Limitations And Accuracies Of Time And Frequency Domain Analysis Of Physical Layer Devices

Transcription

1 Limitations And Accuracies Of Time And Frequency Domain Analysis Of Physical Layer Devices Outline Short Overview Fundamental Differences between TDR & Instruments Calibration & Normalization Measurement Accuracies Measurement Comparisons Page 2 Accuracies of Time & Freq Domain PHY Meas 1

2 TDR and N1930B Physical Layer Test Software DCA and 54754A TDR Modules E8364B PNA with 2-port Testset Page 3 TDR Set-up (Configuration Showing 1=>3, 2=>4) Channel 1 Channel 2 Channel 3 Channel 4 BTL Board Thru Adapters Page 4 Accuracies of Time & Freq Domain PHY Meas 2

3 Appreciating the Complexity of it all Device Length Frequency Range Accuracy Repeatability Reciprocity Normalization SOLT Device Complexity Voltage & Temp Drift Calibration Risetime Number of Points Time Step Averaging Time Base IF BW Instrument Architectures Source Error Source Drift Rcvr BW Source Stability Noise Floor Dynamic Range Signal-to-Noise Ratio Filter Roll-off Page 5 Fundamental Differences between TDR & Instruments Measurement Domains Sources Receivers Architectures and Sources of Errors Calibration and Measurement Summary of how TDR and Measurements Differ Accuracies of Time & Freq Domain PHY Meas 3

4 Time and Frequency Domains All frequencies make up each time point Device Under Test t 0 TDR Time Domain t 1 =delay Device Under Test t m Δφ=2πf*time delay deg Frequency Domain t m = phase offset nsec Phase Group delay = dφ/df = time delay Group Delay Frequency Page 7 Jitter in Time Domain is Phase Error in the Frequency Domain phase(radians) =2*pi*f*delay Page 8 Accuracies of Time & Freq Domain PHY Meas 4

5 C2 CH1 S 11 MARKER GHz 1 log MAG 10 db/ REF 0 d B GHz START GHz STOP G Hz 1 _: db CH2 S 21 log MAG MARKER GHz 1 10 db/ REF 0 db GHz S TART GHz STO P GHz 1 _: db TDR and Measurement Techniques t Incident wave TDR Reflected wave DUT Transmitted wave TDT t Incident wave S 11 Reflected wave DUT Transmitted wave S 21 Page 9 TDR Block Diagram 2 Sources (Step Generators), 2 Samplers, and 2 ADCs Trigger Trigger ADC Channel 1 Channel 2 ADC DUT Clock Trigger Step Generators Device Reference Planes Front Panel Samplers & ADCs TDR TDT Page 10 Accuracies of Time & Freq Domain PHY Meas 5

6 Network Analyzer Block Diagram Reflected Signal S 11 = b 0 /a 0 Transmitted Signal S 21 = b 3 /a 0 Page 11 TDR and Sources Power decreases across frequency band causes loss of accuracy at higher frequencies TDR Power constant across entire frequency band no loss in accuracy at higher frequencies Page 12 Accuracies of Time & Freq Domain PHY Meas 6

7 TDR and Receiver Bandwidths Agilent TDR has 4 Wide-Band Receivers with 2 choices for cut-off Frequency IF Bandwidth has 4 Narrow Band Receivers (definable by setting IF BW) that are swept across the frequency range of interest Loss of gain in the high region with TDR No Loss of gain with Noise Floor Page 13 Sources of Error in TDR Instruments- All significantly reduced with Normalization Oscilloscope Finite bandwidth restricts it to a limited measurable risetime Small errors due to trigger coupling into the channels & channel crosstalk Clock stability causes trigger jitter in the measurement Step Generator Shape of Step stimulus (risetime of the edge, aberrations on the step, overshoot, non-flatness) Cables & Connectors Introduce loss and reflections into the measurement system System_Risetime = sqrt[scoperisetime^2+steprisetime^2+testsetuprisetime^2] Page 14 Accuracies of Time & Freq Domain PHY Meas 7

