Opamp, mu-amp, correct biased transistor, ... they all simply have lower headroom as voltage drops...
Using the log base-emitter response is way too fragile.
I'd do it the obvious way - use something which senses the voltage and lowers gain directly from that. An OTA with its bias current fed from the power supply through a resistor does exactly this.
thx RG, very true. Thought about mulitplier f.e. as well ...
Still there is one disadvantage: bias shift and harmonics
headroom? check Mu-amp's self bias gives same smooth clipping @ less headroom
Compression? check OTA
...
But! harmonic bias shift? Extra second order harmonic @ stronger picking? Log response shifting would be better in that actually along with compression....
bias shift in low gain Darlington tranny stage?
On the other hand, your point about turning to the obvious ... : once you use a stage per goal (a mu - amp, an OTA, ...) you could simply use a sharply biased input stage that gives enormous assymetric gain at stronger picking and then feeds the other stages. The result can be just about "it". Since wanting it simple all-in-one stage brings you directly to ... tubes again. And a bit more complexity is acceptable with solid state.
Quote from: Steben on May 12, 2011, 02:55:01 PM
The result can be just about "it". Since wanting it simple all-in-one stage brings you directly to ... tubes again. And a bit more complexity is acceptable with solid state.
If you are after a solid state tube emulator, there's at least thirty years of smart electronics people that have tried that, with varying success. Anything simple and all-in-one has likely already been tried.
> Using the log base-emitter response is way too fragile.
? ? ? After 12-cent resistors, it is the most predictable thing in electronics.
(http://i.imgur.com/MamDL.gif)
Assuming ZERO source impedance, gain is proportional to supply, over w-i-d-e range. Supply higher than Early Voltage (~~50V) runs into finite plate resistance. Supply comparable to Vbe, gain falls off. Between, it is pretty dead-nuts proportional to supply.
This is a simple form of a multiplier widely used for its predictable action over a wide range of gain.
You also note the problem: the input impedance drops as gain rises. If source is NON-zero, this negates some gain-change. If source is high impedance, the total voltage gain is constant (ratio of load to source times hFE). You can hide that with a stiff buffer (emitter follower or opamp, not JFET) in front.
Also note the input overload voltage is smaller than most hard-played guitar. NFB would spoil the gain change. You must attenuate in front. You must accept marginal S/N (60db-70db).
There's an oversight in the penciled numbers: input impedance is about half that shown due to 1Meg/Gv.
> enormous assymetric gain at stronger picking
Fiddle the "1Meg" to set collector low, about 1/4 of supply. Gain at 30mV peak is very asymmetric, approaching 26% pure 2nd harmonic.
> tubes again
Tube gain does not change so much with supply. 400V to 100V gives maybe 20% change, not 400% as with a transistor over 4:1 range of bias.
Nice PRR ! thx
Perhaps that 30mV can be more if you use a darlington?
Tubes: yes true. Yet, a classic tube amp has several triode stages (a plexi 4 in front of end pentode). Each stage has a bit cathode shift (grid-cathode bias).
Thinking about a circuit which uses a 386 chip ... "excellent" dynamic current draw ....
(http://i31.photobucket.com/albums/c356/StebenDerMeister/ampsim.jpg)
ampsim diagram
one tranny stage only in front, 386 stage with relative low input impedance (to lower output of tranny stage) and low voltage zener diode clipper with compensating cap (more clipping at higher freq).