Only NMOS output stages - [scientific investigation]

Thread Starter

dumbinvestor

Joined Feb 17, 2020
27
Hi Everyone!!

I'm currently trying to design an analog audio amplifier using only type n (nmos) transistors. My problem is regarding the output stage. The common-emitter with a diode-connected load in parallel with the speaker allows the signal to pass but it is very energy inefficient. Are there any other topologies that would allow for a more energy efficient circuit?
1628102199525.png
Really struggling out here, any help would be appreciated!!
 

Audioguru again

Joined Oct 21, 2019
5,162
Your circuit is a class-A heater and has DC across the speaker which is bad. The Mosfets heat a lot all the time even when the amplifier is not producing sounds. The DC in the speaker causes the speaker coil and cone to be pushed to one side and heats it all the time.
Most audio amplifiers are class-AB push-pull. The speaker gets the normal AC audio but gets no DC. The output transistors draw a very low current with low heating at idle.
Modern amplifiers are class-D switching push-pull. They heat only a little.
 

Audioguru again

Joined Oct 21, 2019
5,162
The book in post #4 has no page numbers.
Section 6-97 begins with audio amplifier circuits.
They are much more complicated than the very simple circuit shown in post #1.
 

crutschow

Joined Mar 14, 2008
30,774
Those Siliconix circuits take an interesting approach to solving the problem of properly biasing the output stage of MOSFET audio amps, since the approaches used for biasing bipolar output stages don't work well (or at all) with MOSFETs.
 

Thread Starter

dumbinvestor

Joined Feb 17, 2020
27
I should also add some info for context.
My limitations are the following. I can only use IGZO TFTs.
Why? Trying to achieve a transparent/foldable audio circuit.
 

Ian0

Joined Aug 7, 2020
6,286
I should also add some info for context.
My limitations are the following. I can only use IGZO TFTs.
Why? Trying to achieve a transparent/foldable audio circuit.
I presume the IGZO TFT's are quick? If so, all modern Class-D amplifier use a Half-bridge running at a few hundred kHz, which is generally all-N-channel. Then you have to make a transparent foldable output inductor!
If you want Class-AB then, apart from Siliconix's autobias circuit, I can't think of a design which doesn't require a preset to set the bias.
 

Thread Starter

dumbinvestor

Joined Feb 17, 2020
27
Your circuit is a class-A heater and has DC across the speaker which is bad. The Mosfets heat a lot all the time even when the amplifier is not producing sounds. The DC in the speaker causes the speaker coil and cone to be pushed to one side and heats it all the time.
Most audio amplifiers are class-AB push-pull. The speaker gets the normal AC audio but gets no DC. The output transistors draw a very low current with low heating at idle.
Modern amplifiers are class-D switching push-pull. They heat only a little.
I actually made a mistake in the design that I showed you. The speaker isn't connected to vss but to ground. But it won't matter. When compared to other topologies is still very inefficient.
 

Ian0

Joined Aug 7, 2020
6,286
I actually made a mistake in the design that I showed you.
We guessed!
The constant current load circuit you have will be about 25% efficient.
You can get to about 50% and remain in class A by using semiconductor versions of White cathode follower and SRPP valve circuits.
You can get to 78% using class B (just a bit less for class AB, which is Siliconix Autobias etc.)
Class D will take you into the >90% region.

I remember taking a lecture course on amorphous silicon back in my 3rd year at university back in 1986 - I remember it being scheduled for 9am on Saturdays.
What I don't remember from it (maybe due to what I was doing on Friday nights) is whether the FETs were enhancement or depletion or whether you could make both, and whether you could construct P-channel devices.
 

Audioguru again

Joined Oct 21, 2019
5,162
The Mosfet at the top in your extremely simple amplifier does the opposite to what is needed:
1) When it tries to pull the speaker voltage positive towards +Vdd then it produces less and less current instead of producing the required more and more current into the speaker.
2) When the lower Mosfet tries to pull the speaker voltage negative towards -Vss then the top Mosfet conducts more and more power supply current instead of less current.
 

Thread Starter

dumbinvestor

Joined Feb 17, 2020
27
With this simple circuit, I'm getting around 2% energy efficiency. I'm assuming it's destined to fail so I won't even try to produce it. There's no point. Another Limitation that I have is that I can't use many resistors since they take a lot of area. It is even possible to produce a audio circuit only using one type of technology?
 
Top