Hi, I have am looking for an oscillator architecture to drive a variable inductance tank. I want to be able to quantify a changing inductance. I have tested the colpitts architecture below by observing peak voltage however my concern is of the temperature effects of using 2 capacitors for the tank - the 2 capacitors may have different drift coefficients making compensation more challenging. Is there a preferred oscillator architecture for an application like this using 1 capacitor and 1 inductor? Unrelated, is there anyway to maintain constant amplitude of the tank and measure the current required to maintain the constant amplitude and inductance changes?

 

What frequency range are we talking about?
Is the variable inductance tank shown in the diagram or is it the load for Vout?
What is up with using a 50 year old ultra crappy opamp for this project?

 

I do not recommend the shown circuit.
The problem is as follows: When changing/tuning the inductance, the oscillation condition will be changed as well. Hence, you must adjust the gain correspondingly.
Why not using the classical LC-tank configuration (bandpass, instead of 3rd-order lowpass as in your circuit) ?
Design a positive feedback path consisting of a resistor R2 in series with an LC parallel block with a parallel resistor R1.
At resonance the pure resistive pos. feedback factor R1/(R1+R2) must be compensated by a negative feedback factor R0/(R0+RF).
That means: The opamp must have a gain of (1+RF/R0).
(In reality, somewhat larger because of oscillation start and due to tolerances ).
Any change in the L value will shift the resonaance frequency but not the oscillation condition.

 

What frequency range are we talking about?
Is the variable inductance tank shown in the diagram or is it the load for Vout?
What is up with using a 50 year old ultra crappy opamp for this project?

The variable inductance will be the L in the tank. I also am not using this opamp I just found this photo to show the general configuration.

Why not using the classical LC-tank configuration (bandpass, instead of 3rd-order lowpass as in your circuit) ?
Design a positive feedback path consisting of a resistor R2 in series with an LC parallel block with a parallel resistor R1.
At resonance the pure resistive pos. feedback factor R1/(R1+R2) must be compensated by a negative feedback factor R0/(R0+RF).

Thanks I have just worked up this simulation in single supply configuration. For parallel resistor R1, I am considering the parallel resistance of the coil for this. This parallel resistance will vary as the inductor varies and with temperature. Is there some condition tat has to be met or I just need to ensure some gain factor of (1+ Rf/R0), i.e. 2 to 3?

 

The variable inductance will be the L in the tank. I also am not using this opamp I just found this photo to show the general configuration.



Thanks I have just worked up this simulation in single supply configuration. For parallel resistor R1, I am considering the parallel resistance of the coil for this. This parallel resistance will vary as the inductor varies and with temperature. Is there some condition tat has to be met or I just need to ensure some gain factor of (1+ Rf/R0), i.e. 2 to 3?

WHY will you rely upon a resistance that has value that is unknown?
Is there any specific reason? Use a resistor in the lower kOhm range in parallel - and you do not depend on such unknown quantities...

 

So what opamp will you use and what frequency ranges will this be for?
How do you plan to very the inductance?

 

I have some of these lying around that I am going to use. My tank will be less than 250kHz

https://www.analog.com/en/products/ada4522-2.html

It is an eddy current sensor. So from what I'm thinking I can either sense the parallel resistance change of the tank or measure the inductance. Simplest way to measure I believe will be to measure the oscillation amplitude loss with a precision peak detector. Resonant frequency shift will be another method. If inductance and Rp changes over temperature and with eddy current effects, I imagine that I will have to fix (compensate) one of them and measure the other

 

I have some of these lying around that I am going to use. My tank will be less than 250kHz

https://www.analog.com/en/products/ada4522-2.html

It is an eddy current sensor. So from what I'm thinking I can either sense the parallel resistance change of the tank or measure the inductance. Simplest way to measure I believe will be to measure the oscillation amplitude loss with a precision peak detector. Resonant frequency shift will be another method. If inductance and Rp changes over temperature and with eddy current effects, I imagine that I will have to fix (compensate) one of them and measure the other

Your opamp is going to run out of gain at the top of that range.

