Power_Buffers.pdf
(
391 KB
)
Pobierz
<< Prev
TUBE CAD JOURNAL
Next >>
An E-Mail
tube fancier after all.
I have hooked up headphones to the line amplifier
and the sound is just about perfect, so how do I
preserve as much of that sound as I can? One idea I
have is to build a solid-state amplifier that adds no
gain or feedback, which means that the line amplifier
will do all of the voltage amplifying and the power
amplifier will do all of the current providing.
The attached schematic shows two circuits: the
first is a pure solid-state unity-gain buffer and the
second is a hybrid buffer. What do you think of
them? Which would be the better path to follow?
Thanks in advance for your kind reply and the
possible publishing my letter ;)
Tony F.
United States
Subject: Why Just Tubes?
This is my third e-mail to the Tube CAD Journal
and I thank you for your kind replies to the first two e-
mails, which were quite lengthy and informative.
However, those e-mails were never published in your
journal. Was it that they were not deemed interesting
enough to include or were they excluded because
the circuits contained were pure solid-state? If the
latter was the reason, then I must make a plea on
behalf of the solid-state half of the world: your journal
is one of the few places that audio electrical
engineering is explained carefully and fully. Bless
you and your writers. But limiting the journal to just
tube audio is probably a big mistake. Here is why:
my guess is that more than half of the journal's
readers are from the solid-state camp anyway.
I know from my group of audiophile friends--
mostly fellow college students--that none of the tube
guys are technically inclined in the smallest degree.
They love tube sound and spend huge sums of
money on single tubes, but they cannot tell you what
a cathode is. (This deep division reminds me of the
one between Linux and Mac followers, with Windoze
users in somewhere in the middle; never ask a Mac
user which CPU their system uses, for example.)
This is not, however, a plea that you publish only
solid-state articles, but that you just include a few
pure solid-state articles for the other half of your
readers. (I would love to read your explanation of
current-feedback amplifiers, for example.)
Now for the technical part of my letter, I like the warm
and smooth sound from tube equipment, but I do not
like the fat and shallow bass reproduction from tube
gear. (Actually, I am sure the poor bass response
comes from the coupling capacitors and output
transformers and not from the tubes themselves.)
Therefore, some sort of hybrid system would seem to
be best: a tube preamp driving a solid-state amplifier for
instances. This is what am currently using, with a tube
line amplifier (your circuit) driving my 100 watt solid-
state amplifier. This sounds better than either the pure
tube or pure solid-state equivalent systems I have tried.
But the sound seems to be too close to the pure solid-
state side of the spectrum. Hell, maybe I am a closet
Tony, your first two emails were not
published precisely because they were tube free,
not because they failed to provoke interest.
The original intent of this journal was to
mimic the
Math CAD Journal,
a magazine put
out by the MathSoft people to asset users of their
Math CAD
program by providing math-related
articles that illustrated how their software could
be used. For example, users of
Math CAD
who
explain how they found the program useful write
many of the articles.
Well that was the intent, but the
Tube CAD
Journal
went off in another direction altogether.
Part of the reason was that many of the tube
fanciers needed, as you pointed out, to be
technically brought up to speed. Another part
was my hope of expanding the tube-audio
horizon, which I felt was collapsing into a few
simple-minded topologies and practices.
In fact, you are not the first to recommend
widening the journal's range of topics to include
all audio related circuit devices and topologies.
But I cannot see too great a need to cover pure
solid-state topics when
Elektor
,
Electronics
World
,
audioXpress,
and the solid-state
manufacturers themselves (in their data books)
do a good job on these topics. But then, maybe I
am wrong on this. What do you, the readers,
think? As for the circuits you submitted, you
have an entire article as a reply...
www.tubecad.com Copyright © 2002 GlassWare All Rights Reserved
< PREVIOUS
Pg.
1
NEXT >
<< Prev
TUBE CAD JOURNAL
Next >>
Local Feedback Buffers
The local feedback buffer has no global
feedback loop, relying instead on degenerative
feedback at each active device’s output to keep
the output in line with input voltage. Yes, this is
the same mechanism used in a cathode follower,
which keeps both its output distortion low and
its output impedance low. (Solid-state devices
offer the added advantage of allowing a
symmetrical topology not possible with tubes, as
there is no P-channel version of a triode.)
Whereas the global feedback buffer is usually
run in lean Class-AB mode, in order to reduce
power consumption, the local feedback buffer is
usually run in beefy Class-A mode, as the higher
currents both enhance the linearity and extend
the frequency bandwidth.
IC examples of both types are readily found in
the National Semiconductors catalog. The
LH4004, LH4006, LM102 and LM310 are of the
global feedback type and the LH0002, LH0033,
LH0066, and LH4001 are of the local feedback
type. (Linear Technology also makes some
excellent buffers.) Examining the schematics for
these buffers is a good source for gaining insight
into the designing of a buffer, as the schematics
are often reduced to idealized versions that are
much clearer than their actual implementation.
