Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

[Moved]: High power 440 KHz oscillator design

Status
Not open for further replies.

Rushy

Junior Member level 1
Joined
Jan 25, 2018
Messages
16
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Location
Dorset
Activity points
148
My first post! ..and my first electronic project since the 1980s!

I want to design an oscillator running at 440 Khz to experiment with an inductive coupling circuit I have. Basically I have a small coil that resonates at 440 Khz and I want to drive a much larger coil with this frequency to see if I can increase the coupling distance to around 6 inches everywhere inside the larger coils diameter. I don't know what current I will need but I'd like to try up to 1A as a start. I've looked at standard circuits such as Colpitts, Hartley & Wien and I wonder if a transistor of sufficient power rating and freq. will work, and if such a transistor exists. Some info. I've read suggests I may need to use MOSFETs? I have no knowledge of them or what circuit I could use.

To summarise my requirements:

Freq. 440 KHz (trim +/- 10 KHz)
Waveform: typically sine or pulse but quality not so important
Load: large coil air cored. e.g diameter 12 ~ 18 inches.

Any pointers to where to start would be much appreciated!

Rushy
 

As the waveform isn't important, use any small square wave oscillator and feed it to a MOSFET to switch the current on and off. Think of a power MOSFET as an almost ideal switch, when off the leakage is negligible, when on a good MOSFET is like a <1 Ohm resistor. Although they draw virtually no DC current into their gate pins, they do have a high capacitance (possibly >1nF) between the gate and source pins so you need to be sure you can produce enough current to lift the gate voltage when driven at 440KHz.

Beware that driving with anything but a pure sine wave will produce lots of harmonics, some of them falling inside the AM broadcast band so be prepared for complaints if the neighbors find out. Driving 440KHz sine wave would be considerably more complicated.

Brian.
 

As the waveform isn't important, use any small square wave oscillator and feed it to a MOSFET to switch the current on and off. Think of a power MOSFET as an almost ideal switch, when off the leakage is negligible, when on a good MOSFET is like a <1 Ohm resistor. Although they draw virtually no DC current into their gate pins, they do have a high capacitance (possibly >1nF) between the gate and source pins so you need to be sure you can produce enough current to lift the gate voltage when driven at 440KHz.

Beware that driving with anything but a pure sine wave will produce lots of harmonics, some of them falling inside the AM broadcast band so be prepared for complaints if the neighbors find out. Driving 440KHz sine wave would be considerably more complicated.

Brian.

Thanks for the advice Brian. I think I will try a 555 timer circuit to generate a square wave. I believe they can run astable up to 2 MHz so 440KHz should be fine. Then I will add the MOSFET device and see how it goes. Harmonics may well be a problem so I will measure the spectrum and if they are really excessive I may have to rethink this.

Rushy
 

As the waveform isn't important, use any small square wave oscillator and feed it to a MOSFET to switch the current on and off. Think of a power MOSFET as an almost ideal switch, when off the leakage is negligible, when on a good MOSFET is like a <1 Ohm resistor. Although they draw virtually no DC current into their gate pins, they do have a high capacitance (possibly >1nF) between the gate and source pins so you need to be sure you can produce enough current to lift the gate voltage when driven at 440KHz.

Beware that driving with anything but a pure sine wave will produce lots of harmonics, some of them falling inside the AM broadcast band so be prepared for complaints if the neighbors find out. Driving 440KHz sine wave would be considerably more complicated.

Brian.

Thanks for the advice Brian. I think I will try a 555 timer which should run astable up to 2 MHz so I will look at a standard circuit and get my square wave running. I have a signal generator so I presume I could play with a MOSFET device and feed it a square wave and see how it behaves with my coil? I'd like to look at the harmonics and decide early on if this is a viable solution. If it is I would probably use a 555 timer to generate my wave and it would be simple. Is there anything I should consider when choosing a suitable MOSFET? I've never worked with them and don't know how to select one.

Thanks- Rushy
 

Driving pulses into an inductor WILL cause harmonics to radiate, the real question is whether you can live with them or not.

For the MOSFET, the main considerations are that it can handle the current and voltage. There are hundreds of types that should fit the bill but go for rated at least twice the supply voltage (you haven't specified it).

The secondary consideration is the gate to source voltage to make it conduct. MOSFETS, at least the enhancement type used for power applications, do not conduct when there is no voltage between gate and source. As the gate voltage is increased, nothing happens at first, then the device starts to conduct until it can't be more conductive. The voltage it starts to conduct is called VGS(th) and the effective resistance across the drain and source when it is fully conducting is called RDS(on). To achieve full conduction you typically have to raise the gate voltage to 10V or more, the data sheet will tell you the maximum the device can withstand.

