March 26th 05, 11:56 PM
>>Nick's Audio - Page 2
Gainclone - Technical
This page is not so technical, but it summarises the engineering points
that I consider to be significant, and why I had no hesitation in
building my 'final' version immediately, rather than going through
several iterations of GC.
Firstly, we should look at what amplifiers the GCs are being compared
to. Up till now, it has been generally acknowledged that the most
superlative audio amplifiers in the World, are tube amplifiers, and
that (in all cases I believe ?) the ones right at the top of the tree
are SET (single-ended triode) amplifiers. So then, why is it so
incredibly expensive to produce these wonderful amplifiers ? Here are a
few reasons:
* all tubes require HIGH voltage, and in the case of powerful SET
amplifiers these voltages can be deadly - as much as a KiloVolt. Thus
ALL components have to be built to accommodate much higher voltages
than one would require for a normal solid state (SS) amp
* all tube amps have to have an output transformer between themselves
and the loudspeaker. Transformers have to not only handle the highest
of the voltages, but have to transfer the audio signal with zero
deterioration in that signal - an impossible task of course, but the
best transformers have (perhaps) 99.9% silver windings and 99.9% pure
iron cores. They have three difficult jobs to do : 1. voltage
conversion, from maybe several hundred volts of ac to a few tens at
the most 2. Impedance matching - difficult to match an anode resistance
of Kilohms to a typical loudspeaker impedance of 8 ohms or so, and
both being somewhat variable under dynamic conditions and 3. Voltage
isolation - one can't allow 1 Kilovolt dc across 8 ohms, can one ?
* capacitive coupling (almost certainly) between the high voltage
anodes of one stage to the low voltage grids of the next, though
interstage transformers (of the highest quality) are sometimes used.
* an exotic power supply, usually a tube rectifier, requiring high
voltage high grade reservoir and smoothing capacitors, and high grade
inductor(s)
* A well designed heater supply (usually dc). Remember that tubes are
just like very dim light bulbs - they have a filament, and heating this
causes the electrons to flow, and it is they that cause the tube to
amplify the signal
* a chassis or maybe more than one chassis to carry all the ironmongery
- these large amplifiers often weigh more than 25Kg (about half a
hundredweight)
* signal connections that have to traverse quite large distances in
these giants, while not allowing the signal to pick up any mains hum or
interference, and not allowing the signal to deteriorate in quality due
to eg the capacitance of the screened cable. The best amplifiers use
silver screened cable internally
The above list is not exhaustive, but is enough to show that it is a
minefield that the poor audio signal has to traverse, and that is why
these exotic amplifiers are so expensive to produce. The only other
thing I will mention however, is "ZERO NEGATIVE FEEDBACK (0NFB)", and
the top quality SET amps all have zero NFB for the ultimate
performance.
Apologies to solid state fans, including those of digital amps, which
are amazing in their own right, but the professional reviewers and
gurus, plus the audiophile forums, answer these things in massive
detail, way way beyond my remit, knowledge or capability.
The Op-Amp Chip
"Op-Amp" stands for "Operational Amplifier" and these devices have been
in use for scientific and industrial purposes for a long time. The
principal function of an OA is precision control and/or measurement,
either with a single input, or maybe 'differential' inputs to react to
the differences between two varying signals, or one signal and a
reference. If one holds one of the inputs at a fixed voltage, eg
zero volts, then the output from the OA will vary according to one
signal only, and this is the way we are using GCs in the audio world.
Why then would an OA chip be superior to a conventional pre-amp or
amplifier ? Well, this is a misleading question, but here are a couple
of websites which point the way, as explained:
http://www.adx.co.nz/techinfo/audio/gainclone1.htm
I LOVE this website, because that is what it is all about, and explains
exactly why these things are what they are - essentially it's all about
THD (total harmonic distortion) and in my technical summary below I
mention the subtlety of why it may have taken so long for such a
discovery to have been made
http://www.national.com/ds/LM/LM3886.pdf
This is the data sheet, direct from National Semiconductor's website.
