Why cascade transistors?

[demonstar]

I've been trying to find out why people cascade transistor sections but haven't found much. Is ts to do with frequency response? I don't really understand why it would be done at the moment. I'm sure I'll be missing something (I'm sure there'll be a very good reason!)! I understand it opens up the options for more feedback configurations. How come you don't tend to see opamps cascaded as often as transistors?

Thanks.

[MetalGod]

Opamps have MUCH more gain available than transistors.  Transistors are generally cascaded to get more gain.

[petemoore]

How come you don't tend to see opamps cascaded as often as transistors?
 
  Take a look at the shematic diagram of an opamp. There are numerous transistors in an OA.

[d95err]

Opamps have MUCH more gain available than transistors.  Transistors are generally cascaded to get more gain.

Yes. For all practical purposes, opamps have virtually infinite gain, so nothing would be gained by cascading them. This is the basic reason for wanting an opamp. The gain is (almost) infinite, so you must have a feedback network to control the gain of the circuit. This has the benefit that the gain will stay the same, regardless of the supply voltage, variations between indiviual chips etc. In a transistor gain stage, gain is much more difficult to predict, and can vary a lot with individual transistors, supply voltage, temperature etc.

So cascaded transistors are often a way to get near-opamp gain and hence, you will usually see a feedback network in such circuits (e.g. like in the Fuzz Face).

[demonstar]

Ok, thanks guys!

I guess this makes me ask if opamps are so great, why do we use transistors? 

[gaussmarkov]

as i understand it,  op amps are too perfect.  interesting sounds often come out of imperfections in the amplification process.  amplifiers based tubes are the leading example of imperfect.  they sag.  they clip softly.

[Steben]

We simply use opamps and transistors for different purposes.
All sound different when pushed out of the linear region.
The first opamps actually were worse than transistors even in nosie and gain stability. But the evolution is still going on. There are opamps today that for instance clip nicely (-> SansAmp, Black cat overdrive, ...).

[Gus]

It is often not just for gain if it is an effect designed by someone that knows what they are doing.

[demonstar]

Thanks again! I decided stuff this. The best way to find out is to get my hands dirty. I'm in the middle of fiddling with the breadboard. I'm just going to try lots of different transistor circuits and mix and match the gain sections of each and see what happens. That way I know first hand if cascading transistors makes any difference and will hopefully learn other useful stuff in the process.

[gaussmarkov]

We simply use opamps and transistors for different purposes.
All sound different when pushed out of the linear region.
The first opamps actually were worse than transistors even in nosie and gain stability. But the evolution is still going on. There are opamps today that for instance clip nicely (-> SansAmp, Black cat overdrive, ...).

i'd love to see more discussion of this, because my limited understanding of op amps (based on the "infinite gain" description) implies that op amps hit the rails abruptly, not in a way that would produce rounded clipping.  any help appreciated!

[petemoore]

i'd love to see more discussion of this, because my limited understanding of op amps (based on the "infinite gain" description) implies that op amps hit the rails abruptly, not in a way that would produce rounded clipping.  any help appreciated!
  Lifting the diodes, Cw-ing the gain, and boosting the input of a Dist+ or Rat demonstrates exactly what opamps [-voicing] sound like when 'railing out'...generally less than smooth to ragged clipping.
 
 

[R.G.]

I'm just going to try lots of different transistor circuits and mix and match the gain sections of each and see what happens. That way I know first hand if cascading transistors makes any difference and will hopefully learn other useful stuff in the process.

There's about fifty-some years of that kind of thing done by experts already documented in books. You might shorten up your time a bit by some reading.

On the other hand, if you refuse to listen to the naysayers, and just go experiment on your own, you may find your own beautiful reality and zvex may adopt you.

