“Plans for extending NPL’s
measurement capabilities using
dielectric waveguide impedance standards”
Nick Ridler1 and Richard Collier2
1NPL, and 2University of Cambridge
Contact: [email protected]
3rd Millimetre-wave Users Group meeting, Friday 4th April 2008
Overview
•
•
•
•
•
•
Why dielectric waveguide?
The story so far
Extending to higher frequencies
Extending to two-port measurements
The key issues
Summary
Why dielectric waveguide?
Compared to metallic waveguide at millimetre wavelengths:
¾ Better Connections – more repeatable; lower loss
• can be non-contacting (to some extent)
• junctions can be inspected visually
¾ Standards easier to characterise (using mechanical measurements)
• defined by external guide dimensions (height & width)
¾ Better electromagnetic properties
• lower dispersion
• lower loss
The story so far . . .
Prototype VNA measurement system developed
– 75 to 110 GHz band
– One-port capability
The system used:
– Dielectric waveguide test port reference plane
– Dielectric waveguide calibration standards
The story so far . . .
Research highlights included:
• Choice of dielectric material
• Design of metallic-to-dielectric waveguide transition
• Design of calibration standards
• Automated calibration
• Typical uncertainty of measurement
Choice of dielectric material
0
Six materials considered:
-1
S 21 (dB)
PEEK – very lossy
PTFE & HDPE – low loss,
but too flimsy
Rexolite & TPX – rigid,
but ‘high’ loss
PP – rigid and low loss
-0.5
-1.5
-2
-2.5
PP
Rexolite
-3
75
80
HDPE
TPX
85
90
95
100
105
Frequency (GHz)
Therefore, polypropylene (PP) was chosen as the dielectric
110
Dielectric waveguide transition
Various taper designs considered:
0
No Taper
(a)
(b)
(c)
(d)
(e)
-10
(b)
-20
S 11 (dB )
(a)
-30
-40
(c)
(d)
-50
-60
75
80
85
90
95
Frequency (GHz)
(e)
(a) H-plane asymmetric worked the best
100
105
110
Calibration standards
Short-circuits, offset by lengths of dielectric waveguide:
@ 75 GHz
@ 110 GHz
110 GHz
75 GHz
1
1
0.8
0.8
0.6
0.6
0.4
2.435 mm
0.2
3.632 mm
0
-1
-0.8 -0.6 -0.4 -0.2
0
-0.2
-0.4
0.2
0.4
0.6
0.8
1
4.561 mm
5.100 mm
5.590 mm
0.2
2.435 mm
3.632 mm
0
-1
-0.8 -0.6 -0.4 -0.2
0
-0.2
-0.4
-0.6
-0.6
-0.8
-0.8
-1
-1
Real
0.000 mm
0.000 mm
Imaginary
Imaginary
0.4
Real
0.2
0.4
0.6
0.8
1
4.561 mm
5.100 mm
5.590 mm
Automated calibration
Computer-driven motorised standards wheel
Containing up to six standards
Standard
waveguide
horn
Dielectric taper
inside this
uniform section
of waveguide
Test Port
VNA
Offset short
circuits
Uncertainty of measurement
Uncertainty estimated from:
•Random errors (connection
repeatability)
•Systematic errors (cal stds
imperfections)
Typical VRC uncertainty ≈ 0.006
(44 dB return loss), 75 to 110 GHz
Uncertainty ‘surface’ for
3-standards calibration
Extending to higher
frequencies
Waveguide bands under investigation are:
¾
90 to 140 GHz (WR-8 / WG28)
¾
110 to 170 GHz (WR-6 / WG29)
¾
140 to 220 GHz (WR-5 / WG30)
Extending to two-port measurements
¾ Measure all four S-parameters
¾ Magnitude & Phase
¾ Use two-port VNA
¾ Dielectric waveguide test ports & standards
The key issues
• The size of the waveguide (getting very small)
• Availability of VNA systems (there’s not many
around)
• Choice of cal routines – extend to two-port
• Choice of cal standards – include two-ports
Waveguide size
Aperture dimensions are
getting small . . .
Size
Width (mm)
Height (mm)
WR-08 / WG28
2.032
1.016
WR-06 / WG29
1.651
0.826
WR-05 / WG30
1.295
0.648
This means that waveguides are more difficult to make accurately
But, since we are interested in external dimensions, these can be
measured (relatively) easily
Therefore the uncertainty due to dimensional imperfections in the
standards can be determined (relatively) easily
VNA systems
•There are not many systems to
choose from:
–OML (www.omlinc.com)
–AB Millimetre
(www.abmillimetre.com)
•NPL doesn’t have any systems at
these frequencies
•Therefore, NPL is looking to
collaborate with owners of such
systems . . .
VNA cal routines
Two-port calibration routines
¾ ‘SOLT’ – extends the ‘three-known-loads’ (SOL) technique using
a known Thru connection
(Our ‘three known loads’ are three offset short-circuits)
¾ ‘SOLR’ – extends the ‘three-known-loads’ technique using an
‘unknown’ Thru (needed only to be a reciprocal device)
¾ ‘TRL’ and ‘LRL’ – uses only a Thru (in TRL) or a Line (in LRL) as
the known standard
VNA cal standards
Standards need to be known either fully or partially:
Standard
SOLT
SOLR
TRL
LRL
(3 known loads)
(3 known loads)
Offset
short-circuits
Fully
Fully
Partially
Partially
Thru
Fully
Partially
Fully
-
Line(s)
(Fully,
if used)
(Partially,
if used)
Partially
Fully &
Partially
Summary
• A plan is now in place to develop dielectric waveguide standards
for mm-wave applications
• This work will include
¾ modelling of standards
¾ practical VNA experiments
¾ VNA algorithm development
• NPL will need to collaborate with an owner (or, owners) of VNA
systems operating at these frequencies
• NPL welcomes such a collaboration as a knowledge generating,
and knowledge sharing, exercise
Bibliography . . .
“Traceable high-precision impedance measurements for the millimetre-wave band using
dielectric waveguide transmission lines”
J G M Yip, N M Ridler, R J Collier and M-H J Lee
IEE Proc – Sci Meas Technol, Vol 153, No 6, pp 235-240, November 2006.
“New impedance measurement system using dielectric waveguide for the millimetre-wave
region”
J G M Yip, R J Collier and N M Ridler
CPEM 2006 conf digest, pp 282-283, Torino, Italy, 9-14 July 2006.
“Towards a new form of national impedance standard for millimetre wavelengths using
dielectric waveguide”
J G M Yip, M-H J Lee, N M Ridler and R J Collier
BEMC 2005 conf digest, pp 21-22, NPL, Teddington, 14–17 November 2005.
(Also published at the ARMMS Meeting, 24 & 25 April 2006, Milton Hill House Hotel, Steventon, Oxfordshire.)