1. Transmission Line Studies Need to understand data transmission at 640 Mbps along bus tape. Analytic formulae FEA calculations from Roy. Network analyser measurements from Roy Possible improvements? WP4 February 20161

2. Transmission Lines - 101 Two main sources of attenuation in transmission lines – Dielectric loss: parameterised by loss tangent . Attenuation: typically ~ 1 dB for our application. – Resistive loss: depends on skin depth WP4 February 20162

3. Carbon Fibre Poor conductor at DC R~ 1  /square. Should get better at high frequency because fibres can couple capacitively if they are close but not actually in contact. Need to measure R vs . Feed this into FEA? WP4 February 20163

4. Other Sources Attenuation Surface roughness: – Increases effective path lengths for currents if undulations comparable to skin depth. – Could increase attenuation ~ 20%. Radiation – Should be small at our frequencies? WP4 February 20164

5. Back of the Envelope Results Consider 1.3m long tape at f=320 MHz (640 Mbits/s). Dielectric loss in range 0.4 to 1.0 dB Resistive loss harder to calculate What is effective width w for return current? Large losses ~ 10 dB. Need FEA  Roy’s results with ANSYS. W Cu Polyimide Cu WP4 February 20165

6. ANSYS 2D Field Solver Calculate E and B fields  C and L  Z0 Stripline geometry WP4 February 20166 E

7. 7 B

8. Impedance vs frequency: ATLAS tape Single ended: Z0 ~ 25  Aimed for 50   expect large reflections and bigger losses. Frequency 1 GHz WP4 February 20168

9. ATLAS Tape 1000 mm 25  source impedance 100  m track 81  m substrate 18  m track thickness Single ended calculations up to now. Transmission & Reflection vs frequency Frequency1 GHz WP4 February 20169

10. S parameters with carbon fibre. Conductivity adjusted to give 1  /square Need more realistic model of carbon fibre? WP4 February 201610

11. 50  track 1mm wide 4.018mm substrate 1000 mm long WP4 February 201611 Good Transmission Line example

12. Current Density Current asymmetric because of geometry WP4 February 201612

13. Stripline Add Cu ground layer to replace carbon fibre Single ended losses still very large … WP4 February 201613

14. Options to improve transmission Avoid losses in carbon fibre by adding extra Cu/kapton screen layer. – Lamination done, tape to be tested. – Unlikely to be sufficient. Wider tracks – 200 um tracks – Need extra 120 um to control Z0  0.1% X0 per stave Middle of stave – Reduce length for longest data line. WP4 February 201614

15. Wider Tracks: Impedance ANSYS: 2D field solver for E and B  C and L  Z0. Z0 = 32 . 200  m track and gap. 91  m dielectric above and below strip  separation between Cu ground planes of 200  m. WP4 February 201615 Current distribution Impedance vs frequency Current

16. Wider Tracks: Attenuation Attenuation 5.9 dB @ 320 MHz WP4 February 201616

17. Testing Set-up Bus tape on wooden board Bus tape grounded to scope Agilent 86100C TDR used to measure S- Parameters 17WP4 February 2016

18. Measurement configurations noCF = nothing on back of tape CF = carbon fibre laminated on back of tape With ground = carbon fibre connected to shield layer with copper braid 18WP4 February 2016

19. 19WP4 February 2016

20. 20WP4 February 2016

21. CF m6h1 with ground 21WP4 February 2016

22. CF m6h1 with gnd, 3dB pre-emphasis 22WP4 February 2016

23. 23WP4 February 2016

24. CF m0h0 24WP4 February 2016

25. CF m0h0, 7dB pre-emphasis 25WP4 February 2016

26. CF m6h1 with gnd, 3dB pre-emphasis 26WP4 February 2016

27. 27WP4 February 2016

28. No CF m0h0 repeat 28WP4 February 2016