High-Power Buffers
Although buffers might be new to many
audiophiles, they are a central part of the analog
electrical engineering practice. In short, a high-
power buffer is a special type of power
amplifier: it relays its input voltage to its output
un-amplified and it delivers the needed current
increase into the load. In other words, like a
conventional amplifier, the buffer can deliver
power into a load, but the buffer does not add
any voltage gain; instead, it only provides the
required current gain. For example, a good tube-
based line stage can usually put out at least 30 to
50 volts of peak output swing, which into an 8-
ohm load would equal 50 to 150 watts of power,
which if the line stage could deliver 4 to 6 amps
of current, it would deliver that much power.
Buffers come in two basic styles: the global
feedback buffer and the local feedback buffer.
Global Feedback Buffers
The global feedback version is simply a
conventional power amplifier with all its gain
being returned to its input stage. A typical power
amplifier has a voltage divider in the form of a
feedback loop, bridging its output to its input,
which defines the gain of the final gain of the
amplifier. The greater the voltage division, the
greater the final gain.
What is happening here is that the feedback
mechanism within the amplifier strives to keep
the voltage divider's center voltage in line with
the amplifier's input voltage. The greater the
voltage division, the greater must be the output
voltage to allow the same matching of
amplifier's input and feedback voltages. If, on
the other hand, we eliminate the voltage divider
and return all of the output voltage to the
negative input, the amplifier's feedback
mechanism will force the output voltage to fall
until it matches the input voltage. In other
words, we have turned an amplifier into a buffer.
(Not all power amplifiers, however, can be
converted into buffers, as not all amplifiers are
unity gain stable.)
MJ Stereo Technic
The wonderful Japanese audio magazine,
MJ
Stereo Technic
also known as
Audio Technology
MJ
, has for the last two decades run articles that
featured high-power buffer circuits. Usually, the
circuits look like they are just the last half of a
conventional amplifier, but a few have been
more interesting. The aim of these buffers is to
provide no gain, but sufficient current gain to
drive loudspeakers.
Surely, the logic is compelling: tubes cleanly
provide voltage gain, but are current limited;
solid-state devices provide huge current gains,
but are not as linear at voltage amplification. (A
gross oversimplification, but essentially correct.)
So why not use each only for its best use? In
other words, why not have a hybrid system?
www.tubecad.com Copyright © 2002 GlassWare All Rights Reserved
< PREVIOUS
Pg.
2
NEXT >
<< Prev
TUBE CAD JOURNAL
Next >>
Yes, this is somewhat like breaking a hybrid
amplifier into two components, line stage and
buffer amplifier, which makes sense, as making
a typical hybrid amplifier is effectively like
placing two line stages in series with each other
needlessly. On the other hand, if a passive
volume control box is used in the line amplifier's
stead, then the unity-gain buffer will not provide
full output, only producing a few milliwatts of
output power, as the line level voltage are
seldom much over one volt.
Well, so much for the introduction to buffers,
let us now look into actual topologies.
In actual practice, the actual efficiency will
probably come in at 40-45% with solid-state
devices and low DCR inductors.
+200V
10µF
200
input
6BX7
output
+15.5V
470µF
1M
777
1k
0V
Local-Feedback Buffers
This type of buffers allows the greatest
simplicity, as a local-feedback can be made from
as little as one active device. The circuit below
shows a single N-channel MOSFET loaded by
an inductive load.
No-gain triode-based headphone buffer
Using a resistive load would quarter the
theoretical 50% efficiency figure (12.5%) and
thus cannot be universally recommended. The
exception might in the case of a pure tube low-
power buffer for driving low-ohm headphones
(8-32 ohms), as here we only need milliwatts of
output. The circuit above shows a 6BX7 triode
biased a single cathode resistor. The better
approach for driving loudspeakers is to use an
active constant current source, as the maximum
theoretical efficiency falls to only 25%.
+30V
1M
2M
10µF
.22µF
200
input
+4V
output
+1V
10kµF
1M
400k
1k
0V
+30V
This buffer can
only
be biased in pure Class-A
mode, as the inductor must be able to give up
same amount of peak current into the load as the
MOSFET pulls through the load during peaks.
For example, if the idle current equals 2A, then
the peak symmetrical output into a load
impedance is 2A. Given an 8 ohm load, 2A
equals 16 watts of RMS power. (The key word
was "symmetrical," as the MOSFET could
swing much more positive going current into the
load, assuming that power supply can support
the needed voltage swing.) Like all inductively
loaded Class-A amplifiers, the maximum
theoretical efficiency is 50%.
4k
input
200
100k
output
10v
+4V
1
5v
-30V
www.tubecad.com Copyright © 2002 GlassWare All Rights Reserved
< PREVIOUS
Pg.
