Light Bulbs and LDRs in Audio Oscillators

[German_EE]

I'm looking at low distortion audio oscillators and all of them include either a small light bulb or an LED/LDR combination somewhere in the circuit. I'm guessing that it's some sort of level control or AGC because the filament in a light bulb will react too slowly to affect the quality of the output waveform.

Can someone give a more precise explanation of why a light bulb or LED/LDR combination are used here?

[glarsson]

The oscillator must have an amplification of exactly 1. Lower and it will reduce in amplitude until it's dead. Higher and it will increase in amplitude until it over drives and generates square waves. The bulb (or LDR or JFET or...) stabilises the amplitude.

[unitedatoms]

That's how Hewlett Packard started historically, with a bulb in oscillator. Genius. http://www.hp.com/hpinfo/abouthp/histnfacts/museum/earlyinstruments/0002/

[ajb]

Commonly called a vactrol, it's quite a neat way to convert a voltage signal to a resistance.  As you say, a light bulb will have a definite low-pass filtering effect due to its thermal time constant (although very small bulbs might pass some bass frequencies, I imagine), but LEDs will pass well past the audio frequency range.  I've mostly heard of them being used in synth applications to allow, for instance, a low frequency oscillator outputting a control voltage to modulate the frequency of another oscillator in the audio range, or to allow control voltages to manipulate filters.

[Benta]

I'm looking at low distortion audio oscillators and all of them include either a small light bulb or an LED/LDR combination somewhere in the circuit. I'm guessing that it's some sort of level control or AGC because the filament in a light bulb will react too slowly to affect the quality of the output waveform.

Can someone give a more precise explanation of why a light bulb or LED/LDR combination are used here?

This is typical for Wien bridge sine wave oscillators. It's an AGC, as you suspect, and is there to keep the gain loop in the oscillator stable, otherwise it either doesn't oscillate, or the waveform is clipped.
I'm aware of at least three types of gain stabilization:
1: light bulb, due to its non-linear resistance/voltage behaviour.
2: LED/LDR, where the LDR is the non-linear element.
3: self-heating NTC, also non linear in the context.

[ArthurDent]

There were lamp/LDR units made for feedback with isolation, one was the Raytheon Raysistor . The CK 1102 I have pictured here with a date code of 6434 had a response time of about 0.2 seconds so these incandescent lamp ones were pretty slow but if you could design around that, they were good at smoothing out faster changes you might want to filter out. I used one to stabilize the heater voltage in an old WWII receiver.   

On the audio oscillators I had that used lamps it was interesting to switch ranges or twirl the frequency knob quickly and watch the lamps change brightness as the feedback changes damped out over a few seconds.

[Audioguru]

A Wien Bridge oscillator needs a gain of exactly 3 when oscillating and high gain to begin oscillating.
Try it. A light bulb stabilizer cause the amplitude to bounce up and down when it is turned on and when the frequency is changed.
A rectified waveform feeding a Jfet is used today.

Here is an Applications Note for Sinewave Oscillators from Texas Instruments. Caution, some of the opamps are not biased properly (shame on them):
 

[GopherT]

I'm looking at low distortion audio oscillators and all of them include either a small light bulb or an LED/LDR combination somewhere in the circuit. I'm guessing that it's some sort of level control or AGC because the filament in a light bulb will react too slowly to affect the quality of the output waveform.

Can someone give a more precise explanation of why a light bulb or LED/LDR combination are used here?

This is typical for Wien bridge sine wave oscillators. It's an AGC, as you suspect, and is there to keep the gain loop in the oscillator stable, otherwise it either doesn't oscillate, or the waveform is clipped.
I'm aware of at least three types of gain stabilization:
1: light bulb, due to its non-linear resistance/voltage behaviour.
2: LED/LDR, where the LDR is the non-linear element.
3: self-heating NTC, also non linear in the context.

Another option for non-linear device is head-to-tail Zener diodes. Zener do not turn on hard at low currents (10s of micro-amps) so the right voltage of Zener can creat the slight effect needed to turn the slightly-to-much-gain to just-the-right-amount of gain. 

[Zenith]

IIRC, a Wien Bridge oscillator has to have an amplifier with gain of exactly 3 otherwise oscillation dies away or increases until clipping occurs. You can't set the gain exactly right with precision resistors and presets. However, a carefully chosen incandescent bulb has a non-linear characteristic and also its thermal inertia which creates a time lag and putting one in the feedback path causes the oscillator to settle down and have very low distortion. The thermistor was a similar approach which I believe gave slightly better results. However the thermistors were specially made for the purpose. They were a small bead of material at the end of two fine, stiff wires in a glass envelope and are now hard to find and expensive. A problem with these two approaches was amplitude bounce when changing frequencies which could go on for seconds at low frequencies, and also I believe their performance fell off at below about 25Hz.

