And of course, most of this doesn't matter: nowadays, it's typically easier to generate signals using microcontrollers, which are more power efficient, require fewer external parts, and cost next to nothing.
Using that approach, decent-looking square waves are easy well into multi-megahertz frequencies. Arbitrary waveform generation up to about 100 kHz is typically not hard either, be it with built-in DACs or PWM.
Easier it may be, but there are many ways to do it wrong, or at least suboptimally, with microcontrollers. For example, see the discussion of the signal generator on the FNIRSI handheld oscilloscope by Jeff Geerling here https://www.youtube.com/watch?v=IRBT0N67YFA. So yes, microcontrollers make a lot of the analog oscillators obsolete, but you still have to know what you're doing.
HN endorsement (at this point, I'm collecting these):
"It is hard to put into words how so many people 'out there' can print things, have people read them, and they are so worded that the average person is probably reading this as "the truth" ...
That said, I think these circuits are a increasingly just relics of a bygone era of part economy where doing complex math and hand-winding inductors was the way to go.
I was reading through this and it’s wild how sensitive an oscillator can be to tiny changes in parts or layout. It really shows how much of analog design is patience and tweaking. Got me wondering though, is there a simple way to test stability without going deep into fancy gear, or is that just part of the grind?
I’d love to see something on crystal oscillators. What’s the minimum viable (or minimum power) circuit to produce some oscillations (and how do you modulate the frequency)?
And of course, most of this doesn't matter: nowadays, it's typically easier to generate signals using microcontrollers, which are more power efficient, require fewer external parts, and cost next to nothing.
Using that approach, decent-looking square waves are easy well into multi-megahertz frequencies. Arbitrary waveform generation up to about 100 kHz is typically not hard either, be it with built-in DACs or PWM.
Easier it may be, but there are many ways to do it wrong, or at least suboptimally, with microcontrollers. For example, see the discussion of the signal generator on the FNIRSI handheld oscilloscope by Jeff Geerling here https://www.youtube.com/watch?v=IRBT0N67YFA. So yes, microcontrollers make a lot of the analog oscillators obsolete, but you still have to know what you're doing.
I have once tried to understand the Colpitts oscillator, used for example in metal detectors...
HN endorsement (at this point, I'm collecting these):
"It is hard to put into words how so many people 'out there' can print things, have people read them, and they are so worded that the average person is probably reading this as "the truth" ...
This article is blah, and 'meh' at best."
Previous installment: https://lcamtuf.substack.com/p/how-has-mathematics-gotten-so-abstract/comment/161589096
Great article as usual. Would be interested in hearing your take on LC circuits (Colpitts, Hartley).
I should probably cover it at some point; I have a tiny bit about it in the second part of the introductory article on impedance (https://lcamtuf.substack.com/p/impedance-part-2-why-do-lcr-meters), but not much else.
That said, I think these circuits are a increasingly just relics of a bygone era of part economy where doing complex math and hand-winding inductors was the way to go.
True, but LC circuits are also interesting because they can be used to model other resonant systems, like piezo-electric transducers for example.
I had a whole job for years at IBM that was mostly just getting the PLL working. There was a big team of us.
I was reading through this and it’s wild how sensitive an oscillator can be to tiny changes in parts or layout. It really shows how much of analog design is patience and tweaking. Got me wondering though, is there a simple way to test stability without going deep into fancy gear, or is that just part of the grind?
I’d love to see something on crystal oscillators. What’s the minimum viable (or minimum power) circuit to produce some oscillations (and how do you modulate the frequency)?
I have a section on crystal oscillators here: https://lcamtuf.substack.com/p/clocks-in-digital-circuits
It's one of the early posts on this blog and it's gotten very little attention, but maybe it answers some of your questions - LMK!
(It also mentions the op-amp relaxation oscillator, but you can skip ahead.)