When to have only voltage gain in an oscillator

[Cup of Joe]

TL;DR Summary

Is it possible to have only voltage gain in an oscillator and when can you do this? Under what conditions/circuit topology?

What I already know
In general, power gain is desirable for an oscillator in order to make up for the losses and then feedback that gain (amplified signal) into the oscillator for it to keep oscillating. Voltage gain is not generally used for oscillators.

What I want to know
Since power gain is needed in general, I want to know under what conditions or circuit layout for the "ungeneral" or special case where only voltage gain is used. This means that the current decreases as voltage increases and the total power stays the same.

I think my question may be a little strange since power gain is all you need and it is not difficult to do, but I really want to know this.

Thank you.

 

[DaveE]

What I already know
In general, power gain is desirable for an oscillator in order to make up for the losses and then feedback that gain (amplified signal) into the oscillator for it to keep oscillating. Voltage gain is not generally used for oscillators.

Nope.

Voltage, current ,and power, as well as their respective gains are all related by the circuit's impedance. With a fixed impedance if you have one type of gain, you have all three. Depending on the circuit, or amplifier type, it may be simpler to think in terms a particular type of gain though. For example, a BJT amplifier stage will have gain and normally a significant impedance transformation between the base and the output (collector or emitter) circuits. So for this device it is convenient to speak of current gain, which is relatively independent of the impedance change. In RF circuits, people like to deal with power gain because power is more conserved with circuit configuration changes, like impedance transformations.

You can make an oscillator with any sort of gain. You can analyze any given oscillator in terms of any of these gains, although some are easier to work with than others.

 

[berkeman]

https://en.wikipedia.org/wiki/Electronic_oscillator

 

[DaveE]

I want to know under what conditions or circuit layout for the "ungeneral" or special case where only voltage gain is used.

Most any oscillator that uses a conventional voltage in, voltage out, op-amp (like uA741, LM324, TL074 etc.) is best analyzed as a circuit with voltage gain.

 

[Cup of Joe]

Thank you @DaveE and @berkeman for your inputs.

I do, however, need to have a more objective viewpoint on this subject. Could you please take a look at these answers to my question from EE stack exchange to see if they make sense: https://electronics.stackexchange.com/questions/582104/voltage-gain-vs-power-gain-in-an-oscillator?

I am kind of getting different answers from that site compared to here. It's why I would like for you to see them. They both say that power gain is required.

Thanks.

 

[DaveE]

Thank you @DaveE and @berkeman for your inputs.

I do, however, need to have a more objective viewpoint on this subject. Could you please take a look at these answers to my question from EE stack exchange to see if they make sense: https://electronics.stackexchange.com/questions/582104/voltage-gain-vs-power-gain-in-an-oscillator?

I am kind of getting different answers from that site compared to here. It's why I would like for you to see them. They both say that power gain is required.

Thanks.

Yes, I think it is fair to say that all oscillators require power gain. Really, the requirement is unity (or greater) loop gain, usually expressed as voltage or current. But since all passive components in the real world have power losses, you must have power gain to overcome those.

However, I think it is a distinction without much difference. You still need gain, in whatever form you care to describe it. The one value that voltage, current , and power have in common is zero, which is what a lossy circuit will end up with without gain. So, if you pick a location in the oscillator's feedback loop and characterize the excitation there, to see if it's oscillating, it will have a fixed impedance (except strange circuits), and thus voltage, current, and power are all related to each other and result from the same circuit behavior. If you measure one gain at that location, you can derive the other gains from the circuit impedance.

Really, I don't think you are worrying about a very useful fundamental rule; voltage vs. power gain. It all ends up depending on the circuit in question, and the analysis techniques you want to apply. RF guys nearly always look at power, audio guys usually work with voltage or currents. The reason they like different methods is really more advanced than this thread. The correct version is the one that allows you to describe a specific circuit with equations that model it well. You will want to become proficient with all of these formats and understand how they are related. They all always exist in your circuit.