The awesome thing to note when looking at the equivalent circuit
schematic, is that there are no fewer than SEVEN constant current
sources (yup, those things that look like two vertical 'Olympic'
rings), at each stage of the device, right up till the main power
stage. That is the key to why the THD is so low - see later also.
Let's do an initial comparison between the op-amp design and the SET
tube amplifier, using the SAME bullet points identified above:
* NO high voltages - an op-amp chip is balanced (voltage-wise) about
zero volts, requiring + and - supplies of typically 25 to 40 volts
only.
* NO output transformer required. In fact no coupling capacitor either,
because the output normally sits at ZERO volts, so there is no standing
DC voltage across the loudspeaker terminals. Immediately all those
stringent requirements disappear, and in theory a piece of wire MUST
be BETTER than any transformer in the World. I do say in theory,
because impedance matching could be a problem, but I think not a
significant one in the case of power op-amps
* NO exotic power supply required. In fact most GCs have unregulated
power supplies of the simplest type, and the op-amp chips are very
tolerant of supply voltage variations
* NO heaters to worry about
* NO large chassis required - see my pic of the LM3886 chip between my
Audion Silver Night amps
* NO screened lead required - with careful design the signal path can
be made so short that simple connecting wire can be used. Ideally this
should be silver, but if it is only a few cm or inches then provided
the signal and ground wires are similar material and length, there
should not be a problem.
>From the above 'comparison', on paper the op-amp chip looks very
favourable, but as we all know, it is dangerous to come to conclusions
just by looking at specs. Making amplifiers of the most superlative
quality and performance is the province of the experts who have built
up this art over a long period of time, and some of them have 'golden
ears' which can detect differences in sound through the change of a
single component in an amplifier. I can only comment on my single
build of a 'Gainclone', using normal grade components, but do so in
good faith, and on my 'pleasure' page indicate the results compared to
what I know.
In an op-amp chip the crossing point is zero, and the signal is driven
+ or - in accordance with the applied signal, but unlike SET amps which
must have NO negative feedback for the ultimate performance, this
golden rule of zero negative feedback can and must be broken when using
op-amps, which have HUGE open-loop gains. So why doesn't the negative
feedback totally mess things up, as is the case with SET amps? There
are two main reasons: 1. The transient/frequency response/slew rate,
call it what you will, as they are all closely related, is so high that
at audio frequencies, even allowing for the highest harmonic
frequencies, the negative feedback gets back to the input within a
fraction of a cycle of the applied frequency (ie effectively
instantaneously) and 2. Without any capacitive or inductive components
there are no weird phase shifts to cause positive feedback to occur, or
to affect the applied signal in the wrong way. Thus we can, by using
purely resistive negative feedback, set the gain of the amplifier to
more or less anything we want. The other thing to consider is just how
much control has an op-amp chip really got, related to the SE amp ?
Suppose the gain of our GainClone is set exactly to 10. Now we FREEZE
the audio signal in time, to say, one microsecond. At that instant in
time we can consider the input signal to be a DC voltage, which is the
combination of ALL the music components (bass, mid, treble, vocal,
etc) at that instant. Say the voltage is 1.2156 volts, then at that
instant the output voltage will be INSTANTLY driven to + or -
(depending on whether the amp is inverting or not) 12.156 volts. If
we could then FREEZE the input signal at that level indefinitely, then
the output would remain ROCK SOLID at 12.156 volts also indefinitely (I
am assuming dc coupling, but for audio amps this is academic - most GC
designs (?) have no coupling caps so really do go all the way down to
dc in their frequency response)
Without going into a treatise about SET amps, which are not my
province, suffice to list the main 'obstacles' that they have to deal
with:
*low gain zero nfb means LARGE devices to effectively deliver a small
amount of power effectively
*anode resistance of at least several Kohms requiring an output
transformer. Thus the load is highly inductive, and with the
loudspeaker on the end, the load impedance can vary considerably over
the audio frequency range
*in most SE amps there have to be coupling capacitors, and maybe also
an interstage transformer. Any component in the signal path can at
best provide no deterioration in the signal, but even the best
capacitors and transformers cannot be as good as a simple piece of
wire over the entire frequency range.
