The explanations given explain why I never see an amplifier with an inductance , and usually not the load, in series with the feedback loop. In most amplifiers the feedback is taken from a point close to the output, but never past the output. Not even in regulated power supplies is the voltage sensed on the return side of the load. Now you have learned the reason why feedback is not taken from the other side of the load.. That is because usually it does not work.
The whole purpose of negative feedback is to correct distortions at the output.

 

Hello, could you please answer the last part of post 17,i need to know how to see the phase margin situation based on the oscilation i get on the scope, THE AC responce barkausen condition.

Quote: My goal is to learn how to investigate stability issues of a circuit i have built

Because your circuit has no overall feedback, any possible ringing can be caused by the last stage only (as I have mentioned earlier).
This stage seems to be critical because there is an L-C-lowpass in the negative feedback loop which adds negative phase shift to the loop.

This loop has already a phase shift of 180deg (due to neg. feedback). In addition, the real opamp (with finite and frequency-dependent gain) will add further phase shift to the loop. As a result, the total phase shift will probably not reach the critical 360deg limit (oscillation), but it will be rather close to it. That is the reason the step response shows severe ringing.

For a detailed analysis I propose to simulate the loop gain for the last stage only.
Are you familiar with loop gain analysis and the method to find the phase margin?

 

My circuit has a very simple goard i put DC voltage of 10mV and i get 20mA on the inductor.
pure DC signal.
My goal is to learn how to investigate stability issues of a circuit i have built
I can do LTSPICE sinulations to learn how to tune the circuit into stability.
There could be many issues, maybe i have noisy power supply .
Maybe i have bad decoupling circuit.i need to learn how to investigate in real life.

suppose i have oscilations on my scope instead of DC, how can i know the phase margin situation from it?
looking at the AC responce i cant see anywhere a point where i have 0dB and 180 phase(barkausen condition)

Thanks.

View attachment 312409
View attachment 312410

Hello again,

If your goal is to learn how to analyze stability then you should get a book on control theory. That should help. However, there will be times when you run into theoretically stable circuits that do not appear to be stable in the real life bench test. This is where the 'hands on' experience comes into play. You have to figure out what is causing the instability that does not appear on the schematic. This involves various parasitic elements which include the routing of traces and wires and possible the proximity of some parts to other parts. It may not be easy sometimes. The only way to learn how to do this is by experience with lots of circuits and troubleshooting them.
A pretty good way to analyze stability is to do a root locus analysis. This is probably the best way. However, if the circuit is stable in theory and not on the bench again, you have to turn to your hands on experience once again.

Since it looks like the first two stages are stable as is, it is probably that last stage that is causing a problem. One thing you can do is try to replace that stage with a different circuit.

That said, a quick fix MAY be to place a resistor in parallel to the inductor. This in theory, makes the circuit very stable. The value of this resistor would have to be determined by you because it does create a DC offset in the output of the op amp circuit which may or may not be acceptable.
Another idea that often works is to slow down the op amp by making it a slow integrator. That is by placing a capacitor or RC network from the output directly to the inverting input. This slows down the response to the output changing so the op amp has time to catch up. In this circuit though it would be in parallel to the inductor, which may cause an oscillation.
The general method is to use an RC compensation network. Since the circuit does not seem unstable as is though, it may be hard to choose the values except by experimentation on the bench.

I can show you a general method to solve for the output of a circuit like that last stage. It helps in most cases possibly not all.
Do you understand transfer functions like this or similar:
Vout/Vin=1/(s+1)

For example, one technique is to handle the op amp as a voltage controlled voltage source (VCVS).
With this method, you analyze the output transfer function let's call it Tos. Since this will be the network connected to the output, the feedback signal will be:
Vfb=Vout*Tos
The input we can say is just Vin for now. This means the difference signal is:
Vdiff=Vin-Vfb
Pretty simple.
Now to produce the output, the op amp gain multiplies that by its internal gain we can call 'A'. This leads us to:
Vout=A*Vdiff
Writing this out more explicitly, we end up with:
Vout=A*(Vin-Vout*Tos)
With that we can see that Vout appears on both sides, so we need to solve for that explicitly. We get:
Vout=(Vin*A)/(Tos*A+1)
Now take the limit as A goes toward infinity and we get:
Vout=Vin/Tos

This looks simple, but Tos is function of 's' and can be very complicated with both a numerator and denominator. It can however then be analyzed in the frequency domain by replacing 's' with "j*w". In the time domain, we have to take the Inverse Laplace Transform.

If this looks like something you can do I can show an example.

Another example handles the op amp as a voltage controlled current source (VCCS) with one pole and low output impedance.

 

Once again, the source of the instability is the incorrect application of feedback. The load is not part of the amplifier, and yet it has been included in the feedback loop, where it provides a phase shift that assures instability. If you move the feedback point to the junction of the emitters of Q1 and Q2 the amplifier will be quite stable with no other changes. To control the current in the inductor you can add a current sense resistor in series and used a differential amplifier to provide a feedback signal referenced to the circuit common.

Often the simplest scheme is not used for the reason that it does not work.

