Purely passive RC Phase Shift Oscillator?

[Cup of Joe]

TL;DR Summary

It is possible to build a purely passive RC phase shift oscillator?

It is possible to build a purely passive RC phase shift oscillator with 2 separate (in the future) RC stages like this?

Where there would be 2 RC networks that each provide 180 degrees of phase shift. Of course, there would be a buffer in between the 2 RC networks so that they behave as 2 distinct RC networks (each RC network has 3 stages of R and C). And I was thinking that since this is a purely passive oscillator circuit, then the buffer would have to be a boost converter connection that amplifies the signal in between the RC networks. The boost converter would have a wire attached in between the RC networks in the middle of the picture above.

Moreover, I would add a boost converter connection to each point of the circuit where it needs gain in order for the oscillation to occur. The output would be on the right side of the image above. In this way, I hope to create a purely passive RC phase shift oscillator.

I have attached the relevant part of the circuit above from my circuit simulation because the other part I am still working on and does not work. The feedback seems to be difficult to do. Also, I added the diode as a temporary element so that the +5V is forced through the RC network instead of directly going to -5V.

Please note that I am a beginner at designing circuits. My intention with this circuit is to reach the GHz range of frequency, to learn how to design my own circuits, and to learn more about how RC networks work as a hobby project. I want to reach very high frequencies, which is why I cannot have transistors or any other active element in my circuit because transistors and other active elements will not work at these frequencies.

Is this kind of circuit possible? Can anyone help me with designing this kind of circuit? Thank you.

 

[Baluncore]

Summary:: It is possible to build a purely passive RC phase shift oscillator?

Is this kind of circuit possible? Can anyone help me with designing this kind of circuit?

If there was a buffer it would not be passive.

The presence of resistors in the circuit is a sure indication that there will be losses. Each phase shift element also involves a significant attenuation. That loss must be recovered by gain, which you do not have in a passive circuit.

You could use a unity gain buffer driving a step-up transformer, but that is still an active network.

 

[Baluncore]

I want to reach very high frequencies, which is why I cannot have transistors or any other active element in my circuit because transistors and other active elements will not work at these frequencies.

How many hundred GHz do you need ?
https://en.wikipedia.org/wiki/High-electron-mobility_transistor

 

[Cup of Joe]

If there was a buffer it would not be passive.

The presence of resistors in the circuit is a sure indication that there will be losses. Each phase shift element also involves a significant attenuation. That loss must be recovered by gain, which you do not have in a passive circuit.

You could use a unity gain buffer driving a step-up transformer, but that is still an active network.

I meant something like this:

Where the +20V source (just as an example) would be the buffer (i.e. boost converter or something) and it would provide all the gain you need. And you would have this voltage source attached to wherever you need gain in the circuit. This voltage source would be outside of the circuit and not inside it so it does not interfere with the oscillating signal.

This is what I meant by passive.

 

[Baluncore]

You are generating a DC offset voltage.
Where does the AC signal that is to be phase shifted come from ?

 

[Cup of Joe]

You are generating a DC offset voltage.
Where does the AC signal that is to be phase shifted come from ?

From what I was thinking, it should come from the RC network, no?

 

[berkeman]

Where does the AC signal that is to be phase shifted come from ?

From what I was thinking, it should come from the RC network, no?

No, I think he's asking what your signal source is. What do you use to generate the AC signal? What frequency is it, and is it ground referenced or does it have a DC offset? Why are you showing DC voltages labeling some of the nodes in your AC filter schematics?

 

[Cup of Joe]

No, I think he's asking what your signal source is. What do you use to generate the AC signal? What frequency is it, and is it ground referenced or does it have a DC offset? Why are you showing DC voltages labeling some of the nodes in your AC filter schematics?

I intend to have a DC power source that will become an AC signal via the RC networks. Then, I will feedback that AC signal back into the RC network to continue the waveform.

 

[Baluncore]

I intend to have a DC power source that will become an AC signal via the RC networks. Then, I will feedback that AC signal back into the RC network to continue the waveform.

That is not the way electronics works. See if you can simulate it.

To make an oscillator you need a frequency selective network. That network will have losses so you will need an amplifier to make up for the loss. The more R you have in the circuit, the more will be the loss, and the more amplification you will need to sustain oscillation. If you make the tuned circuit from an LC there will be less losses and it will take very little gain to keep it running.

 

[Cup of Joe]

That is not the way electronics works. See if you can simulate it.

To make an oscillator you need a frequency selective network. That network will have losses so you will need an amplifier to make up for the loss. The more R you have in the circuit, the more will be the loss, and the more amplification you will need to sustain oscillation. If you make the tuned circuit from an LC there will be less losses and it will take very little gain to keep it running.

