Why collector/drain output stages are so rare???

[Christer]

Originally posted by Ed Holland

I built my first Hi fi amp from two of these modules, at age 17. I'm now 36 and still own and use that amplifier - it has a very clean sound and seems well behaved and "musical". However on one occasion, it did receive Swedish International Shortwave radio on 9 MHz. I was alerted (and startled) by the time signal, heard between tracks on the CD to which I was listening. The speaker leads seemmingly acted as antenae! Perhaps this pickup was an artefact of the common emitter circuit design, gain in the output stage?

It will have a higher open loop output impedance and thus relies on feedback to get a low output impedance. However, that also means that the output impedance will get significantly higher at HF than for emitter followers, so I guess that makes it easier for the amp to pick up radio signals which sneak into the feedback loop. Maybe you didn't have any Zobel filter at the output to guarantee a low output impedance at HF (which is anyway not the main reason they are used)?

[Ed Holland]

Originally posted by Christer


Maybe you didn't have any Zobel filter at the output to guarantee a low output impedance at HF (which is anyway not the main reason they are used)?

There is a resistive/inductive Zobel network, as I recall, though this seems to have been ineffective at 9 Mhz

The pickup was so feint that it was barely audible if it did occur. Most of the time, it was not evident, but could be provoked with an RF signal generator!

Ed

[AKN]

Hi edl,

QSC used this output topology.

Which one(s) of the QSC schematics are you refering to?

[Piercarlo]

Originally posted by Ultima Thule

but on the other hand common emitter/source have much more gain added to the overall open loop which by the means of higher GFB rate also will add up to decrease output impedance/increase the damping factor where an all else equal common collector/drain will instead have initial lower output impedance and higher damping factor before the GFB is wrapped around, but also less open loop gain...

Just a little correction. Sziklay compounds (another name for collector/drain output) *don't* increase overall open loop gain but establish a second local loop inside to the general NFB loop. And open loop decrease of output impedance is not automatically guaranteed from the Sziklay compounds: this should be a explicit goal to be obtained by the designer. The real advantage of Sziklay is a better behaviour of equivalent output transistors (which are, for the remaining amplifier, emitter followers again).

This "equivalent device" may be optimized for a more "ideal" beta behaviour and improved crossover distortion (in theory at least) that may permit a less stressing (and more linear) loading of the VAS stage preceding the output section, which, as knew, is the "real" output stage from a output voltage generation standpoint. Output impedance *may be* reduced by the Sziklay but it's not by default: it's an option of the designer which may decide of investing some of extra current gain to do so instead of merely stabilize it (and, in the same time, stabilizing the "multiplier factort" which define the output load reflected on VAS stage and also, in some limits, the dependence of VAS open loop gain from the output loading).

However Sziklay (and more complex topologies as the triplet output stage - substantially a Sziklay buffered by a *real* emitter follower) pay their electrical improvements with an increased uncertainty on the thermal stability side, due to the difficult of tracking thermal behaviour of a complex stage from a single point of control (the usual Vbe multiplier). On paper is all wonderful but non reality the "wonderful" come with some odds which, often push back the designer to a more collaudated simple emitter follower stage.

Hi!
Piercarlo

[AKN]

Originally posted by 4fun
Hi edl,



Which one(s) of the QSC schematics are you refering to?

edl,

Ok, found them.
When you mentioned QSC i was thinking about their grouded collector designs and forgot their other designs with collector output.

[edl]

OK,

Cheers

[Ultima Thule]

Originally posted by Piercarlo


Just a little correction. Sziklay compounds (another name for collector/drain output)...

Piercarlo,

I agree with you a lot, but still fail to see the "little correction" in your context while I was explicitly talking about common emitter/source VS common collector/drain solely without complicating the subject with nested loop schemes.
But if we would like to add up with outputstages deviating from above said we can even add in the current dumper which could be seen as a "double emitter" output device for a BJT setup in comparison to the CFP which could be seen as a "single emitter + single collector" output device.

Cheers Michael

[Piercarlo]

Originally posted by Ultima Thule



Piercarlo,

I agree with you a lot, but still fail to see the "little correction" in your context while I was explicitly talking about common emitter/source VS common collector/drain solely without complicating the subject with nested loop schemes.
But if we would like to add up with outputstages deviating from above said we can even add in the current dumper which could be seen as a "double emitter" output device for a BJT setup in comparison to the CFP which could be seen as a "single emitter + single collector" output device.

Cheers Michael

Ok! . You intend "raw" collector outputs, just as it was a VAS with some beefing up in current without a real buffer current stage (as usual emitter followers are); it's right? If so i agree with you: we talking about different things. A tit for tat due to my not-well-running english! :-).
However, if are this kind of ouput stage in discussion, the answer to the original question is very simple: collector or drain output are "quiet" output stages only with resistive load. With any other kind of load (and with the presence of NFB!) is a something which span from an headache to a deep nightmare! :-).

Ciao
Piercarlo

[ilimzn]

Simple answer: OL gaindepends heavily on load. Consequently, CL gain and stability depend on it also. Reactive loads can lead to all sorts of interesting effects

Still, these topologies are useful if the effective number of stages can be kept at a minimum, normally one or two. Still, some form of feedback (local, global, nested) is needed to give the design stability, if only at DC.

There are also advantages to these topologies, usually the ability to swing very close to output rails, when used in complementary fashion.

[Eva]

The huge pitfall of collector/drain output stages is the complete lack of quiescent current and cross-conduction control when large high frequency AC signals are amplified. Even popular circuits like SKA suffer very badly from that phenomena, but it seems that in audio world nobody is competent enough to take the effort to insert a small shunt in series with one of the drains or collectors and see actual current waveforms with oscilloscope (like I did).

Anybody that cares to take a look will find unexpected current tails in these output stages, that in some circuits may be large enough to become already destructive when playing just large amounts of trebble.