It is also important to consider that the gate to source pins effectively look like a capacitor to the circuit driving them. Almost no DC flows in to or out of the gate but it might take quite a lot of current to charge the gate voltage up and discharge it afterwards. While not conducting the MOSFET does not dissipate any heat because it passes no current, when fully conducting it dissipates 'I x RDS(on)' which is also very small but in the linear region where it is partially conducting the dissipation could be high. This is why high current drivers are normally used to drive the gate in switch mode power supplies, it is to minimize the time the device is partially conducting so it looks more like a perfect switch.

Brian.
 

I've looked at standard circuits such as Colpitts, Hartley & Wien and I wonder if a transistor of sufficient power rating and freq. will work, and if such a transistor exists. Some info. I've read suggests I may need to use MOSFETs? I have no knowledge of them or what circuit I could use.
Oscillators which use their MOSFETs in the saturation region are very inefficient e.g. Colpitts, Hartley (around 50% efficiency).
For high efficiency, MOSFETs are used in their triode region (where betwixt is suggesting).

This guy did the same thing you are trying to do (I think) and many other nonprofessionals used the same circuit because it is easy to implement and get it going.
https://www.youtube.com/watch?v=t7Elo6IjPN0
 

H-bridge sends current back and forth through series LC. Resonant frequency is detected automatically. Sine waves pass through the coil.
Some imbalance is needed for oscillations to begin.

I've tested this design in hardware at low power, low frequencies. My simulation includes a low ohm resistor in series. It's needed to detect zero crossings. Value should be high enough to generate a voltage, sufficient so as to bias the H-bridge first in one direction, then the other.

Schematic with scope traces, showing over 1 Ampere going through the coil at 440 kHz:

astable H-bridge provides AC sine thru series RLC.png
 
  • Like
Reactions: Rushy

    Rushy

    Points: 2
    Helpful Answer Positive Rating
Driving pulses into an inductor WILL cause harmonics to radiate, the real question is whether you can live with them or not.

The usual solution is a series resonant topology which achieves almost sinusoidal coil current despite of being driven by a square wave voltage. You can review e.g. Qi wireless charge circuits for reference.

- - - Updated - - -

Also nicely demonstrated in Brads self resonant circuit.
 
Thank you all for the informative replies. Special thanks to Brad for drawing my attention to H-bridge with cct diagram, I had not come across that design and it does seem suited for what I want so I think I will proceed with this idea now.

- - - Updated - - -

That is very interesting, I'd like to try H-bridge but would want to run with a 12V supply. Could you recommend a transistor type for the current / freq I need?
 

I'd like to try H-bridge but would want to run with a 12V supply. Could you recommend a transistor type for the current / freq I need?

Probably a TO-3 package and heat sink. Notice the sine wave suggests the half-bridges have moments of shoot-through (although there may be times the transistors switch clean On and Off). It's hard to be sure how to calculate dissipated watts, but at 12V 1A it could generate a lot of heat. Since I have several metal 2N3055 (a common type) I would start by using those for the N-device.
The PNP counterpart is MJ2955A or TIP2955.
This family should be able to switch at 440 kHz.
 
  • Like
Reactions: Rushy

    Rushy

    Points: 2
    Helpful Answer Positive Rating
Unlike the demonstration circuit in post #7, the final circuit will rather generate square waves at both H-bridge legs and have the transistors controlled with required dead time. For currents up to 5 or 10 A, a 12 V supplied H-bridge can use transistors without heat sinks at 440 kHz.
 

Unlike the demonstration circuit in post #7, the final circuit will rather generate square waves at both H-bridge legs and have the transistors controlled with required dead time. For currents up to 5 or 10 A, a 12 V supplied H-bridge can use transistors without heat sinks at 440 kHz.

To minimise harmonics, a sine wave at the bridge legs would be preferable and your post #8 suggested that series resonant topology would achieve this? That attracted me to this idea, but I presume there are other factors to consider that I don't yet understand?

- - - Updated - - -

Probably a TO-3 package and heat sink. Notice the sine wave suggests the half-bridges have moments of shoot-through (although there may be times the transistors switch clean On and Off). It's hard to be sure how to calculate dissipated watts, but at 12V 1A it could generate a lot of heat. Since I have several metal 2N3055 (a common type) I would start by using those for the N-device.
The PNP counterpart is MJ2955A or TIP2955.
This family should be able to switch at 440 kHz.

Thank you Brad. I am familiar with the good old power transistor 2N3055 from my youth. I made several projects with them - happy days :) I haven't done any electronics since the 1980s and things have moved on just a bit! Could you (or anyone reading) advise me of any tools I may use to calculate required components values etc. this would be a revelation to me.
 

At sine wave at the amplifier output is only possible with a linear amplifier achieving maximal about 75% efficiency, thus I think it's not the preferred option. Current waveform of the series resonant circuit is almost sinusoidal, you can add LC filters to further reduce harmonics if it's an actual problem. According to post #1 it shoudn't.
 

