Prev: It's WAR! EEVblog ripoff
Next: PIC18 Timer 2 weirdness
From: Morris Slutsky on 11 Feb 2010 00:18
On Feb 9, 7:18 pm, "Paul E. Schoen" <p...(a)peschoen.com> wrote:
> "JJTj" <up yers.con> wrote in message
>
> news:5u43n55kd1gd8k7k1v7fhjgmdlvcg9ggt5(a)4ax.com...
>
>
>
>
>
> >>But a choke load is much more efficient than a resistor load.
>
> > Yes, but more hum and magnetic interference.
>
> >>And twice as efficient as an active constant current source.
>
> > I'd bet you don't know how to design CCS(s) that
> > interact with the circuit.  Didn't Mr. Pass teach
> > you anything?  Maybe it was over your head.  Was for
> > me till I built a few.  Once you see what happens
> > with SE designs, and how current is used, you get the
> > picture.  I'm not talking single device CCS devices,
> > which are little better then resistors, I'm talking
> > controlled transistor/Mosfet devices that read the amp
> > and adjust to the need.  A choke just sits there doing
> > the best it can.  Fighting what the PS feeds it.
>
> >>Only the choke load can actually give back the idle current
> >>as necessary.
>
> > See what I said above.  A choke is a one trick pony, often
> > done well, but a interactive CCS reads the circuit and does
> > what is needed to keep the current flowing full tilt.
>
> > Yeah, more parts.  But better operation.
>
> I was able to make an LTSpice simulation that gives about 37% maximum
> efficiency with a 6 VDC power supply, 1.94 VAC input, and 3.13 VAC output
> into an 8 ohm load, at 200 Hz. It appears to work from 20 Hz to 20 kHz. But
> the biasing is very critical, and probably will be unstable with power
> supply, temperature and device variations. I used an STD30NF06L MOSFET
> which has a low turn-on voltage. R3 in the schematic is not needed, but the
> V2 signal source should be low impedance.
>
> I have some 100 mH chokes rated at 10 amps that I found at a Hamfest in a
> pile of freebies. But they are probably 30 lb each.
>
> I agree that this is not really a very practical design, and much better
> alternatives exist.
>
> Paul

Not practical but fun perhaps. Probably I really need a fancier
circuit that separates DC bias from signal feedback. I'll look into
the CH-1 and it's relatives (the CH-2 might actually be better for
this).

Thanks for all your advice.

From: Les Cargill on 11 Feb 2010 19:28
Morris Slutsky wrote:
> On Feb 9, 12:05 pm, JJTj <up yers.con> wrote:
>
<snip>
>
> Maybe it was over my head, maybe it wasn't, you decide I guess. Sure
> I read his website already. And I guess I see his Aleph-type current
> sources as not really single ended amplifiers. In fact, I know
> someone with a Pass power amp and it sounds very very good. I could
> almost see switching from a choke design to an Aleph type thing. But
> that's really, really push-pull from my point of view.

Huh. No, I'd just say he's using the CCS as an "active anti-resistor."
It's just how the JFET is biased. I think JFETs switch current,
not voltage - like tubes do.

> Nothing wrong
> with that, I could have more than 1 transistor per channel - it'd be
> so hilariously cool to do it with 1, though, you can understand that.
>
> Efficiency is something that single ended designs often suck at. Push-
> pull stages - ideal class B is 78.5% efficient, ideal class A push-
> pull is 50% efficient, class AB lies in between. Ideal single ended
> class A with an ideal choke or transformer is also 50% efficient.
> With an ideal CCS load, 25%, and with a resistor load 12.5%.
>
> Pass has a few designs up on his DIY site that lie between these
> situations. He has, for example, resistor-loaded single ended amps
> (with horribly low efficiency). He also suggests use of incandescent
> lamps as loads, which have a strong temperature coefficient and could
> perhaps even be seen as something BETWEEN a resistor and a CCS load.

I saw that, and I am surprised they would not add horrible
nonlinear distortions. But then again, they've used filament
light bulbs as current limiters on PA speakers for years.

> He has CCS loaded single ended class A, the "Zen" amplifier.
>

That thing is awesome looking.

> And the Aleph is also sort of an in between thing, between CCS load
> and push-pull class A.
>
>> I'm not talking single device CCS devices,
>> which are little better then resistors, I'm talking
>> controlled transistor/Mosfet devices that read the amp
>> and adjust to the need.
>
> What the Aleph current source does, as far as I can tell, is sense the
> current into the load and adjust itself accordingly. It's responding
> to the bottom FET that actually is providing the transconductance
> which drives the load. The amount of current sensing/adjustment seems
> to be variable - from pure CCS on one side, to fully compensating load
> current swings on the other - at that point, what you've got is a
> pretty clever way to do push-pull without an explicit phase inverter
> stage.

Okay, then why does it look like it's just a bias element to me? Is
it actually "sensing", or it just literally adding a DC bias? A
CCS is like a voltage source with a Zener somewhere ( wherever
it makes the thing act as a CCS). It's just
essentially a regulator - but a current regulator. Voltage can vary;
current is constant.

> A lot like an SRPP or some push-pull pentode stages that don't
> require a phase inverter - works so long as you stay in class A. And
> so you can push it up to the same efficiency level as a choke loaded
> stage if you want to.
>
> I will consider such designs. I suppose chokes being expensive and
> heavy and all. But there's something kind of elegant and brute force
> about a single MOSFET just sitting there heating away.
>

Oh yeah. Cool as all get out. And I could be wrong, but it looks
like the thing swings *current* more than it swings voltage.

> Assuming I can keep a stable bias on the damned thing and get a decent
> choke in the first place, which might be pretty hard actually.

Heh. I don't understand how the choke helps much at all. Not in a power
path like that. I always think of chokes as filters, as big ferrite bead
arrangements.

--
Les Cargill
First  |  Prev  | 
Pages: 1 2 3 4
Prev: It's WAR! EEVblog ripoff
Next: PIC18 Timer 2 weirdness