At your own risk. Take all safety precautions even when experimenting with low voltage.
This is on using just one transistor to drive directly a loudspeaker.
_____________________________
| | 2N3053 |
73KΩ [] | c |
in ----==-----------|--------- b < 800Ω [ ]<
4K7 | e |
______________________|________. |_________| |^| 6.5mA
' | 10.3V
The Pleiades 2N3053 power amplifier schematic (73KΩ is a typical value, Rcb is adjusted by ear-brain)
Pleiades experiments with the 2N3055 driving directly an 8ohm voice coil have been previously described.
This is on driving a 800Ω Philips AD5046N full range speaker.
Before attempting to drive this speaker with a triode electron tube such as an WE417, 6C4, triode connected EF183, for example it was desired to try a transistor. This has the advantage of no heater voltage which is important for an amplifier designed to operate from a battery and still be class A. Class A means electrons flowing all the time.
The 2N3055 experiments gave a fine sound. The problem is impedance mismatching making an amplifier that consumes 100-200mA but does not sound as loud as it should. And when a 9V battery is used only a few songs can be played before the battery is discharged.
The 2N3055 would have prefered to see a load of 80Ω instead of 8Ω. In fact supplying with 12V, only 1 or 2 volts are dropped in the speaker. (The speaker is directly connected to the collector on these experiments, so electrons flow all the time through the speaker too, the cone can be seen to be somewhat deflected to a new equilibrium).
A speaker of 800Ω is good news and there are many transistors that can be seen in old germanium datebooks to drive 600Ω.
However when the drawers were opened a nice transistor on a round heat sink caught the attention of the eye immediately. It is a 2N3053. And it's datasheet shows good news. Current about 30mA. The impedance seems to be very close to the speaker.
So by impedance matching an efficient transfer of energy without reflecting back will take place.
It was tried and yes it works beautifully. And the current consumption is about 10 times less while the sound is as loud. And it is louder, as the Philips with its AlNiCo magnet is more efficient and sounds louder.
The sound is amazing. While writing this Madonna can be heard coming from a point source as clear and smooth.
Signal path, setup:
To Have and Not to Hold - Madonna - Sony D-EJ758CK CD Walkman at its 300Ω line out impedance - a series resistor of 4.7KΩ - 2N3053 with 73ΚΩ bias resistor from collector to base - Philips AD5046N connected to collector, 12 V power supply, emitter connected to minus or gnd
More batteries in series were also tried for louder sound so the best conditions (using a variable bias resistor while listening) are as follows:
Vb=36V, Ic=21mA, Rcb=120KΩ
Vb=24V, Ic=14mA
Vb=10.3V, Ic=6.5mA, Rcb=73KΩ, Vc=5.8V
It is nice that the voltage is now shared and that about half of Vb appears at the collector as about half drops through the voice coil itself.
Power consumption is very low now by a current as low as just 6.5mA. Strictly speaking power is voltage times current. The sound is world class with just one transistor and 2 resistors from source to speaker.
It is nice that now most of the power input is not converted to heat, (the 2N3055 got hot), but much of the power is converted to music. To calculate power the P=VxI relation is used. The power input is 10.3Vx6.5mA=67mW. The power desipation at the transistor is 5.8Vx6.5mA=37mW. The rest of power (10.3V-5.8V)x6.5mA=30mW is desipated at the speaker voice coil. Of course when signal exists the transistor cools down. The voice coil moves and radiates acoustic power etc. For an excellent description of power desipation in class A circuits see T.S. Gray.
The Pleiades power amplifier is based on the circuit presented by Jones. And the attempt of course is to further simplify it by removing as much components as possible!
So how does it sound like?
At the moment the least power version was chosen as it sounds very nice. The sound is smooth. With amazing midrange. Extended treble but not hyped, not harsh at all. And great bass as the amp being direct coupled goes down to as close to 0 Hz as possible. The almost 0Hz component can be seen by the permanent deflection of the cone.
At peaks there is of course some clipping, soft clipping distortion. But it is the kind of distortion that mastering studios charge many hundreds of dollars per hour to put into the final product by class A operating electron tube amplifiers, Fairchild compressors etc.
It is great that Madonna comes from a single point in space in this mono arrangement. It is close to her being in the room at some point in space.
The amplifier is enough loud to listen to great quality music and being able to keep a conversation.
And nobody will tell you to turn it down.
How could the electron tube version sound like?
The electron tube's anode can be connected directly to the 800Ω voice coil. Rk can be variable for max gain and best sound quality. Or cathode connected to gnd and high Megohm anode to grid resistor may make operation of the electron tube at a relatively low anode voltage. (See Pleiades V4, V5, V6 for example). This electron freeing resistor conserves the value of Gm. Since μ is dictated by the geometry of the tube it follows that ra will be lower. Or since we can have high Ia with low Va this means we have a lower Ra or electron tube output resistance. With a WE417A it should be possible to have an Ra of 800Ω, if this is so it will be a perfect match in terms of max power transfer to the 800Ω voice coil.
References:
2 Transistor Miniature A.F. Amplifier by K. Jones - page 41 - Audio Amplifiers - Edited by J. R. Davies - Data Publications Ltd - London
Applied Electronics - T.S. Gray - MIT
On Preserving Electron Tube Transconductance at Low Anode Potential - euroelectron
Tubes vs Transistors (vs op amps), Is there an audible difference? - Russel O. Hamm - Journal of the Audio Engineering Society
No comments:
Post a Comment