3
NEXT >
<< Prev
TUBE CAD JOURNAL
Next >>
The constant current source also allows for
DC coupling at the output (with a bipolar power
supply). The simplest circuit topology would
require only two MOSFETs. (Transistors would
not work as well because of the transistor's low
input impedance, which would drag down the
line stage output stage.) Added features would
include a DC servo loop.
One feature, which is not optional, is a
carefully designed fusing arrangement. The
danger of placing the all of the biasing resistors
on the amplifier side of the fuses lies with
blowing a single power supply rail fuse, which
would protect the output devices, but would not
protect the loudspeakers, as a steady 2A current
draw (from the top MOSFET or the constant
current source) into loudspeaker would vaporize
all but the most buffed voicecoils. Sown below
is a push-pull power buffer with complimentary
output MOSFETs. Should any power supply
fuse, the output will remain close to ground and
not slam to the opposing rail.
The downside to driving the output
MOSFETs directly is the MOSFET’s high input
capacitance.
SE Amp CAD
SE Amp CAD
Successful design and analysis of a
single-ended amplifier output stage
requires an accurate model of the tube's
plate curves. SE Amp CAD is a tube
audio design program that has a library of
30 tubes and over 100 output
transformers and SE Amp CAD knows
how these tubes really curve in a singled-
ended amplifier.
DC Offset
+30V
10µF
200
input
.22
Windows 9x / Me / NT / 2000
10
.22
10
100k
For more information, please visit our
Web site or write us at:
GlassWare
PO Box 231
Fenton, MI 48430 USA
www.glass-ware.com
200
10µF
-30V
Push-pull buffer with an intelligent fusing topology
www.tubecad.com Copyright © 2002 GlassWare All Rights Reserved
< PREVIOUS
Pg.
4
NEXT >
<< Prev
TUBE CAD JOURNAL
Next >>
Added Complexity
The Class-A mode of operation and the
follower configuration both make the single-
ended buffer amplifier a perfect candidate for
adding a DC servo-loop. At all times, an IC
strives to keep its positive and its negative inputs
at the same voltage, which moves its output to
swing positively or negatively in response.
Consequently, when the buffer's output drifts
too positive the IC's negative input pin will see a
greater positive voltage than the ground
referenced positive input, which will provoke
the IC's output to go negative until the its input
once again match in DC voltage. DC is a key
adjective here, as the .22µF capacitor effectively
absorbs the AC part of the signal presented to
the IC's negative input; if a scope's probe is
attached to this juncture, all that will be seen are
the frequencies below 1 Hz.
Similarly, the Op-Amp does not need to swing
anymore than a few volts positively at its output
to keep the buffer’s DC offset in line with
ground. For example, a DC offset would likely
never exceed 100 mV and the top MOSFET
would at most only require 5 volts DC to set idle
current to that of the bottom MOSFET.
+30V
+30V
1M
10µF
200
.22µF
input
.22
.22
100k
10k
+30V
200
10µF
15v
-30V
+15V
-30V
input
.22µF
200
Push-pull buffer with a DC servo loop
100k
1M
The next addition might be the creation of an
active current source made up of the bottom
MOSFET and an additional IC. In the circuit
below, we see the current flowing through the 1-
ohm resistor being monitored by the IC's
negative input, which it will compare it to its
positive input and thus work to keep the voltage
across the resistor constant. A constant voltage
means a constant current. The current is easily
set by dividing the reference voltage by 1 ohm,
thus in this example, the constant current source
is set to 2.5A, which equals 25 RMS watts and
20 volts into an 8 ohm load, whereas increasing
the current to 3A will increase the wattage to 36
watts and the peak output voltage to 24 volts.
This constant current source is a truly active
one, with the IC
fully in the circuit
, as its
feedback loop encompasses the bottom
MOSFET.
10v
0V
output
.22µF
1M
+4V
LF412
-15V
1
-30V
15v
Single-ended buffer with a DC servo loop
The two 15 volt zeners are there to displace
half of the 60 volts made available by the bipolar
power supply, as the IC is voltage limited (a
high-voltage IC would not require these zeners,
but such ICs are expensive.) Basically, the Op-
Amp does not need to see at its input the full
output voltage swing of the buffer’s output, only
the few millivolts of DC offset at the output.
www.tubecad.com Copyright © 2002 GlassWare All Rights Reserved
< PREVIOUS
Pg.
5
NEXT >
Plik z chomika:
fred1144
Inne pliki z tego folderu:
Power_Buffers.pdf
(391 KB)
gainclone_2007.zip
(408 KB)
Zen-Lightenment.pdf
(582 KB)
100W Guitar Amplifier (Mk II).mht
(185 KB)
Zen-V8.pdf
(532 KB)
Inne foldery tego chomika:
1
1.files
11.files
2008.04.28
2008.05.04
Zgłoś jeśli
naruszono regulamin