Depends entirely on whether you want to experiment or want to know. Education is always expensive. You're choosing to pay for it with parts cost and time. Sometimes money and time spent on books is cheaper and faster. But have a good time!

i'd love to see more discussion of this, because my limited understanding of op amps (based on the "infinite gain" description) implies that op amps hit the rails abruptly, not in a way that would produce rounded clipping

The issue is somewhat complicated.
(1) Opamps are merely arrays of transistors connected up in both direct gain-providing and supporting-role circuits. It is possible to build opamps out of discrete transistors if you will be careful about matching the transistors where they need matched. So Opamps are not distinct from transistors, just a complex mix all in one small package.
(2) The *entire* point of opamps is to make what happens inside the opamp itself not matter to the outside world. This is done by using the massive gains in a feedback arrangement to "hide" whats going on inside.
(3) The imperfections of the insides intrude in two ways. (a) the way we keep it stable - not oscillating and (b) when feedback can no longer cover up the insides.
(4) To keep it stable, we must decrease its gain at some high frequency where a gain of > 1 would inevitably make it oscilate. So we have to get from a gain of 10 zillion at DC down to less than 1. This has to be done with one capacitor for advanced reasons outside this discussion. Because 1 cap can only reduce gain by 20db/decade, the cap rolloff of gain has to start down around 1 Hz. So the open loop gain of almost every opamp is falling by 20db/decade throughout its useful range.
(5) Gain also decreases when the amp starts to bang into the power supplies. Like all other amplifiers, this is not perfectly abrupt, there is some rounding when you look at it open loop. When the open loop gain decreases, it can no longer hide the internal operation as well, so the banging into the power supplies slows down as you get close to the power supply.
(6) The higher the closed loop gain (i.e. Rf/Ri is big) the less internal gain is available to hide the internal operations. An opamp used open loop has a gain of a zillion that reduces with that internal capacitor, but you can see the internal wiggles and imperfections now.
(7) The lower the closed loop gain, (i.e. a follower or inverter with Av=1) the more gain is available to hide the internals and the more abrupt the clipping against the power supply looks.
( the slow-down cap in side also limits how fast the output can move - it's slew rate is limited. So an opamp may not be able to go toward the rails as fast as the signal driving it to the rail. It's slew rate limited.  Slewing distortion is one that was hard to find.

The first opamps were OK for DC, dreadful for audio, because they had low open loop gains to hide internal distortion, low slew rate so they distored on treble signals, had lots of noise (which feedback neither helps nor hurts), and could "stick" at a rail if driven into it. The 741 was an *improved* opamp that cleaned up some of the problems of early ones. And it's dreadful.

Opamps are not good for rounded clipping unless the feedback is set up to tell them to do rounded clipping. Some are less bad than others.

[Steben]

Opamps are not good for rounded clipping unless the feedback is set up to tell them to do rounded clipping. Some are less bad than others.

exactly, but acceptable hard clipping (as in "distortion to fuzz") is possible with TLC2262 for example or even a TL07x series (as used in SansAmp or XXL).
especially when there is a decent high roll-off feedback.
They are designed to "control themselves" in the saturation area. Which means they clip harsh, and that's it. Standard opamps or older ones tend to go crazy rather than to clip, I mean they suffer from more than harsh clipping (bias drifting, latch-up,...).

[gaussmarkov]

thanks very much.    

[demonstar]

Depends entirely on whether you want to experiment or want to know. Education is always expensive. You're choosing to pay for it with parts cost and time. Sometimes money and time spent on books is cheaper and faster. But have a good time!

I try my best to balance experimentation and reading books. I'm not planning on going totally experimental and blocking the world out. I am more thn happy to swallow chunks of information from books.. I've been reading a good book lately called "understanding amplifiers". I just feel it is a bit general and fails to address everything I'd like. I would recommend it for beginners in this area.

Thanks.

[SISKO]

I have another question.
Say im working in an common collector amp whit a voltage divider at its base and an emmiter resistor (10V power supply, no need for amplitude distortion). Im working on it in the linear region (obvious). The input voltage is 100mV peak to peak.

So, in order no to distort the signal i shold have a gain of.. say 90.
What should i choose, 1 transistor satge with gain= 90 or another arrengment (two stages 1gain= 4.5  2gain=2   i.e.)?

Im thinking of a filter problem formed by Rc and the coupling cap (yeah! thinking while writing )

Is there any problem in asking too much gain to a transistor? (i know, 90 its not THAT much gain, but in other cases)
And the last one (sorry) what happens when we, say a transistor wiht beta=80 ask for a gain of 120?

I know, those are silly questions, but i just cant get em out of my mind.