> a different method of setting bias...by using an O'scope rather than calculations...
Sure, for DIY or boutique-build.
In mass production, and especially in ICs, you don't have time to put a scope on every unit. And IC currents tend to vary from wafer to wafer.
You need a self-bias scheme.
In an "inefficient" Class A amp, the minimum idle current is half the peak output current. The maximum current is the size of the heatsink or power supply. In most headphone cases, the power supply isn't a real concern. So your minimum current is say 125mA (1 watt in 32 ohms), your target current is maybe 180mA to cover wafer baking variation, and routine maximum current might be 250mA. If a 250mA part is fed 30V and put on a 20 deg C/Watt heatsink, it will run VERY HOT, too hot. But it is easy to have temp-sensing on a chip. So if temperature hits 100 deg C (which is tolerable for modern silicon if it is steady, not varying like a Class B design), turn-down the bias current to hold the termperature. Simple thermostat, except proportional not on/off like home-heat thermostats.
Yes, if used with a 50 deg C/Watt heatsink, or a big sink in 50 deg C ambient, maximum allowable dissipation will be like 1.5 watts. On +/-15V supply, bias current will have to turn-down to 50mA. It won't be full Class A with loads under 100 ohms. In 32 ohms, above 150mW (pretty loud) it will work as Class AB. But that is fine for most uses. "Class A" power amps are more a fad than an essential.
We could have a Bias pin and program the desired operating point. But I'd like to see a simple 5-pin TO220 package, which can't spare a pin for bias.
Oh, and logically the bias current should vary with supply voltage. Use a resistor across the rails instead of an FET. At +/-5V we only need about 45mA for class A into 32 ohms. This also gives "cheap box" makers the option to simply reduce rail voltage if they don't want to feed it the current and power needed for full-power operation.
Sure, for DIY or boutique-build.
In mass production, and especially in ICs, you don't have time to put a scope on every unit. And IC currents tend to vary from wafer to wafer.
You need a self-bias scheme.
In an "inefficient" Class A amp, the minimum idle current is half the peak output current. The maximum current is the size of the heatsink or power supply. In most headphone cases, the power supply isn't a real concern. So your minimum current is say 125mA (1 watt in 32 ohms), your target current is maybe 180mA to cover wafer baking variation, and routine maximum current might be 250mA. If a 250mA part is fed 30V and put on a 20 deg C/Watt heatsink, it will run VERY HOT, too hot. But it is easy to have temp-sensing on a chip. So if temperature hits 100 deg C (which is tolerable for modern silicon if it is steady, not varying like a Class B design), turn-down the bias current to hold the termperature. Simple thermostat, except proportional not on/off like home-heat thermostats.
Yes, if used with a 50 deg C/Watt heatsink, or a big sink in 50 deg C ambient, maximum allowable dissipation will be like 1.5 watts. On +/-15V supply, bias current will have to turn-down to 50mA. It won't be full Class A with loads under 100 ohms. In 32 ohms, above 150mW (pretty loud) it will work as Class AB. But that is fine for most uses. "Class A" power amps are more a fad than an essential.
We could have a Bias pin and program the desired operating point. But I'd like to see a simple 5-pin TO220 package, which can't spare a pin for bias.
Oh, and logically the bias current should vary with supply voltage. Use a resistor across the rails instead of an FET. At +/-5V we only need about 45mA for class A into 32 ohms. This also gives "cheap box" makers the option to simply reduce rail voltage if they don't want to feed it the current and power needed for full-power operation.
