Tuesday, 15 March 2016

555 Flyback Driver and Plasma Speaker Part IV

I have been working more on the Plasma Speaker.  A few things have become apparent....The addition of the class A amplifier is a significant improvement as is the use of a switch-mode high current power supply.

It has also become apparent that the high current power supply not being current limited has destroyed the +12 Vdc on the PCB at least 5 times!  This is probably due to poor constructional technique (my bad soldering) and the tracks not being thick enough to carry the amount of current present in normal operation.  I was playing with the circuit last night after having put the high voltage probes into a laser cut case (which worked perfectly) I managed to catastrophically damage components...Standard form for me to be honest....that is why I call this prototyping!

The issues I'm having are all to do with construction and design choices that I made when I initially designed the plasma speaker PCB.  I'm going to design a new version which improves the situation:
  • Increase current rating of all tracks carrying +12V or GND.
  • Incorporate the Class A amplifier into the circuit so that everything is on one PCB
  • Increase the gain on the Class A amplifier as it still (in my opinion) is not loud enough
  • Move the components which get incredibly hot off the PCB and onto the heatsink.
First off I think I really need to find out just how much current the main switching transistor is subjected to.  This is going to involve checking the datasheet for the transistor and performing some ohms law.  I hate maths but in this case it has to be done...I cannot assume all will be well if I don't check the current requirements of the 12 Vdc conductor.

Here is the new schematic with all of the circuitry on one sheet:

The datasheet for the IRFP250 is here:

From the datasheet we can see that the RDSon parameter of the transistor is 0.085 ohms.  If we assume that the dc resistance of the flyback transformer is also quite low then the amount of current constantly present as the transistor switches will be high...The energy from the flyback voltage at the primary must also be taken into account.

In order to discuss this I should really link in a page discussing how flyback transformers function:

EE Times article on flyback transformers

Wikipedia Entry on Flyback Transformers

My schematic diagram has not ever shown the actual flyback transformer which is actually a transformer with 10 turns on the primary core and several thousand turns on the secondary core which is then connected to an internal 'flyback' diode and capacitor.

Basically what this means is that the output of the transformer and associated energy output is related to the winding ratio, the capacitor and the load applied at the output which in this case is an arc through the air.

For the purposes of calculating how much current will be flowing in switching transistor part of the circuit I'm going to simulate the circuit.  The reason for simulating is that it's quicker for me than calculating all of the Ohms law required on paper....Here is the circuit after simulating:

Well....that explains why the MOSFET Q3 got so hot and needed such a large heatsink along with the flyback diode D1 and the 120 Ohm resistor....1.6A constantly flowing in that part of the circuit is a great deal and also explains why the 12 Vdc conductor and the conductors in the FET part of the circuit needs to be as thick as possible.  I have made assumptions on the turns ratio of the transformer but it doesn't really matter as I have seen from the power supply current meter that I'm using to power this circuit that these calculated current values are close enough....Therefore the design needs to account for this 2 Amp current being present - I actually think the instantaneous peak currents will be considerably higher than this and the current is also higher when the audio modulation is applied.

To that end we need to redesign the PCB to take this into account.  Here is the new PCB layout:

Top Layer Of PCB

Bottom Layer of PCB

Both Layers with Dimensions

I then etched and populated the PCB.  I have found that the circuit works better but still had some issues.  I have mounted the 120 ohm 5 Watt resistor on the heat-sink along with the clamp diode which I have swapped for a TO220 packaged version.  I also changed the operating frequency of the 555 oscillation by changing the value of C1 (in the uppermost schematic) to 100 pF.  This changes the oscillation frequency to somewhere always above 20 kHz which removes an annoying high pitch whistle when the circuit is in use.

There are still issues with the circuit but I believe this is now as good as it will probably get.  I need to obtain a suitable high current power supply and mount the circuitry properly to make it easier to move around.  I have had a lot of fun developing this circuit and visually it's really attractive.  It's practicality is exceedingly limited.  A plasma speaker loses a great deal of fidelity with low frequency bass sounds, generates significant amounts of ozone, uses a large amount of electrical power and is fundamentally dangerous because of the high voltage DC that is present.

Here is a video of me playing around with it using an electronic keyboard to provide the audio input.

That's all for now - take care people, especially with high voltage dc circuits and plasma speakers!

Sunday, 6 March 2016

555 Flyback Driver and Plasma Speaker Part III

So here is the complete Plasma speaker circuit in all it's glory!

It actually creates a significant amount of high voltage and works very well.  I would caution anyone else attempting to replicate this circuit to please be very careful.  I haven't given myself a shock yet but it could happen and will hurt if it does....Exercise sensible precautions please!

Here is the previous post in case people need to catch up:

555 flyback driver and plasma speaker part II

I have found that the 3D printed HV probe holders work quite well.  I also have found that setting the distance between the probes is critical to obtaining a reproducible arc and that the constant re-strike of the arc causing the audio to sound terrible.  From experimentation I have found that the audio signal from my mobile phone is more than enough to drive the 555 modulation pin when it isn't capacitively coupled.  When capacitive coupling is added the audio is barely heard.  The capacitor on the audio input reduces the hissing considerably.  Here is a video showing the current audio output of the plasma speaker...it sounds pretty terrible but it does work:

I have decided to do two things....improve the HV probes and provide a simple class A audio amplifier to the pin 5 input of the 555.  This should improve the sound and get rid of the horrible hissing!

So to that end I have designed a very simple single transistor class A amplifier using a BC548 transistor.  Here is the schematic:

In designing the circuit I referred to this website...which is rather useful for this kind of thing:


I knew how to design a Class A amplifier well enough but I had forgotten how to select the components values correctly...in particular I wanted to increase the low frequency response and limit the bandwidth of the amplifier to reduce the high frequency response.

