thebugger

modified 10 years ago

Amplitude Stabilised Phase Shift Oscillator

138

02:37:26

Press the switch shortly. I saw someone looking for a phase shift oscillator. They usually tend to clip the wave, because there's nothing to constraint the gain after the wave has reached the desired amplitude. This does exactly that. It senses when the wave has reached a certain voltage and starts drawing away from the gain, decreasing the amplitude. As it decreases the regulating part of the circuit disables and the rising starts again. It's a cyclical process. Anyway any of these amplitude stabilization circuits introduce nonlinear noise into the wave and increase its THD. Even without measuring it i can say that the THD of the circuit ia very high (maybe 10%). Only wien bridge Oscillators and specifically bulb compensated ones offer low distortion. Any other nonlinear component like a diode for instance will introduce much noise. 

published 10 years ago

hurz

10 years ago

Think about your last statemants and you will find out its not fully correct. The point is, it doesn't matter what the nonlinearity is made with. Could be a diode or it could be a bulb!! Yes the bulb is also nonlinear. A clipping opamp output is also a nonlinear behaviour. So what i would like to say is, don't blame diodes, they can also be used the same perfect way as bulbs. The advatage of bulbs is its Tau filtering heating beside its nonlinearity in resistance. This long tau makes it not nesseary to and extra lowpass filters to settle a fitting feedback gain.

thebugger

10 years ago

Yes i know almost every component is nonlinear but diodes are more sharp in their operation than bulbs. As you said the bulbs slowly stabilise the output while diodes do it more abruptly. Diodes tend to cause higher distortion than bulbs in any linear circuit.

hurz

10 years ago

"abruptly"? Its just a matter were the operating point is and how much you tune on x axis! " Diodes tend to cause higher distortion than bulbs in any linear circuit." This is like a comparison of apple and fish.

thebugger

10 years ago

No it's not. I've tweaked a wien bridge oscillator with diode compensation as much as possible and still got a >1% distortion, even though i tweaked it just above the clipping limit. The bulb influences the circuit more smoothly while the diode does it much more abruptly. Yes abruptly is the word. The diode starts conducting very sharply after you reach its unlocking point, and introduces similar to Class B crossover distortion into the circuit. Even at the slightest level it's gonna be amplified 3 times bringing it to a much larger scale. Bulbs don't cause the same nonlinear distortion.

hurz

10 years ago

I'm pretty sure you can emulate a bulb with a diode. You need some linear additiinal stuff around and you cant see any difference as black box! What you for sure do not can do is replace one with the other.

thebugger

10 years ago

I'm pretty sure you can't emulate a bulb just by using diodes. A bulb activates more smoothly, whereas a diode will activate sharply giving a crossover pattern in the distortion graph. A bulb on the other hand doesn't activate at a specific level, and even if it does it will not be so fast, it'll be more slow and smooth. As you said the longer tau does its job in a very specific way, that neither a diode nor a thermistor can duplicate.

hurz

10 years ago

Ok, first think we need is 'a circuit to emulate a bulb'.

hurz

10 years ago

Any progress on your bulb emulating circuit?

thebugger

10 years ago

No, i haven't tried, but i guess I'd be using an op amp in some configuration to simulate it.

thebugger

10 years ago

But you'll still need some amplitude dependant feedback, like diodes.

hurz

10 years ago

You have a circuit which emulates a bulb? I haven't seen this!

thebugger

10 years ago

No i don't. Where did you get that impression :D you can't emulate a bulb with only linear components, because its characteristic is non linear. You'd still need to use a diode for instance, but you won't get the same result. As i said diodes start conducting very sharply after you reach their cut in voltage, whereas a bulb will transit into a higher impedance state more smoothly, thus not introducing a crossover type distortion in the signal. We can try a JFET as an Automatic Gain Control but i don't think it'll work in EC. I'll try it though.

hurz

10 years ago

problem with you is, you use always words unquantivied. No numbers values! You talk about more "sharply" or less "smoothly" but no values to compare... This is somehow irritating and does endup with no conclusion. Why the hell a diode is more "sharply" then a "bulb"? Sharp in what and which range? because a bulb need 230V and a diode just 0.7V? What if you take a look and just 0.6 to 0.65V at a diode and for the bulb 0 to 230V? What is more "smoothly? VALUEs EAMPLEs NUMBERs!

thebugger

10 years ago

I don't think you get it. Oh well :)

thebugger

10 years ago

Wow why would you use a 230V bulb. We're not talking about a tube oscillator. A 9V bulb should do. Moreover they're not used at they full voltage rating. It'll depend from bulb to bulb but they're often used 1/3 of their maximum voltage rating. If you set everything correct the bulb's resistance will only change in the range of a few ohms but will still influence the bridge enough to stabilise it, whereas a diode will have ≈∞ resistance at the beginning which will sharply fall to (depending how it's designed) very little ohms. So as you can see the diode will influence the circuit much more than a bulb will therefore introducing much more distortion than a bulb will. Also the diode will induce a very specific crossover distortion, whereas the bulb will have almost no such wave deshaping effect.