 

The closed loop gain starts to roll off between 100 and 200 kHz. Assuming I need a gain of 2-3 shouldn't it be OK? I can always lower the tank frequency a bit. Should I be looking for something just a bit higher?

 

The closed loop gain starts to roll off between 100 and 200 kHz. Assuming I need a gain of 2-3 shouldn't it be OK? I can always lower the tank frequency a bit. Should I be looking for something just a bit higher?

Never mind. I did not look closely at the graph labeled "Voltage Noise Density". I thought it was showing Unity Gain Bandwidth.
So back to how you plan to vary the inductance.

 

So back to how you plan to vary the inductance.

Its an eddy current sensor, so the inductance (and presumably inductor parallel resistance) will change based on the distance of a metallic object. It is still unclear to me what the best thing to measure is, resonant frequency shift or amplitude. I hypothesize that if there is enough feedback gain that the amplitude will not drop, only the frequency will shift. If gain is dialed in correctly, then the amplitude of the oscillator will drop as eddy current effect is happening. In either case, I can compensate for Rp shifting over temperature I believe by including a parallel resistive network.

 

Its an eddy current sensor, so the inductance (and presumably inductor parallel resistance) will change based on the distance of a metallic object. It is still unclear to me what the best thing to measure is, resonant frequency shift or amplitude. I hypothesize that if there is enough feedback gain that the amplitude will not drop, only the frequency will shift. If gain is dialed in correctly, then the amplitude of the oscillator will drop as eddy current effect is happening. In either case, I can compensate for Rp shifting over temperature I believe by including a parallel resistive network.

OK. How is the inductor constructed?
BTW - I did not think there was a parallel resistance in an inductor model.

I also don't think the inductance will change, but the impedance will appear to change because of the drop in current when the main field and the eddy current field interfere with each other.

 

OK. How is the inductor constructed?
BTW - I did not think there was a parallel resistance in an inductor model.

I also don't think the inductance will change, but the impedance will appear to change because of the drop in current when the main field and the eddy current field interfere with each other.

magnet wire coiled around a ferrite core. It indeed has parallel resistance tested at AC, in addition to series resistance, confirmed on an impedance analyzer.

 

magnet wire coiled around a ferrite core. It indeed has parallel resistance tested at AC, in addition to series resistance, confirmed on an impedance analyzer.

To the extent that the field of the inductor is affected by a metallic object, I think it will be easier to measure the drop in current as opposed to any other circuit parameter.

I was expecting you to say it was an iron core. AFAIK, ferite cores are used to prevent eddy currents. Am I missing something?

 

One more thing. With the values in post #1 the resonant frequency of the LC combination is just over 10 kHz. Were you planning to use different values in a subsequent design? Maybe you meant 10 μH for the inductor.

 

That circuit was just repsentative of the architecture, not of what I need to use.

Having said that, how would the circuit look to measure the current? In my initial post I was thinking to do that and I would be interested in a clever way to do it.

 

That circuit was just repsentative of the architecture, not of what I need to use.

Having said that, how would the circuit look to measure the current? In my initial post I was thinking to do that and I would be interested in a clever way to do it.

I'm not sure of the magnitudes of the currents or the magnitude of the current changes so you'd have to give me some idea of what we'd be dealing with.

 

Attached is what I have so far. I would say something like this to start with. How to convert to measure and convert to DC current? Looks to be roughly 100 uA per 100 uH change (ignoring any RP change).

TI has a device that seems to measure L and RP. http://www.ti.com/lit/ds/symlink/ld...ll-mousermode-df-pf-null-wwe&ts=1589681810244

 

Attached is what I have so far. I would say something like this to start with. How to convert to measure and convert to DC current? Looks to be roughly 100 uA per 100 uH change (ignoring any RP change).

TI has a device that seems to measure L and RP. http://www.ti.com/lit/ds/symlink/ld...ll-mousermode-df-pf-null-wwe&ts=1589681810244

View attachment 207344View attachment 207345

Cool package. I've never seen one of those before.

 

The 200 ohm and 3k ohm resistor values in your circuit are much less than the 10k resistors used in the datasheet, because your resistors need an opamp that produces much more output current than your opamp and almost any other opamp can do.