The circuit works fairly simply...An audio signal from a suitable source is presented at the 3.5 mm headphone jack input - only one side of the audio signal is provided - this amplifier is mono. This is then passed to C1 - a 1 uF electrolytic capacitor which is used to remove any dc offset and chosen in such a way as to not overly affect the bass response of the amplifier (more on this later).  The next components in the circuit are R3 and R4 which bias the NPN BC548 transistor into constantly being ON.  These values are set by ohms law.  We need at least 0.7 volts to turn an NPN transistor ON. Lets do the maths just for fun:

Ohms Law; V / R = I

In this case:

V: 12 Volts
Rt: R3 + R4 which is 120 kΩ + 10 kΩ = 130 kΩ

I = V / Rt

I = 12 V / 130 kΩ

I = 9.23076923077 * 10^-5 A or 92.3 µA

The voltage applied to the base of the BC548 transistor can be calculated by = I * R4
therefore the voltage applied to the base of the BC548 transistor:

92.3 *10^-6 A * 10 kΩ

The voltage applied to the base of the BC548 transistor is 0.923 Volts or 923 mV

The circuit has been designed so that 0.923 volts is always applied to the base pin of the transistor to 'bias' the transistor ON.  The audio signal applied will increase this voltage and be amplified.  The next components applied to the collector of the transistor are a 10 kΩ potentiometer and a 100 Ω resistor.  At the emitter of the transistor we have another 10 kΩ  potentiometer and a 10 uF capacitor. All of these components combined set the gain of the amplifier. There are formulae that can be applied to calculate the amount of gain.  I guessed at it...It's not particularly important in this case. When the potentiometers are at maximum (according to my simulations) the input signal is amplified roughly 130 times greater than the input...the amount of gain is controlled both 10 kΩ  potentiometers which can be set by the operator.  The 10 uF electrolytic capacitor C3 is known as the emitter decoupling capacitor and is added to prevent any stray audio signal being present on the emitter pin of the transistor.

Finally at the output of the amplifier we have a 1 nF ceramic capacitor C4 and a 10 uF electrolyitic capacitor C2.  The electrolytic capacitor C2 prevents any dc voltage being passed to the next stage of the circuit, in our case, pin 5 of the 555 timer. C4 is used to limit the bandwidth of the amplifier.  In this case I have set all of the capacitor values to set the amplifier's frequency bandwidth to be between 200 Hz and 20 kHz which is roughly the range of human hearing.

I simulated the circuit in order to check what the output would be like and check the gain would be sufficient and to verify the frequency response.  It was helpfully not clipped and gave a good amplified approximation of what was to be expected.

Here are the results of the simulation...I have placed probes at the more interesting points in the circuit:

Simulation Schematic
Here is the simulated oscilloscope output:

The input signal is shown with the blue trace, the red trace shows the amplified output.  The output is inverted but that won't matter in this case.

The really good thing about simulating circuits is that the frequency bandwidth can be checked without actually building the circuit.  Here is the simulated audio frequency response of the amplifier:

If the capacitor values C1, C3 and C4 are changed for different values the frequency response of the amplifier is significantly affected.  C1's value changes the bass frequency responses, C3 changes the treble response and C4 changes the bandwidth of the amplifier.  In this case I have tweaked the values to try to give the best response between 200 Hz and 20 kHz without losing too much bandwidth.

Because its me I've designed a simple single sided PCB for this circuit.  It could easily be made on veroboard (stripboard) or using some other method.

Top Layer of PCB
Bottom Layer of PCB

Here is a render of the PCB to show how it will look once etched and populated:

Top View of Class A Amplifier Render
ISO view of Class A Amplifier Render
Here is the bill of materials:

Part Value Device Description Vendor Part Number Quantity Cost

12VDC_INPUT N/A M025MM Standard 2-pin 5mm screw terminal Farnell 9632972 1 0.245
AUDIO_OUT N/A M025MM Standard 2-pin 5mm screw terminal Farnell 9632972 1 0.245
C1 1uF CAP_POLPTH1 Electrolytic Capacitor Farnell 1236686 1 0.0464
C2 10uF CAP_POLPTH1 Electrolytic Capacitor Farnell 9451056 1 0.034
C3 10uF CAP_POLPTH1 Electrolytic Capacitor Farnell 9451056 1 0.034
C4 1nF CAPPTH1 Ceramic Capacitor Farnell 1141779 1 0.0758
C5 100uF CAP_POLPTH1 Electrolytic Capacitor Farnell 1902882 1 0.0345
C6 100nF CAPPTH1 Ceramic Capacitor Farnell 1141775 1 0.0721
JP1 N/A AUDIO-JACKPTH 3.5mm Audio Jack Farnell 1608405 1 0.534
R2 100 RESISTORPTH-1/4W ? Watt Carbon Film Resistor Farnell 9342397 1 0.0523
R3 120k RESISTORPTH-1/4W ? Watt Carbon Film Resistor Farnell 9342540 1 0.0492
R4 10k RESISTORPTH-1/4W ? Watt Carbon Film Resistor Farnell 9342419 1 0.0523
RV1 10k POTALPS-KIT PCB Mount Variable Resistor Farnell 1191725 1 1.4
RV2 10k POTALPS-KIT PCB Mount Variable Resistor Farnell 1191725 1 1.4
T1 BC549 BC549-NPN-TO92-CBE BC549 NPN Transistror Farnell 2453797 1 0.238

Total 4.5126

Again I haven't factored in the cost of the PCB or it's manufacture but it would be reasonable to estimate the total cost of the project to be around £6.00

Here is a quick video showing the circuit in operation with the plasma speaker.  The audio is very much improved!

Now I need to get back to putting the HV section and the electronics into some sort of casing.  That's all for now - take care people!