Desired value of L and C in colpitts oscillator?

[samu]

Hi everyone,

I´ve been searching in the forum but i didn´t find this information.

I just built a "VLF like" metal detector that works with a colpitts oscillator. It´s not a thunting model, it´s used to detect large metals in a conveyor belt. It usually works fine, but sometimes it makes a fake detection. I´ve tried so many different coils and there are some coils that works fine. Also, if i change the caps, the metal detector works better or worse.
What i want to know, is what are the best values for everything. I mean, do i have to improve Q?, do i have to get less R?, the problem is the frequency?, maybe some spurious frequencies?.
I used the values on the schematic but i don´t understand why this values and no others. I saw a metal detector like this that uses other cap of 5,6nF in parallel with the 10nF one when a larger coil is used.

Maybe the coil value is wrong in the schematic. I calculated it but i didn´t measure it. It´s a 500x250mm elliptical coil made with 9 turns of 0,9mm copper wire.

Thank you in advance, any help welcome, and sorry for my English, i´m still learning.

[Qiaozhi]

Hi everyone,

I´ve been searching in the forum but i didn´t find this information.

I just built a "VLF like" metal detector that works with a colpitts oscillator. It´s not a thunting model, it´s used to detect large metals in a conveyor belt. It usually works fine, but sometimes it makes a fake detection. I´ve tried so many different coils and there are some coils that works fine. Also, if i change the caps, the metal detector works better or worse.
What i want to know, is what are the best values for everything. I mean, do i have to improve Q?, do i have to get less R?, the problem is the frequency?, maybe some spurious frequencies?.
I used the values on the schematic but i don´t understand why this values and no others. I saw a metal detector like this that uses other cap of 5,6nF in parallel with the 10nF one when a larger coil is used.

Maybe the coil value is wrong in the schematic. I calculated it but i didn´t measure it. It´s a 500x250mm elliptical coil made with 9 turns of 0,9mm copper wire.

Thank you in advance, any help welcome, and sorry for my English, i´m still learning.

What you have posted here is a strangely drawn Colpitts oscillator.

Do you know the frequency at which it is supposed to oscillate? The values you have given should result in a TX frequency of approximately 182kHz, which I would have thought was somewhat high.

This only appears to be the transmitter circuit. Where is the rest of the design?

[ivconic]

Samu's post gives me excellent preamble for next question;
(we can talk here or start new thread. But i think this is closely tied to his question). Consider this, which situation would be better; larger capacitance&smaller inductance or smaller capacitance&larger inductance?
I met some texts where author say one thing and in other another.
Larger capacitance&smaller inductance seems to give better results latelly at newer designs...or i just pulled out wrong conclusion?
This is laic question, to generalized.
But...would be good to explain phase&amplitude relation to LC relation at one known oscillator, as an example. For example we can take....let's say White's Classic oscillator.

[Qiaozhi]

Samu's post gives me excellent preamble for next question;
(we can talk here or start new thread. But i think this is closely tied to his question). Consider this, which situation would be better; larger capacitance&smaller inductance or smaller capacitance&larger inductance?
I met some texts where author say one thing and in other another.
Larger capacitance&smaller inductance seems to give better results latelly at newer designs...or i just pulled out wrong conclusion?
This is laic question, to generalized.
But...would be good to explain phase&amplitude relation to LC relation at one known oscillator, as an example. For example we can take....let's say White's Classic oscillator.

Here are my thoughts on this subject:

For a given TX frequency, you are limited by the transmitter circuit. If C is made too large, or L is made too small, it will fail to oscillate. Therefore, within this range of values, which is the best compromise?

Using a larger C will help to eliminate capacitive problems with the ground. This is because the ground capacitance will be swamped by the large value of C. On the other hand, a larger value of L will provide a greater electromagnetic field density around the coil, that will help to improve detection depth.

Personally, I think the safest and most reliable solution is to use the mid-range values for L and C, that are suitable for the particular TX oscillator design. Of course, making a few different coils, and doing some real world testing, is the only way to really find the answer.

[samu]

Hi,

First of all thanks to Qiaozhi and Ivconic for the answers.

I just realized that the type of metal detector is not "VLF like" but "BFO like", sorry. Actually, the oscillator out signal is integrated and then the changes are measured in a comparator.

The oscillation frequency is between 120Khz and 200Khz depending on the coil and caps used. For the caps and coil posted, the frequency is about 160 KHz.
I think that depending on the coil, there is a best value for the caps. I saw a metal detector like mine with a bigger coil that use a 5,6nF cap in parallel with the 10nF one. I´m sure that somebody tuned the circuit looking for anything i don´t know. The frequency in that case was 120 Khz.

My biggest problem is that with some coils (for example 11 turns) the metal detector make fake detections sometimes. Maybe is more sensible to EMG noise when the circuit is not perfectly tuned. Also there is another problem because the fake detections sometimes spend a lot of time in appear. For example, the MD works fine and one month after it starts to fail.
I don´t have a coil factory so i can´t build all the coils exactly equal, so the coil that works fine once, next time works bad.
I have an oscilloscope with FFT, ¿do you think the problem are the spurious frequencies?.

