# اهمیت ارتباط Power Supply با مدار تقویت یک آمپلی فایر از دید طراح Metaxas

## اهمیت ارتباط Power Supply با مدار تقویت یک آمپلی فایر از دید طراح Metaxas

چهارشنبه 6 ژوئن 2012
/ / /

برندی هست به اسم Kostas Metaxas که آمپلی فایر ترانزیستوری میسازه و مطلب جالبی در مورد ارتباط منبع تغذیه با استیج های تقویت داره. قبلا هم اشاره ای داشتم به نظر nutshell در مورد وضعیت نویز سوئیچینگ درمنبع تغذیه آمپلی فایر ها .

تجربه هایی که من در شنیدن صداها داشتم نشون میده هم منبع تغذیه و هم وضعیت برق سیستم خیلی مهم هست.

مجله تی ان تی هم مصاحبه ای داشته با طراح این برند:

http://www.tnt-audio.com/intervis/mase.html

متن زیر نوشته این طراح در مورد منبع تغذیه هست :

PDF Format

The most significant difference between VALVE and TRANSISTOR circuits is the amplifier/power supply regulator circuit interaction which is far more critical in Transistor circuits.
To illustrate this phenomenon using the the most basic amplifier gain stage, let us assume that our input signal is a 1.0 Volt peak to peak, 1 kHz sine wave and the amplifier stage has a gain of 10 so that the output voltage is 10 Volts peak to peak. For simplicity, resistor R is the load resistor which dictates the overall gain.
If this was a VALVE amplifier, the high voltages which are typical in valve circuits (from 200-400 Volts DC) would result in a valve of around 50,000 to 100,000 Ohms for resistor R.
The equivalent transistor amplifier using much lower voltages (from 12-15 Volts) would have a substantially lower value of R between 200 Ohms-100 Ohms. Therefore the power supply used in the transistor amplifier is virtually directly linked to the transistor amplifier circuit compared to the isolation of over 50,000 Ohms in the Valve circuit.
If we study this basic circuit to analyse the behaviour of the input signal we can see that the 10 Volt output signal not only appears on the output of the amplifier circuit, but in fact also travels through resistor R towards the power supply.

