high output power colpitts oscillator

[lulezo]

Hi guys..
i am designing a colpitts oscillator and having some problems in the desired output power, the actual power i need is about 100 watt, surely i have to do some amplification, but the current is in the range of micro amperes !
so what am i suppose to do to increse both the voltage as well as the current in the circuit to minimize the ampilfication stages.

the attached file shows the circuit diagram and the output waveform.
please help.

 

[WimRFP]

Hello,

Does the frequency to be very stable, is a pure sine wave required?

What about the load, and does it vary under operation?

When frequency stability is not of importance, some distortion is allowed and you can live with about 70..80% efficiency, you can make a >100W Class C oscillator with a single transistor (did this several times) or 2 transistors (balanced oscillator).

When higher efficiencies are required, you can make a class D or E oscillator, this requires some more components.

When the impedance of the load varies strongly with frequency, you may get some problem when trying to generate the power with an oscillator only.

 

[lulezo]

many thanks for reply
yes the frequency must be stable all the time, because am trying to generate frequency to be the same as of the resonant frequency of a coil attached to the circuit, probably I'll couple the L1 (100uH) inductively with another coil.
no need for a pure sine wave, i may even use a square wave, so it doesn't matter.

 

[WimRFP]

Hello,

Some more info on the load would be nice (series resonant, parallel resonant)? What is consuming the power (just a resistor or sort of induction heating)?

The advantage of an oscillator is that when the coil detunes somewhat (for example due to load variation), the oscillator changes frequency automatically so that the LC circuit resonates again.

When you make a high efficient power amplifier (zero voltage or zero current crossing), you can destroy it with a wrong load

 

[lulezo]

ok then I'll illustrate my project in details:
am trying to transfer power wirelessly for about 2 meters (an experiment once done in MIT), the basic idea consists of a primary generating a magnetic field which in turn received by a secondary generating a current (as in transformer),.

Here the two coils (primary and secondary) resonating at the same frequency (working at resonant frequency makes optimum power transfer), and this frequency should be high (in the range of 3-10 MHz) because there is a relation between the wavelength and the radius of the aperture (coil) as in optics and antenna optics, working at that range of frequencies increase the quality factor.
The 100 watt needed because of the excessive losses which inevitably will occur.

the attached circuit wil explain exactly what i mean.

 

[FvM]

As a general comment, it seems unlikely to me, that you will be able to handle the project without some deeper knowledge of
electronic circuits, particularly RF related.

2N3055 in your first circuit has a gain bandwidth product of 2.5 MHz, it cna't work as a 3-10 MHz power oscillator. Although it's
not impossible to design a one-stage power oscillator with 100 W output, it's much more promising to use separate oscillator and
power amplifier. It's the only way to achieve stable output frequency and precise power control.

 

[karesz]

... because there is a relation between the wavelength and the radius of the aperture (coil) as in optics and antenna optics, ...

Hi,
Its relative unusual to make a coil with extreme high parasitic capacitances-but not inpossible.
I see the bigger problem on the secondary site/load coil, then it will maybe mobile and so are the "self capacitances" changing from all in the environment; self from your (a person) presence and moving between the two coils. Another question is the right rectifying at this frequency & power_ok for a bulb its not needed, but by another loads?...
Basically;
1,
I can not believe that so a system is realistically usable (I linked MITs script to an Eda-comment 2-3monts ago too), and its very poor as energy transfer efficiency_at 2 meter workings distance in all cases :-(...
2,
For usable work distance ,with realistic couplings efficiency, is the radius of transmitter coil to see/take!
Otherwise_I think_you must see a RF-energy transport system as antennas on both "coils"...
K.

 

[lulezo]

@FVM :
you are absolutely right about my poor knowledge in electronics, but when i chose this project, I was concerned only with the power transfer mechanism, convinced that I'll have a suitable oscillator in the lab. but since I discovered that all the signal generators I have, was of a very poor power output, I decided to build my own oscillator.
I chose the 2N3055 because it can handle up to 150W ,and i may use a frequency of 2 MHz ,if i find no other suitable transistor.
i mentioned that i have to use a seperate power amplification circuit, but I am not good in it, that s why i needed your help.

