It has always been an ideal of design engineers to amplify audio in a simple way. It started off in the 19th century, when no electronics were there (!), to try and have control over the current in a wire by applying a magnetic field to it. That didn't work. Much later, in the 1970's, we succeeded in controlling current by a voltage through newly designed mosfets. Those devices do the job in a simular way by creating a kind of ‘field' as a result of the applied voltage on the gate and this field controls the current going from the drain to the source.
The 'inaudible' wire
Another way of looking at the problem occurred in audio electronics. When trying to get hold of the specific quality of an audio circuit several types of measurement equipment were designed such as voltmeters, current meters, distortion meters, signal generators etc. The result of all this was that we could measure the various capabilities of an amp and put the results in a specification sheet. This seemed a proper way and we succeeded in creating amplifiers with very low distortion and high output power. Some engineers, Peter Walker from Quad for instance, then thought they had achieved the optimum possible sound quality and described this amps as behaving like a ‘wire with amplification'!
Human perception
Nowadays we know better. We know that we don't know a lot of what's happening in audio. We also know that the human hearing is more complicated then what we presumed before. The way our brain behaves when listening to music cannot be described as listening to specific tones with specific harmonics etc.
Distortion is a factor to be considered but below 1 percent harmonic distortion there's hardly a human person who is able to hear it. Other kinds of distortion like cross over and transient intermodulation may be perceived at a level of 0,1 per cent. But then again there are other factors which deteriorate the sound quality in a more serious way.
'Time' is an important factor while a part of our audio perception as this is based on discerning time differences and even the perception of pitch has to do with a function of our brain which measures the wavelength (or time!).
Some designers created amplifying circuits with significantly higher harmonic distortion but with lower or even without ‘feed back'. Any part in a circuit shifts time. So feed back (from output to input also called ‘overall feedback') may introduce a ‘time' problem.
Electronic parts
A lot of research has been done on improving the quality of passive components such as resistors and capacitors. Those components are always in the signal path and contribute to the overall ‘sound quality'.
A capacitor introduces a specific problem, it has a ‘memory'. It behaves like a battery, after being loaded with a voltage and thereafter unloaded one can still measure some voltage. (If you leave the headlights of your car ‘on' during a night the battery will go down and you're not able to start up the engine. Then, when you switch everything off and leave it for a couple of hours, you might be able to start the engine again! That's ‘memory'.)
This ‘memory' causes a kind of time ‘smearing'. Nowadays there's a lot of difference in capacitors. Socalled ‘audiophile' capacitors may give a better sound quality then the average polyester type. Those components are mostly specified as being ‘paper-in-oil' ones. Funny thing is that the old oil-filled capacitors as used in the fourties still have a remarkable good sound quality.
Considerations
At APN we have been busy on developing amplifiers for over thirty years. During the nineties we came to the conclusion that there are three configurations possible which have ‘High End' audio properties. On the other hand we think that some configurations deteriorate sound quality as such. Let's start off with the latter ones.
1. Differential or balanced configurations. This requires extra electronics in the signal path and some times twice as much!
2. Configurations with lots of active components (be it transistors or tubes) and high levels of overall feedback.
3. Operational amplifiers (op amps) without specific measures around it to control it's behaviour.
The better ones should be configured as follows:
1. Very simple configurations with transistors such as Jean Hiraga's ‘Le Monstre', the later ones with just one power fet and/or our own A-18.
2. Single Ended tube amplifiers
3. Hybrid amplifiers
Some other things should also be taken in consideration:
1. The power supply may also be a major factor. With most tube configurations the power supply is part of the signal path! When using semiconductors (diodes) for the rectifier something should be done to suppress the ‘sparks' resulting from those devices.
2. As a result of the development of ‘switching' power supplies in computers new electrolytic capacitors where designed which have a lower impedance over a wider frequency range. Those capacitors behave ‘better' in audio power supplies.
