1. Interpretation of resonant wire measurements on the TCSPM
N.Biancacci
ABP impedance meeting
13 September 2017
Acknowledgements: F.Caspers, G.Mazzacano, E.Métral, B.Salvant

2. Outline Introduction Resonant wire measurements on TCSPM
Analysis of Impedance
Conclusions and outlook

3. ICFA Impedance Workshop, 20 September 2017
Introduction
Resonant wire measurements
VNA
DUT
Collimators are not easy to be measured with classical stretched wire method.
Matching cannot be ensured for jaw and tapered sections.
The change in gap change the line impedance as well.
Resonant methods are “automatically” matched at the resonance frequencies.
Losses can be inferred from the Q factor of the resonances [*].
Works well for broadband resistive wall losses, i.e. far from resonances.
[*] Details in T.Kroyer, F.Caspers, E.Gaxiola CERN AB-Note
ICFA Impedance Workshop, 20 September 2017

4. ICFA Impedance Workshop, 20 September 2017
Introduction
TCSPM: HL-LHC prototype collimator with three stripes of different material on the main jaws.
𝜌 𝑐 =0.4 𝜇Ω𝑚
𝜌 𝑐 =53 nΩ𝑚
TiN Mo
𝜌 𝑐 =1𝜇Ω𝑚
MoGr
The resonant wire method was applied to the TCSPM in order to check
if we can distinguish different losses on different stripes.
if the Molybdenum effectively reduces the impedance.
Measurements done for different jaw gaps and wire horizontal displacements.
ICFA Impedance Workshop, 20 September 2017

5. Resonant wire measurements on TCSPM
S21 and S11 raw data.
Low coupling through the capacitors
Resonances in transmission multiples of the device length.
Lowest resonance at 76 MHz, i.e. 2m long DUT
ICFA Impedance Workshop, 20 September 2017

6. Resonant wire measurements on TCSPM
x displacement
Worked out the measurement post-processing in detail:
Accounted for gap-varying characteristic impedance
Fitted resonance with Lorentzian as only few points for the 3dB bandwidth determination
𝑍 𝑐 =60 ln full gap ∅ 𝑤𝑖𝑟𝑒

7. Resonant wire measurements on TCSPM
Half gap: 5mm
TiN
MoGr Mo
TiN
MoGr Mo
Q factor variation vs wire position at two different frequencies
Higher Q -> Lower losses
Already from the raw Q’s it is counter-intuitive the behavior of MoGr w.r.t Mo

8. Resonant wire measurements on TCSPM
Half gap: 5mm
TiN
MoGr Mo
TiN
MoGr Mo
Q factor variation vs X position at two different frequencies
Higher Q -> Lower losses
Already from the raw Q’s it is counter-intuitive the behavior of MoGr w.r.t Mo
This is also evident looking at the reconstructed impedance.
Are these Q’s representative of resistive wall losses?
→ We decided to study the variation of impedance vs gap as 𝑍 𝑙 ∝ 1 ℎ𝑎𝑙𝑓 𝑔𝑎𝑝

9. Impedance vs gap at x=0 mm displacement
Analysis of impedance
Impedance vs gap at x=0 mm displacement
As measurements were done also at larger gaps (up to 30mm):
Fitting the impedance behavior Vs half gap, we can select only those measurements “best fitted” by
𝑍 𝑙 ∝ 𝑎 ℎ𝑎𝑙𝑓 𝑔𝑎𝑝 +𝑏
The coefficient 𝑎 is then considered representative of on jaw resistive wall losses.

10. Impedance vs gap at x=0 mm displacement
Analysis of impedance
Impedance vs gap at x=0 mm displacement
1.2 GHz
938 MHz
591 MHz
67 MHz
NB: Method assumes to be far from resonances!

11. Impedance vs gap at x=0 mm displacement
Analysis of impedance
Impedance vs gap at x=0 mm displacement
1.2 GHz
938 MHz
591 MHz
67 MHz

12. Analysis of impedance ~0.5 Ω ~1.5 Ω Example at 1.2 GHz
TiN
MoGr Mo
Example at 1.2 GHz
Close to resonance -> Bad fits.
Dependence not only on 1/gap -> additional losses on top of the ones due to the jaw broadband impedance.
Dependence of impedance vs position reflects additional losses.

13. Impedance vs gap at x=0 mm displacement
Analysis of impedance
Impedance vs gap at x=0 mm displacement
1.2 GHz
938 MHz
591 MHz
67 MHz

14. Analysis of impedance x displacement ~0.5 Ω ~1.5 Ω Example at 962 MHz
TiN
MoGr Mo
Example at 962 MHz
The impedance is best fitted by 1/half gap dependence.
Mo is clearly showing less resistive wall impedance.

15. Analysis of impedance x displacement ~0.5 Ω ~1.5 Ω 0.4 1.4
TiN
MoGr Mo
Example at 962 MHz
The impedance is best fitted by 1/half gap dependence.
Mo is clearly showing less resistive wall impedance.
The relative difference is in good agreement with respect to IW2D .

16. Impedance vs gap at x=0 mm displacement
Analysis of impedance
Impedance vs gap at x=0 mm displacement
1.2 GHz
938 MHz
591 MHz
67 MHz

17. Analysis of impedance x displacement Example at 591 MHz
TiN
MoGr Mo
Example at 591 MHz
The impedance is best fitted by 1/half gap dependence.
The difference is above the measurement resolution

18. Analysis of impedance x displacement 0.7 0.7 Example at 591 MHz
TiN
MoGr Mo
Example at 591 MHz
The impedance is best fitted by 1/half gap dependence.
The difference is above the measurement resolution
The relative difference is in good agreement with respect to IW2D .

19. Impedance vs gap at x=0 mm displacement
Analysis of impedance
Impedance vs gap at x=0 mm displacement
1.2 GHz
938 MHz
591 MHz
67 MHz

20. Analysis of impedance x displacement Example at 67 MHz
TiN
MoGr Mo
Example at 67 MHz
The impedance is best fitted by 1/half gap dependence.
The difference is within the measurement resolution

21. Analysis of impedance Example at 67 MHz
TiN
MoGr Mo
Example at 67 MHz
The impedance is best fitted by 1/half gap dependence.
The difference is within the measurement resolution
Expected as well from IW2D (0.1/0.2 Ohm difference)

22. Simulations
Simulated the resonant wire setup to asses the effect of a possible tilt in the wire sweeping path -> little effect!
Angle wire = 0deg
Angle wire = 5deg
x [mm]
x [mm]
0 deg
5 deg

23. Conclusions and next steps
Analyzed the TCSPM resonant wire measurements in detail.
Few points over the resonances -> fit to get Q’s. Not very good practice, needs for automatized acquisitions (VBA scripts available in BE-RF VNAs)
Conversion as-it-is from Q to Z gives results inconsistent with theory: other kind of losses are gathered into the measured Q factor overcoming the effect of the impedance.
Accounting only for the impedance which exhibits a 1/half gap dependence, a reasonable agreement with expectations and IW2D is reached.
Impedance dependence over position is more complicated close to resonances: we cannot infer only the broadband impedance there.
Next steps:
Simulations of the resonant method for a given gap show the expected impedance dependence on horizontal position but:
Much higher resolution in simulations than measurements (ideal Q determination)
Only losses on the jaws (perfect contacts)
To be done:
Full sweep on frequency to check the impact of resonant modes.
To check the impedance dependence versus gap as done in measurements.
And why not… if the collimator will come back we could redo some nice measurements!