Class D amplifier with maximum power adjustable from 20W to 300W
This is a powerful class D amplifier easy to implement. Designed with the-shelf components, It can reach up to 300W RMS with a supply voltage which can vary from +/- 30V a +/- 60V depending on the desired power. To climb a few values of the components beyond that power should be revisited, for both maximum working voltage for the power dissipated. I set this limit to avoid complicating the diagram unnecessarily, thereby increasing the manufacturing costs and overall dimensions.
This is the complete diagram of an amplifier channel:
parts List
Principle of operation of the class D
This is a self-oscillating class-D amplifier. This provides excellent performance thanks to the dependent variable frequency from the input signal. Each amplifier section is described below on the basis of function scolta.
Each component has more than one function in order to limit the complexity of the circuit and consequently the costs, it is the result of successive simplifications dates from a long study design.
Input stage
U1a is an operational amplifier connected as an inverter.
R1 and C1 form a high-pass filter having a cutoff frequency of 7 Hz useless to fall below such frequencies. R1 defines the fixed gain and the input impedance of the amplifier, R2a and R2b secure the amplification with the formula
-(R2A + R2B) / R1
No value is critical, you can choose any value from 10k to 100k for R2.
C2 together with R2 form a low-pass filter which reduces the possible high frequency noise.
Its cut-off frequency is set
Ft = 1 / (6,28xR2xC2)
C2 must be adapted as a function of value R2.
Inverter and integrator
This is the ring of actual reaction.
C3 eliminates the continuous component of the output U1a and again form a high pass filter with R3
always with the formula
1 / (6,28xR3xC3) = 3,4Hz.
C3 can vary from 2.2uF to 10uF without problems.
C4 is used to reduce intermodulation distortion that can generate the integrator U1B. Indeed, R8 falls through the integrator signal. You could omit C4 but, It is frequent to see it mounted in this type of circuitry in class D, Furthermore, down a bit’ the oscillation frequency drive (C4 = 1 nF is 255kHz, It is not 330kHz).
R8 specifies how the integrator stage together with R3 gain.
The gain is -R8 / R3.
Besides 150kOhms R8, the amplifier becomes unstable. R3 is the input impedance of the integrator and is rather low, It is therefore the value of the input stage seen from U1a.
R4 protects U1B in case of default. Indeed, if the saturated output voltage, This is equivalent to the power potential (given by T1 or T2 through R8).
R4 forming a divider with R8 limited to +/- 6VDC maximum voltage on the inverting input. R4 could be replaced by two 1N4148 diodes connected to the operational powers but, with a single resistance is easiest!
In normal operation, the inverting input of the input voltage fluctuates on +/- 100 mV around and R4 is useless (the amplifier operates without R4).
If R4 is reduced (up to 1k) It increases the offset output. The integrator behavior deteriorated. A supply of a +/- 50VDC, the offset output is measured in:
R4 = 10k: 6mV
R4 = 2.2K: 36mV
R4 = 1k: 71mV
It is obtained that it is advantageous to put the maximum value possible for R4 but, touches always consider the maximum allowable excursion +/- 10V on the inverting input. 10k turns out to be a good compromise.
C5 is the capacitor integrator. Its value greatly influences the frequency of oscillation (the heart of the operation amplifier in class D). A +/- 50VDC is measured:
C5 = 220pF: 255kHz
C5 = 470pF: 236kHz
C5 = 1nF: 164kHz
In reality, in the frequency of operation there is also the “slowness” the TL072, years from the internal circuitry with which it is made feel. This is why we have adopted the frequency limit below 300kHz, but sufficient for this amplifier, the standard TL072 is a good budget compromise, with a sufficiently low noise figure to be used in audio and fast enough for the specific use applications.
The integrator output voltage is a triangular signal that goes from + 1.0V a + 4.2V with C5 = 220pF.
The choice of power supply +/- 9,1V is a value more than enough.
total earnings
The total gain of the amplifier is defined by two individual amplifications in cascade: the amplifier based on U1a and the integrator.