I would also add that at the most fundamental level of modeling a circuit's behavior, voltage, current, and impedance are the go to parameters. This is because the fundamental circuit laws, like Kirchhoff or Maxwell, are expressed with these. This is what simulators, like SPICE use. Power is incredibly useful, related to energy conservation, but a secondary parameter that is ultimately derived from the more basic parameters: voltage (e-fields), current (h-fields), and impedance (the relationship between e and h fields).

 

[DaveE]

Bottom line: You need gain. You get to choose whether you call it voltage, current, or power gain.

 

[Cup of Joe]

@DaveE Thank you very much for your insight, I understand the topic better now. I will keep all that in mind.

 

[LvW]

What I already know
... Voltage gain is not generally used for oscillators.

Remember that the oscillation condition formulated by Barkhausen is based on voltage gain only (voltage gain around the loop,loop gain, of unity.)

 

[LvW]

Remember that the oscillation condition formulated by Barkhausen is based on voltage gain only (voltage gain around the loop,loop gain, of unity.)
More than that, speaking about "power gain" and "power amplification" one should know that a real "amplification" of power is not possible. Using these terms, we usually mean a conversion of DC power into an output AC power - steered (controlled) by AC input power.

Having this in mind, we can analyze the AC power distribution within a closed feedback loop, which fulfills Barkhausen`s oscillation condition:

* When we open the loop at a suitable point (without changing the load conditions remarkably) we create two nodes Vin and Vout. The test signal injected into node Vin of the open loop will be amplified with a voltage gain of unity and appears with the same level at the node Vout.

* Of course, we can calculate the signal power which goes into the loop at Vout - determined by the input impedance Zin at this node.

* Because the voltage gain is unity - and under closed-loop conditions - Zin is now connected to the node Vout=Vin and the same signal power as determined before does exist at this node.

* Therefore, we can say that the power gain within the closed loop is unity.

 

[DaveE]

Remember that the oscillation condition formulated by Barkhausen is based on voltage gain only (voltage gain around the loop,loop gain, of unity.)
More than that, speaking about "power gain" and "power amplification" one should know that a real "amplification" of power is not possible. Using these terms, we usually mean a conversion of DC power into an output AC power - steered (controlled) by AC input power.

Having this in mind, we can analyze the AC power distribution within a closed feedback loop, which fulfills Barkhausen`s oscillation condition:

* When we open the loop at a suitable point (without changing the load conditions remarkably) we create two nodes Vin and Vout. The test signal injected into node Vin of the open loop will be amplified with a voltage gain of unity and appears with the same level at the node Vout.

* Of course, we can calculate the signal power which goes into the loop at Vout - determined by the input impedance Zin at this node.

* Because the voltage gain is unity - and under closed-loop conditions - Zin is now connected to the node Vout=Vin and the same signal power as determined before does exist at this node.

* Therefore, we can say that the power gain within the closed loop is unity.

Then you'll learn about duality, and realize that Barkhausen is only half of the story (or less). Whatever thing you manage to feedback in the system is what has to have unity gain.

 

[sophiecentaur]

one should know that a real "amplification" of power is not possible.

This is just a 'words' thing and I'm not sure what you mean by it. Amplification implies conversion of Power from some source to increase the level (or Power) of a signal. There are passive devices (no power supply) which can reduce the level of a signal (resistors etc.) but an antenna, can gather more RF Power from the surroundings than a simple dipole. That is also referred to as Gain because you get more but the gain of an antenna or dish is achieved by re-directing RF power into the feed of a receiver. But I wouldn't say that amplification is involved here.

 

[LvW]

... but an antenna, can gather more RF Power from the surroundings than a simple dipole. That is also referred to as Gain because you get more ...

...more than what? Answer: More than an isotropic antenna would receive. Hence, there is - of course - no amplification effect - and the term "gain",in this case, is really questionable in my view. However, this is not a problem if everybody who is using this term knows about the corresponding definition.
Yes - generally I agree with your comment.

