Topology for cascaded gain stages

[dano12]

Continuing on the month's fetish for low-gain transistor stages, would it be possible to string 3 or 4  stages together, directly coupled, with only gain pots for each stage?

Something like this:



I'd like to start breadboarding various stacked ideas--any feedback on:

1. The lack of coupling caps between stages?
2. Having gain pots instead of the more traditional inter-stage "level" pots?
3. Is this just a bad idea all around?

[gez]

The pots adjust the bias of each transistor, so with direct coupling the next stage's bias will be affected and so on.  It might mean you have to adjust all the other knobs to compensate everytime you change the settings of one.  For that reason, it might be wiser to AC couple the stages or just set fixed values for some of those pots.  Also, with that many stages all at full pelt, you might just find you end up with an oscillator.

Buy a breadboard if you don't already have one.  Try it, measure everything with a multi-meter and see what affects what.  Refine it.  Have fun.

PS  Input impedance will be on the lowish side too.

[Sir H C]

PM me, I have a circuit that does some DC coupling with some gain loss but works pretty well with variances in beta and the rest.  The curse is as your gain goes higher, small offsets become big ones at the end, and with variances over temperature, things can swing wildly.

[Mark Hammer]

In principle, nothing wrong with what you posted.  Keep in mind, though, that multiplicative gain is applied to everything at the input, both signal AND noise.  You really don't want hiss to encounter a gain of x5 four times in a row (x625).  Keep in mind, too, that one eventually starts to run out of clean headroom, especially with a 9v supply, and that with each additional stage of limited-headroom gain, you start to produce harmonics of harmonics of harmonics.  For both those reasons, my instinct has been to limit bandwidth at the high end in the earliest stages, and "spool it out" in later stages.  There might also be a case made for using coupling caps between stages to keep any residual 60hz hum from growing too noticeable as well.

[Sir H C]

Another cool technique pulled from limiting amplifiers in the RF world is to bias the first stage off the last (odd number of stages required), but with a large resistor/cap arrangement to only have this at DC, and otherwise have it go straight through.

[dano12]

In principle, nothing wrong with what you posted.  Keep in mind, though, that multiplicative gain is applied to everything at the input, both signal AND noise.  You really don't want hiss to encounter a gain of x5 four times in a row (x625).  Keep in mind, too, that one eventually starts to run out of clean headroom, especially with a 9v supply, and that with each additional stage of limited-headroom gain, you start to produce harmonics of harmonics of harmonics.  For both those reasons, my instinct has been to limit bandwidth at the high end in the earliest stages, and "spool it out" in later stages.  There might also be a case made for using coupling caps between stages to keep any residual 60hz hum from growing too noticeable as well.

By "limiting bandwidth" do you mean introducing filter caps (22-100nf) before the transistor?

[dano12]

PM me, I have a circuit that does some DC coupling with some gain loss but works pretty well with variances in beta and the rest.  The curse is as your gain goes higher, small offsets become big ones at the end, and with variances over temperature, things can swing wildly.

Wait until I'm done swing blindly at this pinata. Then I'll ask for real help

[petemoore]

  looks somewhat similar to a BMP, a distorter, but using the rails as the clipping ceiling..different transistors and bias etc.
 Since it'll be clipping anyway, throwing some diodes in to clamp some of the later stages to X amount of output, [to be input to the next gain stage] so the rest of the signal swing is clipped off is one way to reduce the 'multiplicitave multiplying' factor of the cascaded stages.
 But breadboard makes perfect sense for assembly of the 'first draft'.

[Sir H C]

To limit bandwidth, you can put a cap in parallel with either the collector resistor or the base-collector resistor (but realize that this is multiplied by the stage gain).  Rolloff is set where the impedance of the cap = the resistor load there.

[Mark Hammer]

By "limiting bandwidth" do you mean introducing filter caps (22-100nf) before the transistor?

I was deliberately vague because while I know it can be done, I didn't know how in the discrete context.  Sir HC has pointed out one way.

Note as well that one can use a combination of collector caps and emitter caps in parallel with emitter resistor, to achieve emphases within the passband.

[dano12]

Thanks for the ideas and advice.

I've added Ca, Cb and Cc between the stages,  C1a-C1d as bandwidth limiters and C2a-C2d as emitter caps.

Not so clean anymore



I'll breadboard it tonight.

[Mark Hammer]

Way back in one of the exchanges at the dawn of history, I think that Jack Orman and RG came to some kind of consensus that one started to run out of tangible reasons to cascade any more stages after the 3rd gain stage.  I may remember that wrong, but it sounds close to true.  Got that ring of "truthiness" to it.

Note that emitter caps can be used to achieve high-frequency boost in multiple ways.  For instance, if I run the outside legs of a 1k pot between the emitter and ground, and tie the .47uf cap between the wiper and ground, my full-passband emitter resistance remains constant.  What I vary is how much resistance is imposed between emitter and cap before the path splits.  In that way, you can get graded amounts of high-end boost.

Of course, that's 8 trimpots.  Perhaps it might be wisest to either drop a gain stage or install emitter trimpots in only one or two stages, or do both.

[Ben N]

Way back in one of the exchanges at the dawn of history, I think that Jack Orman and RG came to some kind of consensus that one started to run out of tangible reasons to cascade any more stages after the 3rd gain stage. 

Tell that to Mike Soldano!

[TELEFUNKON]

to me, 47n for C1abc looks a BIT big.

[aron]

Check this out:

http://www.diystompboxes.com/wiki/index.php?title=Simplemods#Simple.2C_Easy_Mods.2FTips_.26_Techniques

Secret of stacked stages.

[dano12]

I managed to get the first schematic on the breadboard before running off to kid duty.

As predicted, it sounded like ass, with fantastic oscillations, noise, hiss and general tonal mayhem.

Tonight I'll start adding the improvements one by one.

Interesting about the conventional wisdom of diminishing returns after the third gain stage.

[coxter]

Any updates?

[Gus]

looks like a mess to me.

Bunch of thing I don't like
do you understand the complex interactions going on in the schematic?

You have the open loop gain changing

The feedback C to B interacting with the stage before collector output R(changes both output R and bias point and open loop gain with turning  the gain pots) then the power supply current waveform and output Z with high gain and cheap batterys can cause oscillations

DC on pots is not a good thing

HP, LP .........

I don't like trim pots in the collector leg you are using Si transistors one can design you often don't need trims with Si if you know what you want.  (My schematic bias pots are not for finding one bias point but to allow different sounds)

[johngreene]

Since you are intending to adjust the 'gain' with the pot in the collector, why not just have the wiper connected to a capacitor that progressively bypasses the rest of the pot? That way you don't change the biasing, just the amount of AC resistance in the collector. The effect would be more pronounced if you didn't AC bypass the emitter resistors too.

$.02

--john

[mac]

What about fixing the collector resistors and replace those 680R with 1k pots a la Fuzz Face?

For both those reasons, my instinct has been to limit bandwidth at the high end in the earliest stages, and "spool it out" in later stages.

I agree. IMHO the best high cut filter is the guitar tone control. In effects like this I like the tone after the first stage, but this is my taste.

mac