Class A Amplifier 2004.
Review
My last amplifiers:
I have already
create in the past several amplifier, by remaining rather close
to circuit seen this or there. It was two stage circuit in
general (2 differentials, 2 mirroirs, output stage) such as we find it
on Internet or books. They are complex, compromised
stability/speed is not very simple to acquire, high open loop gain, and
one powerful feedback reaction. But finally sound is good, and I
thought it was quiet good for a lot of year until this new try.
This type of circuit answers to a logic of full control: good control of gain, good control of output impedance.
They are in general very good. But...
The doubt:
Why ampli with tubes is so good?
Why we sometimes speak to prefer weak feedback rate instead strong feedback rate?
The only way to understand, it is to tread underfoot in the dish...
New idea:
Basic
idea is to test what the extreme simplicity of a circuit can give.
As I liked to make things good, I put in it means, I mean
a budget of about 3000FF (500$).
Then here is how it is conceived:
- Simple circuit, inspired by assemblages in tubes, but with MOS transistors.
- No feedback.
- Supply perfectly filtered.
Choose schema:
-
The amplification of the signal is made via a single transistor MOSFET
of Internal Rectifier (or several in parallel to divide potency): irfp140.pdf
- The load will be simply a resistance of 10Ω for this first version.
- No feedback, who gives us an output impedance of 10 ohm.
-
And the signal will go through an unavoidable serial capacitor. But to
put the maximum of chance on my side, I chooses mark Black Gate.
And here is the result:
Photo
Explanation of schematics:
Amplifier:
Schema is following:
Schema
of amplifier. I simplified a bit , by representing only a transistor and a resistance.
In reality there is two transistor in parallel for N1 (the
mosfet which amplifies the signal), and 3 resistance of 33 Ω in
parallels for R1 (resistance of load).
An important thing also, it is that resistance R1 and R2 must be not inductive.
Components are besides according to:
Photo
Photo
Photo: le transistor mosfet
on left
(*2), the load resistance on right (*3).
But for simplified, let us do as if there was an only one of each.
To
work well, the gate must have a continuous voltage of about 4 in 5V,
while the apparatus on the which it is connected up (a CD reader for
instance) does not have contniuous voltage. I use therefore a polypro
capacitor (C2) of 2.2μF. Then a means is needed that this 4~5V
voltage auto-regulate itself. I use therefore a voltage divisor
with bridge resistor (accomplished with R4 R5) to have good
continous voltage. But so that there is no interference
between bridge and audio signal , I abolish the AC
signal which come from the bridge R4 R5 thanks to a very stupid method
which is a resistance (R3) and a capacitor (C1).
Therefore the
signal arrives on the gate of the mosfet N1, then the signal goes out
amplified between the drain (N1) and supply to give the output signal
on resistance R1.
But there is a strong continous voltage, about
24V for the 39V supply. It is therefore necessary to eliminate it, what
I makes with a big capacitor (C2). But instead of ruining mewith
the polypro, I preferred capacitor Blackgate bought in Rubycon
To
linearize a bit the signal and reduce the distorsion, I add a
resistance of 0.5Ω in the source of the transistor MOSFET.
The voltage in the drain is about 40 % of tension of supply.
Output Impedance:
The output impedance is therefore 11 Ω.
This can pose problems with speaker where impedance varies too much.
At the beginning I use only one amplifier for bass and treble, as everybody made.
I
could resolve the problem of variation of impedance by knocking up a
filter with resistance and of capacitor which I connect between this
amplifier and the source, a CD reader for instance. It is sufficient
just like that, because I conceived my loudspeaker so that it is there
not too much variation in impedance.
Later I use two amplifier by loudspeaker, and I reserve this amplifier only for the mediums and treble.
Then
it is not necessary to me anymore to envisage a filter in front of
this amplifier, because the impedance of the loudspeaker is not much
bumped.
Photo
Photograph: view on one of the amplifier, with transitors and resistance fixed on the radiator.
Supplies:
It is a simple supply 39Vdc.
There is one by channel, the building of the amplifier is therefore double-mono.
The schema is the following:
Fig:
Schema of supply.
It is a wholly basic supply, where I add an inductance between two capas of filtration.
The
output of the transformer provides a 50Hz alternative signal. I use a
classic bridge D1... D4 to rectify the voltage, then the capacitor
C1 is use as tank during the idle times of rectifier (due of in the
sinusoidale form of inut voltage). I prefer using a capacitor with
strong value here, this allows to restrict the snore produced by the
rectifier.
But often the voltage on the capacitor C1 is still
loud, there is the traditional 100Hz (double frequency due to
rectifier), with high frequency noise which lugs around the sector
and that is not always fully eliminated in the first stage.
Then
I use an inductance (L1) of good value, rather 1mH, followed by a
second wave of capacitor (C2 C3) like 20000uF to have a second big tank
of energy. It allows to abolish definitely all noise which could come
from sector supply. This circumspection is seldom taken by constructor,
it is definitely damage.
Another circumspection consists in putting capacitor C2 and C3 (capacity FRS) directly against the electronics of the ampli.