 

[sophiecentaur]

Antenna ‘gain’, as a name is dodgy until you realize it’s a ratio of powers. The reference power is what’s harder on the brain. It assumes a certain cross sectional area which catches the energy going past. Practical Engineers tend to use a half wave dipole because an isotopic radiator doesn’t exist and you can use a home made dipole (with your fingers crossed) as your reference standard.

Antenna gain in dB fits into all your link calculations, along with inverse square loss, feeder loss, absorption loss when you use the appropriate signs. So I can’t think of a better name for it.

 

[LvW]

Practical Engineers tend to use a half wave dipole because an isotopic radiator doesn’t exist and you can use a home made dipole (with your fingers crossed) as your reference standard.

Antenna gain in dB fits into all your link calculations, along with inverse square loss, feeder loss, absorption loss when you use the appropriate signs. So I can’t think of a better name for it.

Yes - of course, an ideal isotropic antenna does not exist . But that is not the question, I think.
As you know, the gain of an antenna is NOT given in dB but in dBi (that means: Gain above isotropic radiaton).
Am I wrong?

 

[Averagesupernova]

As you know, the gain of an antenna is NOT given in dB but in dBi (that means: Gain above isotropic radiaton).
Am I wrong?

Yes you may be wrong. I've seen it spec'd as dBd also. As in dB gain over a dipole (real world, not isotropic). That may be a ham thing, I don't know just how accepted that is.

 

[LvW]

Yes you may be wrong. I've seen it spec'd as dBd also. As in dB gain over a dipole (real world, not isotropic). That may be a ham thing, I don't know just how accepted that is.

I do not blindly trust wikipedia in any respect - however:
https://en.wikipedia.org/wiki/Antenna_gain

 

[sophiecentaur]

If dBi is stated then that's what it is. Without that 'i', I'd tend to assume that it relates to a dipole because improvements are instantly recognisable for what they actually are in practice. I would expect there to be a declaration of units for all variables in any case.

 

[DaveE]

I would expect there to be a declaration of units for all variables in any case.

Yes, this $\uparrow$ $\uparrow$
In my experience dB is often used with some implicit agreement about what the ratio is or what you are taking the log of, which people sometimes gloss over. It's really helpful when people put an m, i, c, v after it, but there are still questions about what those are. What power? what voltage? What antenna polarization? etc. You simply have to have the context to get it right, although sometimes it's really easy to guess correctly. In common usage it is way too sloppy to be taken as a definition of things. It's more like shorthand for math and even then it comes in two flavors.

 

[Baluncore]

My pet peeve is any plot of “gain” from 0 dB at the top to –120 dB at the bottom, then wrongly naming that axis “attenuation”.

The gain of an antenna should be specified in dBi because it is the effective power gain in the direction of the main lobe, relative to an omnidirectional radiator.

A dipole has a radiation pattern gain of approximately +2.15 dBi because there is no energy radiated in the axial direction, so more is radiated broadside, which is where dipole gain is measured. Gain in that case is not a measure of asymmetry, since the presence of an axial null would give it infinite relative gain.
dBd can only have meaning in the direction of the assumed boresight.

 

[Cup of Joe]

Man, I didn't think I would keep the conversation for this long! I must be a very interesting person...

 

[Baluncore]

Man, I didn't think I would keep the conversation for this long!

The perennial interest in oscillators obviously has a gain significantly greater than +1.
That leads to thread distortion, pulling the topic off-subject.

 

[LvW]

Man, I didn't think I would keep the conversation for this long! I must be a very interesting person...

Why are you surprised? It is quite normal that in a conversation - even if it is about technical questions - some words or sentences appear that do not directly belong to the topic, but are also worth discussing.
Sometimes, moderators try to stop such discussions - unfortunately!

 

[LvW]

The gain of an antenna should be specified in dBi because it is the effective power gain in the direction of the main lobe, relative to an omnidirectional radiator.

Yes - this makes much sense because in the design of a communication link we always start with the EIRP (Equivalent Isotropic Radiated Power) which is delivered to the antenna.

 

[sophiecentaur]

Yes - this makes much sense because in the design of a communication link we always start with the EIRP (Equivalent Isotropic Radiated Power) which is delivered to the antenna.

