FAQs on insertion loss testing with Polar Atlas using SET2DIL and SET2SEIL
Application Note AP9017 


1. How does Bidirectional SET2DIL improve on SET2DIL?

Bidirectional SET2DIL launched in June 2013 and overcomes repeatability problems that may arise in previous versions, notably imbalance of traces on the test coupon and any asymmetry present in the TDR samplers.

2. What is SET2SEIL and where is it used?

SET2SEIL is a single ended (S21) insertion loss test and is a modification of the SET2DIL test method which uses a coupon designed with uncoupled traces for single ended measurements.  The measurement math is also optimised for single ended measurements.

3. What influences the maximum usable bandwidth of a SET2DIL measurement?

Cables, probes, low loss cables and quality coupon design all help achieve the maximum bandwidth possible from a given rise time TDR.

4. What is SKEW in relation to a SET2DIL measurement?

SKEW adjustment takes place to ensure that the forward crosstalk energy is added back into the transmitted pulse at the correct time in order to most faithfully reproduce the differential insertion loss calculation.

5. Do coupons need backdrilling? 

It depends. For thinner boards it is good practice to probe on the long side of the via to minimise stub length, but for thicker boards backdrilling is advisable.  As with all new techniques it is worth investigating the limits using a qualification coupon both with and without backdrilling to investigate the improvements it offers.

6. Why is line fit used – when should I choose root over linear line fit?

Linear fit usually gives more credible results with high loss base materials as the cross over between copper loss and dielectric loss dominating happens well below 2 or 3 GHz – with lower loss base materials this crossover occurs at a higher frequency and a root fit to the raw data may give more credible results.

7. In early Atlas versions the EOP display was held on the oscilloscope screen at the end of the EOP setup – why has it gone from recent versions?

The live EOP display left the probe outputs exposed to ESD so the later versions of Atlas suppress this final screen to give the TDR heads maximum protection. There is no material impact on testing or EOP setup.

8. Surface trace loss measurements seem to vary from day to day on the same coupon – why?

Humid air is very lossy and any moisture ingress into the soldermask can make significant impact on insertion loss with microstrips. Usually from a design perspective loss is far more predictable if traces are kept on inner layers. Should there be a need to use surface traces the user should be aware of the soldermask susceptibility to moisture absorption.

9. What are typical test times for SET2DIL measurements? 

This depends on the number of averages used – for BiDirectional SET2DIL it is generally accepted that 256 averages gives a good compromise between test speed and repeatability. Depending on the speed of the controlling PC the SET2DIL measurement should complete in around 16 to 20 seconds per test.

10. How does the MATLAB script compare with Polar Atlas?

Prior to the availability of BiDirectional SET2DIL a MATLAB script was in circulation allowing manual processing of the raw TDR files necessary for SET2DIL testing.  Polar Atlas math is also MATLAB based – however it incorporates all the latest thinking from the authors of the SET2DIL and SET2SEIL methodologies – including BiDirectional SET2DIL. The script does not support BiDirectional and is now regarded as legacy and no longer recommended for commercial testing.