Aleph current source: huh?

[pixpop]

I'm trying to understand how this cct works, and mostly I do. But there's one part that puzzles me.

Referring to figure 2 of the Aleph 30 schematic, from the Pass website, there's a biasing network consisting of R17, R19, and R20 whose job is to reduce the amount of voltage that needs to be dropped across the current sensing resistors (R34 - R36).

Since this network is referred to the positive rail, it seems to me that it injects power supply noise directly into the base of Q5. The noise will be multiplied by the gain of Q5 and will then modulate the gates of the current source mosfets.

I'm wondering why it would not be better to refer the bias network to a constant voltage, instead of the power rail.

Is there something I'm not understanding, or is there some reason why this doesn't matter?

[acaudio]

Hi,
check the another way:

http://www.diyaudio.com/forums/attac...amp=1110824921

with them you can only increase bias, pure regulation.

Regards
Adam

[metalman]

Think of the R17,19,20 arrangement as a voltage divider network around Q5, with power supply noise injected at the top of R17, and the Aleph current source (ACS) control signal injected at the right side end of R20. With a good power supply, the noise on the voltage rails will be in the few mV range or less, whereas the ACS control signal is in the volts range. For demonstration lets say PS noise = 10mV, and ACS control signal is 1V. The effect of the voltage divider reduces the power supply noise to ~0.2mV at the base of Q5, and the ACS control signal to 0.98V, putting the noise -75dB relative to the signal.

Now keep in mind that this description is an oversimplification of what really happening, and when you include the stabilizing effects of C9, R18, & R21, the difference increases further, probably into the range where noise is -100dB vs ACS signal.

So to answer your question in your own words, because of the circuit design, the fact that the system is referenced to the positive rail doesn't really matter.

Cheers, Terry

[Nelson Pass]

Actually the circuit is referenced to the output, via the bootstrap
capacitor C9. Whatever little noise is injected by the supply
through the top resistor R17 is divided out by the output
impedance by about 70 dB.

When building this circuit into a current source amp with a
high output impedance, this small noise current can become
a measurable part of the output noise, so I decouple the supply
voltage to R17 with an RC network.

[pixpop]

Ahh, that makes sense. It's quite clever, really.

Actually, when I look at it now, I see C19 as behaving like the constant voltage I was thinking about.. because it tends to keep the voltage at the bottom of R17 constant WRT the output.

An actual constant voltage device such as a zener referred to the output would have a similar effect, but might behave worse by introducing a noise source at the base of Q5.

[lumanauw]

When I dont understand about Aleph schematic, I always wonder what R124,125,126,127 doing in output (Aleph3 cct). It's unusual to see a Resistor put in series with output, it can higher output impedance, and why put it there?

But that R position may have give birth to another smart idea. Look at patent #5,019,789. I change from "Why put it there?" to "That's smart"