alephx dc offset

[lumanauw]

I'm sorry, but after reading about alephx thread(s), I still cannot understand why alephx design needs the 30ohm resistor at each output shunted to ground to maintain low DC offset. I know that aleph is very stable, why if we built 2 alephs back to back, it becomes unstable in dc offset? Could anyone explain the reason behind this in alephx, is it because we're using IRF9610 for differential input? (what if using bipolar)?

[Edwin Dorre]

Hello,

It is to stabilize the absolute DC offset. With an Aleph-X you have two types of DC offset, the one the loudspeakers 'sees'; DC offset, and the absolute DC offset, which is the one the loudspeaker does not see.

You measure the normal DC offset between the OUT+ and the OUT- and you measure the absolute DC offset between OUT+ and GND and OUT- and GND. Any absolute DC offset between 1V is OK! When you have a to high absolute DC offset the amp will not give its full power potential because you limit the swing of your amp. It will not sound better or worse when you have a 0V absolute DC offset! Ofcourse your normal DC offset has to be low; anything lower than 100mV is considered ok, anything lower than 30mV is perfect!

The 30ohm resistors are to stabilize this absolute DC offset!

Edwin

[grataku]

These resistors need not be 30 ohms I used as high as 200 ohms without seeing a change in behavior. It can be argued that if the mosfet are matched well enough and the chassis temperature is kept under control, one could probably get rid of them altogheter. In my experience with Grey's circuit, the absolute DC offset tracks the chassis temperature anyways, so the whole amp doubles as a thermometer even with the resistors in place. It is reproducible but everything needs to be iteratively tuned in place and after the amp is fully warmed up.


Nelson must have something different in his circuit that improves on the abs DC somehow. Counting on the fact that the temperature is the same in all the customer's listening rooms would be too low tech a solution.

[Nelson Pass]

It's just a case of belt-and-suspenders. The output resistance
augments the DC stability, but the amps do not rely on that
alone.

[GRollins]

As for the 'why' of it, it's a function of the fact that the feedback crosses over from one side to the other. In a standard Aleph, the feedback enters the "back" side of the differential and causes a compensating shift in the output of the "front" side. In the Aleph-X, the feedback is going in on the "front" side to begin with, so the compensation appears on the wrong side.
As an analogy, think of the new generation of fighter jets that are ultra-maneuverable. Generally speaking, the reason they're so eager to stand on one wingtip is that they're aerodynamically unstable. So how do they fly? They use computers to correct the handling on a moment-by-moment basis. The result is that the plane does what the pilot tells it to do instead of spinning off into the fifth dimension.
When all else fails, sit down with the schematic and imagine that one side of the output drifts a little bit positive. Then trace the DC offset through the feedback to the front end and over to the other side...then back to the original side. I once used the analogy of a Moebius strip to describe the topology of the Aleph-X. I still think it's a workable analogy. You have to look at it a while, but eventually you'll get it.
No, it's not related to the kind of gain device in the front end. Bipolars, JFETS, MOSFETs, or tubes would all do the same thing.

Grey

[lumanauw]

So, it is because of a "positive feedback" that the differential have on alephx dc offset? If we doesn't use at all the 30ohm resistor, would it be uncontrolable?

Is there any "brilliant" idea how to get rid of this dc offset problem, while maintaining the quality of reproduction of alephx? (not using servo offcourse). Super matching the 9610?

The DC offset is the "negative side effect" that alephx have. The main attraction of alephx is the quality of the reproduction. I have long suspiction that the differential plays the most important role in audio power amp. Alephx can be soo good (looking at the comments from those who has built it), because it make use both of the differential transistors. Q1 motors the right side aleph and Q2 motors the left side aleph. Both Qs are in use.

I post a schematic on http://www.diyaudio.com/forums/showt...&postid=283929 to make use both of the differential transistors. Will this results in better performance than ordianary 3 stage power amp?

[grataku]

The problem really is not much of a problem. Just something to be aware of.
I tuned my proto on the bench channel by channel, everything was mounted on the heatsink so the unit was essentially complete as it is in the assembled chassis. Under those condition everything was tuned for 0 DC offset. After putting everything together in the chassis the absolute DC became something like 3 V and the only reason that would be is the different chassis temperature.
Adjustment takes a while because any adjustment to the front end bias causes a variation whose effect won't be immediately seen until everything gets to a new steady state point, and the direction and magnitude of the change wasn't always predictable.
To make the story short, I test the DC on three days span after consecutive turn on and the thing always settled back to the point I had adjusted it so that was good enough for me. I plugged in my $3.5k speakers and never measured anything since. I am so worried about the health of my speakers that I removed the gain stage protection circuit, too.

[jarek]

Nelson must have something different in his circuit that improves on the abs DC somehow. Counting on the fact that the temperature is the same in all the customer's listening rooms would be too low tech a solution.

I think that the key to stabilise absolute DC offset is holding differential pairs at constant temperature. This is easier to achieve in closed chassis than on the bench. In this case temperature of the differential pair is almost independent on ambient temperature.
Regards

[lumanauw]

What if we insert a current mirror between 9610 and 392 resistors?

[Nelson Pass]

Originally posted by lumanauw
So, it is because of a "positive feedback" that the differential have on alephx dc offset

No. The circuit has no more intrinsic offset than a normal
Aleph, but the X circuit does not in itself provide common mode
feedback, only differential. In an ordinary Aleph you have the
matching of the devices, the intrinsic loading of the output to
ground through the load, and the standard feedback, which
reduces the absolute DC offset.

The concept Aleph X (or XA in Pass lit) does not have the common
mode feedback to ensure low absolute DC offset, although what
the speaker sees for DC may be quite low. For aesthetic and
practical reasons, we provide matching of devices, loading to
ground at both outputs, and a small amount of common mode
feedback (known elsewhere in this forum as "magic resistors")
to ensure low absolute DC offset.

Absolute DC offset is only an issue in turn-on/turn-off thumps
and in extreme cases, the maximum amount of power that can
be delivered to the load. Differential DC is what the speaker
sees, and is the truly important parameter.