In the case of R2 = 47K the amplifier has gain, U1a: -R2 / R1 = -47k / 22K = -2.14 (variable according to the table)
Gain Supplement: -R8 / R3 = -100k / 4.7K = -21,3 (fixed)
The total gain of the amplifier D is then -2.14 x (-21.3) = 45, with such amplification are obtained in output of 250W effective 4 Ohm with an input standard signal 2 Vpp or alternatively we can still say 0,707 Veff.
Transistor level translator
The transistor T3, a PNP allows the “shift” the integrator output voltage to -Vcc. Indeed, the current through R5 it's the same (neglecting base current), the current flowing through R7. Putting R5 = R7, then the voltages at the terminals of R5 and R7 are the same. We would then R7 in the same integrator output voltage, subtracted the Vbe (error of about 0,6 V). Given that use a triangular shaped signal, you do not need a very fast switching transistor. The potential of its collector varies little and, T3 does not saturate during normal operation contributes to good linearity.
R6 It limits the current that can enter at pin1 (IN) of the IR2184 assuming T3 conductive and there was a saturation amplifier U1B or a default potential.
Its main constraint is to support at least Vce = Vcc (60V). The choice fell on the classic BCX42 (125V, 800mA, 330mW) perfect for this use.
Control of the power transistors: IR2184
A circuit made of discrete components that had made this function, It would have been far more expensive in terms of overall dimensions and distortion of the output signal as well as monetary.
A specific integrated circuit makes it very easy control of the two amplifier MOSFET, I chose the IR2184 (half bridge driver) della International Rectifier . The MOSFETs are driven with the phase delay with a time of about 0.4us.
If the input IN (leg 1) It is compared to 0V WITH (leg 3 which it is -Vdc), T2 it is on, T1 it is blocked, which ensures a low output level (-Vdc) at the exit of the final.
If the potential at the pin IN It is between 3V and 5V with respect to pin WITH, T2 is blocked, T1 is on but, It may remain so only for 10 O 20 ms because its control is powered by the bootstrap capacitor C12. If IN remains permanently 5VDC, T1 and T2 are blocked and the system does not oscillate. The output of the transistors is 0V (connected to ground through R4 + R8 or speaker).
The IR2184 is fed to 12 V (between 10V and 15V is the typical value) and consumes about 30mA operating at 250 kHz with due transistor IRFB5620 come “load”.
D4 e R14 dealing with load C12 (bootstrap capacitor). This capacitor provides the T1 when this command is run. T1 may remain so for a few tens of milliseconds, but it is adequate for this amplifier.
The IR2184 must be positioned very close to the T1 and T2.
It should be expected a special track distinct from that which feeds T2 that goes to pin 3 IR2184 worth of its destruction, for the same reason the two electrolytic C7 e C8 They must also be mounted as close as possible to the final T1 and T2.
To reduce the switching losses of the transistors,using components R11, R12, Dl e D2. The diodes allow a rapid opening of MOSFET transistors with fast discharge of the gate capacitance.
The IR2184 is able to provide a current greater than 1A for this purpose. R11 and R12 provide a little extra downtime, which avoids the risk that a MOSFET enters into conduction before the other has interdict.
Output stage
The output stage is constituted by two identical N MOSFET transistors T1 and T2 and by decoupling capacitors C7 and C8. The transistors are sized as follows:
VDS = with +/- 60supply V, It must be 120V to which must be added a 30% – 40% approximately margin.
Then, scelgo VDS = 200V.
ID = 15A (worst case Vcc = 60V / Load = 4 ohm) The advantage of having a high ID to a MOSFET is the Rdson resistance is low, with consequent dissipation (conduction losses) low. Then, I choose ID = 25A
They should be chosen transistor with low gate charge Qg otherwise for charging will take more time degrading the final signal and passing much time in the linear zone, that would be deleterious to two factors, an increase of the final distortion and an increase in operating temperatures.
For these reasons I have chosen IRFB5620: 25A 200V 60 mOhm specially designed for class D amplifiers.
Output filter
The LC low pass filter in the output calculation by the following formula
L1 = RL x 1,41 / 6,28 x F
Considering the speaker load RL 4 Ohm F and the filter cutoff frequency that must be at least a couple of octaves below the switching frequency and at least one octave above the maximum reproducible frequency I arbitrarily taken as the value of frequency 43KHz
Applying the formula I then
L1 = 4×1,41/(6,28×43000) = 20,95 uH
I choose a standard value of 22uH and recalculating the frequency value with the formula
F=RLx1,41/6,28xL1 = 40,8KHz still within the limit of one octave with respect to the maximum frequency.