That attracted me to this idea, but I presume there are other factors to consider that I don't yet understand?
I think that FvM was saying that instead of using the self oscillating circuit from post #7, you can get rid of the "R" resistor and drive the transistors fully into saturation thus achieving high efficiency and square wave at both legs. You still need the LC series resonant circuit.
Could you (or anyone reading) advise me of any tools I may use to calculate required components values etc. this would be a revelation to me.
There is not tool that designs an H-Bridge by itself specifically for this application, you need to use math as usually and then simulate it.
Just use the first harmonic approximation (fundamental harmonic approx), which gives a rather good approximation and is the way this circuit is designed in many applications e.g. EV charging.
 

Ok yes I realise Brad has chosen L & C values in #7 to give the resonant frequency of ~ 440 KHz I wanted, but in my application I am using a large copper coil of unknown inductance and capacitance that I made as my primary. The secondary is a commercial small coil (taken from a cheap wireless charger kit) and I found by trial and error that it resonates when I feed 440 KHz into my primary. At least the amplitude of the induced wave seen in the secondary increases very drastically only at this particular frequency so I presume that is the resonance! I am trying to achieve Resonant inductive coupling (2nd resonance technology) without necessarily having resonance in the primary which should be possible. So I guess I can use H bridge with the output driving a switching circuit to power my primary. This is what I am thinking now.

Resonant inductive coupling
 

using a large copper coil of unknown inductance and capacitance that I made as my primary.

My simulation does not make it obvious that my 420 uH value is selected so that it restricts current below 2 Amperes (with a 3V power supply). The capacitor is selected to determine resonance at 440 kHz.

Since your power supply spec is 12V, you'll need a larger Henry value than 420 uH, and a proportionally smaller Farad value. For a simplified schematic, here an op amp sends more than 1 Ampere through the series LC. A mere .07 ohm resistor can detect zero crossings, generating 0.1V swing. This is sufficient to cause a change of state from the op amp's output.

op amp 12V drives LCR auto detect resonant 440 kHz.png

However the above is only theory and I suspect that it breaks down somewhere. My simulation develops half a million volts across the coil and the same across the capacitor!
I doubt that happens in real life. Only experimentation can confirm what is or is not possible.
 
  • Like
Reactions: Rushy

    Rushy

    Points: 2
    Helpful Answer Positive Rating
My simulation does not make it obvious that my 420 uH value is selected so that it restricts current below 2 Amperes (with a 3V power supply). The capacitor is selected to determine resonance at 440 kHz.

Since your power supply spec is 12V, you'll need a larger Henry value than 420 uH, and a proportionally smaller Farad value. For a simplified schematic, here an op amp sends more than 1 Ampere through the series LC. A mere .07 ohm resistor can detect zero crossings, generating 0.1V swing. This is sufficient to cause a change of state from the op amp's output.

View attachment 144244

However the above is only theory and I suspect that it breaks down somewhere. My simulation develops half a million volts across the coil and the same across the capacitor!
I doubt that happens in real life. Only experimentation can confirm what is or is not possible.

Thanks for the info. Brad. Half a million volts! I guess that is theoretical of course. At the moment I am using a single transistor in cutoff / saturation to switch my square wave and supply ~ 1A through my coil. I've had to put a 10 Ohm resistor high wattage in series with the coil to limit the current to 1A with a 12V supply. Check my schoolboy knowledge - Could I have biased the transistor to limit the load current without the need for a resistor? Then the transistor would be dissipating around 12W whereas the resistor is doing that now and the transistor is switching and not working so hard.

This has highlighted a new query : I have a very low voltage across my coil as nearly all the volts are dropped across my resistor. Does this matter for the production of magnetic flux around the coil? I presume the current flowing through the coil is the key thing? If I had a high impedance source and used a higher voltage and lower current would that make any difference to the flux?
 

A unit which expresses flux intensity is Webers. Calculated as Henries * Amperes.

You installed a resistor, which works to reduce current. And yes you can use a transistor or mosfet to achieve this. The method is similar to resistive drop. It generates heat the same as the resistor does. Nevertheless the method is used in many simple power supplies, including my homebrew supply.

It's worth a try for you to test high and low amounts of current through your coil. Be aware that by abruptly shutting off current in an inductor, it often generates voltage spikes. The greater the current, the greater the spike.
 

A unit which expresses flux intensity is Webers. Calculated as Henries * Amperes.

You installed a resistor, which works to reduce current. And yes you can use a transistor or mosfet to achieve this. The method is similar to resistive drop. It generates heat the same as the resistor does. Nevertheless the method is used in many simple power supplies, including my homebrew supply.

It's worth a try for you to test high and low amounts of current through your coil. Be aware that by abruptly shutting off current in an inductor, it often generates voltage spikes. The greater the current, the greater the spike.

Good point about the voltage spikes. I think I should include a protective diode to protect the transistor from this.

Paul
 

I propose adding a diode to protect against coil back emf. See link for diagram. I/P Signal source is 0-12V peak square wave at ~ 150KHz. L1 is hand made large coil unknown value. D1 TBD.

**broken link removed**
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top