@Karesz:
the parasitic capacitance is very low (in the range of pF) and it is computed by the relation:
C = 2Ha (pf) (Medhurst's formula)
where a is the radius of the coil used in centimeters, H is a factor based on a table attached.
the experiment i shall perform wont include moving load (i.e mobile), the load will be static, and about the usability of the project, i agree with you, but thats the way all great discoveries develop, you have to start with a very low usability.
the radius of the coil used in my experiment is about 30 centimeters, using lower frequencies will increase that radius, because of the relation i mentioned above (to maintain a good quality factor).

 

[WimRFP]

Hello,

Fully agree with FvM. This is a double tuned circuit. When you use a single oscillator, it is likely that the oscillator will "jump" between two frequencies because of the double tuned circuit.

Regarding 3..10 MHz and 2m distance, you will very likely exceed EMC regulations and maybe radiation safety limits (ICNIRP guidelines).

I assume that the distance is large with respect to coil diameter. You are right, using resonance in the receiver coil greatly enhances the "capture area" of the coil, hence the overall efficiency.

Before spending lots of time experimenting, I would suggest you to first calculate the power transfer between the setup that you have in mind. When you are familiar with AC circuits and transformers this can be done. Maybe you want to do some actual Q factor measurements to calculate loss resistance for in your simulation.

Regarding frequency, when you stay on the low side, you can use regular switching mosfets. I made a full bridge power amplifier at around 700 kHz where I used a 10W oscillator to inductively drive the 4 mosfets. I have doubt whether this is easy to do at 3..10 MHz with standard SMPS mosfets.

IXIS RF and Microsemi (acquired APT) ) make special HF capable (expensive) mosfets. They have several application info that may give you some idea about the way to go. Forget bipolar devices, unless you have some bipolar RF power devices at hand.

Also radio amateurs (HAM) are using mosfets in their power amplifiers at the lower end of HF, based on easy to get standard SMPS mosfets. Probably you will not get >80% efficiency, but >100W output is possible. When you go for the highest efficiency, load mismatch is a problem.

My first thoughts go to a push-pull approach (with N-chan mosfets with source connected to ground, transformer coupling, Class C to E operation).

Added after 1 hours 55 minutes:

Hello,

Without the power amplifier you can do the analysis in combination with small signal measurements (except for the rectifier efficiency).

To quess the H field at certain distance you can use:

H = I*N*A/(2*pi*dist^3).
Only valid when distance >> diameter of coil
Product of I*N*A is called "magnetic moment [Am^2]

Vtx = 2*pi*f*L*I

Dist = distance between coils facing each other
A = surface area of coil (0.25*pi*diameter^2)
I = current through coil
N = number of turns.
f = frequency
Vtx = transmitter voltage across transmitting coil
u0 = permeability of vacuum


For the reception coil:

EMF[V] = N*A*uo*H*2*pi*f

When you use same size coils (so same inductance) for transmitting and receiving, the ratio EMF/Vtx equals the flux linkage factor that you can use in the model for the coupled inductors in a PSPICE program. Now add the series loss resistances based on the assessed quality factor of the coils and capacitors, and you can start a simulation.

For the RX coil, add a series capacitor that cancels the inductive reactance and add the load resistance. When load resistance equals series loss resistance of RX coil, your efficiency will never exceed 50%.

For the TX coil, drive with a sinusoidal current source, don't add a capacitor and "measure" the voltage across it. By doing this you can ignore the detuning effect of the mutual coupling. This means when you change the distance, the receive coil will not detune in simulation

The input power you can calculate from Vrms*Irms*cos(phi). When you do a time domain simulation, just multiply V(t) and I(t) and integrate the result (low pass filter will also work, output of LPF equals real input power).

When loss in TX coil turns out to be less then 50%, you can increase the load resistor at the receiver.


As there is 1/dist^3 in the H-field, efficiency decreases rapidly with increasing distance. You will notice this as increasing the distance means that you have to reduce the coupling coefficient in the inductor coupling.