This circuit by T. Stewart from EE stack exchange made an oscillator with just DC inputs. I am just trying to make pretty much the same circuit, but without the BJT (another RC network). So I know that it can be done with DC inputs.

 

[Baluncore]

I am just trying to make pretty much the same circuit, but without the BJT (another RC network).

You are eliminating the gain, by eliminating the bipolar transistor, and you really think it will still work?

 

[Averagesupernova]

I must say this idea is the most bizarre thing I have seen yet on PF.

 

[Baluncore]

I must say this idea is the most bizarre thing I have seen yet on PF.

The original circuit presented by T. Stewart from EE stack exchange works well.
It is the Cup of Joe gain-free version that is the most bizarre.

Maybe Cup of Joe could build me a car without an engine, that rolls up-hill for ever and ever, like an M. C. Escher waterfall.
Or would that be prohibited by the PM ban on PF.

 

[DaveE]

A self-sustaining oscillator without gain (at the frequency of oscillation!) sounds a lot like a perpetual motion machine to me. Let me know when you've done this, I'll want to buy stock in your company.

 

[Cup of Joe]

A self-sustaining oscillator without gain (at the frequency of oscillation!) sounds a lot like a perpetual motion machine to me. Let me know when you've done this, I'll want to buy stock in your company.

The circuit is not self-sustaining and the gain will be provided with voltage sources attached to the circuit from the outside. I already provided a quick diagram above. I would laugh at your sarcastic humor, but, sadly, there is no joke here.

 

[Cup of Joe]

I do not understand why you think there is no gain in the circuit. I clearly stated that I would provide the gain via external voltage sources (i.e. boost converters or something like that) attached to the circuit from the outside like this (for argument's sake):

Where the +20V source (just as an example number) would be the voltage source that would provide the gain. And you could add this to wherever this circuit needs gain so it can oscillate. Although, I could be wrong about this and it might not work.

 

[analogdesign]

If there is gain, then the circuit it is not passive (this is how you confused a lot of folks). Instead of the mythical gain at the +20V node (you don't want a boost converter, btw, those are DC devices and you want an AC gain) why not use a transistor? That's what they are for. In fact, the stackexchange circuit you linked to would work fine.

You're going to struggle getting a GHz working on a PCB (especially as a beginner), but it could be a fun simulation study. If you need an oscillator to drive another circuit, there are plenty of high-speed commercial oscillators you could use.

 

[DaveE]

I do not understand why you think there is no gain in the circuit. I clearly stated that I would provide the gain via external voltage sources (i.e. boost converters or something like that) attached to the circuit from the outside like this (for argument's sake): View attachment 287632
Where the +20V source (just as an example number) would be the voltage source that would provide the gain. And you could add this to wherever this circuit needs gain so it can oscillate. Although, I could be wrong about this and it might not work.

As others have said, and I'll repeat: that doesn't provide gain at the frequency of the oscillation. It is just a DC offset. It doesn't even provide gain at DC. Injecting that voltage is essentially addition in your equations, gain is multiplication of the input signal, not adding stuff to it.

Let's say you have a 1GHz, 1mW signal in that circuit right before you inject the 20Vdc, how do you think that will increase the magnitude of the AC signal after that point? I think a bit more Math and circuit analysis is in order here. There is a very fundamental difference there that you will need to understand. If you don't pursue a more analytical path, you won't make any progress.

Note that the signal S₁(t) = sin(ωt) has the same AC signal strength as the signal S₂(t) = 1 + sin(ωt).

 

[Cup of Joe]

If there is gain, then the circuit it is not passive (this is how you confused a lot of folks). Instead of the mythical gain at the +20V node (you don't want a boost converter, btw, those are DC devices and you want an AC gain) why not use a transistor? That's what they are for. In fact, the stackexchange circuit you linked to would work fine.

You're going to struggle getting a GHz working on a PCB (especially as a beginner), but it could be a fun simulation study. If you need an oscillator to drive another circuit, there are plenty of high-speed commercial oscillators you could use.

What I meant by passive was that all the components in the circuit would be passive components (R and C) but no active components like a BJT. So, yes, I guess from a bird's eye view, this circuit would be an active one. I already have built the circuit that I showed everyone from EE stackexchange by T. Stewart and it works. I just want to design my own RC phase shift oscillator and not buy one (it takes the good experience away!).

 

[Cup of Joe]

As others have said, and I'll repeat: that doesn't provide gain at the frequency of the oscillation. It is just a DC offset. It doesn't even provide gain at DC. Injecting that voltage is essentially addition in your equations, gain is multiplication of the input signal, not adding stuff to it.

Let's say you have a 1GHz, 1mW signal in that circuit right before you inject the 20Vdc, how do you think that will increase the magnitude of the AC signal after that point? I think a bit more Math and circuit analysis is in order here. There is a very fundamental difference there that you will need to understand. If you don't pursue a more analytical path, you won't make any progress.