The later way of doing it was with a CdS cell and LED, which I think was capable of even better results but with more complex circuitry.

There were also another couple of ways of doing it, which allowed fast settling at the cost of performance. One was using back to back diodes and the other was to use an FET as a voltage controlled resistor. However, the CdS cell is an ohmic resistor at a given light level, whereas a FET has quite a small linear region. I've got a Philips oscillator with a switch to select low distortion or fast settling and it either uses a thermistor or diodes accordingly.

The CdS cell/LED was also used in analysers to measure very low distortion as a way of making the notch filter auto tune, as past a certain level it's hard to tune the filter accurately and it rapidly drifts.

I think there are even more elaborate methods using digital feedback etc these days.

There are lots of good articles on Wien Bridge oscillators on the WWW. One is  Linear Technology application note 43 by Jim Williams. The WB section starts at page 29.
 https://www.analog.com/media/en/technical-documentation/application-notes/an43f.pdf

[Benta]

Thank You for summing up everything that's been said earlier in the thread.   

[Zenith]

I wrote it and by the time I'd found the reference and checked it through there were several more posts. There seemed no harm in posting it and it never occurred that it might for instance, call forth a response from a rude idiot.

[unitedatoms]

I wrote it ..

Right. I remember by being surprised how slow Cadmium-Selenide sensors are. That genrators are perfect application, even better than incadescent bulbs

I remember that class of inventions is named Parametric Components. For example DC -DC converters based on thermometric effect will give very low noise with very few components. Another example is matter phase state based thermal ovens like Metcal soldering iron or Oven for frequency standard crystal on old satellites, where transition of liquid polymer to solid is a stable point.

[Benta]

I wrote it and by the time I'd found the reference and checked it through there were several more posts. There seemed no harm in posting it and it never occurred that it might for instance, call forth a response from a rude idiot.

Frankly, with 12 posts, you're not really entitled to call long-time contributors "rude idiots". I'd say that you need to get a bit of experience with this forum. Just check the thread before you post.

[Richard Crowley]

[David Hess]

Jim Williams went into detail on this subject in his Linear Technology application note 43.

Light bulbs and LDRs have the advantage over FETs in low distortion gain control circuits because the drain voltage modulates the conductivity of the FET slightly.  The result is that oscillators with an equivalent low distortion using FETs for gain control are much more complicated than the light bulb or LDR alternatives.

[T3sl4co1l]

Any gain control medium will work.  In the traditional (distorted) oscillator, average gain drops as clipping occurs.  This need not manifest as obvious distortion.  Example: a diff pair driving a resonant tank.  The current waveform becomes distorted as the diff pair is driven into saturation (Vin > 30mV say), even if the output voltage never reaches saturation (due to an adequately low load resistance, but not so low that oscillation is hindered outright).  In that case, output distortion is given by the current waveform as filtered by the resonant tank, which will be -20dB/dec typically.

Another example is a single transistor oscillator with the right biasing scheme, like this: https://www.seventransistorlabs.com/Images/100MHzOsc.png note V(R15) and R19 set bias current, and V(R15) sets peak oscillation voltage (or something to that effect).  At least, from my testing, this oscillator has a surprisingly wide and consistent range of bias versus output level, without breaking into squegging and without weird kinks and hysteresis in the transfer curve.  (But not particularly low distortion, not at VOSC; that's what the bottom filter section is for.)

In a gain controlled oscillator, we simply have a more refined version of the traditional oscillator.  All gain-control elements distort (heh, well, it may be more accurate to say all things distort and just leave it as a matter of degree!), we're just looking for elements which distort less than the usual case, and, preferably, which are dependent so that we can wrap a control loop around it to get arbitrarily precise amplitude as well.

To that end, we have the diode or BJT (exponential [trans]conductance), FET (using the "triode" range of output curves, or the exponential or quadratic range of the transfer curve), and any kind of DAC, the most important being the multiplying DAC (switched R-2R ladder or "digital pot"), and the PWM DAC (which is identical to an analog mixer overdriven to give a square wave "LO").

We might exclude the (quantized) DAC due to steppiness of the amplitude response.  Namely, if we put it in a loop, at best, it will dither between two nearly-perfect settings, never quite settling down.  (In control terms, the incremental gain between steps is infinite, so there's a RHP pole, and it reduces to a hysteretic controller in the small-signal limit.)  We can produce continuously variable PWM, which is nice, but the PWM ripple must be filtered arbitrarily well, which is awkward.