Regarding losses in the coil, you have the skin effect issue, but also a proximity issue, so use the correct formulas. If in doubt, make an inductor and determine the quality factor (for example by using resonance with a known capacitor).

 

[lulezo]

@WimRFP: many thanks for your useful informations, and so sorry for the latency.
now please lets focus on the oscillator circuit, am really no good at amplification circuits, so need a little help in designing the amplifiers, as i mentioned i need the output of the circuit attached above (oscillator) to be 100 watt (5A 20V). so for a quich start what am i suppose to do ?, it will be nice if any one could modify my circuit to obtain a larger output power, and then attach it.
thanks again WimRFP.

 

[WimRFP]

Hello,

What frequency would you like (please don't go too high)?

Before making the final decision on the oscillator approach, please read the previous posting carefully.

I use power oscillators frequently, both class C types and high efficiency types.

When you don't exactly know what you are doing, you will get the oscillator running (with a resistive load), but when adding your type of load, it is likely that you destroy your active device.

 

[lulezo]

ok let the frequency be 700 KHz, and which circuit do you recommend ? using a transistor or op-amp ?

i read your previous post and it was very helpful !

When you don't exactly know what you are doing, you will get the oscillator running (with a resistive load), but when adding your type of load, it is likely that you destroy your active device.

couldn't get it, but look at the attached load connection approach, it may help.[/quote]

 

[karesz]

Hi,
"Quote:
When you don't exactly know what you are doing, you will get the oscillator running (with a resistive load), but when adding your type of load, it is likely that you destroy your active device."
It means:
If you connect a practical load (mostly mismatched!) to a power semiconductor stage: it will be possibly (and rash) killed! :-(...
K.

 

[WimRFP]

Hello,

You were talking about 100W output. This means when using a 24V DC supply, you need at least 4.5A DC input to your circuit.

Your circuits in the annex will only provide mW, with very low efficiency. These are typically low topologies for low power. The have some chance on positive outcome off the experiment, I would suggest you to do the calculation/simulation of the complete inductive coupling (see my posting from 1 May). My first guess is that you need a high number for the product "Ampere*turns" in your 30cm coil.

What output power (of receiver coil) would you expect?

The inductive coupling from the secondary coil (receiver) to the output coil that will drive the load, is a good idea as you can change the coupling between the two very easy to get best output. I think you need a schottky rectifier circuit, as a LED may not be a good rectifier at 700 kHz and above.

Regarding the 700 kHz, this is a frequency where you can use standard SMPS mosfets to generate the required output power. Though it is also possible with BJT, I would not recommend this, as you need provisions to avoid deep saturation of the BJT. Though the radiation loss is negligible for a 30cm coil at 700 kHz, you may expect some interference to AM wave receivers. If you expect this, you may use a frequency around 500 kHz, this is below the AM broadcast band and further reduces radio frequency radiation.

When you accept low efficiency (say less then 20%, that is <20W output with 100W oscillator output), the power oscillator approach is feasible as the receive coil in that case will not result in wild impedance behavior of the transmit coil in the oscillator.

Added after 4 hours 19 minutes:

Hello,

To give you some idea of a power oscillator that can put over 90Arms through a 0.3uH coil, see the attachment. A wide flat strip formed to a 300mm diameter loop has about 0.3uH of self-inductance. Note that about 90Arms runs also through C1 and C2. Capacitors losses are present, but not accounted for in this simulation.

Dissipation in the inductor is about 80W in this simulation. Rload is the additional load because of the inductive energy transfer to the RX coil and dissipates about 70W. dissipation in the two mosfet is about 40W (20W in each IRF540).

Please note that this is just an example to give you some idea about the difference between a small signal and power circuit. The advantage of such class C oscillators with parallel resonance is that the current reduces automatically when removing the load. Of course a short circuit will destroy the mosfet when there is no supervisory circuitry present. power supply decoupling is not shown.

When you want to build such power circuits, you need to seek local advice from somebody that has experience with power circuuits.