 

Once again, the source of the instability is the incorrect application of feedback. The load is not part of the amplifier, and yet it has been included in the feedback loop, where it provides a phase shift that assures instability.

Hi,

That's not theoretically accurate. Do a complete analysis to explore this idea.
Perhaps you can explain why you came up with this idea in the first place. There is no theory I know of that says anything about where the load should be placed, but perhaps I misunderstand your idea.

 

Once again, the source of the instability is the incorrect application of feedback. The load is not part of the amplifier, and yet it has been included in the feedback loop, where it provides a phase shift that assures instability.
....................................
The explanations given explain why I never see an amplifier with an inductance ,
.................................
That is because usually it does not work.
The whole purpose of negative feedback is to correct distortions at the output.

We should not forget what the TO wrote: My goal is to learn how to investigate stability issues of a circuit i have built

For my understanding, this means that we should not discuss if the circuit will work or if there are "better" circuits.
The TO gave us a circuit - and based on this example he wants to study some questions related to "stability".
In this context Ithink
1) It is not clear that the circuit will be unstable - instead, it will most probably exhibit a small phase margin with a step response that will show some ringing. Under worst case conditions it will continuously oscillate (depends on the opamps specification)
2.) What is the purpose of feedback? I think, there are several purposes:
* Better linearity (less distortions)
* Adjustment of gain
* Creation of complex poles (filter theory)
* Fundamental concept of control loops

 

Stability issues are certainly related to an amplifier being useful. In addition, the application of feedback connections not being in series with the load has been used for at least the past few decades for some reason, such as producing the desired results.
Certainly the TS was able to observe "stability issues", and certainly the explanations of the cause, admittedly without the detailed math explanations, explained the issue.

 

Hello ,The heart of the story is the inductor of the YIG.
I need to construct a device in which we input 1mV and we get of the coil output 20mA.
The LC meter shows L=6.04uH and Cx=0.015uF.
The input to the device is DC there is no need for bandwidth. (because YIG works purely on DC)
What OPAMP do you reccomend me to use, so it will be low noise and stable?
given my goals Could you please reccomend me some structure i can try and simulate with this opamp?
Thanks.

 

What is a "YIG" ?
What does it do ?
.
.
.

 

A "YIG" is a sort of tunable microwave oscillator that is frequency controlled by a magnetic field. So the plan is to set the current in the control electromagnet to a specific value to establish a specific frequency. I have seen circuits of current amplifiers that use an op-amp and a pass transistor, they do not look at all like the circuit in this post. One thing about the electromagnet in a "YIG" device is that it is only one polarity. So one would never drive it with an audio amplifier. Often an arrangement called a Current Follower is used, and probably that would work very well for the application, whatever it is. Unfortunately I was not able to locate a current follower in my recent search of my archives.

 

Then this Thread would appear to be all about the
Frequency-Response of the YIG's change in Output-Frequency
that can be generated with reasonably low-distortion.

And it took ~30 posts to finally get close to this point.

Why can't a reasonably accurate statement of the
required specifications ever be made in these Forums ?,
must everything be "Top-Secret" ?,
or, "You don't need to know anything more than I've told You, just answer my questions" ?

This happens way too often.
.
.
.

 

In this particular instance it probably would not have made much difference, except that probably may folks had assumed that it was some sort of audio transducer. That had been my guess because usually that is what one drives with an audio amplifier. AND have you EVER seen audio negative feedback in series with the speaker???
FEEDBACK is to remove amplifier distortion. Also, feedback is fed back toward the input of an amplifier.
To control the load current a separate system will need to generate a signal voltage proportional to the load current and then feed it to a summing input where the set-point is fed in. Classical servo system arrangement.

 

In this particular instance it probably would not have made much difference, except that probably may folks had assumed that it was some sort of audio transducer. That had been my guess because usually that is what one drives with an audio amplifier. AND have you EVER seen audio negative feedback in series with the speaker???
FEEDBACK is to remove amplifier distortion. Also, feedback is fed back toward the input of an amplifier.
To control the load current a separate system will need to generate a signal voltage proportional to the load current and then feed it to a summing input where the set-point is fed in. Classical servo system arrangement.

Hello again,

I think you may be on to something here, but the load is not in series with the feedback really. It's in series with the lower resistor which is 100 Ohms.
What this means is that when the current in the inductor is 10ma DC, the voltage across the resistor is 1 volt DC. That 1v feedback is compared to the input and an error signal is generated internal to the op amp. The amplification then attempts to make both inputs equal by changing the current though the inductor. If the non-inverting input happens to have changed to 2 volts, the op amp will force more current though the inductor and the feedback voltage (previously 1v) will be raised to 2 volts, as the op amp slews up.

The overall effect is that the current in the inductor is low (presumably) and the op amp is regulating the current in the inductor.

Maybe that's not the best way to do it but that is what is being attempted so far. It could be a pole of the op amp messing things up so that should be checked too.

 

I've been thinking about how I would go about accomplishing this project
but there is a tremendous amount of vital, missing information, that MUST be provided first.
.
.
.