None of the above problems are experienced by the GC op-amps
Another points of comparison:
Heat - the large SET amps have to deliver FULL power at all times, even
when NO signal is applied. They are thus a useful supplement to
Central Heating in winter time, dissipating in some cases many hundreds
of Watts in heat energy. By contrast the op-amp gives ZERO output when
no signal is applied, and under dynamic conditions applies most of the
energy to the load (the loudspeaker). My GCs run totally cool, even
after listening to very loud music for several hours.
The above suggests that SET amps will become obsolete, taken over by a
humble op-amp chip, but NOT SO. I have to say that easily the best
sounds I ever heard were from the SET amps, but I qualify this by
saying that my own GCs are the only GCs that I have heard, and from my
Page 1 I identified areas in which my audio set-up needed to be
improved. I hope that after making such improvements the sound will
really be as good as the best SET amps, but I cannot say or make that
claim at this point in time. In fact I wonder if there is some
undefinable 'magic ingredient' in the Single-Ended Tube amps which
means that op-amp designs will not quite get there - it would be nice
to think so.
My belief is, that like the wonderful Steam Engines of the past, SET
amps will endure for ever, and even if the op-amp chips do prove to be
a match in the long run, they will never fully replace SET amps, as the
true audiophile will continue to take great pleasure in getting equally
the best sound in the World, while enjoying the massive
directly-heated Triodes glowing gently in the dark (best way to listen
!!!!)<<
To sum up even though he is technically full of ****, there is a
certain fascination with steam engines and vacuum tubes. Steam engines
never fascinated me that much but the Paxton Phoenix would have been a
hell of a car had it been built. Turbine and Stirling engines would be
more fun in my opinion just as output tubes glow just as nicely in
push-pull IMO.
Gainclone - Technical
This page is not so technical, but it summarises the engineering points
that I consider to be significant, and why I had no hesitation in
building my 'final' version immediately, rather than going through
several iterations of GC.
Firstly, we should look at what amplifiers the GCs are being compared
to. Up till now, it has been generally acknowledged that the most
superlative audio amplifiers in the World, are tube amplifiers, and
that (in all cases I believe ?) the ones right at the top of the tree
are SET (single-ended triode) amplifiers. So then, why is it so
incredibly expensive to produce these wonderful amplifiers ? Here are a
few reasons:
* all tubes require HIGH voltage, and in the case of powerful SET
amplifiers these voltages can be deadly - as much as a KiloVolt. Thus
ALL components have to be built to accommodate much higher voltages
than one would require for a normal solid state (SS) amp
* all tube amps have to have an output transformer between themselves
and the loudspeaker. Transformers have to not only handle the highest
of the voltages, but have to transfer the audio signal with zero
deterioration in that signal - an impossible task of course, but the
best transformers have (perhaps) 99.9% silver windings and 99.9% pure
iron cores. They have three difficult jobs to do : 1. voltage
conversion, from maybe several hundred volts of ac to a few tens at
the most 2. Impedance matching - difficult to match an anode resistance
of Kilohms to a typical loudspeaker impedance of 8 ohms or so, and
both being somewhat variable under dynamic conditions and 3. Voltage
isolation - one can't allow 1 Kilovolt dc across 8 ohms, can one ?
* capacitive coupling (almost certainly) between the high voltage
anodes of one stage to the low voltage grids of the next, though
interstage transformers (of the highest quality) are sometimes used.
