High speed buffer for measurement

[hosseineslahi7]

Dear experts,
I am trying to design a buffer after an up-conversion Mixer that can drive 50 ohm load for measurement. This buffer is supposed to be used for Mixer output measurement. However, we know that there is a huge load capacitance imposed by off-chip probes and wiring. This unwanted capacitance at the load can be in order of a few pF, disturbing the Mixer performance during th measurement process.

I saw common-source topologies with capacitive cross-coupling neutralization technique in literature, but it may not be a suitable structure for high frequency measurement under high capacitive load.

Does anyone have any suggestion about a robust buffer structure for up-conversion Mixer measurement?

 

[Easy peasy]

freq range ?

 

[hosseineslahi7]

freq range ?

Between 6-8GHz

 

[FvM]

I would expect 50 ohm matched transmission lines and impedance matched load at 8 GHz rather than probes with input capacitance.

3 pF in 50 ohm system makes 1 GHz low pass.

 

[hosseineslahi7]

I would expect 50 ohm matched transmission lines and impedance matched load at 8 GHz rather than probes with input capacitance.

3 pF in 50 ohm system makes 1 GHz low pass.

Thanks for this recommendation. My knowledge in terms of transmission line is low, so, my answer may not be correct. However, I am looking for a design that should be fully integrated into my 22nm layout. We will think about the transmission line option if we do not find any other solution.

 

[dick_freebird]

You might want to look at having a calibration structure on the chip,
where probe capacitance effects can be quantified for post-test
de-embedding.

A GSG probe trio might let you keep more transmission-line-y
(or at least, well known and well calibratable) qualities from chip to
test head, though that may be fancier than the plan, from the probe
card vendor's point of view.

For product acceptance, and maybe even design validation, you
may want to impose the figured loading on your design and find
the test limits / expected outcomes at a tester-friendly location,
which correlate back to the element of interest. Like maybe all
you need to prove the mixer, is amplitude and spurs from a
set of test tones, with limits appropriate to the lineup.