Request For Comments …
Your Comments ….
Regarding this new idea of mine:
a Dual Voltage Low Noise Differential Darlington …
The concept is as follows: ”LOW NOISE” transistors tend to operate off low voltages and can not deliver any significant current output. So there is a natural “need” to protect them:
a). from the very high voltages of the supply rails of the power amplifier,
b). from excessive current draw from the Voltage Amplifier stage.
Then again, it would be “nice” to have them as the very first stage of the amplifier.
Dual differential, at that (i.e. both horizontally differential, as well as vertically differential).
So the idea is the following: we build a “Darlington” where each transistor of the pair is supplied from a “Different” voltage level (protective cascode for the Low Noise control transistor) but at the same time they are “fed” from the same “current mirror”, so that jointly, the Darlington pair suffices the demand for “equal” currents within each branch of the differential amplifiers, that are built from such funny Dual Voltage Low Noise Differential Darlingtons
Actually, the circuit as show below is comprised of two complementary sets of differential amplifiers, each respective to “opposing” voltage rails, so as to create a vertical mirror.
But the core concept falls back to the basic question:
Can I “split” the current stemming from each of the respective branches of the current mirror, and take a small part of this current “aside”, so as to feed the LOW NOISE input transistor, (albeit indirectly, through a cascode that lowers it’s collector voltage to a safe level,) and let the other part, the “main” part of that current flow go to the executive transistor of the Darlington pair, and then “join” these currents together again (at the emitter), as if the currents ‘originated’ from a single, super, Low Noise, High Gain, High Voltage, High Current Input transistor ?
The basic differential amplifier block is comprised of the two branches of such a special configuration. I highlighted the main building blocks of the idea, for easier reference:
RED: The Constant Current Drain.
GREEN: the “Basic” Differential Amplifier, made of “Normal” transistors. Although a matched pair of transistors is used here, degeneration resistors are included so as to even out the symmetry of the circuit even more.
YELLOW: The current mirror, based on matched transistors. The role of this circuit is to guarantee an EVEN split of the current, as defined by the (RED) Constant Current Drain, into two equal halves. “No matter what”. This current mirror has two equal valued degeneration resistors in the emitter circuits of the transistors. Although a matched pair of transistors, such degeneration resistors are included to provide for an even more “equal” split of current, into exactly similar valued HALVES.
BLUE: These are LOW NOISE, LOW VOLTAGE and LOW CURRENT transistors and together with the aforementioned “Green” transistors, they form a pair of “Darlington Transistors” of sorts. The blue ones constitute the driver transistors of the Darlingtons. But these Darlingtons are a bit … odd, as they share the same current, which is sourced from the (yellow) current mirror, but each of the transistors within such odd Darlington works off a different collector supply voltage. This is possible due to a cascode transistor that lowers the voltage potential on the collector of the (BLUE) driver transistor. The current path via such (BLUE) driver transistor “steels” or sidetracks just a tiny bit (due to high beta/hfe, less than 1%) of the total balance of the current that is made available by the (YELLOW) current mirror. The “rest” of the current (99%+) flows unobstructed to the “normal” (GREEN) differential transistors which act as the executive/output transistors of the Darlingtons, but also, at the same time – they constitute the main differential amplifier. It is assumed that the Low Noise (BLUE) driver transistors have a beta (hfe) of at least 100. This is a reasonable assumption, as most of the time they tend to have their amplification factor in the multiple hundreds.
PINK – Now, this is the “Ziggy” thing. I came up with this cascode, as a means of lowering the voltage, on the output of the (PINK) cascode transistors, so that the Low Noise transistors do not blow up. As the voltage of the emitters of these transistors is 0,7V lower than on their base, it suffice to provide a stiff voltage reference, with reference to the original rail voltage, but pulled down by a significant offset value. This can be done by a hushed up Zener diode. The voltage at the zener is significantly decreased (for example: by 24V or more). I shall try to keep the zener quiet by means of a set of capacitors (C filter, or better yet, a CRC filter). Here is a depiction of the “silent zener” concept:
The Zeners are placed symmetrically with respect to both of the high rail voltages and share a common load resistor (very similar to the current sinks – these also share a common load resistor to bias the bases of the current sink transistors). From each respective high voltage rail side, these zeners reduce the rail voltage by the said 24V and “impose” such a voltage reduction (specifically, by 24,7V, to be precise) upon the collectors of the LOW NOISE transistors …
Will such a “Half Cascoded” Setup UPSET (pun intended!) the workings of the amplifier ?
Please bear in mind that the “total current” of each branch remains unchanged. From the perspective of the differential amplifier, we “still” have two equal currents. The differential amplifier may say: “Who cares that those currents were split up, since they came back together anyway” ?
From the perspective of the blue transistors, the voltage variations on the collector of the green transistors does not make a difference, since the cascode “protects” the collector from their impact. And besides, if the Zener diode is set to half of the value of the rail voltage, the signal swings of this stage not, under any conditions, have a reasonable chance to reach a level of signal voltages comparable to the mighty value of the zener voltage. Do not see any risks here either.
The current is defined by the Constant Current Sink (at the red bottom), but at the same time, it is split exactly into two equal halves at the very (yellow) top. So in essence, the “total current” flowing into the green transistors of the Differential Amplifier is “exactly” the same as it was defined by the yellow mirror. OK, it “splits up” on the way, but then “joins” back together, in the form of the base current of the blue transistors…
Needless to say, the WHOLE of the “LEFT” hand side of the picture is the “complementary” double, or “exactly the same” setup, but organized “upside down”, whereby in place of every PNP transistor there is an NPN transistor and vice versa.
The whole LEFT side is “referenced” to the “OTHER” supply rail (standing “upside-down”). This technique is normally used to provide for “vertical symmetry”. It allows to reach low THD levels of the third harmonics. It also allows to achieve even values for the the rising edge slew rate and the falling edge slew rate, keeping both of these maximum achievable slew rates at at very high speeds (i.e. lots of delta Volts per microsecond).
So … returning to the bizarre differential darlington idea … will it work ?
(C) 2014-05-20, zjj_wwa of hiend-audio.com
P.S. Many kind thanks to Andrew Sparks for is valuable insights and for the “go ahead”.