C and L coefficients

were derived for the short and open. For the load a model was also created and the parameters derived.

Such models are used by the professional VNA’s where a delay and L and C values with coefficient and frequency

dependency are entered in calibration settings.

Below is the derived data and for the short, load and open, also presented as traces, to verify the simulated trace for

s_11 is precisely on top of the S11 trace meaning a perfect match. See also the data for the 9 markers and how

exactly the fitting is performed by the Optimizer. It shall however be seen that when using calibration standards

derived from adaptors not being 50 ohm that these arbitrary calibration simulations create problems. The model is

the delay transformed to Z and added to the inductance expression and the sum transformed back to s parameter

32

Please observe that for the short the models do not include any RealZ and thus 0 ohm. As less than 20 mohm is not

significant and was not taken into consideration. We are not dealing with rocket science

Else the fitting seems OK.

Regarding the load standard we obtain precise fitting and the method OK to use be used. The model consist of the R

value in parallel with the fringe C multiplied by the delay.

The figure of merit is very low proving a good fitting

33

The “ugly” open is not well suited to model the described way, as the fitting requires to manually adjust the delay

keeping it from being involved in the fitting process and by several attempts (by changing the delay scrolling its

value by the mouse wheel) an acceptable fitting was found.

Figure of merit is not low enough for a perfect fitting

34

above seen the poor fitting …

The manual modified delay prevented a calculation of useable figure of merit

35

As demonstrated we are in trouble with this method as the derived delay actually is the apparent delay so a better

method is to avoid delay in the model and only use R, C and R components. We just add C, L and R in series

Figure of merit superb. The fitting of the R value is also very good bearing in mind its influence is insignificant

36

Net result is that with these data entered into arbitrary calibration setting the flange short open and load are

excellent calibration standards. The optimizer does a perfect job and it is not possible to see the under laying trace

and figure of merit is low as it should be.

A repetition of fitting avoiding delay for the short and load are done below just to investigate if improving the

overall over result

37

Model used is L, C and R in parallel and figure of merit just fine

38

For the load we are using a more sophisticated model consisting of a L C transmission line modelled as series

connection of L and C transformed to transmission with z2t (the section from reference plane to where the load

resistors are mounted) and added to the load R shunted with fringe CF transformed to transmission with s2t. The

sum transformed back to S with t2s.

Figure of merit excellent and so is the fitting as seen below.

Just observe the numeric data how perfect the fitting is

39

All over the task to produce calibration standards for the UHF/PL259 is definitely possible and done with success

The models avoiding the use of delay are absolutely to prefer:

Loading the found data into a calibration file and running a Sweep for the 45cm semirid cable will prove the validity

and so done by calibrating with the Reference calibration standards and change the flange Open short and load

setting one at a time.

First the open flange calibration standard

Calibration setting for open according to the L, C model

Result of sweep after recalibration with semi rigid cable OK S11dB saved in Plot1

40

Then the Short after L C model

Calibration setting for short

S11 dB deviates slightly from Plot1 when running semi rigid test after recalibration.

Enabling the realtime recalibration allows us to analyze the reason for the slight deviation which can be corrected by

a delay of 5ps as seen on next screen plot

41

Next for the flange load the L C R did not gave good results (not documented) whereas the first method with delay

gave excellent result. Remember the delay much be the complementary delay meaning negative.

42

The S11 dB trace so to speak unchanged

Changing the C value from 2.5124pF to 2.4124pF (100fF) bring the S11dB trace to perfection

43

However the measurement and the creation of the document took so long time that the open calibration has drifted

so a recalibration brought following (amazing) result. All the calibration setting 100% according to the fitting done by

the Optimizer and a perfect result obtained.

Conclusion:

By careful attention to all details it seem feasible to produce a reproducible UHF calibration kit based on standard

flange adaptor provided purchasing the same adaptors from same supplier PTFE (Teflon) isolation and strictly follow

the guidelines for producing the adaptors. The background for being able to do so it the development of the

“Reference” calibration kit, as else the derivation of the L C R and delay data is not possible. Likewise, it is essential

to possess a semi rigid test cable (See below).

44

What remains is a simple male calibration kit and that might be done nicely with a combination of a male male

adaptor the female calibration standards and let the Optimizer do the hard job to find the data to use for arbitrary

calibration. The procedure is exactly the same as already described for the female kit

What about using an empty female flange adaptor as male open ?

At first we measure the male adaptor at the end of the test cable in unterminated condition, just leave it as open.

Below us shown the phase plot and it deviates slightly above 200Mhz. As there will be variation from adaptor to

adaptor and the outer fastener will slide in random positions it is not a reliable method. The Z0 is higher than 50

ohm as well.

The solo PL259 male adaptor as open with apparent delay of 30pS which can be used as the adaptor is in calibrated

50 ohm condition

Adding an empty female flange adaptor or an empty female female adaptor does not change the situation but just

adds a further 5.4ps delay, but the male center pin length varies from one adaptor type to another, and creates

uncertainties.

45

The PL259 male adaptor as open covered by 27mm long empty female female adaptor or an empty female flange

adaptor has an apparent delay of 35.5ps.

The female female adaptor

To characterize the female female adaptor we measure it terminated with the 0ps short and model it with optimizer

followed by a fitting. The L and C value to be used for entering in the arbitrary calibration for thru. The R value that

high that not relevant to implement. I even did a custom trace simulation so all three traces for Imag Z on top of

each other.

46

Excellent fitting as seen below. VNWA calibrated with the female flange calibration kit.

L= 4.333098631737E-009 Henry and C= -1.488767741539E-013Farad is to use and using the formula 𝑡 =𝐿

𝑍0 we

calculate the delay to 86.66ps and can be used in the simple model of the calibrations setting. The C value correct

first at very high frequencies.

Initial setting without usage of C value

47

A T-Check demonstrates very good calibration as below 2%.

Adding the impact of C the calibration setting changes to the following

86.66e-12 is the delay and y2s(-i*w* 1.488767741539e-013) is the s parameter for the C value.

The T-Check now below 2% all over and excellent

48

A Warning: Do not expect to see your own T-Check as fine as above. The load used at the third leg of the T adaptor

must be pure 50 ohm, and the T adaptor itself must also be first class 50 ohm and that is not likely yours are in the

class. I did spend much time to investigate my T-adaptor and the load for such results as shown above. If running up

to 15% at 500 MHz is still an OK calibration.

Final summary:

The described method for fabrication of calibration standard is providing reliable calibrations. It is of course quite

time consuming to go all the way as described. The home made female flange is just fine provided you use a

permanent Ext. port 1 delay of 44.2ps if your adaptors are identical to mine. Else the method to find yours delay is

described.

I have just ordered some male and female PL259 to SMA adaptors where I am able to characterize with the VNWA

tool “Optimizer” based on a Touchstone file saved for measurements of these adaptors terminated by e.g.

Rosenberger SMA calibration parts.

When these adaptors arrive I will make a reports on the results.

July 27 2015

Kurt Poulsen de OZ7OU