* an exotic power supply, usually a tube rectifier, requiring high
voltage high grade reservoir and smoothing capacitors, and high grade
inductor(s)
* A well designed heater supply (usually dc). Remember that tubes are
just like very dim light bulbs - they have a filament, and heating this
causes the electrons to flow, and it is they that cause the tube to
amplify the signal
* a chassis or maybe more than one chassis to carry all the ironmongery
- these large amplifiers often weigh more than 25Kg (about half a
hundredweight)
* signal connections that have to traverse quite large distances in
these giants, while not allowing the signal to pick up any mains hum or
interference, and not allowing the signal to deteriorate in quality due
to eg the capacitance of the screened cable. The best amplifiers use
silver screened cable internally
The above list is not exhaustive, but is enough to show that it is a
minefield that the poor audio signal has to traverse, and that is why
these exotic amplifiers are so expensive to produce. The only other
thing I will mention however, is "ZERO NEGATIVE FEEDBACK (0NFB)", and
the top quality SET amps all have zero NFB for the ultimate
performance.
Apologies to solid state fans, including those of digital amps, which
are amazing in their own right, but the professional reviewers and
gurus, plus the audiophile forums, answer these things in massive
detail, way way beyond my remit, knowledge or capability.
The Op-Amp Chip
"Op-Amp" stands for "Operational Amplifier" and these devices have been
in use for scientific and industrial purposes for a long time. The
principal function of an OA is precision control and/or measurement,
either with a single input, or maybe 'differential' inputs to react to
the differences between two varying signals, or one signal and a
reference. If one holds one of the inputs at a fixed voltage, eg
zero volts, then the output from the OA will vary according to one
signal only, and this is the way we are using GCs in the audio world.
Why then would an OA chip be superior to a conventional pre-amp or
amplifier ? Well, this is a misleading question, but here are a couple
of websites which point the way, as explained:
http://www.adx.co.nz/techinfo/audio/gainclone1.htm
I LOVE this website, because that is what it is all about, and explains
exactly why these things are what they are - essentially it's all about
THD (total harmonic distortion) and in my technical summary below I
mention the subtlety of why it may have taken so long for such a
discovery to have been made
http://www.national.com/ds/LM/LM3886.pdf
This is the data sheet, direct from National Semiconductor's website.
The awesome thing to note when looking at the equivalent circuit
schematic, is that there are no fewer than SEVEN constant current
sources (yup, those things that look like two vertical 'Olympic'
rings), at each stage of the device, right up till the main power
stage. That is the key to why the THD is so low - see later also.
Let's do an initial comparison between the op-amp design and the SET
tube amplifier, using the SAME bullet points identified above:
* NO high voltages - an op-amp chip is balanced (voltage-wise) about
zero volts, requiring + and - supplies of typically 25 to 40 volts
only.
* NO output transformer required. In fact no coupling capacitor either,
because the output normally sits at ZERO volts, so there is no standing
DC voltage across the loudspeaker terminals. Immediately all those
stringent requirements disappear, and in theory a piece of wire MUST
be BETTER than any transformer in the World. I do say in theory,
because impedance matching could be a problem, but I think not a
significant one in the case of power op-amps
* NO exotic power supply required. In fact most GCs have unregulated
power supplies of the simplest type, and the op-amp chips are very
tolerant of supply voltage variations
* NO heaters to worry about
* NO large chassis required - see my pic of the LM3886 chip between my
Audion Silver Night amps
* NO screened lead required - with careful design the signal path can
be made so short that simple connecting wire can be used. Ideally this
should be silver, but if it is only a few cm or inches then provided
the signal and ground wires are similar material and length, there
should not be a problem.
>From the above 'comparison', on paper the op-amp chip looks very
favourable, but as we all know, it is dangerous to come to conclusions
just by looking at specs. Making amplifiers of the most superlative
quality and performance is the province of the experts who have built
up this art over a long period of time, and some of them have 'golden
ears' which can detect differences in sound through the change of a
single component in an amplifier. I can only comment on my single
build of a 'Gainclone', using normal grade components, but do so in
good faith, and on my 'pleasure' page indicate the results compared to
what I know.