 

With feedback circuits there is that nasty variable of "Phase Margin", which if it is not adequate allows instability and even oscillation. That is a very good reason to avoid having an inductance in the feedback loop. The inductance of that coil is not mentioned, it might even be unknown. If the feedback voltage sample were taken upstream of the inductance the effect would be different.
I looked at the "schematicsforfree" site under power/ circuits/current sources and there were several current sources shown, both sinking and sourcing, and in every one of them the current sample for feedback path did not include an inductor.
Thus it seems that even if the load is an inductor, the feedback path should not include that inductor.

 

The Proposed-Schematic shows an Inductance of 1.6uH,
but doesn't explain much else.
.
.
.

 

Hello Mr Bill, could you please reccomend me a current follower like you said where the inductor is not in the feedback?
Thanks.

With feedback circuits there is that nasty variable of "Phase Margin", which if it is not adequate allows instability and even oscillation. That is a very good reason to avoid having an inductance in the feedback loop. The inductance of that coil is not mentioned, it might even be unknown. If the feedback voltage sample were taken upstream of the inductance the effect would be different.
I looked at the "schematicsforfree" site under power/ circuits/current sources and there were several current sources shown, both sinking and sourcing, and in every one of them the current sample for feedback path did not include an inductor.
Thus it seems that even if the load is an inductor, the feedback path should not include that inductor.

 

With feedback circuits there is that nasty variable of "Phase Margin", which if it is not adequate allows instability and even oscillation. That is a very good reason to avoid having an inductance in the feedback loop. The inductance of that coil is not mentioned, it might even be unknown. If the feedback voltage sample were taken upstream of the inductance the effect would be different.
I looked at the "schematicsforfree" site under power/ circuits/current sources and there were several current sources shown, both sinking and sourcing, and in every one of them the current sample for feedback path did not include an inductor.
Thus it seems that even if the load is an inductor, the feedback path should not include that inductor.

Hello again,

In short, it is a current regulator. Replace the 100 Ohm resistor with a 0.1 Ohm resistor and see if it makes more sense to you.
That's a typical way to regulate current.
This is actually a way to regulate a buck converter for really fast response using state variable feedback. State variable feedback in a buck means measuring the current though the inductor, and as you put it, that would be part of the feedback signal.

If this still does not make enough sense, replace the inductor with a capacitor, and you get an integrator which is very common. As you put it, then the capacitor would be part of the feedback signal. Apparently, that's not a problem as it is used in a lot of circuits.

Now as amazing as this may sound, a theoretical analysis with no parasitics produces a response for the inductor circuit that is very similar to the capacitor circuit. The only difference is the constants but not the general shape of the response.

Could it be that the parasitics we don't have information about is interfering more with the inductor circuit than the capacitor circuit. That is a possibility.

Another way to regulate current is to use a differential amplifier and sense resistor. That's the long way of doing it, but a lesson in theory.
In most circuits on the web we see a current regulator regulating current through a simple resistor. It's rarer to see a circuit regulating current though an inductor.
Maybe we can come up with a different way of doing it entirely and see how it works.

Maybe an idea is to create a buck circuit where we are more interested in the current though the inductor than at the output. That way we could regulate the current in a different way. The average DC current though the inductor may be determined by the output voltage and dummy load. I may look into this. That MUST be stable. Here we would be regulating the output voltage as usual, and inferring the current from that.

 

Hello ,I am Tuning a YIG FM coil with this device.
The frequency range of the YIG is about 12MHz at 9.5Ghz certer frequency.
i was wrong when thinking that my analog driver needs 0 bandwidth( DC input only)
I dont understand how the BW of the driver has anything to do with the YIG FM coil?

The FM coil data:
sesitivity: 450 KHz/mA
3dB bandwidth 2.2MHz
resistance 2Ohm
inductance 1.5uH

How this data influnces of the requirements i need to implement on the driver?
Thanks.

 

Hello ,I am Tuning a YIG FM coil with this device.
The frequency range of the YIG is about 12MHz at 9.5Ghz certer frequency.
i was wrong when thinking that my analog driver needs 0 bandwidth( DC input only)
I dont understand how the BW of the driver has anything to do with the YIG FM coil?

The FM coil data:
sesitivity: 450 KHz/mA
3dB bandwidth 2.2MHz
resistance 2Ohm
inductance 1.5uH

How this data influnces of the requirements i need to implement on the driver?
Thanks.

Hi,

Are you saying that you have to pass a 9.5GHz signal through this circuit to operate the 'coil' ?
Could it be that you just need a DC signal to alter a magnetic field in order to operate the YIG? If this is a more typical device then you only need to apply a DC signal. You may have a second coil for adjustments also.
If you had to pass a 1GHz signal through an amplifier (driver), that amplifier would have to be capable of working well at that frequency.

Also, why did you think you needed a push/pull output driver? What is wrong with an emitter follower for example?

Another note, it could be that the 100 Ohm resistor used with the coil is too high in value. Using a low noise op amp would allow using a much smaller resistor for the current sense I think.


Here are some other circuit ideas.
The lower circuit is a little more conventional if you can get away with a unipolar drive, which you probably can.