Transistor Output Capacitance and signal integrity.

[Oneminde]

"The best amplifier is a wire with gain.. everything else is deductive".​

Lets start with I am not an engineer, so there might be things that elude me, but are apparent to someone who is designing amplifiers and boy am I in the right place 🙂. The talent this site contains are enormous amount of talent,

The triode is for most part the best amplification device we have and very few sand devices come close. The only transistor that I know of is the JFet and one of the reasons can be found the internal capacitance. We could argue that Pentodes like the KT88 also is a very good device.

My topic for the day is Output Capacitance and signal integrity - if all else is equal, if we have component A and its gate to anode/drain/emitter is 25pF compared to component B which is half at 12pF. Would it stand to reason that this difference should be more favourable for the 12pF if we want to maintain the highest signal integrity.

A 300B tube have roughly Cg-a of 7.2pF. A KT88 on the other hand is 2.5pF and we know that similarly designed amplifiers will sound a bit brighter or more detailed on the KT vs the B amp.

The questions is simple: In an audio amplifier, is a low gate capacitance always favourable ?

 

[Zen Mod]

one step aside
suddenly you realize that amplifier is more made of compromises than of parts

so, just one tiny detail is hardly useful to make generalization of

anyway - in general - multiply cojones of driver stage, no worries of doubled interelectrode capacitances

but then , you must think of current demands of Gate itself, not just of interelectrode capacitances

etc.

etc.

etc.

 

[Mark Johnson]

Several of the undesirable results of grid-to-plate capacitance (AKA base-to-collector capacitance ; AKA gate-to-drain capacitance) can be reduced or eliminated by employing a cascode circuit. After cascoding the plate or collector or drain is held at a constant voltage, so there is no Miller multiplication of the grid-plate / base-collector / gate-drain capacitance.

 

[TNT]

Even more reason to stay in the digital domain as absolutely long as absolutely possible and the transition from digital to analog should be as absolute short as absolute possible 🙂

Thats how to remove a problem instead of solving it.

//

oops, I have stumbled into the Lab agin without noticing - sorry...

 

[Oneminde]

shortest path for sure, but say we already have a short part and want to bump the performance. We could have a similar debate about foil resistors vs std thin film. One is better than the other 🙂

 

[william2001]

It's out of my league but I remember seeing something in one of the Blowtorch threads (skim through those real quick.. 😆 ) They were talking about this undesired device capacitance. The discussion was of Toshiba audio jfets at the time.. In addition to cascoding JC also mentions that the Vds matters here also. My takeaway from it was that we want this on the high side, I seem to remember a number like... 15Vds as a reasonable target. The front end I'm about to start looks to run them at ~18.5Vds, if my figures are right. But cascoding is the main solution, probably, if it's causing terrible insomnia and you just have to do... something🙂. FE currently in my sig has both "suggestions", cascodes with values chosen to put 15Vds across the K170's..

 

[Oneminde]

So the overall consensus is that architecture matters more than utmost component performance. A good architecture can be helped to a degree by a good component, but a bad circuit is FUBAR no matter how good the component is 🙂

 

[jan.didden]

Yes, there are three important things: architecture, architecture, and architacture.
Also called 'topology'.
The process of putting together the best topology for the purpose and available budget is called 'circuit design'. 😎

Jan

 

[Oneminde]

I intended to use the word topology but decided upon architecture ... same same but different 🙂