Hot Dang! I was just going to introduce EIRP and you got in there first.

 

[alan123hk]

For voltage, the oscillating condition of the oscillator is $~\beta A=1$

For power, the following equations can be defined for the oscillator

$P_i = \frac {V_i^2} {R_i^*} ~~~~~ P_o = \frac {V_o^2} {R_o^*}~~~~\Rightarrow~~~~\text{Power Gain}~ (G) =\frac {P_o} {P_i} = \frac {V_o^2} {V_i^2} \frac{R_i^*}{R_o^*} = A^2\frac{R_i^*}{R_o^*} ~~~~\Rightarrow~~~~A^2 = G \frac{R_o^*}{R_i^*}$

In the same way, the equation of $\beta$ can be obtained, which is $~~\beta^2 = \alpha \frac{R_i^*}{R_o^*} ~~~$, where $\alpha$ is power feedback coefficient.

Then the situation becomes as follows.

$~\beta A=1 ~~~~\Rightarrow~~~~ \beta^2 A^2= 1~~~~\Rightarrow~~~~ \left( \alpha \frac{R_i^*} {R_o^*} \right) G \left( \frac{R_o^*}{R_i^*} \right) =1~~~~\Rightarrow~~~~\alpha~G=1$

So I believe $~~\beta A=1 ~~$ and $~~\alpha~G=1 ~~$ are equivalent.

 

[VirginiaGuy]

Bottom line: You need gain. You get to choose whether you call it voltage, current, or power gain.

and the input, output, and load impedance all determine what 'flavor' you want.
If you are feeding a high impedance load from a lower impedance source, voltage gain is generally what you want.

Low impedance loads often rely more on current gain in the preceding stage
to ensure adequate signal power transfer into the load impedance.

Driving a 50 ohm load typically requires a 50 ohm source.
Driving a 1 mega ohm load more typically requires a voltage source.

Things can get rather strange when power levels are very high.
Like driving an antenna with megawatts fr broadcast.

We used to drive self protection jammers on aircraft at such a
high level the waveguide needed SF6 pressurization to limit arcing.
Air at 60,000 ft has some very different characteristics than at 30,000 ft.
We needed to be able to generate 100s of MW pulses.
Capable of damaging the radar receiver of whatever illuminated you.

 

[sophiecentaur]

@VirginiaGuy if gain is required then it’s Power Gain. However you look at input and output signals and their impedances you need Power Gain.
People can worry too much about what to call things. It’s Energy that makes things happen.

 

[DaveE]

People can worry too much about what to call things.

Absolutely. It's understanding what's happening that matters much more than putting things into one or another pigeon hole.

if gain is required then it’s Power Gain.

It’s Energy that makes things happen.

Yes but sometimes energy isn't a limited thing. Sometimes one representation is easier to deal with than another, or illuminates the underlying behavior better. For example an audio voltage amplifier, where the power supply can supply whatever energy the feedback system requires. It would be rather tedious and not very illuminating to think of this circuit's operation in terms of power, when it is voltage that is input, fed back, and output. In this case, everyone will calculate the power after they have found the voltage and impedance information.

Since the OP asked about oscillators, and oscillator behavior is all about feedback, I don't think it makes sense to think in any terms other than the simplest format for whatever signal you are using in the feedback loop. This is sometimes, but not usually power, even in RF designs. Unity feedback is the benchmark, however you measure it.

 

[sophiecentaur]

This is sometimes, but not usually power, even in RF designs.

If you include Impedance in that statement then it's got to be Power gain. A transformer doth not make an oscillator oscillate.
And, on the subject of gain; it's very naughty of people to talk about voltage gain in dB unless the impedances are explicitly quoted. It accounts for a load of confusion and misunderstandings when people try to follow the 'formula' for a voltage ratio and include the "20" instead of the "10" without having a clue why.

