1
\$\begingroup\$

In the ARRL (2011) section 9.3.1 about LC oscillator circuits it says that "[t]he N-channel JFET source follower shown appears to be the most popular choice nowadays", but does not explain the advantages of JFETs over BJTs.

All I can think of is the fact that BJTs need extra components in order to be biased, and JFETs have a higher input impedance.

What would be the advantage of a JFET (e.g. J113) over a BJT (e.g. MMBTH11) for this application? And why the preference for an emitter follower topology?

EDIT: Here are the schematics from the ARRL (2011)

enter image description here

And my simulation which works well when tested on breadboard

enter image description here

\$\endgroup\$
  • \$\begingroup\$ No, I don't happen to have the 2011 edition of the ARRL handbook handy, and can't look at section 9.3.1. Show the schematics. \$\endgroup\$ – Olin Lathrop Jul 6 '18 at 13:19
  • \$\begingroup\$ A mistake often made is ignoring the path from collector/drain back to the resonator (ground). The ARRL "bypass" capacitor makes this path low-impedance. Your simulator forces this impedance to zero via "V1", but a breadboard may not. \$\endgroup\$ – glen_geek Jul 6 '18 at 15:46
  • \$\begingroup\$ @glen_geek That was a question I hadn't looked at yet - what is the purpose of this capacitor? Is it something to do with the Miller capacitance that andyaka mentioned? \$\endgroup\$ – talikarng Jul 6 '18 at 22:31
1
\$\begingroup\$

What would be the advantage of a JFET (e.g. J113) over a BJT (e.g. MMBTH11) for this application? And why the preference for an emitter follower topology?

I've run common-collector colpitts oscillators using BJTs from sub VHF to 600 MHz and there is no great problem to be found. I prefer the BJT because there are more options to choose from and JFETs appear to be not as much in favour as they used to be. But, the reduction in component count will be significant to some designers so it shouldn't be ruled out.

As to why the common-collector colpitts oscillator is preferred is down to Miller capacitance - the collector is not used hence it does not force negative feedback to the gate and cause problems. Loop gain comes from the voltage amplification due to C3 and L (Colpitts A) and the oscillation frequency sits on the slope of the resonant peak.

It's the same answer for a JFET - miller capacitance is constant and drain amplification is zero.

\$\endgroup\$
  • \$\begingroup\$ "I've run common-collector colpitts oscillators using BJTs from sub VHF to 600 MHz" - Am I right to presume that >600MHz the sizes of the inductors and capacitors become the limiting factor in how high the frequency can go. \$\endgroup\$ – talikarng Jul 6 '18 at 22:30
  • \$\begingroup\$ Based on what you quote from my answer, no. \$\endgroup\$ – Andy aka Jul 6 '18 at 23:55
1
\$\begingroup\$

I suspect it was practical: They worked reliably with acceptable phase noise. Before the present era, things that worked properly became more common by evolutionary processes.

With a bipolar at VHF it is quite easy to not have good Q due to the impedances, and thus poor PN performance.

Alternatively you use a nice high gain transistor, and it oscillates at 900MHz because of parasitics. Hams didn''t have specans to debug that sort of thing.

A bipolar transistor can make a working oscillator at several times its FT - thats well into the GHz for an MRF901, so plenty of space for an unwanted oscillator mode to exist.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.