8 Sources of Error in Instruments (2 Port) Random Errors: Instrument Noise, Switch Repeatability, and connector repeatability Systematic Sources of Error Directivity & Crosstalk errors relating to signal leakage Source & Load Impedance mismatches relating to reflections Frequency Response errors caused by reflection & transmission tracking issues within receivers 6 Forward & 6 Reverse Error Terms 12 Terms total for 2 Port Device & 48 Terms for a 4 port Device Reflection Tracking (b 0 / a 0 ) Transmission Tracking (b 3 / a 0 ) a 0 b 0 Source Match Directivity Crosstalk DUT b 3 Load Match Page 15 Loss of Source Power Affects Accuracy of TDR as Compared to the ~1 db diff ~2 db diff ~4-6 db diff The higher the frequency band the less accurate the TDR measurement Measured with normalization and additional correction TDR Norm@20pS Error: /TDR Page 16 Accuracies of Time & Freq Domain PHY Meas 8

9 & TDR Rough Comparison of Dynamic Range Dynamic Range Approx. TDR Dynamic Range Dynamic Range = MaxSignal-Noise Floor with wider Dynamic Range allows measuring signals that are 40-80dB down TDR Noise Floor Noise Floor Page 17 & TDR Attributes at a Glance Accuracy Source Power Dynamic Range Noise Floor Receiver Bandwidth TDR 0.5-5% 200mV 45dB 30-40dB GHz Depends on calibration & freq range or 0-18GHz 0.2-1% -5 dbm 90dB dB 10MHz-20GHz Depends on calibration Disclaimer: Values appearing in this table are estimates and do not infer or imply any guarantee by Agilent Technologies as to actual results Page 18 Accuracies of Time & Freq Domain PHY Meas 9

10 Calibration & Normalization Overview of TDR & Calibration Calibration File & Data Storage Differences Levels of Calibration with TDR TDR Normalization vs. Calibration Risetime Effects on Accuracy The Normalization Process Advantages of Normalization Fixture Error Correction Techniques Most Accurate S-Parameter De-embedding Line-Reflect-Match (LRM) Thru-Reflect-Line (TRL) Short-Open-Load-Thru (SOLT) Normalization Reference Plane Calibration Port Extension Time Domain Gating = Pre-measurement error correction = Post-measurement error correction Easiest Page 20 Accuracies of Time & Freq Domain PHY Meas 10

11 Levels of Calibration with TDR NONE: No calibration is performed Hardware Service Calibration assumed MIN: Module Calibration & RPC (Reference Plane Calibration) Module Calibration calibrates the gains and offsets of the data acquisition channel. RPC removes the delay of the cables and de-skews edges for the Diff & Common Mode launches. MAX: Normalization (Agilent only) Normalizes to standards placed at the Reference plane. Takes out delay and the loss associated with the cables, the reflections due to mismatch of the source, and improves the pulse edge and flatness. Allows for maximum accuracy and is heavily risetime dependent. Page 21 Comparing TDR Calibration Methods with a SOLT Calibration of a Thru Adapter Magnitude TDR Waveforms Norm@20pS Norm@30pS Phase & Page 22 Accuracies of Time & Freq Domain PHY Meas 11

12 Improved TDR Calibration Additional correction Save the frequency response of thru and reflect calibration standards Correct the measured response with the frequency response of calibration standard. The frequency and time domains are not simply related with FFT and IFFT With additional Correction Page 23 How Risetime Affects Noise in the Time Domain Noise in the TD increases with faster risetimes Normalization at 10pS can be achieved but it is extremely noisy Normalization (2046 pts) is a good balance between frequency domain accuracy and low TD noise TDR TDR TDR Page 24 Accuracies of Time & Freq Domain PHY Meas 12