 

[Averagesupernova]

If you include Impedance in that statement then it's got to be Power gain. A transformer doth not make an oscillator oscillate.
And, on the subject of gain; it's very naughty of people to talk about voltage gain in dB unless the impedances are explicitly quoted. It accounts for a load of confusion and misunderstandings when people try to follow the 'formula' for a voltage ratio and include the "20" instead of the "10" without having a clue why.

This may be a matter of semantics and wording. Of course, we cannot take a transformer and expect it to function alone as an oscillator. But to say we cannot take a transformer and use it in the feedback path to step up the voltage and get an oscillator to run would be wrong. Of course it is implied we are using a valve or transistor of some sort.
-
Now as to this voltage/db thing, that's a slippery slope. Decibels are commonly used with voltage and I know it ticks some people off severely. The unit dBmV is a reference of zero dBmV is 1mV into a given load. It is commonly used in the cable TV industry. If the power goes from zero dBmV to 10 dBmV the power has gone up 10 times.
-
Voltage itself is spec'd in dB increase or decrease, but I don't find it all that common. At lease not in amplifiers. One place I can think of dB is used to spec in the form of voltage is crosstalk for instance. Take a compact disk player with line level outputs. We don't really care if they are terminated at exactly 10K ohms or whatever. Many amplifiers that a CD player would plug into vary within a range. As long as both channels are terminated the same, that's fine. We measure the voltage, do the math with '20' and state the spec.

 

[sophiecentaur]

But to say we cannot take a transformer and use it in the feedback path to step up the voltage and get an oscillator to run would be wrong. Of course it is implied we are using a valve or transistor of some sort.

Of course. You need an active device in there somewhere and it provides extra Power somewhere. All a transformer can do is to change the impedance without amplification.

One place I can think of dB is used to spec in the form of voltage is crosstalk for instance.

(again) of course. You are measuring a ratio of signal levels at the same point so the impedance that you measure the crosstalk can only be the same.
Insert a 6dB amplifier or a 6dB attenuator and you can only be sure that it will do what the label says if you are using the standard impedance throughout.

Decibels are commonly used with voltage and I know it ticks some people off severely.

Yes, it does tick them off because, as you know, people are horribly confused because they don't know what's going on and you hear "Is that Voltage dBs??" The proof of the pudding and all that. So best to steer clear and describe voltage ratios in plain numbers.

 

[Averagesupernova]

So best to steer clear and describe voltage ratios in plain numbers.

Agreed. But, sometimes crosstalk or separation is over 100 db. That's a bit much to chew off in a voltage ratio. What you said about the transformer changing impedance is dead on correct. That can be said of any situation involving transformers unless isolation is the objective.

 

[DaveE]

If you include Impedance in that statement then it's got to be Power gain.

OK, go for it. There's nothing wrong with expressing everything in terms of power. Occasionally you will annoy people, but you won't be wrong. OTOH, "If you include impedance" then voltage, current, or power are all acceptable. Not necessarily equally optimum for analysis, but acceptable. In that case it clearly doesn't have to be power. You don't have to be an expert EE to know what the voltage or current is for 10dBm into 50 ohms.

I still contend that having the flexibility to choose the representation that is most clear, most standard in industry, or easiest to manipulate is the best option. There are many signal processing applications, like control systems or audio, where power is relatively insignificant. Just as there are other systems like RF receivers and transmitters where power is of utmost importance. It is myopic to insist that various engineers all use the same tools or have the same preferences.

And, on the subject of gain; it's very naughty of people to talk about voltage gain in dB unless the impedances are explicitly quoted. It accounts for a load of confusion and misunderstandings when people try to follow the 'formula' for a voltage ratio and include the "20" instead of the "10" without having a clue why.

Yes, as I previously said, there is a burden on people to communicate clearly the assumptions made or salient circuit details required to know what ratio is being discussed. Otherwise, it sounds to me like you are describing people that don't know enough about dB's to be using them at all. I agree just "following a formula" is a horrible idea in this regard.

 

[Averagesupernova]

Otherwise, it sounds to me like you are describing people that don't know enough about dB's to be using them at all.

Bingo!