# Why does this power opamp have MOSFETs at the output stage

I saw a post a few days ago.

Power Dissipation of Op-Amp where I was actually curious about the circuit where they use BJTs all over the circuit but use 2 MOSFETS for the output stage.(also why are they using JFETs?)

In most designs that I have seen online, only BJTs are used.

Is there any advantage of using MOSFETs instead of BJTs for the output?

Are MOSFETs not less linear than BJTs? If so what is their advantage?

It would be great if someone can shed some light on it.

• It must have seemed like a good idea at the time ... Seriously, it must have been the best way to meet some set of criteria, which you haven't specified. The datasheet suggests that it is intended for delivering high pulse currents with high slew rates, such as required in industrial inkjet printers. Note that this is NOT a monolithic opamp; it's a module built from discrete parts. – Dave Tweed Nov 5 '18 at 22:48
• I thought people used MOSFETs for power applications because they are more efficient that BJTs. The wasted power in a MOSFET is Rdson*(output current) and Rdson is only limited by how wide you make the transistor, while in a BJT you have the saturation voltage no matter what so you will be dissipating Vcesat*(output current). I guess that's for switching applications though, and this is linear. – DavidG25 Nov 5 '18 at 23:05
• @DavidG25 Yes, this is a linear application, so there is no efficiency difference between the two technologies. (incidentally, it's $R_{DS,on}·I_{out}^2$, not $R_{DS,on}·I_{out}$) – Hearth Nov 5 '18 at 23:11
• As for the JFETs, they're used to have a higher input impedance than BJTs can provide, though given that they used MOSFETs elsewhere, I don't know why they didn't just use MOSFET input transistors, which provide an even higher input impedance than the JFETs. – Hearth Nov 5 '18 at 23:13

The reason could be partly due to the absence of second breakdown. In BJTs the power dissipation capability is often reduced at high voltage as the current tends to be concentrated in localised hot spots. Some MOSFETs do not suffer from this. With this device the safe operating area graph shows a straight line up to 100 V when the device is dissipation limited.

MOSFET's are not ideal output transistors but they have 2 major advantages over bjt's.

1. They do not have the same type of thermal runaway issues like bjt's do, but so many MOSFET's are designed for switch mode apps only. Still pins 27 and 28 are used to sense output drive current and clamp it to safe levels.

2. MOSFET's have a very low ON resistance especially when saturated ON. Thus the Vce drop at high current levels is a fraction of what bjt's have, so you can drive very low impedance loads with better efficiency than bjt's.

3. When I mentioned that MOSFET's are not ideal output transistors for power amplifiers I was referring to the fact that the bias voltage for the output stage is 8 volts or more (Q14 and IC1 handle that), and matched P-Channel and N-Channel MOSFET's are hard to make due to their atomic structure. N-channel MOSFET's work with electrons while P-Channel MOSFET's use 'hole' carriers. Also most MOSFET's made today are for switching apps, so being ON 50% in a linear mode can make them heat up in a big way.

4. At low voltages under 100 volts this is not much of an issue, but a 2,500 watt amplifier with +/- 120 volt supply rails makes implicit use of 8 to 10 pairs of matched NPN-PNP power transistors, rated for 250 VDC. With so many in parallel the very low output impedance required is met, and I have found no P-Channel MOSFET's with anything close to a 250 volt rating. This is a technical issue that maybe solved one day.

5. High powered type AB amplifiers waste 10% of full power just sitting idle with no sound injected. This includes MOSFET types because they also need a high bias current to keep THD to a very low level. The fix for this issue is a digital power supply designed around a Texas Instruments DSP engine that feeds just the voltage the output stage needs within 10 uS. It is much more efficient but much more expensive-for now.

• Statement 1 seems sweeping. In 1982 I built an audio amplifier with Hitachi MOSFETs 2SK175 and 2SJ55. Transfer curve shows a sweet spot at about 80 mA where the temperature coefficient of Id versus Vgs is approximately zero. This made biasing easy. No second breakdown either, adding to robustness. These days most MOSFETs are aimed at switching applications. They tend to have a positive tempco, making thermal stability a problem in class AB. They suffer an effect similar to second breakdown, which limits Pd at high Vds. There are particular MOSFETs aimed at linear applications these days. – Steve Hubbard Nov 6 '18 at 2:08
• @SteveHubbard Agreed that technology is changing fast. I am sure there are labs researching into high voltage matched P-Channel and N-Channel MOSFET's. In theory they should be the outputs of choice, but we are not there yet. I will modify line 1 thanks. – Sparky256 Nov 6 '18 at 2:25
• great explanation! – Navaro Nov 6 '18 at 23:43