13 Why Risetime Increases Noise in the Time Domain Risetimes determine the bandwidth (BW) of the Normalization filter Sharpening the risetime increases the frequency response but also the noise in the higher band This results in a distorted noise floor and a limit to your risetime beyond which there will be diminishing returns 0dB -3dB Log (Amplitude) 0dB -3dB Log (Amplitude) Noise Floor Log (Frequency) Basic System Response fc Basic System Response Noise Floor Normalized System Response Log (Frequency) Initial fc New fc Page 25 Summary of Good TDR Calibration The real advantage of Calibration (and more particularly Normalization) is that you can remove unwanted effects of cables and connectors leading up to your device. Magnitude and Phase (S-parameters) of thrus will show error increasing as a function of frequency. A good calibration at a reasonable risetime will show acceptable noise in the time domain. Checking Reciprocity is a way to verify a good calibration has been performed. Page 26 Accuracies of Time & Freq Domain PHY Meas 13

14 Measurement Accuracies: Reciprocity, Repeatability, Drift Using Reciprocity to Check Measurement Credibility Reciprocity with a TDR & Repeatability with a TDR & Drift with a TDR & Summary of Calibration & Measurement Accuracy Using Reciprocity to Assure Good Calibration Reciprocity is the constraint that for passive devices S12=S21. In measurements S12 virtually overlays S21 when a Thru path is measured. For TDR measurements the alignment is not as good. Be aware when exporting data that some tools may require a certain level of reciprocity (eg. HSPICE). to Port 1 Reciprocity on a Thru Adapter SE Measurement to Port 3 to Port 2 S13(Mag) = S31(Mag); S13(Phase) = S31(Phase) to Port 4 S24(Mag) = S42(Mag); S24(Phase) = S42(Phase) Page 28 Accuracies of Time & Freq Domain PHY Meas 14

15 TDR Reciprocity of BTL Board TDR Reciprocity: +/- 25 degrees Phase +/- 4 db Magnitude SDD12/SDD21 SDD12/SDD21 Page 29 Reciprocity of BTL board Reciprocity: +/- 2 deg Phase +/ db Magnitude SDD12/SDD21 SDD12/SDD21 Page 30 Accuracies of Time & Freq Domain PHY Meas 15

16 Repeatability with Normalized TDR measurements TDR Repeatability: +/- 60 degrees Phase +/- 4 db Magnitude Magnitude Repeatability Difference Phase Repeatability Excellent Good Difference Cumulative Error 2 measurements Normalized at 20pS (with new calibration) 5 days apart Page 31 Repeatability with with SOLT cal Repeatability: +/- 2 deg Phase +/- 0.5 db Magnitude Magnitude Repeatability Difference Phase Repeatability Excellent across entire range Excellent across entire range 2 measurements also 5 days apart Page 32 Accuracies of Time & Freq Domain PHY Meas 16

17 TDR Source Drift Over the Course of a Day Thru adapter over a 12 hr period at 4 hr intervals with the same calibration 8pm 8am 8am 12pm Sequence pm 4pm 4pm Range of Drift 8pm 13pS of drift corresponds to 95degs phase difference (drift is always bounded) Page 33 & TDR Attributes at a Glance Reciprocity Repeatability Drift TDR* 2-4dB dependent on calibration 3-6dB dependent on calibration Magnitude within noise of instrument +/- 25 degrees +/- 60 degrees 20GHz 0.25dB Magnitude +/- 2 degrees +/- 0.5dB for Magnitude +/- 2 degrees Magnitude within noise of instrument < 5 degrees * TDR values may seem larger than expected. It should be noted that these values are at the high end of the frequency range Disclaimer: Values appearing in this table are estimates and do not infer or imply any guarantee by Agilent Technologies as to actual results. Page 34 Accuracies of Time & Freq Domain PHY Meas 17