For the use capacity instead the formula
C = 1/(6,28xFxRLx141) = 1/(6,28x40800x4x1,41) = 690nF
approssimerò that the standard value 680nF, from field trials the addition of resistance R13 in series with the capacitor C14 linearizes the response of speakers.
Auto Start
As previously said, if the input “IN” the IR2184 is high statically, T1 and T2 are turned off (C12 is discharged and simultaneously will lock D4).
You must put a few millivolt signal (a little 'music) input of the amplifier to start the self-oscillation.
In normal use of an audio amplifier, This poses no problem. This Class D amplifier is simply “in sleep” before the small musical stress.
Supply
Whatever dual voltage between +/- 30V e +/- 60V
To overcome these limits in both the top and bottom must be replaced resistances limiting the zener with constant current generators. Excluding the Solution this case only to simplify the final schema, a resistance even if power, It is certainly less bulky than two resistors a diode and a transistor that carry out the same work.
Values to be assigned to R2 to the desired power
A parity of the input signal is possible by varying the value of the resistors R2a and R2b in accordance with the table below to obtain the desired power without changing other values if not the dual supply voltage to be supplied.
practical realization
Since I started to make a compact circuit I created two distinct unravel the first with discrete components that allowed me to make all the adjustments necessary in the function test and a final with SMD components for those who love ultra-compact professional achievements.
Below the routing for tests with approximate measurements of 10cm x 6cm
side components and copper side
Then once everything was calibrated and adjusted for best operation I passed the final version with the use of integrated SMD power components that are still remaining discreet.
The size I have not been able to reduce further as the two electrolytic and the coil are very bulky items.
Despite this, the final measurements have been reduced to 99mmx43mm measuring 10cm side is given by the size of the fin and below would not be able to descend not too increase the operating temperatures.
If we consider, however, that in such a strip cropping it entered an amplifier can deliver to the load of 300W effective 4 Ohm it is remarkable.
Salve, table’ extending the circuit to create an OTL tube / mos class d amplifier. The C12 bootstrap is not reported as a value. What do you recommend? 1uF is enough? Greetings, Cicognani engineering
cicognani
Salve, having to send variable frequency signals of 22KHz to about 27KHz in input … first, if possible with the values given in the schema, according to how I can transmit in the air +/ the same input frequencies but amplified , taking into account that of course I cannot use a normal loudspeaker but rely on ultrasonic capsules perhaps in arrays? thank you.
The passband of this amplifier allows the use of those frequencies
For the speaker to use, I don't know what to tell you.
The output filter is rated for impedances of 4/8 ohm and you must try to get close to these values for your transducers to have good efficiency
Thanks for the reply Amilcare , and I apologize for the delay .. something more detailed ? I really need to amplify, certainly not at 300W … 30 would be enough… 50W but I find major problems precisely with adapting the output impedance with the ultrasonic transducer that I can still measure while working with the generator at 22KHz 27KHz it could be that the normal bf amplifier is not at all suitable for these frequencies above 20KHz?
At this point I suggest you search the Nuova Elettronica's LX945 network
An amplifier with disarming simplicity and low cost but, with a bandwidth that reaches 100KHz
With certain data on the transducer, the shot could be further adjusted
good morning,
I tried to simulate the circuit under consideration with multisim, from the simulation a continuous signal is obtained. Someone can give me an answer and resolve any errors present??
if you are willing to help me write me at this email address ” diego.002.elettronica@gmail.com”
thank you so much
so I will provide you with the multisim project
I’m very interested I also simulating this in multisim. Do you still have the files? If you dont mind I’m at support@easyamp.com. Thank you
wanting to try this amp : I ask if you can have email GHERBER the files of both SMT and traditional solutions
They are no longer in possession of the file you ask, if there are still in the hands of those fine gentlemen who stole my PC, the pattern is correct and you know works, if you want to make a printout with data Article.