Hello,
I have to make a class AB amplifier for high frequency applications (400 kHz - 700 kHz).
There is almost no literature in that frequency range so I started to learn about audio architectures to improve them.
I made a simple CFA schematic but I have not so much experience.
The amplifier has to deliver 15/20 W to a 20 ohm load (the load should be capacitive because is a piezotransducer).
Please give a look to my schematic to see if it shoud be ok or if there are little or big mistakes.
Any opinions to improve this design should be appreciate.
I have to make a class AB amplifier for high frequency applications (400 kHz - 700 kHz).
There is almost no literature in that frequency range so I started to learn about audio architectures to improve them.
I made a simple CFA schematic but I have not so much experience.
The amplifier has to deliver 15/20 W to a 20 ohm load (the load should be capacitive because is a piezotransducer).
Please give a look to my schematic to see if it shoud be ok or if there are little or big mistakes.
Any opinions to improve this design should be appreciate.
I have never driven a piezo-transducer with a current feedback design, only voltage feedback, I generate trapezoidal waveforms requiring slew rates around 100V/µs. Typically I prefer a MOSFET output stage biased by about 3-5V so that output stage power dissipation is very low when my waveforms are idle.
I am also not a fan of the diamond configuration that you have used because Q3 and Q2 have the full rail-to-rail voltage across them, I wonder if it would be worth tying Q2 and Q3 collectors together instead of to the opositie rails, this would reduce the power dissipation of the input stage but I do not know the affect that this would have on amplifier distortion or stability without building and testing the design.
I am also not a fan of the diamond configuration that you have used because Q3 and Q2 have the full rail-to-rail voltage across them, I wonder if it would be worth tying Q2 and Q3 collectors together instead of to the opositie rails, this would reduce the power dissipation of the input stage but I do not know the affect that this would have on amplifier distortion or stability without building and testing the design.
Can I recommend an article by Peter Blomley: ‘New approach to class B amplifier design’ Wireless World, Feb. 1972, pp 57-61 and Wireless World, Mar. 1972, pp 127-131. In this, Blomley claims that his proposed architecture is suitable for high-frequency use. Although old, it is an enlightening read.
Biggest issue is that the output transistors are way too small. For 20 Watts output, you need 50+ watt transistors. Remember that the power rating is for a case temperature of 25C, which is not going to happen, and potentially the transistor dissipates 2x the output power. Silicon is cheap today so I would use a typical 2SC5200/2SA1943 pair, so that failure is unlikely, or use MOSFETs.
Then there is the issue of voltage swing efficiency. This circuit will not get close to rail-to-rail output, so you may want to run the VAS at a slightly higher voltage than the output or use a different IPS design. Symmetric IPS require serious emitter resistors to control the bias currents, which wastes voltage swing. BTW, tying the collectors of Q2 and Q3 together makes them a pair of diodes, not a good idea. You could drop the collector voltages, but it should not be a problem as is. For speed, R4 and R14 need to be as small as possible. 1K is just small enough for audio, depending on Q1, Q11. Looks like this is a simulation, so did it give the required slew for 700KHz? +/-30V is probably not enough to get 20 Watts into 20 Ohms.
Then there is the issue of voltage swing efficiency. This circuit will not get close to rail-to-rail output, so you may want to run the VAS at a slightly higher voltage than the output or use a different IPS design. Symmetric IPS require serious emitter resistors to control the bias currents, which wastes voltage swing. BTW, tying the collectors of Q2 and Q3 together makes them a pair of diodes, not a good idea. You could drop the collector voltages, but it should not be a problem as is. For speed, R4 and R14 need to be as small as possible. 1K is just small enough for audio, depending on Q1, Q11. Looks like this is a simulation, so did it give the required slew for 700KHz? +/-30V is probably not enough to get 20 Watts into 20 Ohms.
Apex makes hybrid chip https://www.apexanalog.com/resources/products/pa09u.pdf that will fit your need.