18 Measurement Comparisons Single Ended Comparisons of TDR & Measurements Balanced (Differential) Comparisons of TDR & Measurements 25 Ohm Mismatch Airline (3.5 mm connectors) Device Characteristics: Device that is traceable to NIST Insertion Loss Low loss with known variations Return Loss Known resonance pattern over wide dynamic range Page 36 Accuracies of Time & Freq Domain PHY Meas 18

19 25 Ohm Mismatch Airline: Time Domain Comparisons mvolts Very good agreement for length of the device (regardless of calibration) Normalization corrects for cable losses in RPC measurement Typically Ref. Plane Calibration will not show correct voltage or impedance values Normalization will need to be used. Since Ref. Plane Calibration does remove the DELAY of the cables (but not the loss) it can predict approximate device length TDR TDR with RPC Page Ohm Mismatch Airline: Insertion Loss Comparisons Comparison of - TDR measurements using different calibration methods Phase Magnitude Norm@20pS with correction RPC Only Mismatch causes this standing wave pattern TDR Norm@20pS TDR with RPC Page 38 Accuracies of Time & Freq Domain PHY Meas 19

20 25 Ohm Mismatch Airline: Time Domain Comparisons Impedance (ohms) Excellent agreement for length of the device Impedance discrepancy w/rpc Good agreement on length of device for low loss controlled impedance DUT 10% variance (2.5 Ohms) on impedance between and TDR with Reference Plane Calibration < 1 ohm discrepancy between and TDR calibrated at 30pS risetime TDR Norm@30pS TDR with RPC Page Ohm Mismatch Airline: Return Loss Comparisons Good correlation with Normalized TDR and across a wide dynamic range: Resonances (due to length of mismatch line) Magnitude for normalized measurements Resonance Magnitude TDR Norm@20pS TDR with RPC Page 40 Accuracies of Time & Freq Domain PHY Meas 20

21 Single Ended Summary With Normalization and correction the TDR can be an effective strategy for measuring impedance and obtaining frequency domain characteristics of moderate loss devices at frequencies below 10 GHz. Reference Plane Calibration without correction should only be used for estimating the length of a device and getting and idea of the response of the device. Page 41 BTL Example: Differential Return Loss Details ~ 1 db diff ~ 2 db diff 2-6 db diff First 3 divisions - look very good Mid 4 divisions - errors increase Top 3 divisions - getting worse Norm@20pS corrected Difference Page 42 Accuracies of Time & Freq Domain PHY Meas 21

22 BTL Example: Differential Insertion Loss Comparisons Start at 10GHz Norm Tr=20pS Norm Tr=30pS RPC weak midband Highband is poor for TDR Without additional correction Page 43 BTL Example: Differential Insertion Loss Details ~1 db diff ~2 db diff ~4-6 db diff First 3 divisions - look very good Mid 4 divisions - errors increase Top 3 divisions - getting worse Norm@20pS corrected Difference Page 44 Accuracies of Time & Freq Domain PHY Meas 22

23 Overall Summary The simplicity of the TDR is useful for lower data rates and for obtaining an intuitive understanding of signal integrity effects The Vector Network Analyzer provides more accuracy and repeatability than the Time Domain Reflectometer TDR normalization provides data closer to a than that derived from a TDR utilizing only RPC calibration Frequency domain data derived from TDR data is less accurate at higher frequencies This inaccuracy leads to error that can be interpreted as pessimistic insertion loss data and optimistic return loss data for frequencies greater then GHz. The accuracy provided by data will be required for data rates above 6.25 GB Page 45 Resources Websites Software Physical Layer Test System (PLTS) Advanced Design Software (ADS) Hardware N Vector Network Analyzer 86100C Time Domain Reflectometer PLTS Studio - Analysis Only Software N1930B-1NP networked license N1930B-1FP fixed license N1930B-1TP USB key license Physical Layer Test System (PLTS) Configurations Page 46 Accuracies of Time & Freq Domain PHY Meas 23