RadioBanter - selecting bias inductors for amps

wrote in message
ps.com...
If I am designing an amplifer that is to operate over say 3.4 to
4.2GHz, should I be looking for an inductor with self resonance above
that band, or at centre frequency, and why?

The "safe" advice is to choose one where the SRF is above the band you're
designing for, but strictly speaking you can usually get away with SRF within
the band or even a little below.

Your inductor can be modeled as an "ideal" inductor with a capacitor in
parallel. Up until resonance, the reactance of the physical inductor is
positive and varies (roughly) linearly with frequency. At resonance, the
reactance of the capacitor is equal but opposite that of the inductor so you
have Yin = 1/sC + sL = 0 -- an open circuit. Above resonance, the physical
inductor still has positive reactance but now it's decreasing with
frequency... not what you want!

(I know that at self resonance, the inductor will be resistive, below
inductive, and above capacitive, but I don't know what I should be
looking for.)

You're looking for an impedance that's "significantly" (conservately, at least
10x but realistically often 4x is fine -- the more conservative you are, the
less insertion loss you'll get) greater than the characteristic impedance of
your transmission line. (If you run it through a calculator, you'll see that
50 ohms in parallel with, say, j*50*10 is still *very* close to 50 ohms, and
even 50 ohms in parallel with j*4*50 really isn't half bad).

Should I be looking for high Q or low Q, and for what reasons would I
choose each.

About the only upside of a high Q inductor for biasing is slightly lower
insertion loss, and a very significant downside is that the inductor's
resonances above the band you're interested in will be "strong" and can cause
large (1dB), narrow perturbations (resonances) in the system's insertion
loss. A lower Q will tend to "smooth out" those resonances.

Indeed, if you're looking to build a wideband bias circuit, you often use
several inductors in series (small-medium-large, say) *with resistors in
parallel that purpose de-Q the inductors*. Without the de-Qing resistors, the
high-Q inductors have nasty interactions (two or three together will conspire
to appear as something approaching a short circuit at some frequency!)...

What numerical values actually equate to High Q and Low Q?

I'd call "high" Q100 and "low" Q10. High Q inductors are generally not
obtainable in "chip" form; you need air cores or exotic dielectrics to get
such high Q's at VHF/UHF. (For that matter, at least in my mind, high Q
inductors generally shouldn't be used for biasing -- they're used for filters.
:-) )

Is Q of 9 low, and Q of 35 high, for example?

35 is "middle of the road, drifting towards low." Q=35 is a decent inductor
to use for biasing... not too lossy, but lossy enough to avoid the resonance
problems discussed above.

---Joel