At present we have our ‘A-50' design which was originally developed as an ‘A-30' in 1989 and later on improved several times. The current A-50 design has several advantages over other designs:
1. At the input a tube amplifies the voltage to the desired value. A second tube is a special type of cathode follower with a very low output impedance. Thereafter the signal is coupled to the output devices which act as current amplifiers. The coupling is done by two polypropylen capacitors.
2. The input has a high impedance and is directly connected to the first tube.
3. The output has a very low output impedance and may deliver peak current values of over 100 Amperes. The latter means that almost ANY loudspeaker is driven correctly, may it be a loudspeaker with a very complicated filter, an electrostatic one or a magnetostatic one. So you never have to worry about the amplifier when purchasing new loudspeakers.
This A-50 is a ‘class-A' design. The advantage then is that voices and woodwind instruments are reproduced with a more natural character. A disadvantage is the price tag because a huge power supply is needed and the heatsink is quite large to keep the temperature at a reasonable level.
When thinking about this design it is obvious that there are some disadvantages. The cabinet is quite "big" when compared with class-B amplifiers. When switched into class-A it consumes a lot of power from the mains connection and this will cost more money, again when compared with class-B types. Another disadvantage is that an extra control amplifier is needed in order to drive the circuit in a proper way.
When starting off from scratch there are some things we want:
1. A tube at the input as this is the only device which in it's way is able to amplify very low signal levels without adding a threshold.
2. The sensitivity should be around 300 mV so there's no need for a pre-amp.
3. As in our other amplifiers we don't want any "active" protection circuit, the amplifier should protect itself.
4. When configured in ‘Class-B' it's a lot cheaper and a bit more dynamic in it's behaviour.
Some tubes work quite linear with an anode voltage of 40 Volts or higher. Some people use tubes at quite low voltages like 24 Volts and that's going fine but less linear and one needs feed back to cut down the odd harmonics. In the current design we have a DC power supply of + and - 30 Volts, alltogether 60 Volts. The minus connection from the supply is needed as a reference for the tube, but at that supply there's always some "hum" occurring. So we added a voltage stabilizer in order to suppress that hum.
Have a look at the schematics.
A disadvantage now is that the "ground" level at the input differs from the ground level for the loudspeaker connection. Those two should never be connected to each other or we would have a massive short circuit. This could be avoided by adding a big capacitor between the output and the loudspeaker. Although the latter is much "safer" we didn't do so because that capacitor deteriorates the sound quality overall and more specific the output impedance at low frequencies will go "up" and the behaviour of the loudspeaker at the lowest octaves will be less "controlled".
The new amplifier is available in two versions, as a HB-20 giving 20 Watts into 8 ohms and (with doubled output devices and a bigger mains transformer) as a HB-30 giving 30 Watts into 8 Ohms and 50 Watts into 4 Ohms.
SPECIFICATIONS
| |
HB-20 |
HB-30 |
| output power at 8 Ohms |
20 Watt |
30 Watt |
| output power at 4 Ohms |
35 Watt |
50 Watt |
| output power at 2 Ohms |
32 Watt |
72 Watt |
| input sensitivity |
300 mV |
300 mV |
| bandwidth -3 dB |
12 Hz - 180 kHz |
12 Hz - 180 kHz |
| distortion + noise (1 W) |
-65 dB |
-70 dB |
| distortion 20 Hz - 50 kHz -3 dB |
<0,5% |
<0,5% |
| decrease at 1 W vs 8 Ohm: |
|
|
| at 4 Ohm |
-1,5 dB |
-0,5 dB |
| at 2 Ohm |
-3 dB |
-1 dB |
| overshoot at 8 Ohm//2 µF |
15% |
15% |
| overshoot at 4 Ohm//2 µF |
10% |
10% |
| input impedance |
100 kOhm |
100 kOhm |
| number of inputs |
4 |
4 |
The amplifier is unconditionally stable
