The Impact of Metal Trays on Unshielded Data Cables

Transcription

1 The Impact of Metal Trays on Unshielded Data Cables Andreas Barchanski

2

3

4 Less Weight 10 Gbit/s

5 Cable in a Metallic Pipe

6 1 Gbit/s achievable? 100 m reference length unpractical

7 Model Curves

8 Cable Layout

9 Cable Setup: Insulator Outside

10 Cable Setup: Cross

11 Cable Setup: Wire Pairs Wire-pairs are unshielded, all different in lay lengths and two different pairs of wireisolation diameter Twist length Wire-Isolation Ф CBLS Name blue: 9.06 mm 1.03 SW_0, SW_1 orange: mm 0.99 SW_2, SW_3 green: 9.65 mm 1.03 SW_4, SW_5 brown: mm 0.99 SW_6, SW_7

12 10 m Cable Cable Setup for a 4Port VNA: converters are used to split up into common and differential ports 10m results: The blocks are added to the same schematic for comparison reasons

13 Differential Impedance dd Open/short Zdd= (Zo*Zs) bl=gr 1.03 mm isolator or=br 0.99 mm isolator bl or S-parameter (Zo*(S11-1)/(S11+1) blue orange

14 Differential Impedance Open/short Zdd= (Zo*Zs) bl=gr 1.03 mm isolator or=br 0.99 mm isolator blue orange brown

15 Differential Insertion Loss blue orange Using a material extraction procedure the loss tan-d was predicted between 1e-3 and 2.e-3 (dispersive behaviour). Tan-d was adjusted such that it fits to measurements (calibration)

16 Differential Return Loss measurement results at higher frequencies show a kind of low pass behaviour

17 FEXT / NEXT FEXT blue Interesting simulation-results: FEXT level and trend is ok NEXT is much too idealistic! NEXT blue (too idealistic)

18 Transverse Conversion (Transfer) Loss blue orange TCTL: far (too idealistic) blue orange TCL: near

19 Overview Z IL RL FEXT NEXT TCL TCTL just cable

20 Test: Artificial Unsymmetry εr=1 εr=3 NEXT A wire pair is assigned to a smaller dielectric constant than others. At an eps-ratio of 3 a significant change could be observed, which is unrealistic FEXT Distance between wirepairs is reduced and didn t show any effect

21 Test: Coupling by a Transformer A slight imbalance is introduced by a transformer coupling between wire-pairs

22 Differential Return Loss As expected, the small inductances act as a low pass filter leading to a worse matching at higher frequencies

23 FEXT Just cable The introduction of the transformer shows no effect on FEXT +transformer

24 NEXT Just cable The impact can be seen nicely on NEXT +transformer

25 Transverse Conversion (Transfer) Loss The introduction of a transformer shows no effect on TCL and TCTL blue orange

26 Overview Z IL RL FEXT NEXT TCL TCTL Just cable + transformer

27 Setup

28 Deembedding Introduction of a parasitic inductance of approx. 2nH, 6mm length of a wire Deembedding using a network

29 Deembedding Compensating the inductance effect (negative L) Before Compensation After Compensation

30 Conical Transition in CST-MWS

31 10 m Cable As described before, the actual cable is embedded into the transitions and the serial inductances to verify the measurements.

32 Differential Return Loss blue blue Without L (-2nH) orange

33 NEXT Just cable Correctly modeling the measurement set up results in excellent agreement of simulation and measurement +connector

34 Transverse Conversion (Transfer) Loss blue orange TCTL: far (too idealistic) blue orange TCL: near

35 TCTL/TCL blue orange TCTL: far blue TCTL and TCL: still too idealistic! orange TCL: near

36 Overview Z IL RL FEXT NEXT TCL TCTL Just cable + transformer + connector model

37 TCTL/TCL blue orange TCTL: far blue Another mechnism included? orange TCL: near

38 Imbalance The imbalance shows it even better: there is a too ideal balance inside the single wire-pair. (imbalance: ratio S1,5/S1,6)

39 Introduction of a Twist-Center Offset Twist center 0/0 Twist center 0/0.2 The twist center is offsetted by 0.2 mm and the TCTL is greatly improved

40 Introduction of a Twist-Center Offset Since the offset =0.2 mm was too high, its value can adjusted such that it meets the measuremnet results. The offset value is surprisingly very small, only 5 µm! blue blue TCTL TCTL TCL TCL

41 Imbalance The balance is related to TCTL and TCL, so here an improvement can be seen as well! blue Magnitude / db blue Phase / deg

42 Overview Z IL RL FEXT NEXT TCL TCTL Just cable + transformer + connector model + imbalance

43

44 s Copper-Pipe with outer diameter =12mm and a wall thickness of 1mm

45

46

47 Differential Return Loss blue orange

48 Differential Insertion Loss blue orange bl or

49 NEXT / FEXT NEXT FEXT blue

50 blue TCTL/TCL orange TCTL: far blue orange TCL: near

51 Differential Return Loss Da=9mm Da=12mm blue The cube is touching the cable, representing the worst case scenario orange

52 Differential Insertion Loss blue orange

53 NEXT / FEXT NEXT FEXT blue

54 TCTL/TCL blue orange TCTL: far 9mm 12mm 12mm 9mm 12mm 12mm blue orange 9mm 12mm 12mm TCL: near 9mm 12mm 12mm

55 IL comparison

56 RL comparison

57 Summary Model curves to quantify cable properties Fully parametrized cable model From step by step matching simulation and measurement Influence of the surroundings Excellent prediction