So that leaves the pure-analog methods.  The lightbulb of course is a case of this, except for self-heating rather than an external control loop (or, you could use a bias tee to light it somewhat independently of the oscillator's signal, or any kind of external heating, like, uh... sure, let's just say laser-heating the filament, why not?).  More practical to control, is the transistor-based method.  Most FET controls are based on the triode region, so there you have it.  BJTs are usually used in a more complicated circuit, such as an OTA, or a full on mixer or multiplier.

Drawback to these methods is probably split between distortion on one hand (simpler circuits) and noise on the other (complicated circuits with many transistors = noise sources).

FWIW, LDRs are surprisingly resistive, as far as I know -- despite being a semiconductor material.  As mentioned, they do indeed have nasty time constants, in fact several piled up (or diffusion behavior, or some combination even).  The main drawbacks basically come down to: CdS isn't RoHS, and the control loop can't run very fast (which probably makes an RF oscillator impossible to stabilize in this way).

Tim

[Zero999]

A couple of ordinary silicon diodes will do. My favourite method is to use a J-FET.

Here's a thread showing the lamp, diode and J-FET options simulated in LTSpice.
https://www.eevblog.com/forum/projects/wien-bridge-oscillator-ltspice-simulation/msg1079076/#msg1079076

[Zenith]

[
Frankly, with 12 posts, you're not really entitled to call long-time contributors "rude idiots". I'd say that you need to get a bit of experience with this forum. Just check the thread before you post.

Well since there are now several more posts, saying exactly the same thing, including quoting the Jim Williams ap note, no doubt, in your capacity as self-appointed board mother, you'll be pitching into them. After all, they can have far less excuse for not checking the thread? No?

If the cap fits, wear it.

[Conrad Hoffman]

Many viewpoints are a good thing! My particular interest is in the old RA series evacuated glass thermistors. They were used by General Radio Corp in various oscillators. It's hard to find a suitable lamp to put in the opposite leg, and the thermistors seem to be non-existent these days. You can find a data sheet, but not anybody that sells them. I bought a box of micro-bead thermistors and some glass tubing and (at some point) intend to try making them.

[SeanB]

Got a box of RA53 thermistors around that I got years ago somewhere in a steel trunk, but never used them, instead using the small grain of wheat 28V lamps as a control. You run them at around 1V, they do not glow, but are still going to have the non linear resistance. The small wire lead lamps are still available from RS I know, and best to try the lowest current 28V ones you can get, they are quite bright with only 20mA through them, though of course that is going to give you a 100 to 1000 hour lifetime, depending on the actual lamp and the stiffness of the power supply in starting it. Running at under 5V they last till the leadout wires corrode off the glass.

[Zenith]

The thermistors for this purpose I've seen were marked ITT,  (which was a UK electronics company) and either R53 or RA53. They were in a glass envelope with getter which was maybe an inch by 3/8ths and was wire ended. They also came in a smaller envelope, about 1 inch by 1/8th sometimes marked ITT, sometimes completely unmarked. The bead is tiny.

Until about 15 years ago they were stocked by RS Components and cost about £6. There was a similar one ,it may have been the RA43, but the 53 was always recommended for oscillators. They both suddenly went off the market. The usual places which stock hard to find vintage components such as obscure transistors and diodes, don't seem to have them. It's worth looking out for vintage equipment which is completely beyond repair and selling for pennies or being thrown out. Here there was an Open University oscilloscope and function generator, very low spec and now rarely seen, but it did contain an RA53 thermistor. Then there's ebay.

Years ago I fixed a Far Eastern made audio signal generator which looked as if it was intended for school and hobbyist use. It had surprisingly low distortion. It had thermistor stabilsation but the thermistor was like an AC126 transistor, that is in a TO1 can. I don't know what modern cheap Chinese audio sig gens use for stabilisation.

Best of luck with your efforts to make one. That shows dedication.

[chris_leyson]

Then there is the Meacham bridge stabilized oscillator published in the Bell System Technical Journal in 1938.
https://ia801604.us.archive.org/21/items/bstj17-4-574/bstj17-4-574.pdf
Maybe Bill Hewlett got the idea for the stabilized Wien Bridge oscillator from Meacham's work.
https://en.wikipedia.org/wiki/Wien_bridge_oscillator.
Also, for best spectral purity and lowest phase noise the oscillator needs needs to be amplitude stabilized and linear, I think Ulrich Rohde mentions this in "Communications Receivers" 1st ed but then again it might have been Manassewitsch in "Frequency Synthesizers Therory and Design". Amplitude stabilised oscillators are always good whether RF or audio.

[Benta]

Yes, the R53 and RA53 are classics for oscillator stabilization.
SeanB, instead of leaving them in a box, I'm certain you can make a handful Rands selling them.