# Audio Amplifier With MOSFET

I am working on a new project that seeks to implement a three-stage audio amplifier. One of my stages consists of a CASCODE circuit using the BF998 NMOS transistor. For this stage I am struggling to mathematically build a circuit using this transistor and generate a gain with a specific harmonic distortion.

1. Initially I have chosen a region of operation for my circuit. By observing the graph Id x Vgs, I noticed that Vth is approximately -0.6 V and I chose to operate at VGS=0.1 V (Consider my input signal equal to approximately 260 mV)
2. My output impedance (other stage) is equal to 2367 Ω. Therefore, I have determined the minimum current required to keep my signal from clipping (Iout= 1.4364 mA)
3. Furthermore, I used a resistor bias and considered that the current passing through the resistor branch is 2000 times smaller than the current passing through the drain of the common gate.

From this, I don't know how to control the gain and set harmonic distortion for this stage. I would like to learn how to design these two quantities with mathematical relationships and then implement the circuit in Advanced Design System (ADS). What I have built so far can be seen in the image below: For additional questions, I am always available to answer them. I am very interested in learning how to design this stage of my project.

• Can you provide a link to the datasheet for the transistor? When I searched for BF998 all I could find was a depletion mode dual gate RF MOSFET. Does not seem very suitable for your purpose. May 30, 2021 at 5:06
• Here is a datasheet for the BF 998 docs.rs-online.com/2c78/0900766b80026afc.pdf. The question I would like to answer is the following: How do you set a gain and get a bias for a given gain? May 30, 2021 at 17:02
• Use the datasheet forward transconductance (valid only at the stated conditions) to set the gain in conjunction with the load impedance. May 30, 2021 at 22:53

BF998 is a rather odd device, being a dual gate MOSFET, like a cascode in a single device. Note that if you can achieve higher than the required gain with the configuration shown, you can then control the gain by reducing Vg2 (actually, over an impressively wide range, like 40 or 50dB). This also affects distortion in difficult ways to model, but you may find a Vg2 value that minimises distortion. All in all, it's an impressively versatile device.

I have more experience with the older BF981/983 devices but I seem to recall that 6V drain supply is a little low, up to 10V (and up to 10mA Id) may improve performance.

And normally, Id is controlled by a source resistor (for a given Vg1) which may be decoupled for high AC gain, like biassing a (depletion) JFET. If you have an inductive load : as it is, without any resistor in the source circuit (RS), there is nothing to limit Id and prevent device destruction.

HOWEVER this circuit, for audio, would require an inordinately high inductance (like Henries) in the drain circuit. I'm guessing you're trying to adapt a test circuit from the datasheet, intended for VHF or UHF frequencies, not audio. That won't work.

Use a constant current source instead, leaving R7 as the effective load impedance. (then RS can be used to set Vg1s at a given bias point : this too can affect THD)

I doubt you'll find useful analytical functions but older literature (1970s/1980s) on dual gate MOSFETs may have something : I would plot characteristics at different Id (and/or Rs), Vg2, Vg1 to get an idea how it behaves, and design to your requirements from that.

One final note : while these devices can achieve astonishingly low noise figures, that's at VHF or UHF. At audio, there may be unexpectedly high (because undocumented) LF noise aka flicker noise. If you use a device outside its normal use cases, such things may not be guaranteed.

• Through your tip I was able to determine the transconductance and by simulation, determine the open circuit gain and with that I determined the impedance rds. Finally, I was able to determine the gain for a given load. But I cannot through gain determine the transconductance and ultimately my bias point. However, I would like to through a gain of my choice, obtain a bias point. Would this be possible? If so, how could I do this? You gave me the hint to use a current source, but where would I put this current source? Would I replace the RFC of my drain? May 31, 2021 at 4:27
• Yes the current source replaces the RFC (which is useless at audio by definition). Setting bias to determine gain : start by reading and understanding the characteristic curves in the datasheet and (if you can find a BF998 model) reproducing them in simulation. May 31, 2021 at 12:44