If we assume that the regulator impedance at V+ is around 2 Ohms just for the purpose of this illustration, then let us study the amplitude of the 10 VOLT sine wave as it goes through R and returns back to the OUTPUT of the TRANSISTOR circuit and VALVE circuit.
In the VALVE circuit, the 10 VOLTS goes across the 50,000 Ohms R towards the power supply impedance of 2 Ohms and the 10V signal is attenuated 50,000/2 = 25,000 times.
Therefore 10V/25,000 = 0.0004 Volts of 1,0kHz sine wave.
On its way back to the OUTPUT of the circuit it is attenuated by the impedance of the amplifier (say 100 Ohms) which then: 0.0004 Volts/50,000/1,000 = 0.000008 Volts.
So therefore, a 0.000008 VOLTS of out of phase sine wave accompanies the 10 Volts sine wave as out-of-phase distortion on the VALVE CIRCUIT.
In the TRANSISTOR circuit, the 10 VOLTS going across the 200 Ohms resistor R would be attenuated only 10/200/2 = 0.1 VOLTS.
Then on the way back to the output, the voltage is attenuated by: 0.1V/200/1000 = 0.05 VOLTS of out-of-phase sine wave added to the 10 VOLT output sine wave.
If you compare the TRANSISTOR and VALVE circuit, the ‘phase distortion’ is 0.5% for the TRANSISTOR as compared to 0.000008% for the VALVE which clearly demonstrates one of the major reasons for the difference in sound between VALVE and TRANSISTOR circuits.
If we monitored the V+ point of the transistor circuit using an oscilloscope, we would notice this 0.1 Volts, 1.0 kHz signal. If we were to increase the frequency to 10,000 Hz and up to 1.0 MegaHertz the speed of dynamic behavior of the power supply becomes critical. Using a normal I.C. regulator would result in the signal at V+ actually increasing in amplitude as the frequency increases to that at 1.0 MegaHertz the 1.0 Volt sine wave is not over 1.0 Volt!
To fully understand this interaction between the amplifier an power supply, it is necessary to understand the operation of a voltage regulated power supply.
A voltage regulated power supply is essentially a D.C. amplifier (not unlike a normal power amplifier) which instead of having an audio signal at the input which is then amplified to become a larger audio signal at the output, has a fixed D.C. voltage reference at the input which is then amplified and becomes a larger DC voltage of at the output. The output impedance of the regulator, not unlike the output impedance (or “Damping Factor’) of a power amplifier is less than one ohm at D.C.
If we use a 2.0 Volt zener diode as our fixed DC voltage reference at the input of the D.C. amplifier which has a gain of 10, the resulting output voltage is 20 Volts D.C.
The negative feedback loop of the amplifier which fixes the gain of 10 times the 2.0 Volt zener reference is very important because it maintains the output voltage irrespective; of an increase or decrease in the power supply voltage to the amplifier as long as there is a minimum voltage for the regulator circuit to operate (for a 12 Volt regulator, the minimum voltage is 15 Volts).
This is the STATIC performance of a voltage regulator which although important, does not affect the overall sound of the amplifier as much as the regulator’s DYNAMIC performance which is influenced by the speed and ‘open loop gain’ of the regulator.
To understand why the Dynamic performance of a voltage regulator is so important, we need to go back to our basic amplifier circuit and investigate what happens to the 1.0 Hz, 10 Volt output signal as it goes across resistor R and encounters our voltage regulator.
To ensure an absolutely stable D.C. at V+ the residual of the 10 Volt sine wave at the OUTPUT is fed through the negative feedback loop of the regulator to force the amplifier to correct this error by applying an inverted signal identical to the residual sine wave to totally eliminate the residual sine wave at V+. A high speed regulator would therefore treat a signal 1.0 Mega Hertz in the same manner as a signal at 1.0Khz.
The ultimate voltage regulator would effectively have a theoretical output impedance (or ‘Damping Factor’) at V+ of zero ohms at all frequencies as a result of its wide bandwidth before the addition of negative feedback.
In this way, the attenuation of the 10 Volts across the resistor R residual would be complete, and no attenuated component of the 10 VOLT sine could be deflected and return to the OUTPUT of the circuit and cause severe phase anomalies by adding to the new signal presented at the output remember that it would take a few microseconds for the signal to go through the resistor and come back.
This extraneous out-of-phase information which adds to the new OUTPUT signal then destroys TIME/PHASE characteristics of the amplifier circuit.
In real world power supply circuits, the impedance of the power supply increases as the frequency because the open loop gain is reduced at high frequencies and the amount of feedback used to linearise the amplifier circuit and maintain the low output impedance is substantially reduced.
If we go back to our basic circuit and analysed the performance of an I.C. positive voltage regulator (say a LM78LXX from NATIONAL SEMICONDUCTIONS) it would have an output impedance at the pin of its output lead of around 0.2 Ohms from DC to 10kHz, and then an increase to 0.4 Ohms at 20kHz, then 4.0 Ohms at 1 MEGAHERTZ which clearly illustrates the open loop frequency response has a turnover point around 10 kHz.
When you add the normal distance between the regulator output and amplifier circuits which may be as little as 60mm to as much as 200mm in many circuits, the overall impedance in creases 5 to 10 times. Also, to stabilise the operation of this I.C. regulator, it is essential to use an output capacitor for stability.
Clearly, this is not adequate for high performance, high speed transistor circuits. For this reason, we have approached the design of our regulators as PART of our amplifier circuits, rather than make the fastest amplifier circuit and add a slow I.C. voltage regulator with an output capacitor and call it a finished design. Our discrete voltage regulators are designed to have the absolute lowest noise, reject mains ripple, but more importantly to have a speed (1000 V/microsecond) which is a result of its wide band design (an open loop frequency response greater than 500kHz) and output impedance which is an order of magnitude better than any I.C. The regulator stability is achieved without ANY capacitors by varying the ratio between the local and overall feedback of each device.
We position the regulators within inches of the active circuits (in the case of our OPULENCE, the regulator is 3mm! from the active circuits) and the regulator impedance is flat from DC to beyond 5 MegaHertz at less than 0.05 Ohms.
Beyond this electrical design aspect, we listen to the sound of our regulators whilst developing each amplifier circuit to ensure that every component change or substitution produces an audible improvement from the selection of transistors to best biasing currents , choice of voltage references zener and degree of local feedback.