 

[lulezo]

Thanx Karesz for illustration
@ WimRFP : the effeciency expected is about 40% ,therefore the expected power at the receiver coil is 40 watt.

i designed another circuit with a higher collector current, but the oscillations are declining although i made sure that:

gm*Rc > C2/C1 a condition for oscillations to start
please look at it and tell me whats wrong since am not familiar with such circuits.

 

[WimRFP]

Hello,

Your circuit dissipates all power in the collector and emitter resistor.

First, the voltage amplification of the amplifier is just one (because of the emitter resistor, this reduces effective Gm), so your loop gain is <<1

The loaded Q factor of the LC circuit is <<1, 250 pF is 600 Ohms and this is parallel to 3.3 Ohms. So it will not act as a 180 degr phase shifter. you can only have benifit from L1, C2 as a series circuit as this provides voltage gain (not power gain of course).

2N3055 is not capable for 1 MHz operation as ft is too low. This results in an enormous capacitance parallel to the BE junction.

There is no dc decoupling between collector and base, so Vb is Vc. This results in a low output swing (if the circuit oscillates).

Even with another (fast) transistor, you will waste all your power in resistors, so the overall efficiency of the oscillator will be very low (even if there are no losses in capacitors and inductors).

Most power circuits do not have resistors in their high current path (look to my oscillator example). Single transistor devices will have Vce = Vsup (for the DC case). So your output voltage swing will go from 1V (saturation) to twice the supply voltage.

Because of absence of current limiting resistors in the main current path, the difficulty is to make sure that there is no run away and you get stable output amplitude.

Class C oscillator (sinusoidal output) tend the have instability of the envelope when you make the conduction angle of the active device too small. This chance on envelope instability can be reduces by accepting some collector (or drain) saturation. Problem with BJT is that when It saturates, it will not come out of saturation quickly. In the spice BJT model, the saturation behavior is modeled with the TR (reverse transit time) parameter.

If you want to use frequencies in the 500 kHz and above range, I would suggest to use mosfets (cheap, and available).

 

[lulezo]

Wow that was fabulous !
in the class C oscillator circuit, the inductive coupling is between L1 and L Gate, right ? so what about the arrow in the L2 inductance ?
Gate and Tx, are labled for the gate and the tank circuit only or have any other different meaning ?
what is the 78.37m rectangle ?
many thanx..

 

[WimRFP]

Hello,

Correct, the Gate and Tx labels are just to make reference to the wave forms shown below. the rectangle is the Ampere meter for indicating the total drain current (Brown/yellow trace).

L1 is just a choke and has no coupling, so the arrow has no meaning.

In reallity Lgate will be a small single turn indutor close to the main coil. The output voltage of the gate coil is about 8Vp, while L2 "carries" 120Vp, so this indicates little coupling (surely far below 0.5 as used in the simulation).

If you ever plan to go into such circuits, make a current limiting circuit first (with a very heavy p-channel mosfet). Provide it with an out of saturation timer to shut down the current when it is too long in over current status. It will save you lots of destroyed devices. These type of current limit / shut down circuits are also used in hot-swap applications.

Now you have an example circuit. The best is to put it in your simulator and play with the components to see what happens. You need to know the circuit before building something. Highest efficiency you will get when the "weight center" of the drain current coincides with the lowest value of drain voltage (so current is in opposite phase with drain voltage.

 

[lulezo]

hi
i couldn't make the inductive coupling in the simulator, so i tried to use a transformer with values attached, but the circuit didn't work properly.
please look at the attached file, it contains all the problems i have faced.
i think the problem is in the transformer, but i have no other choice, please help, am using multisim 10, where can i find inductive coupling without using transformer ?

since i have a very poor understanding of your circuit, please send me a brief description for all the components use and how that circuit differ from ordinary colpitts oscillators.

 

[WimRFP]

Hello,

For the diode you must select an ultra fast recovery diode with the requried reverse voltage rating. current rating for the diode: 100mA. 1N4148/1N914, etc works also fine (and is easy to get).

for the transformer I used a coupling factor of 0.5. Do NOT use a transformer model where you have to input core data also, there is no core in this design, just air coils with inductive coupling.

I don't know your simulator, so you have to find out yourself or ask somebody else.

I do not have a circuit description, I designed it yesterday.