Note that the signal S₁(t) = sin(ωt) has the same AC signal strength as the signal S₂(t) = 1 + sin(ωt).

Thank you for your reply and for letting me know this. I think I understand what you are saying. I will look into other options that will provide AC gain, such as transistors.

 

[DaveE]

Maybe our misunderstanding is with your schematic which has everything cutoff below the 1K resistors. They way I see it, you don't even have feedback. Better questions (including formatting) will get better answers.

You absolutely can have gain without BJTs. You could use other active devices like MOSFETs, Vacuum tubes, klystrons, etc. But, by definition, if your device has gain, it isn't passive. Passive devices only have losses.

 

[analogdesign]

What I meant by passive was that all the components in the circuit would be passive components (R and C) but no active components like a BJT.

You've mentioned a couple of times you would have a "boost converter" and that would indeed have active components.

A truly passive oscillator is only possible using ideal components. Imagine you magically have a working oscillation. In your proposed circuit, the amplitude of the oscillations would be reduced a little bit each cycle because of losses. Eventually it would peter out to nothing. That is why practical oscillators have gain in the loop, to regenerate the amplitude and compensate for loses.

Your "boost converter" would need to be an amplifier, and, as such, would need an active device. And if you try to "get around" this requirement using a transformer, you can have voltage gain, but no power gain (the current is reduced by the same amount as the voltage is increased) so you are in the same situation.

What you are proposing is either a perpetual motion machine or you are not realizing your boost converter is really an amplifier.

 

[Cup of Joe]

Maybe our misunderstanding is with your schematic which has everything cutoff below the 1K resistors. They way I see it, you don't even have feedback. Better questions (including formatting) will get better answers.

You absolutely can have gain without BJTs. You could use other active devices like MOSFETs, Vacuum tubes, klystrons, etc. But, by definition, if your device has gain, it isn't passive. Passive devices only have losses.

I was still working on the feedback but I found it difficult. I stopped working on my circuit because I wanted to ask if this kind of circuit was even possible so I do not waste time building something that is impossible. It is why I asked here.

 

[DaveE]

I was still working on the feedback but I found it difficult. I stopped working on my circuit because I wanted to ask if this kind of circuit was even possible so I do not waste time building something that is impossible. It is why I asked here.

The way you've described it, and the (portion?) of the circuit you've shown, is impossible. Impossible at a very fundamental level. We don't have to write down any equations to tell you that, it is obvious to EEs that do this sort of thing for a living. Like asking if a meadowlark could fly to the moon if it just flapped its wings really fast.

OTOH, I'm still not 100% sure what you are talking about, since I can't see the entire schematic, I think.

Unless you can ask your questions with a more clear format, and with more detail, I think we are done here. You have your answer, from several different experts.

Edit: BTW, yes feedback is a complicated subject for beginners, but absolutely worth studying. One of the most important things to understand in electronics. All self sustaining, stable, oscillators have gain and feedback; all of them, every one.

 

[DaveE]

Also, a tip: study amplifiers first, then feedback, then oscillators. This is a natural progression. You really need to know the more basic stuff before you do the harder stuff, that's how electronics (and life) works.

 

[Cup of Joe]

The way you've described it, and the (portion?) of the circuit you've shown, is impossible. Impossible at a very fundamental level. We don't have to write down any equations to tell you that, it is obvious to EEs that do this sort of thing for a living. Like asking if a meadowlark could fly to the moon if it just flapped its wings really fast.

OTOH, I'm still not 100% sure what you are talking about, since I can't see the entire schematic, I think.

Unless you can ask your questions with a more clear format, and with more detail, I think we are done here. You have your answer, from several different experts.

Edit: BTW, yes feedback is a complicated subject for beginners, but absolutely worth studying. One of the most important things to understand in electronics. All self sustaining, stable, oscillators have gain and feedback; all of them, every one.

Yes, we are done here. I have learned what I needed to learn and know about. Thanks.

 

[Averagesupernova]

The circuit you have provided @Cup of Joe would be a serious problem if it oscillated since it would mean MANY MANY tried and true circuits that are not intended to oscillate would then oscillate.

 

[berkeman]

Yes, we are done here. I have learned what I needed to learn and know about. Thanks.

Thank goodness.

Also, a tip: study amplifiers first, then feedback, then oscillators. This is a natural progression. You really need to know the more basic stuff before you do the harder stuff, that's how electronics (and life) works.

This ^^^^

@Cup of Joe -- As you can tell we really want to help you learn this stuff. Unfortunately so far you are working with some serious misconceptions and simplified reasoning. Please study the topics recommended by @DaveE to try to get to the next level, then feel free to start a new thread where you post links to the reading you've been doing and ask questions about that reading. We are here to help.