It allows to decrease the parasitic resistance of supply.
The
radiance of the inductance has a low impact if it is put far
enough from electronics (20cm). It gives a powerful and clean
supply. It is just like that that I like them!
Photo
Photo: view of an end of the transformer, diodes rectifier, with C1 and L1.
Dissipation:
We attack the delicate party of building.
I put two radiators DXC622-70 (Sélectronic) of each side of the box.
There are 2 transistors IRFP140, as well as 3 resistance MPR 220 in TO220 package.
Resistance are isolated from the metallic tongue, this allowed me to
screwed them directly on the radiator without other ceremony. I
don't have problem, this held the temperature without problem.
For transistors, this was a bit complicated. Here is review:
- I first used IRF540 (box not insulated TO220). To fit them, I first
used the kit of classical insulation which we found everywhere.
But the temperature which I choose gives problems. I aimed 80°C for
radiator temperature with 20°C in the room. In practice, I have
definitely had 80°C, but different thermal contacts gives in
fact 110°C / 120°C on on tongues metallic of transistor.
The problem at this temperature is the cannons of insulation: they not
hold it. They got soft, even if I use the better quality series in
black polyamide loaded with glass fiber. This caused me a delay, but I
have it solves in the following way:
- I aimed at transistors in boitier TO247. This package has advantage
to have plastic on above the transistor, with hole for
fixing. The screw of clamping become absolutely isolated electrically
from the metallic tongue, what allows to remove the insulating cannons.
Interface area is bigger than the one of TO220, what reduces the
temperature a bit. Mica stays. It cashes the temperature very well, I
never have had of problem with.
Now the ampli has already several hour of functioning in this
state, without particular problem. However I added a fan for box
cooling. I see temperatures close to the
limit of 85°C of the chemical capacitors, I therefore reduce every
risk by ventilating. You are going to say to me that this
fan makes noise, then I answer you that we reduce the noise by
decreasing the fan voltage (so the fan speed), while keeping a
speed still sufficient for box cooling. In future, it would
definitely be better to find components which can hold 100°C at
least
without problem, to remove the fan.
Used wire::
For supplies, normal cable.
For entrance signal, I decide to use enamelled copper wire of about 0.5mm, that I twisted with the drill.
For speaker output, same idea, but by using several enamelled copper wire in parallel to acquire about 2*1mm2.
Results:
First listening.
Of course, it is necessary to not look especially the yield
of such system. The consumption is high, and besides it is necessary to
target a thermal dissipation about equal to 15 times the maximum
output power. This give to me 150 watt of dissipation for 2*5
watts of audio output.
It has needs to construct a compensation filter to compensate the
impedance variation of the speaker for the first tries.
Once makes, the result is sublime. A neatness in timbre, ways, and
in orchestras absolutely brilliant. And a complete absence of noise! I
thought that it would be not too poor, but there this nailed me the
ass! This structure buried every amplis of class AB which I knows
and that I conceived before, and which however was already great.
Photo
Photo: Global view.
Constitution:
With a little detachment, I shall summarize its tempéramment in a following manner:
- Very good definition of mediums and acutes. Many details, and it
is very difficult to find a trace of aggressivity, even when a
symphonic orchestra charges thoroughly!
- Medium held of the bass. Because of an high output impedance, the low
miss impact, maybe demand a correction filter between
amplifier and the source (ex: CD reader). Even with this, there is a
deterioration of impulse answer.
- Weak available power. This demands high sensitivty speaker, or a use with multi-amplification with active filter.
- It finds its place in an active filter system (for instance with a
classical amplifier for bass ways and this amplifier for
the medium/acute way) where all previous problems is finally
solves. It is what I have made since mid-May 2004, by putting a 3rd
order 300Hz filter in (attenuates faster the bass than
a basic filter RC) in front of this amplifier.
Two small defect that I envisage to correct (2004):
- The output impedance of 11 Ω. I plan to reduce it to 5 Ω,
by changing the box, radiators, and increasing the number of power
components. By the way, this would also increase the output power
from 5W to 8.5W on 8 Ω load!
- The impulse current on startup due to the the load of the output
capacitor. It is not very strong, the loudspeaker seems endure it well,
but however I plan to top up an intermediary relays to short-circuit
the output on startup.
- L'impédance
de sortie
de 11Ω. J'envisage de la réduire à 5Ω, en changeant le boitier, les radiateurs, et en augmentant le
nombre de composants de puissance. Au passage, ça augmenterait
aussi la puissance de sortie de 5W à 8.5W sur 8Ω de charge!
- L'impulsion de courant continue
au démarrage du au chargement
du condensateur de sortie. C'est pas très fort, l'haut-parleur
semble bien le supporter, mais cependant j'envisage de rajouter un
relais en interne pour
court-circuiter la sortie au démarrage.
personal realization of an innovating pure class A amplifier -
new class A amplifier - Amplifier without feedback - creation of
amplifier - creation of schema - try of new schematics of amplifier.
Created : 01 june 2004.
Last update :
11 november 2007.