PSPL Model 4022 High Resolution 15ps Risetime TDR Measurements at the End of 1-meter Coax Cable

Time Domain Reflectometry (TDR) has become an indispensable tool for designers of serial data systems. Recently at Picosecond Pulse Labs (PSPL), engineers demonstrated a 15ps risetime TDR step reflected from a short at the end of a 1 meter coax cable. The combination of the 9ps TDR Step from the PSPL Model 4022 Source Enhancement Module and recently improved Normalization software from Agilent Technologies provides high resolution TDR of singleended and differential devices. 10ps TDR (10%-90%) is possible when the Device Under Test (DUT) is attached directly to the 4022 Remote Pulse head and a 70GHz sampler. Some oscilloscope models place the samplers at the end of umbilical or extender cables, but owners of some sampling plug-ins do not have that option. Often it is physically impossible to attach a DUT to the front panel of a TDR plug-in. This test showed that a high-quality coax cable can be used to extend the reference plane 1 meter away from the scope without grossly degrading the TDR risetime.

X min( m ) where

c * Tr 2*

, r

estimate the minimum separation needed to resolve equal amplitude reflections. In air, this minimum spacing is about 5mm for a 35ps risetime step. Using the 4022 with 10ps step, this spacing becomes 1.5mm. Resolutions improve when the signal is traveling on a low-loss dielectric. The Model 4022 Source Enhancement Module and an Agilent 86100A Sampling Oscilloscope with TDR and 50GHz plug-ins were used in this test. While only one channel was used, this setup supports differential TDR and TDT as well. The 1 meter cable, PSPL part number 5015-40, was a typical microwave-grade coaxial cable with SMA connectors. The setup is shown in Figure 1. Agilent 86100A, 54754A TDR Plug-in and 83484 50GHz Plug-in

TDR 50GHz

1m Coax 2-wire Cable Reflection 5015-40 Standard Pulse Head

Abstract: This paper describes how the Picosecond Pulse Labs Model 4022 9ps TDR module may be used with a 1 meter coaxial cable to extend the TDR reference plane away from an oscilloscope. With this implementation, extender cables or sampling heads with umbilicals are not needed. This is particularly useful for test setups using probes to deliver the TDR signal. Measured results show that 15ps TDR is possible at the end of a 1 meter coaxial cable when using normalization (measured with a 50GHz sampling plug-in in an Agilent oscilloscope). Results also show that even with a TDR system that does not offer normalization (such at Tektronix or LeCroy), significant TDR resolution is maintained.

Ref. Plane 50Termination 4022 Source Enhancement Module

Figure 1: TDR setup

Normalization is Agilent s term for the deconvolution process that corrects for imperfections in the TDR signal. The reflections from a short and a 50-ohm termination at the reference plane are measured. Later when the DUT is attached at that reference plane, the oscilloscope displays a calculated response that is corrected for the characteristics of the input signal. The normalized reflection from a short is a smooth gaussian step without significant perturbations. The normalization software includes a digital filter to set the risetime of the response. Ripples appear on the response when too fast a filter is chosen. Typically, a 10ps-normalized risetime can be achieved using the 4022 when the pulse head and DUT are attached directly to a 70GHz plug-in. Reducing the plug-in bandwidth or adding cables between the pulse head and DUT slow the risetime limit.

Tr is the risetime of the TDR step, r is the relative dielectric constant, 8 c = 3*10 m/s

Equation 1: Relationship between minimum spatial resolution and risetime

The risetime of a TDR step limits the accuracy and spatial resolution of a measurement. With a conventional 35ps TDR step, it is impossible to separate the effects of closely spaced features. Equation 1 can be used to

Figure 2 gives measured results for the TDR system shown in Figure 1. These results compare the nonnormalized reflection from the short at the reference

Picosecond Pulse Labs

Page 1 of 2 www.picosecond.com

plane with the normalized reflection using a 15ps filter. The falltime of the signal without normalization was 56ps. Normalizing improved the falltime to 15ps.

that normalization is not necessary to discriminate the two distinct peaks (recall a standard 35ps commercial TDR is not able to resolve these two wires).

Note that even though the 10%-90% falltime for the nonnormalized signal was measured at 56ps, a significant part of the edge is much faster and only a slower dribble on the lower part of the reflection s edge causes the falltime to measure 56ps. Effectively the nonnormalized reflection falltime has to portions, a slow part and a fast part. As will be demonstrated later, since the reflection from the short has a fast part, its TDR resolution is significantly better than one would expect from a more Gaussian 56ps falltime edge.

Figure 3: 2-Wire Reflection Standard, reflection coefficient = 0.30. Two wires are soldered to the top of a 50-ohm transmission line. The 3.5mm-long wires are spaced 2.8mm apart. The connectors are 2.4mm.

No normalization, falltime = 56ps No normalization at end of 1m cable With normalization, falltime = 15ps With normalization at end of 1m cable Figure 2: Model 4022 TDR pulse reflected from a short at the end of a 1 meter coaxial cable Vscale: 50mV/div, Hscale: 50ps/div

We constructed a 2-wire reflection standard to show the benefit in using 4022 with its fast TDR edge. Two wires were added perpendicular to the 50-ohm microstrip transmission line. Figure 3 shows a photo of this test fixture. Traditional TDR products with a 35ps risetime pulse cannot differentiate between the two wires. The reflection looks like a single distributed discontinuity with smaller amplitude. Using the 4022, two distinct reflections from the wires can be seen in Figure 4. The top trace shows the TDR response of the reflection standard placed at the end of the 1m cable when normalization is not used. The middle trace shows the normalized TDR response at that reference plane. The lower trace shows the most accurate TDR with the reflection standard connected directly to the 4022 pulse head and using normalization. We see that the TDR measurements at the pulse head and at the end of a 1m coax cable are virtually identical. The top trace shows

With normalization At end of 15cm coax

Figure 4: Reflections from the 2-wire reflection standard Vscale: 20mV/div, Hscale: 50ps/div

In summary, the Model 4022 Source Enhancement Module enables high-resolution TDR measurements at the end of a 1 meter coaxial cable. This capability provides great flexibility to users of sampling oscilloscopes. Users of Agilent TDR systems without umbilicals may utilize the PSPL Model 4022 with normalization to achieve high-resolution measurements with the TDR reference plane at the end of a 1 meter cable. Users of Tektronix or LeCroy TDR systems (without normalization) who do not have extender cables may also realize the same test set-up with only a modest decrease in resolution.

Picosecond Pulse Labs

Page 2 of 2 www.picosecond.com