Discrete audio amplifier basics - Part 2: JFETs, MOSFETs and other circuit configurations

Because of the much more elaborate construction of the MOSFET, in which a multiplicity of parallel connected conducting channels is fabricated to reduce the conducting "on" resistance, the chip size is larger and the device is several times more expensive both to produce and to buy. The excellent HF characteristics of the MOSFET, especially the N-channel V and U MOS types, can lead to unexpected forms of VHF instability, which can, in the hands of an unwary amplifier designer, lead to the rapid destruction of the output devices.

However, this excellent HF performance, when handled properly, makes it much easier to design power amplifiers with good gain and phase margins in the feedback loop, where overall NFB is employed. In contrast, the relatively sluggish and complex characteristics of the junction power transistor can lead to difficulties in the design of feedback amplifiers with good stability margins.

Also, as has been noted, the power MOSFET is intrinsically free from the problem of secondary breakdown, and an amplifier based on these does not need the protective circuitry that is essential in amplifiers with BJT output devices if failure is to be avoided when they are used at high power levels with very low impedance or reactive loads. The problem here is that the protective circuitry may cut in during high-frequency signal level peaks during the normal use of the amplifier, which can lead to audible clipping. (Incidentally, the proponents of thermionic valve-based audio amplifiers have claimed that the superior audible qualities of these, by comparison with transistor-based designs, are due to the absence of any overload protection circuitry that could cause premature clipping and to their generally more graceful behavior under sporadic overload conditions).

A further benefit enjoyed by the MOSFET is that it is a majority carrier device, which means that it is free from the hole-storage effects that can impair the performance of power junction transistors and make them sluggish in their turn-off characteristics at high collector current levels. However, on the debit side, the slope of the V_g/I_d curve of the MOSFET is less steep than that of the Vb/I_c curve of the BJT, which means that the output impedance of power MOSFETs used as source followers is higher than that of an equivalent power BJT used as an emitter follower.

Other things being equal, a greater amount of overall negative feedback (i.e., a higher loop gain) must therefore be used to achieve the same low amplifier output impedance with a power MOSFET design than would be needed with a power BJT one. If a pair of push"pull output source/emitter followers is to be used in a class AB output stage, more forward bias will be needed with the MOSFET than with the BJT to achieve the optimum level of quiescent operational current, and the discontinuity in the push"pull transfer characteristic will be larger in size, although likely to introduce, in the amplifier output signal, lower rather than higher order crossover harmonics.

9.10 U and D MOSFETs
I have, so far, lumped all power MOSFETs together in considering their performance. However, there are, in practice, two different and distinct categories of these, based on their constructional form, and these are illustrated in Figure 9.22.

Figure 9.22: MOSFET design styles.

In the V or U MOS devices - these are just different names for what is essentially the same system, depending on the profile of the etched slot - the current flow, when the gate layer has been made sufficiently positive (in the case of an N-channel device) to induce a mobile electron layer, will be essentially vertical in direction, whereas in D-MOS or T-MOS construction the current flow is T shaped from the source metallization pads across the exposed face of the very lightly doped P region to the vertical N-/N+ drain sink.

Because it is easier to manufacture a very thin diffused layer (=short channel) in the vertical sense than to control the lateral diffusion width, in the case of a T-MOS device, by surface masking, the U-MOS devices are usually much faster in response than the T-MOS versions, but the T-MOS equivalents are more rugged and more readily available in complementary (N-channel/P-channel) forms.

All power MOSFETs have a high input capacitance, typically in the range of 500"2500 pF, and because devices with a lower conducting resistance (R_ds/on) will have achieved this quality because of the connection of a large number of channels in parallel, each of which will contribute its own element of capacitance, it is understandable that these low channel resistance types will have a larger input capacitance.

Also, in general, P-channel devices will have a somewhat larger input capacitance than an N-channel one. The drain/gate capacitance - a factor that is very important if the MOSFET is used as a voltage amplifier - is usually in the range of 50"250 pF. The turn-on and turn-off times are about the same (in the range 30"100 nS) for both N-channel and P-channel types, mainly determined by the ease of applying or removing a charge from the gate electrode. If gate-stopper resistors are used—helpful in avoiding UHF parasitic oscillation and avoiding latch-up in audio amplifier output source followers—these will form a simple low-pass filter in conjunction with the device input capacitance and will slow down the operation of the MOSFET.

Figure 9.23: MOSFET symbols.

Although circuit designers tend to be rather lazy about using the proper symbols for the components in the designs they have drawn, enhancement-mode and depletion-mode MOSFETs should be differentiated in their symbol layout, as shown in Figure 9.23. As a personal idiosyncrasy, I also prefer to invert the symbol for P-channel field effect devices, as shown, to make this polarity distinction more obvious.