Leaving Class A

[cfcubed]

Wanted to add my thanks for the article too... Being new to this, articles like that sure help.

Wonder if the "Our meters don’t go to zero." bit is akin to the Spinal Tap volume 11 bit

[MEH]

Is there a general principle that can be stated about the relative quality of generating overall bias current X through n devices vs. through m devices where n < m?

[EUVL]

I found Fig.4 particularly interesting, and presume they showed the distortion of a single pair of complementary devices at the said bias current.

I wonder what the distortion would look like if the same bias current is shared between say 4 pairs of the same devices in parallel, each device only take 1/4 of the bias current as in Fig.4.

Would you perhaps care to comment on this ?


Patrick

[MEH]

Hey! That's my question!

[Nelson Pass]

This question is commonly asked. Certainly the performance is
not identical for both cases, but there is a cancellation factor.

With parallel devices the transconductance as a whole goes up,
lowering the distortion because the variation per device is lower
by virtue of sharing the current.

At the same time, if the bias is divided between devices, the
distortion per device goes up.

The two effects do not give an exact cancellation, but it's close
enough that for most applications you can decide on the number
of parallel devices based on dissipation needs.

One factor that tends to favor parallel devices is that the
total bias dissipation can be larger even for the same size heat
sink, since there's less aggregate thermal resistance from the
junctions to the sink.

[Nelson Pass]

Originally posted by cfcubed
Wonder if the "Our meters don’t go to zero." bit is akin to the Spinal Tap volume 11 bit

Exactly where it came from.

[EUVL]

> The two effects do not give an exact cancellation, but it's close enough that for most applications you can decide on the number of parallel devices based on dissipation needs.

Perhaps the distortion level is comparable, but how about the spectrum -- does the ratio between say 2nd, 3rd, 4th remain largerly the same, or do they shift one way or another ?


Patrick

[Vix]

"Klunk!" -Your amplifier has just left class A operation! Consequently, small aliens will invade the transistors in your amp, generating distortion. Hence you will experience an instant decrease of sound quality, depriving you of listening pleasures, which may have side effects, such as insomnia. Consequently, you will have only two choices. First, you may turn down the volume (that you don't want), or, burn mo' power, to keep the green aliens away. If this statement does not make any sense to you, please go back to post nr.1 of this thread and read the paper. "Klunk"!

No, seriously, nice paper. By the way, if we connect more devices parallel, the capacitance also increases, making the output more difficult to drive?

[Manu]

Mr. Pass, dear Papa,
referring to page 7, you suggest a ccs as best SE Bias for PP topology. What would you advocate instead of a Res to V- solution ...?
Manu

[Nelson Pass]

The resistor works fine as long as the supply isn't too noisy.