thebugger

10 years ago

And as you said, a bulb has a very large time constant which avoids distortion, and may introduce some subharmonics, but a diode has a much higher time constant, and will attempt to control the amplitude within the time of one cycle thus introducing much more distortion. The crossover distortion that comes in a package with the diodes, will further introduce distortion.

hurz

10 years ago

diodes does have a very SHORT time constant! Who tries to teach who? Again no values.

thebugger

10 years ago

In the circuit they'll have a short time constant :D they by themselves can't have a timeconstant. Actually they'll have some (dynamic resistance + junction capacitance) but it's neglectable.

hurz

10 years ago

Now your are right. Diodes in this topic do NOT have a time constant. And it cant that be larger the a a bulb time constant. Its really difficult with you. Back to 230V. Why not a 230V bulb, i published a few days ago a working wien bridge with a 7W 230V bulb. So please dont tell me it doen't work.

thebugger

10 years ago

No it shouldn't be a problem. At that rating the nominal resistance would be 7k so a a properly designed circuit should be able to utilise it. The problem is that it's gonna be very hard to tweak because the bulb has a higher power rating therefore won't have a high PTC at such low levels, but if it's set just right it should work perfectly normal. You know what, I've just decided to build a tube wien bride oscillator. I think i have a 220V/7W bulb somewhere, I'll try to utilise it.

thebugger

10 years ago

Hey i was wondering. I have some constantan. Does it have PTC. I can replace the bulb with it

hurz

10 years ago

No, its almost constant over a wide range of temperature. Thats why the name.

hurz

10 years ago

I think a Tube wien bridge wont need an extra PTC. Just control the gain close to 3+ and the amplitude will be stable cuz of nonlinear soft clipping what you get from tubes. Tubes are to nonlinear to need a regulation. If you emulate a opamp with tube it might be different and the regulating bulb makes sense again.

thebugger

10 years ago

I don't think tubes are as nonlinear as you think. They often have lower THD than transistors with significantly simpler design. And the clipping is as any other. I measured some clipping of the amp i made last night (2xEL84/15W) and it wasn't as round as i thought it to be. Plus the bulb method was created by Hewlett Packard to work with tubes. There must have been a reason. Anyway it didn't work. I'll try it again. It didn't work with a 220V/10W bulb, I'll try a more powerful one.

thebugger

10 years ago

I was wondering something. Think I'll try it. A vacuum tube uses a heater to produce thermionic emission right. So theoretically the heater itself could act as a PTC control right. It's made of tungsten after all. So what if i stabilise the heater voltage with a transistor, and return the output of the wien bridge to the stabilised heater. Theoretically an increasing output may result in a decreased thermionic emmision if properly designed, which in turn may stabilise the amplitude. Just brainstorming here. How does it sound to you?

thebugger

10 years ago

Of course I'll make it with a larger time constant somehow. The idea is that we won't be influencing the signal directly in any way, only the gain of the amp. An Automatic Gain Control feedback to vary the heater power, thus the gain itself.

hurz

10 years ago

I do not know if you know what nonlinear is all about. Tubes are nonlinear enough and do a nice soft clipping. Even simple class A opamps (differential BJT amplifier) do behave nonlinear enough and dont need extra regulating gain control. AGC makes only sense in perfect opamp circuits!

thebugger

10 years ago

Every component besides the resistor capacitor and inductor behave nonlinearly, i know that. I also know that most of them have a part of their characteristics in which they behave somewhat linearly. I'm talking of using exactly that portion. Now to the question. What's your opinion on my previous comment. You think it'll work? The most important question is does tube gain vary with heater voltage. If yes, than it's surely possible. If not, it'll be much more tricky to do.

hurz

10 years ago

The heater just generates electrons which need to be accelerated to an anode voltage. Important is this anode voltage and the gate voltages. The heater must run at a sufficient amount of power to produce enough electrons but does not have direct impact to gain /slope.

thebugger

10 years ago

Shame. It must have some impact on the tube. Shouldn't less electrons mean less gain :?

hurz

10 years ago

The number of electrons which arrives at anode are much less then generated at heater. And the pressure why electrons are moving to the anode is the anode-voltage. Not because they are so many and bored "so hey ho lets go to the anode" ;-)

thebugger

10 years ago

Yes i know that. Negative goes to positive. That's physics 101. But when there isn't so much negative electrons the plate shouldn't be so ,, influenced'' by the stream?

hurz

10 years ago

If this is boring physics for you, then you know electrons which do not end up at the anode are going/falling back into the heater. Actually most of them are quickly returning. Only a little part of them feels attracted by the anode. Anyway, don't stress the anode with to much current/electrons. This easily kills your tube.

thebugger

10 years ago

Triode speaking yes, but tetrodes and pentodes have this problem fixed.

thebugger

10 years ago

I even have a few pentagrid tubes, but i don't see any application for them except in some radio circuits, like a superheterodyne receivers and such.

add comment in editor