Negative feedback in LC oscillators

Ivan wrote:We have another saying in Czechia:"Young boys want toys and old boys want them, too. But the older the boy is, the more expensive toys he wants."  

VBR from Ivan

Hi Ivan,
We have exactly the same saying in English, but fewer words:
"Boys never actually grow up their toys just get bigger and more expensive."

This morning I came to work 1/2 hour earlier than normal so I could head out on Almungavägen, past Arlanda, then into the countryside to pick up Slottvägen leading to Roserberg where I work. Sunny, and the roof off  
I normally drive from home to Rosersberg (9km) but I just wanted to play with my toy

The car has some unusual features: the hand brake is a long way from the driver, beside the passenger's thigh!! I wonder if that was deliberate? Says a lot about the designer(s).

BR Harry - sm0vpo

We have another saying in Czechia:"Young boys want toys and old boys want them, too. But the older the boy is, the more expensive toys he wants."  

VBR from Ivan

Your new car looks great!
Here in Holland (maybe in other countries too) we have a saying: "The older, the crazier"...
Well, sometimes you have to give yourself a little present as a compensation for some bad times.
In my vision you have to make life pleasant for yourself. I do it myself (67 y.) too!
(If you wait for someone else to make your life pleasant, you usually need to have a lot of patience!)

Hi Harry,
IMHO this topic can help more hams, not John only.

Congratulations to your new car!

VBR from Ivan

That was a good response, Ivan.

When I made the retro wein-bridge audio oscillator I use a tungsten lamp because of ecconomy.

The old R53 10mW thermistor now costs something like $25 but the MES lamp costs $0.50. Another point is that the 6V 0.04mA MES lamp sucks 240mW when lit, but in this feedback circuit it is run at a much lower power level, typically 20mW to 50mW. So little that you can not see it lit in a dark room.

30 years ago I would have uses the R53, which was as common as dung in a cow field, but now they are antique rarities.

It is a pity that it is a bit too late to help John.

Very best regards - Harry - sm0vpo


PS - I just bought a new car. thought I should cheer myself up after such an awful year.
The Nissan Juke was high so Maj-Lis could just lean into the seat and swing her legs round. The back had space for the wheelchair.
now i have decided to buy a 46 year old dream. probably an old man trying to look young, but who cares?


You are not old until your dreams turn into regrets.

Hi Harry,
I did not perform real tests, but I am going to add a bit of theory.  

Everyone knows, that a positive frequency-dependent feedback is required to make a circuit oscillate. It is hard to keep the feedback just at the edge of oscillations: such oscillator might not start at switchon or stop oscillations with slightest changes of temperature or supply voltage. Therefore the feedback ratio must be bigger than one. In this case, the amplitude of oscillations tends to grow, until clipping takes place.

Another, negative amplitude-dependent feedback loop must be added to keep the active element of the oscillator in the linear region. A tungsten filament lamp is a possible solution, but it eats quite a lot of power. I suggest a very small thermistor instead, or a vactrol (a low-current LED shining on a photoresistor). Maybe a thin wire thermocouple might do, too. And maybe a sort of AVC, e.g. the rectified output voltage controlling the gain of a dual gate MOSFET.

VBR from Ivan

I know that this topic is more than half a year old, but I was surprised that there were no other responses to it.

I have noticed poor waveform on oscillators in the past, but I try to adjust the amplifier gain so that the positive feedback and the gain just about balance.
If there is excessive feedback then the amplitude increases until the amplifier begins to limit. This appears as distortion. Reducing the feedback cleans up the waveform considerably. Tuned circuits in oscillators also give more cleanliness to output wave-forms than audio/LF circuits where there is no tuned circuit to resonate.

The trick is to get the feedback sufficiently high to ensure reliable starting, but not high enough to cause excessive clipping. This is always the goal of oscillator design.

Negative feedback cancels the positive feedback, which is another way of achieving exactly the same result.

In audio circuits I frequently use some form of non-linear negative-feedback, where the output power causes, say a thermistor, to heat up and change the resistance, which changes the feedback. This is fantastic since the amplifier never gets anywhere near clipping point, and so clipping never occurs. If the oscillator output falls due to insufficient feedback, then the thermistor cools and the feedback increases.

I have never used this technique for RF oscillators, but now I wonder if it could be used to make an ultra-clean oscillator? At RF, probably up to 100MHz+ a simple 6v 40mA MES lamp could probably be used. In audio circuits this is practical, but the lamp burns typically 30mW (the lamp never actually lights up). RF oscillators are also run at much lower power levels, but with modern transistors it may be practical to have an oscillator running at, say, 50mW, as long as device temperature did not alter the frequency stability.

A thermistor would be useless for RF since high values of resistance and other circuit components would make it very frequency conscious. But the MES lamp technique could be well worth a try. It could be put in series with a capacitor across the emitter resistor of a transistor oscillator. If the lamp received too much power then the resistance would rise, reducing the decoupling (adding negative feedback) to the oscillator device, thus lowering the gain. If this works then the oscillator waveform would always be linear.

Could be a challenge for a free-running oscillator, but a controlled oscillator, say in a PLL synthesiser, could give the best of both worlds. The PLL would be real simplicity - no divider, just a frequency/phase comparator to lock the oscillator to a low-power stable oscillator on the same frequency.

This is all "food for thought"

Sorry for the delay responding but A lot has happened in the past year, so I was not able to check the board so often. I was just sifting through unanswered topics and found this one.

Very best regards - sm0vpo

After experimenting with various oscillator circuits (as in my other recent post) I am often frustrated by a poor waveform when breadboarding. I want this (left) but I get this, or usually even worse (right): 



This has led me to experiment with adding some negative feedback with a couple of different configurations with promising results so far: