transistor biasing

[KevinIV]

What is the lowest current a 200mA transistor should be operated at to prevent distortion caused by bias drift?

[audioguru]

What is the lowest current a 200mA transistor should be operated at to prevent distortion caused by bias drift?

Low current in a transistor does not cause bias drift. Improper design does.

The worst possible way to bias a transistor is to ground the emitter and feed current to its base with a resistor from the positive supply. Each transistor has a different current gain so some will be saturated and some will be cutoff. The few that work will have their operating point driffting all over the place due to temperature changes.

A transistor is supposed to have its base fed a current from a voltage divider that has a current of about 10 times the base current.
The transistor should have an emitter resistor with a value of about 1/10th the collector resistor value.
The transistor should have plenty of negative feedback to avoid too much distortion.

[KevinIV]

A voltage divider bias at 10 times the current creates a low impedance for the signal source. An emitter resistor 1/10 the collector resistor is only a gain of 10. A single pole RC will reduce the critical frequency %70.7 and the bandwidth small.

But given the inherent voltage changes in a circuit, isn't 1mA too close to cutoff and the beta at that current isn't linear, but becomes linear at higher currents?

[audioguru]

The datasheet for a 2N3904 has a curve of its typical input impedance (its hfe x its internal emitter resistance). At 1mA its input impedance is only 3.9k ohms when its emitter resistor is bypassed for very high voltage gain.
Its base current is typically 4.3uA so the biasing voltage divider should have 43uA.
If the supply is 9V then the divider is 27k to ground and 160k to the positive supply which in parallel are 23k which is much higher than the input impedance of the transistor.

Of course the transistor should have its emitter resistor unbypassed for negative feedback to reduce distortion. Then the transistor amplifies its input impedance and emitter resistor value.

The hFE and hfe (beta) are its current gains, not voltage gains. It doesn't matter that they drop at low current. The distortion of a transistor is caused by its exponential (not linear) response of its transconductance. Negative feedback makes the transconductance more linear.

[KevinIV]

Okay, I understand the stability with that biasing. Does that mean that there is little versatility with a 2N3904? Signal amplifiers require several stages of varying amplitudes and load impedances. Do you recommend different transistors all operated at their data sheet typical values?

[audioguru]

The beta of a 2N3904 doesn't change more than 3 times from its spec'd minimum of 100 to its spec'd maximum of 300. So if the base bias voltage divider has a current of 10 times the typical base current then the divider current is still 5 times the minimum base current which is fine.

[KevinIV]

If a 2N3904 has a base bias bias current of 4.3uA, then it would need a high impedance load. A base current of 500uA would be needed for a lower impedance load. So isn't 4.3uA base current a bias intended for high impedance loads?

[audioguru]

The current gain of a 2N3904 transistor is typically 230 from a collector current of 0.1mA to about 10mA. Then with a base current of 4.3uA and a collector-emitter voltage of 5V its collector current is 1mA.

If its supply is only 5V then its emitter resistor can be 470 ohms and its collector resistor can be 2k ohms. It is not a high impedance.

If the supply is 30V then the resistor values can be 6 times higher but are still not a high impedance.

[KevinIV]

Since the collector current reduces with an increasing voltage, the load current increases. Can't a 2N3904 handle at least a 100mA load at all? So the base current would at least be 500uA.

[audioguru]

The 2N3904 can drive a collector resistor (not a load) with a current of 100mA but its current gain is less and it might melt.
The range of current gain is guaranteed at a collector current of only 10mA. With a collector current as high as 100mA then its current gain is a minimum of only 30.

Maybe you are looking at the graph of "typical" transistors. But you cannot buy a "typical" transistor because some are better and some are worse.

[KevinIV]

A 2N3904 at 100mA may have a beta of 30, but I guessed it to be closer to 100. The power dissipation was within 100mW of it's maximum. A transistor with a 3A maximum current rating can have a minimum beta of 30. Doesn't that rating apply to any bias voltages?

[audioguru]

A 2N3904 at 100mA may have a beta of 30, but I guessed it to be closer to 100.

A 2N3904 transistor will melt if it has 100mA of current and a collector-emitter voltage higher than only 6.25V. if you want power then use a power transistor mounted on a heatsink. For the 2N3904, the datasheet spec's a minimum hFE of only 30 and a typical hFE of 150 at 100mA.

The power dissipation was within 100mW of it's maximum.

Then it is too hot and it will fail in summer.

A transistor with a 3A maximum current rating can have a minimum beta of 30. Doesn't that rating apply to any bias voltages?

The datasheet will tell you the details.

[ece_jm]

thanks for this wonderful discussion.

just a question:
for instance  the datasheet shows 2 gains, max and min, which gain should we use for safer design?

[audioguru]

the datasheet shows 2 gains, max and min, which gain should we use for safer design?

If you design a circuit using maximum gain then the circuit will work properly only if it uses transistors that have maximum gain but they might not be available and you must test many transistors to look for them.
But if you design a circuit using "worst case" transistors having minimum gain then all transistors with that part number will work perfectly.