Talk:Crossover distortion

I think the article needs a diagram of a "practical" class AB amplifier. This means:

1)Include the base-bias resistors and diodes.
2)Include collector resistors (for protection against thermal runaway).
3)Show optional Neg feedback via op-amp.

I think the above is outside of the scope of the article. This information is already present in other articles here. Alfred Centauri (talk) 03:30, 19 November 2007 (UTC)

Scope of article?

Should the lead clarify that the subject is "crossover distortion in amplifiers" ? I mean, crossover distortion also occur in purely passive components, when weak currents pass through asymmetric welds between different materials. It is measurable and it is also called crossover distortion, but it's a different subject. ??

Cheers, East of Borschov 08:19, 13 November 2010 (UTC)

Example audio?

Could someone please include a sample showing this type of distortion (and ideally also a "clean" sample that shows the same audio without the distortion for comparisson) ? --TiagoTiago (talk) 09:38, 27 June 2013 (UTC)

gm doubling

Should the section on AB amplifiers mention gm doubling as it occurs near crossover and results from two devices being on at the same time? ie. a fix for one problem creating another problem.

https://www.diyaudio.com/community/threads/gm-doubling.249483/ (unless a class A is used) the 26mv across the emitter resistor is considered the best. Self and Cordell discuss this in detail in their books.

Why 26mV? For a small current, a BJT transistor output impedance is nearly that of a resistor on which the current produces 26mV (at room temperature) voltage drop. When the current of one transistor is big, it works alone, the other transistor is in cut-off region, and the output impedance is nearly that of the emitter resistor; when both transistors in the output complementary pair pass equal currents, each has its output impedance near that of emitter resistor, and it is connected via the resistor - the output impedance is made of parallel connection of two serial connections of two nearly equal impedances each, so it is the same as for large current. So, both for a small and for a big current the impedance is the same.

How the impedance affects the signal? The amplifier has some load impedance, and the output impedance of its final stage with the load one form a voltage divider which makes the output voltage smaller than the voltage driving it; if the output impedance varies within the signal voltage, it introduces a nonlinearity to input/output voltage relation of the stage; making the impedance independent of signal voltage removes the nonlinearity.

A practical values: assume these emitter resistors to be 0.22Ω (common in amplifiers for 4Ω load), the 26mV condition requires 118mA of the final stage quiescent current; it is rather large value, especially for a battery-operated amplifier (commonly about 3mA is used); but this all is for a single pair, not for Darlington transistors that are commonly used - for them the calculation isn't so simple. Some idea: driving transistor may provide smaller impedance to compensate the higher impedance of the power transistor.

--JerzyTarasiuk (talk) 06:42, 1 May 2022 (UTC)

Using class AB and feedback

Since most practical audio amplifiers use both feedback and class AB operation, should it be mentioned that both can (and often are) used to combat crossover distortion?

Methods used for reducing cross-over distortion

Two methods are already mentioned in the article: AB class and negative feedback; one more was suggested the talk and I added more information on it: utilize gm doubling to reduce distortion in a stage containing emitter resistor by a "26mV" technique.

Yet another method, not mentioned here, is using a current-driven output stage - solution used e.g. in IC-s like TBA800, TBA810, TBA820. In many other audio amplifier IC-s, there is an NPN driver transistor which has emitter connected to ground, and its collector gets supply current via a resistor (usually connected via bootstrap circuit); the current-driven solution has a PNP transistor instead the resistor, providing high impedance of the driver, which causes driving the output stage by current forced into it. IC-s utilizing the technique have almost no cross-over distortions (see these TBA8xx distortion/power graphs). Disadvantage of the technique is increased susceptibility to NPN/PNP current gain mismatch.

Both the "26mV" and the current driving techniques are applicable to BJT (bipolar junction transistor) output stages only.

Vacuum tube amplifiers used careful choosing of their quiescent current value (a technique similar to the "26mV") in a way that difference of characteristics of two output stage valves working in opposite phases was nearly a straight line (for example: assuming the anode current is 10mA*(Ug/1V+5.5)² for Ug in range -5.5V..-4.5V and 20mA*(Ug/1V+5) for Ug in range -4.5V..0V, and sum of Ug of both valves is -10V, we get difference of their anode currents depending linearly on Ug difference). — Preceding unsigned comment added by JerzyTarasiuk (talk • contribs) 09:38, 1 May 2022 (UTC)