In an op-amp chip the crossing point is zero, and the signal is driven
+ or - in accordance with the applied signal, but unlike SET amps which
must have NO negative feedback for the ultimate performance, this
golden rule of zero negative feedback can and must be broken when using
op-amps, which have HUGE open-loop gains. So why doesn't the negative
feedback totally mess things up, as is the case with SET amps? There
are two main reasons: 1. The transient/frequency response/slew rate,
call it what you will, as they are all closely related, is so high that
at audio frequencies, even allowing for the highest harmonic
frequencies, the negative feedback gets back to the input within a
fraction of a cycle of the applied frequency (ie effectively
instantaneously) and 2. Without any capacitive or inductive components
there are no weird phase shifts to cause positive feedback to occur, or
to affect the applied signal in the wrong way. Thus we can, by using
purely resistive negative feedback, set the gain of the amplifier to
more or less anything we want. The other thing to consider is just how
much control has an op-amp chip really got, related to the SE amp ?
Suppose the gain of our GainClone is set exactly to 10. Now we FREEZE
the audio signal in time, to say, one microsecond. At that instant in
time we can consider the input signal to be a DC voltage, which is the
combination of ALL the music components (bass, mid, treble, vocal,
etc) at that instant. Say the voltage is 1.2156 volts, then at that
instant the output voltage will be INSTANTLY driven to + or -
(depending on whether the amp is inverting or not) 12.156 volts. If
we could then FREEZE the input signal at that level indefinitely, then
the output would remain ROCK SOLID at 12.156 volts also indefinitely (I
am assuming dc coupling, but for audio amps this is academic - most GC
designs (?) have no coupling caps so really do go all the way down to
dc in their frequency response)
Without going into a treatise about SET amps, which are not my
province, suffice to list the main 'obstacles' that they have to deal
with:
*low gain zero nfb means LARGE devices to effectively deliver a small
amount of power effectively
*anode resistance of at least several Kohms requiring an output
transformer. Thus the load is highly inductive, and with the
loudspeaker on the end, the load impedance can vary considerably over
the audio frequency range
*in most SE amps there have to be coupling capacitors, and maybe also
an interstage transformer. Any component in the signal path can at
best provide no deterioration in the signal, but even the best
capacitors and transformers cannot be as good as a simple piece of
wire over the entire frequency range.
None of the above problems are experienced by the GC op-amps
Another points of comparison:
Heat - the large SET amps have to deliver FULL power at all times, even
when NO signal is applied. They are thus a useful supplement to
Central Heating in winter time, dissipating in some cases many hundreds
of Watts in heat energy. By contrast the op-amp gives ZERO output when
no signal is applied, and under dynamic conditions applies most of the
energy to the load (the loudspeaker). My GCs run totally cool, even
after listening to very loud music for several hours.
The above suggests that SET amps will become obsolete, taken over by a
humble op-amp chip, but NOT SO. I have to say that easily the best
sounds I ever heard were from the SET amps, but I qualify this by
saying that my own GCs are the only GCs that I have heard, and from my
Page 1 I identified areas in which my audio set-up needed to be
improved. I hope that after making such improvements the sound will
really be as good as the best SET amps, but I cannot say or make that
claim at this point in time. In fact I wonder if there is some
undefinable 'magic ingredient' in the Single-Ended Tube amps which
means that op-amp designs will not quite get there - it would be nice
to think so.
My belief is, that like the wonderful Steam Engines of the past, SET
amps will endure for ever, and even if the op-amp chips do prove to be
a match in the long run, they will never fully replace SET amps, as the
true audiophile will continue to take great pleasure in getting equally
the best sound in the World, while enjoying the massive
directly-heated Triodes glowing gently in the dark (best way to listen
!!!!)<<
To sum up even though he is technically full of ****, there is a
certain fascination with steam engines and vacuum tubes. Steam engines
never fascinated me that much but the Paxton Phoenix would have been a
hell of a car had it been built. Turbine and Stirling engines would be
more fun in my opinion just as output tubes glow just as nicely in
push-pull IMO.