Page 26 - Chip Scale Review_Jul Aug_2022-digital
P. 26
Since the original deployment of this now have an impedance of 35Ω and this instrumentation on the surface layer
probe card last year, we have evolved required several parts of the interface of the PCB and loopback paths on
the design in response to changing hardware to be redesigned as follows: lower layers of the PCB. The wider
customer requirements and to improve trace geometry for 35Ω required the
functionality in several key areas: • A new probe card PCB layout is position of several other traces to be
required, with trace geometries for adjusted, but overall, the changes could
• The 55GHz mid-band (MB) RF the MB signals matched to 35Ω in be accommodated without affecting the
signals required a change of the the DUT launch area; DUT spacing.
matching impedance from 50Ω to • The probe head cross section However, if only the PCB traces
35Ω; needs to continue to support 50Ω are changed to 35Ω, this will result in
• The alignment mechanism between impedance matching on LB and multiple impedance discontinuities as
the probe head and probe card PCB other RF signals, but also now the signal transitions from a 35Ω DUT
was improved; and requires 35Ω impedance matching to a 50Ω probe, then to a 35Ω PCB
• A feasibility study for implementing for the MB signals; and and finally, back to a 50Ω RF cable
denser multi-site test patterns was • As t he test equ ipment u sed connection. This can be simulated
completed. w il l st il l be 50 Ω , a 35Ω to and displayed using a time domain
50Ω impedance t ransfor mer reflectometry (TDR) plot (see Figure 3).
35Ω impedance matching is required, where the signals The resulting return loss performance
In the original probe card design, on the PCB transition from the i s e v e n w o r s e t h a n t h e s i m p l e
a l l t h e R F s i g n a l s r e q u i r e d a DUT launch area to RF cable “renormalized” case, so additional
matching impedance of 50Ω in the connectors. design changes are required in the other
interface hardware to achieve the best elements of the signal path.
performance. This required tuning Without these changes, test results The probe head and cViper probe
of the cross section of the probe head renor malized to 35Ω would have cross section also need to match the 35Ω
body and the 150μm pitch cViper spring degraded return loss performance and impedance of the DUT MB signals, but
probes to achieve a 50Ω impedance no longer meet the goal of better than the challenge here is to maintain 50Ω
match. The PCB layout was also -10dB up to 55GHz. impedance for all the other signals in the
simulated and optimized to match the To evaluate the feasibility of changing DUT layout. One approach considered
probe head impedance and provide a the tightly-packed DUT signal layout, was to reduce the spacing between the
50Ω signal path from the DUT ball to a simpler coupon board design was 150μm-pitch cViper signal probes and
tester pin. star ted, which implemented new adjacent ground probes by implementing
In a new customer DUT design, the 35Ω trace geometries for both direct a wider signal probe. This was not
MB channels operating at up to 55GHz connection paths from DUT to tester possible, as the DUT layout for the
Figure 3: TDR plots a) before and b) after optimization; and RL plots c) before and d) after optimization.
24
24 Chip Scale Review July • August • 2022 [ChipScaleReview.com]