and the vertical transitions through the
laminate package were accounted for
using EM simulation, and the overall
structure was optimised to compensate
for the parasitics and to ensure a good
RF transition.
The RF interface to all MMICs within
the package was also EM simulated and
compensated to ensure optimum performance
for the packaged MMIC. The
short bond wire lengths and ’V‘ structure
of the RF bonds to each MMIC are evident
from the photograph. This helps to
minimise the parasitic inductance of the
RF bond connections and eases the process
of optimising the RF performance
of the whole FEM.
An evaluation PCB was designed
on a low-cost laminate material, suitable
for the realisation of high-volume
mm-wave modules. The FEM components
were assembled onto the evaluation
PCB, which was attached to an
aluminium carrier to provide ease of
handling. A photograph of the evaluation
PCB is shown in Figure 3. The base
of the FEM is attached to an array of
vias on the evaluation PCB to ensure a
low inductance ground contact. Edge
mounted mm-wave connectors are used
to interface to all RF ports. Three 6-way
SMT connectors were used for applying
control and bias voltages to the FEM.
MEASURED PERFORMANCE
The performance of the FEM was measured
on the evaluation PCB with the
reference ports set at the package. TRL
calibration structures realised on the
same PCB material were used to facilitate
this. The small signal measured performance
of a typical receive path is plotted
against frequency in Figure 4 along
with the simulated performance (dotted
traces). The band-pass shape of the gain
response is predominantly defined by
the bandpass filter. The receive amplifier
(LNA) itself has a much broader band
response.
It can be seen that the measured gain
response is in close agreement with
the simulated response, demonstrating
the accuracy of the filter performance
predicted in the EM simulations used to
design the filter, and confirming the low
loss of the bandpass filter integrated into
the package.
The measured NF of three receive
paths is plotted against frequency,
along with the simulated (dashed red
trace) in Figure 5. The mid-band NF is
3.6 dB, dropping to 3.1 dB at the top of
the band and rising slightly at the lower
end of the band. The switch loss is low
5G Pioneer Band FEM
Figure 7: Measured transmit OIP3 versus frequency.
(around 0.8 dB) and relatively constant
across the band; the NF response is
dominated by the LNA.
Figure 6 shows the measured smallsignal
performance of the transmit path
versus frequency together with the simulated
(dashed traces). As with the receive
path, there is good agreement between
simulated and measured performance.
The harmonic filter starts to roll off significantly
above 40 GHz (providing around
20 dB of rejection at the 2nd harmonic)
and so its rejection performance is not
evident in this plot. The close agreement
between the in-band measured and
simulated gains does, however, confirm
the low insertion loss of the harmonic
filter as predicted by the EM simulations
used in the design process.
The linearity of the transmit chain
was evaluated by measurement of the
output-referred third order intercept point
(OIP3). This measurement was made with
the level of the two intermodulating tones
set to +12 dBm at the common port of
the FEM. The measured OIP3 versus
frequency is plotted in Figure 7, and
varies between +35 dBm and +37 dBm
depending on frequency. The saturated
output power of the transmit chain varies
between 27 dBm and 28 dBm over the
operating band.
CONCLUSIONS
The announcement of the 26 GHz band
(24.25 to 27.5 GHz) as the recommended
pioneer band for mm-wave 5G in Europe
is welcome. However, the availability of
some of the key mm-wave components
required to allow the development of 5G
systems operating in this band is limited.
The work reported here has attempted
to address this by developing an FEM –
comprising LNA, PA and Tx/Rx switch
– integrated in a custom laminate SMT
package and covering the full 26 GHz
5G band. Low loss transmit and receive
filters are integrated into the package
body. The transmit filter provides > 20 dB
of harmonic rejection after the PA and
has a loss of 0.2 dB. The receive filter is a
bandpass structure located after the LNA
and has an insertion loss of 0.7 dB. Measured
performance of the FEM mounted
on a representative PCB has been presented
and shows good agreement with
simulated. Receive path gain is 20 dB
with a NF of around 3.5 dB. Transmit path
gain is 19 dB with an output referred third
order intercept point (OIP3) of +36 dBm.
REFERENCES
1 Components for mm-wave 5G:
https://www.plextekrfi.com/mmwave/
mm-wave-5g/
2 The EU’s Radio Spectrum Policy
Group, “Strategic Roadmap Towards
5G for Europe”
3 Devlin L.M., Dearn A.W. and Pearson
G.A., “Low Loss MM-Wave Monolithic
SP4Ts”, proceedings of the 2001
“Workshop on Design for Broadband
Wireless Access”, Cambridge, England,
3rd May 2001
www.mwee.com November - December 2017 MW 23