I know this metal detector works really well when all is ok, the problem is that i don´t know when all is ok.

[Qiaozhi]

From your description it sounds like this is not even a BFO, but it is relying on a change in coil inductance to signal a target. The false signals are most likely caused by temperature drift. Unless the rest of the circuitry has some auto-track circuitry, you will need to constantly adjust the trigger level of the comparator.

You are probably correct about needing optimum values for the L and C. The amplitude change at the output, when a target is present, will be maximized with these values.

If you think that 120kHz is the correct frequency, then you can try either of these solutions:

1. Replace the capacitors with 220nF and 22nF, keeping the 85uH coil. This will result in a TX frequency of 122kHz.
2. Make a new coil of 193uH, keeping the 100nF and 10nF capacitors. This will result in a TX frequency of 120kHz.

If you have access to a SPICE simulator, you could try various Ls and Cs to find the most sensitive values.

[samu]

The circuit has an auto-tracking system.

I´m trying to simulate the circuit using PSpice. I added a sin (0,05V 50Hz) source serially with the DC source to make the oscillation starts.

The results are not what i expected, the frequency is about 800Khz and the amplitude is only about 50 mV (in the real circuit is about 22V). I changed the values of caps and coil looking for results that seem to be logic, but i found nothing.
I´m using a time domain simulation with 0,01us as max step size.

I also tried an AC analysis connecting a VAC to the transistor base. The result is a peak at frequencies much higher than i expect. For coil=85u and 10n and 100n caps the peak is situated about 1MHz.

I´m using Orcad Capture 9.2.

Do you think the problem can be that the coil is not perfectly in only one plane?. I mean that the coil has some folds like a wave because is not placed in a strong material.

Thank you very much.

[Qiaozhi]

You can start the oscillator in SPICE by setting an initial condition, or by using a pulse source for the power supply, so that the power is supplied after the simulation has started.

Have you included the DC resistance of the coil? This is important. Otherwise you will have an ideal inductance.

[Rov]

Why dont you have a look to calculate the L and C for a colpitts oscillator from the ARRL handbook,its got many examples to research.these guys have a wealth of research on inductance,capacitive,frequency to suit most any purpose.The wealth of info from the ARRL is overlooked so often for solutions,that others repeatedly try to simulate without success,because they dont look beyond the obvious?especially electronics an well known L and C circuits. an Collpits an derivatives.
Use a Decade R or C box in your coil loading to the coil in real time/real situation with your coil an observe the changes,one step at a time ,ON site!as a tech would do!

[Qiaozhi]

Why dont you have a look to calculate the L and C for a colpitts oscillator from the ARRL handbook,its got many examples to research.these guys have a wealth of research on inductance,capacitive,frequency to suit most any purpose.The wealth of info from the ARRL is overlooked so often for solutions,that others repeatedly try to simulate without success,because they dont look beyond the obvious?especially electronics an well known L and C circuits. an Collpits an derivatives.
Use a Decade R or C box in your coil loading to the coil in real time/real situation with your coil an observe the changes,one step at a time ,ON site!as a tech would do!

There is no need for the AARL handbook for this circuit. The calculation is quite simple:

f = 1 / (2pi*sqrt[*L*Ct])
where Ct is the value of the two capacitors in series. i.e. Ct = C1*C2 / (C1+C2)

I would have posted this using Latex, but it appears to be broken on the new website.

[Rov]

Good Work!,you been watching!Samu may well advance now!Thanks!Ive worked in environments with conveyors an live metal detectors for quality control of product an contaminants,an variety of detection circuits ,many proprietry built? an obscure?an had to drag the test gear to site to fix!not always easy!but its a job!Samu your call!

[samu]

I´ve tried thousands of solutions and no one worked. I´m using a 1.2ohm resistance in serial with the coil but this is not the solution. I also changed the power source for a pulse source. Now the oscillator starts but the frequency is about 1Mhz and the amplitude about 50mV what are not the expected. I tried a lot of values for coil and capacitors.

I´m simulating exactly the circuit posted (adding the pulse source and the coil resistance).

This formulas are what i used to calculate the inductance of the coil. I also calculated it with some internet programs and the value is about 50uH to 100uH depending on the web and the exact values. The value of 85uH seems to be correct. What do you think?

I don´t know what else can i try. Perhaps i have to simulate under moonlight or at 12 o´clock the carriage turns into pumpkin, i don´t know.

Thank you everyone!

[Qiaozhi]

Can you post the SPICE netlist here, and I'll take a look.

[chemelec]

Samu, Just for fun, I Prototyped up this circuit with real parts.

I can't get it to oscillate at all, reguardless of any changes I made.

Where did you origionally get this circuit from?

Gary

[chemelec]

This circuit will do what you want.

And it will work with many different inductance coils and the size of the coil shouldn't affect its operation either.

http://www3.telus.net/chemelec/Temp/Det-Osc.png

Take care...Gary