A Wien-bridge oscillator with a couple of FETs for frequency control?
An SA602 with a crystal on the oscillator port and a really well filtered VCO oscillator on the antenna port? With the right filtering that'd give you a nice clean sine wave, and a frequency range from whatever you can call "audio" right down through zero and out the other side. The VCO wouldn't need to be buffered -- the '602 doesn't _want_ much power on it's antenna port, so you could just filter the snot out of the VCO output and tap way down on the output tank circuit. As long as you feed a sine wave in on one port or the other, and stay in the chips linear range, you'll get really nice clean sine waves out.
I do appreciate what you are saying and don't disagree but in this application I want to do some good old fashioned mucky analogue signal processing, not PIC programming (even if I had any PIC experience).
Over a narrow range of frequencies, you can get a really nice sine wave approximation with a square wave generator and a filter, or better yet a counter, a two-bit (that's two digital bits, not $0.25) DAC, and a filter.
Interesting ideas. From memory, doesn't the Wien-bridge oscillator need some form of amplitude stabilisation? WIll look into that one. Re the SA602 - are yousaying this would provide a solution in itself with some external filtering? Will investigate, thanks.
Yes, but if you don't mind a rather distorted sine wave it can be done just by letting the output smack into the supply rails. Since you haven't responded to my questions about what you really need, one can only assume that any old wiggly line on a scope counts as a "sine" wave.
Yes, but you won't find it on any data sheet. You will get a clean sine wave with ease, though.
If you cough up answers to my questions on what you really need, folks will be able to give you more specific help.
I understand not wanting to break the digital barrier -- I tend to cross that particular Rubicon with ease, yet I have circuits out there that are all analog, or that only use "digital" parts in an analog way.
If you used a 555 followed by an active filter with a sharp cutoff, you could get a pretty nice sine wave in a narrow band of frequencies. Since you haven't answered my questions about your needs, I can only say that it'll work great.
There's a bazillion ways to do this to meet your stated requirements -- but only because your stated requirements are _very very broad_.
The sad truth is that square-triangle wave generators are easy to voltage control, and most voltage-controlled options for sinewaves are crufty filtering/shaping addons to the square-triangle generator. An alternative that I particularly like uses programmable impedance amps for a phase-shift oscillator, i.e. figure 17 in
It uses two chips, but they're under a buck. This needs a matched trio of timing capacitors, so is gonna get complicated if you need multiple-range switching.
I'm deliberately trying to make an all analogue design in this case so it's not that I'm particularly averse to taking a digital approach as such.
LF to MF audio band probably 70Hz - 3KHz
Not sure really, I would have thought 2% THD+N would be good enough but I find it difficult to be accurate without having performed any experimentation yet.
I can only answer "nominally" linear. Definitely not log.
10 - 30 degC
Well I suppose the amplitude needs to be reasonably stable with frequency but again only to within say 10%.
I think the XR2206 looks promising so far. I've not yet had enough time to look at your other suggestions but will do. I particularly like the idea of a Wien-bridge if it's easy enough to tune because there is no digital switching going on internally, so should be inherently more quiet.
You are going to find that tuning range to be challenging, both because
70Hz is going to require large capacitances, high circuit impedances, or both, and because you're trying to make the thing work over a 43:1 range.
My Wien bridge idea is _not_ going to work without switched capacitances. In fact, my suggestion for a SA612 with one crystal and one LC oscillator isn't a good one, either -- even a 1MHz LC oscillator is going to want to have more than a 3000kHz range.
If someone were holding a gun to my head to do this with analog circuits, I'd do the SA612 solution with two matched crystals with resonant frequencies in the single-digit MHz region. I'd "rubber" one crystal by putting it in series with a varicap diode, and I'd sweat bullets to make sure that I could get down to a 70Hz difference in frequencies without either crossing my frequencies over (which would get me an audio tone with the voltage to frequency gain reversed) or having some instances of the circuit fail to get down to 70Hz in production.
If a customer asked me to do it, even if they jumped up and down and turned red in the face every time I said "digital", I'd give them one estimate for the analog way, and a second, significantly lower one, for the digital way.
Having tried to rain on your parade _again_, I will say that I think the mixer/RF oscillator approach should work pretty well, and without a whole lot of components.
Take ATTiny13, make a predefined lookup table with pre-distorted sinewave, generate a noise chaped PWM. That would be absolutely the cheapest controllable and fairly accurate generator of a good quality sine wave.
Vladimir Vassilevsky DSP and Mixed Signal Design Consultant
The Wien-bridge does need amplitude stabilisation; it can be as crude as an incandescent filament in small light-bulb (think bicycle lamp) but Jim Williams' various application notes about Wien Bridges for Linear Technology mostly have a proper feed-back-controlled loop to regulate the amplitude.
The Wien-bridge doesn't lend itself to variable frequencies, and I don't think that I've ever seen it used in any kind of programmable frequency generator. I'd be interested to hear how Tim Wescott proposed to make the output frequency continuously adjustable.
Great. Thanks for all the ideas. Looks like I need to compare the
8038 vs XR2206. Dave W.
-----------------------------
I've tested both. On paper they seem similar, but in practice the XR2206 is clearly superior in temperature coefficient, linearity and noise immunity. Also easier to adjust for low distortion. I found no advantage to the 8038 in my application.
Well, the CD4040 is cheap, but a progammable logic device can provide the same funcion - and a lot more beside - in a single chip, which is one of the things the OP is asking for.
And the CD4040 is slow - slow enough for it to be nuisance when I first used the part back in 1975. It looks glacial these days.
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.