I don't want anybody to design it for me, but in general how would you make an oscillator with a sine output of constant amplitude over a range of supply voltage (let's say the operating range of the op-amp). What I have in mind is a battery supply, but with no power wasted in a regulator. Sure the oscillator would also waste power but not as much. Would a square output be simpler?
If wasting power is an issue then you're going about it more or less backwards -- at best, a linear oscillator on an unregulated rail with a constant output is going to use just as much power as an oscillator being powered through a linear regulator -- and it'll probably be worse.
If you can run at a constant amplitude, use a switching regulator.
What frequency range are you looking at?
--
Tim Wescott
Wescott Design Services
http://www.wescottdesign.com
--
| James E.Thompson | mens |
| Analog Innovations | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| San Tan Valley, AZ 85142 Skype: skypeanalog | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
A square wave could be made extremely stable; generate it from CMOS logic powered by a voltage regulator. The supply current would be almost entirely the load current, so the regulator really doesn't waste power.
Easiest is a '555. If you regulate the reference resistor string, the 'triangle' waveform has regulated amplitude. Second-easiest is any oscillator with an AGC component. Consider phase-shift oscillators based on LM13700.
See
Figure 38 oscillator takes two chips ( four amplifiers), but note that A1 has its current-program resistor (30k) free; you can gain-control there.
Miniscule amount of capacitance... in other words... a hack quickly thrown together.
In practice I'd probably use a diff-pair with controlled tail current derived from measuring the amplitude... something simple like...
--
| James E.Thompson | mens |
| Analog Innovations | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| San Tan Valley, AZ 85142 Skype: skypeanalog | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
Age gets better with wine!
The Wien bridge is the classic example - the necessary amplitude control sc heme usually sets an amplitude that's independent of control voltage.
LC oscillators use less current, and there's nothing to stop you using Wien
-bridge style amplitude regulation to get a constant amplitude output indep endent of control voltage. I've got a current mirror variant of the Baxanda ll class-D oscillator which is less efficient than the classic Baxandall - which typically hits 90% efficiency or a bit better - but gets around 50% e fficiency in the configurations I've looked at.
The distortion performance isn't as good as a good Wien bridge - hysteresis in even a heavily gapped inductor means that it's hard to get the harmonic content better than about 100dB below the fundamental, but it's pretty goo d for a low-power oscillator, and way better than you'd get with a Wien bri dge.
Distortion performance is unremarkable. If you aren't pathologically nervous about winding a centre-tapped transformer (or getting it wound) - and a lot of our regular posters are - you can do a lot better with less power dissipation.
I understand what you mean. The idea came that it must be possible, but I've never seen it done specifically, so I got curious about how complex such an oscillator would be. More complex than a regulator it turns out.
Here's an example of an oscillator (for 3.6MHz) that uses AGC. It's not very complicated, though it's unsuitable for "Low KHz" because of the size of the inductor you need.
If you want to run it, you'll need to supply a transistor model, as I didn't include my library part (and you might need to change some resistor values for the AGC to work). But I suspect you'll be able to see how it works without doing that (use LTSpice to view!)
SYMBOL npn 240 -96 R0 SYMATTR InstName Q1 SYMATTR Value CA3046 SYMBOL res 176 -160 R0 SYMATTR InstName R4 SYMATTR Value 100k SYMBOL npn 368 304 R0 SYMATTR InstName Q2 SYMATTR Value CA3046 SYMBOL res 176 -304 R0 SYMATTR InstName R2 SYMATTR Value 47k SYMBOL cap 512 208 R0 SYMATTR InstName C4 SYMATTR Value 2.2n SYMBOL res 288 384 R0 SYMATTR InstName R6 SYMATTR Value 2.2k SYMBOL cap 160 -64 R90 WINDOW 0 0 32 VBottom 2 WINDOW 3 32 32 VTop 2 SYMATTR InstName C3 SYMATTR Value 10pF SYMBOL res 176 384 R0 SYMATTR InstName R3 SYMATTR Value 100k SYMBOL res 416 -64 R0 SYMATTR InstName R8 SYMATTR Value 10k SYMBOL cap 160 96 R90 WINDOW 0 0 32 VBottom 2 WINDOW 3 32 32 VTop 2 SYMATTR InstName C6 SYMATTR Value 3.3pF TEXT 320 576 Left 2 !.tran 0 200uS 0 1nS TEXT -160 576 Left 2 !.inc "..\\Lib\\Transistors\\ca3000.lib"
Yeah I think that's about right. (2mV out of 500 mV) But I use the diodes to trim off the top of the sine wave so it might be worse than that...(or better?) I'd have to do the math.
formatting link
I was thinking you might be able to use Spehro's "ideal diode" trick with an opamp (or two). (Of course it would then need some other voltage reference.)
If you want dead-constant amplitude and lowest harmonic generation, then with or without a regulated rail you need some sort of AGC. Moreover, your AGC itself needs to be well-behaved enough that it, itself, does not cause harmonic distortion.
If you don't just throw your hands in the air and use a DDS, your best bet is probably a Wein bridge oscillator. If you want to go old school with the ADC, use a light bulb.
Can you get by with a squarer wave? If yes, then your easiest and most fuel-efficient way to do this would be to regulate with a switcher and then generate a square wave by any of the many acceptable ways.
Just plain easiest, and not too bad on current consumption if you don't need any appreciable power on the output, would be a linear regulator powering your oscillator.
Maybe tell us what this is for so we can ground our opinions in fact instead of spinning off into our own versions of theory-land?
I've done quadrature oscillators (two opamp integrators) with diodes (or zeners) to limit the voltage. If you put the diodes on one integrator only, the waveform from the other has very little distortion (
It's mainly for general knowledge, since my test signal doesn't have to be that constant. I just got the idea that it must be possible, so, if it was easy then I'd do it, or else just use a regulator. The Wein Bridge is easy enough to be worth a try so I'll breadboard a few and try it with a variable supply to see what happens.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.