SONY TC-D5 PRO serial number 10086 Cabling photos

 

At your risk.

Effort Makes It

 

Προαπάθεια τα καταφέρνεις.

A New Pleiades Meal

At your risk. This is not a nutritionist or a medical advice.

Barley seeds as soaked in double size water.. (The seeds are checked under sunlight so that impurities such as tiny very hard objects that could harm teeth are removed).

Said seeds are left in filtered water for 2 days.

Then at my risk I drink the water which tastes like a very nutritional beer.

Then a spoonful of tahini is added with some new water.

Then royal dates are added for sweetening.

It is ready and it tastes very nice. It feel good eating it and eater eating it.

Design and Investigation of a Triode Single Ended Audio Frequency Power Amplifier with No Negative Feedback

At your risk. Lethal voltage present.

This is my final (3rd) year project for the Bachelor of Engineering in Engineering Electronics at University of Warwick, UK.

Too naive at that time I did not use a PSU with a choke input filter and a small value capacitor.

Also on euroelectron you may find other newer much simpler designs with only one electron tube with the aid of a voltage step up input transformer. These designs can also operate with batteries, at your risk, lethal voltage present.

1992

Design and Investigation of a Triode Single Ended Power Amplifier with No Negative feedback

 

 

Introduction

 

  It has been found that audio frequency amplifiers without negative feedback appearing recently have exceptional sound qualities [][][][][][][]. Most of these amplifiers use triode (valves) operating in their linear region in a single ended configuration.

  This project report describes the design and construction of a single ended no negative feedback amplifier based on one of the most linear audio devices ever made, the Western Electric 300B power triode. Discussion of some aspects of measured and listening performance are given together with considerations as to why these amplifiers sound so different. It also provides a suitable foundation in doing further work in aspects such as time delay in the operation of negative feedback, measured versus auditory perfornance, bandwidth and phase response (group delay) and its relation to delays on the feedback operation mechanism.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

                           CHAPTER I

 

 

 

 

Background             

 

 

  Although electronic engineering is firmly in the age of solid state and advances in many areas are wellcomed, it is not quite the same when it comes to music reproduction. It may be surprising to read in hi-fi magazines published in 1992 and 1993 that amplifiers using valves as active devices like the Audio Note Ongaku are considered state of the art [][][].

  It might be even more surprising to hear that these particular amplifiers are single ended no negative feedback designs using a directly heated power triode (valve) in the output stage. The Audio Note Ongaku appeared in 1990, uses the prewar developed General Electric 211 transmitting tube, quite a lot of silver in its capacitors and output transformers and costs over thirty thousand pounds. This design has more in common with designs of the 1930's than the later Leaks and Quads of the 1950's. Its performance has left UK reviewers amazed.

  They describe their experience [][][][][][] simply by saying that these amplifiers when sourced with a good analogue recording not only do they make the loudspeakers "dissapear", but "the whole system falls away" leaving the listener with the impression of being in the live performance. The dynamic range is discribed as phenomenal. The sound stage is as wide, deep and high as the recording allows and no one can beleive that a 25 Watt amp can sound so loud and dynamic.

 

1. MEASURED VERSUS AUDITORY PERFORMANCE

 

  It happens quite often especially with valve amplifiers that they have awful measured performance but sound remarkably good. When the fist studio consoles using op-amps appeared, some audio professionals [] measured         but when they put some music through it everybody looked horrified at each other.

  When it comes to harmonic distortion low quality commercial amplifiers usualy have total harmonic distortion figures less than 0.001% whereas reference amplifiers may exede 1%. It may not be far from truth to suppose that the less the THD of an amplifier (more negative feedback applied) the worst the amplifier will sound. In general negative feedback has a bad reputation in the hi-fi circles [][][].

  Let us try to explain this apparent contradiction. First of all what do we mean by harmonic distortion ?. Harmonic distortion exists in all non linear systems. If we had a hypothetical linear amplifier and we apply to it a sinusoidal voltage we should get at the ouput a sinusoidal voltage of the same frequency and in general different amplitude and phase. But since nothing in nature is linear, amlifiers are not and they will give a distorted form of a sinewave, let as call it y(t). The degree of distortion will depend on the non linearity present in the amplifier. Since this would be a periodic waveform it can be Fourier analysed. The Fourier series will give all the sinewaves of integer multiple frequency that we would need to add to the fundamental of the output to get y(t).

  This is equivelent to say that the amplifier has created frequency components that did not exist in the input. If the amplifier is soursed with 440Hz, say the sound of a A4 tuning fork through a microphone, the amplifier will also create 880Hz (A5), 1760Hz (E5) and so on. In fact listening to this if the fist few harmonics are dominant compared to the higher it may sound like a A magor chord. If we input more than one sinewave voltages there will also be created other not harmonic related frequencies  (intermodulation distortion).

  When music is input in the amplifier it is much more complicated since many frequencies are now input in the amplifier (instruments create many harmonics, many instruments are played together ,notes have small durations i.e. create wide frequency spectrums etc).

  One reason [] for the new breed of "old triode" amplifiers to sound so different could be the fact that triodes produce predominantly second harmonic distortion. The subjective impact of second harmonic distortion is difficult to identify, except large quantities more than 5% or so exist. This is reasonable since second harmonic means an octave above. One can get a feel for two notes one octave apart by striking say A4 and A5 on a piano. Both notes sound the same but one feels higher than the other (hence they have the same name). On the other hand 3rd, 5th, and 7th harmonic distorthion that op-amps produce is far less musicaly related when considering what goes in an aplifier. Especialy when overdriven a bit they sound more harsh and tiring 0.001% or so 9th harmonic distortion can be identified easily.

  This also gives a reason why good guitar players use valve amplifiers to get clear distortion. These amplifiers when overdriven distort in the biginning with low order harmonic distortion making the sound of the electric guitar fuller and warmer. Good examples of this are groups of the 70s (eg Pink Floyd), Eric Clapton etc.

 

2. WHAT IS WRONG WITH NEGATIVE FEEDBACK

 

  Negative feedback is a method whereby the signal from the output of an amplifier is fed back and subtracted from the input. In this way an error signal is created which tends to correct the amplifier.

  But this is not so simple since time is taken for the feedback to arrive []. Delays exist both in the feedback loop and the amplifier itself. The hypothesis is that by the time the correcting signal arrives the input in the amplifier has changed since music is going on (harmonics of instruments may have decayed, other notes may have been added etc).

  A situation is apparent in op-amps which use large ammounts of feedback. In transient conditions i.e. music the amplifier is momentarily open loop and is clipped because of the tremendus open loop gain. This results in 100% transient intermodulation distortion.

 

3. THE TRANSFER FUNCTION OF AN AMPLIFIER

 

  A linear system is defined as one for which the principle of superposition applies. This means that if the response of the system is known for any two exitations applied seperately, the response to the sum of these will be the sum of each response.

  Linear systems can be characterised by their transfer function. The transfer function maps a complex number at any frequency f. The physical meaning of this complex number can be shown to be [][] as follows. Its magnitude is the ratio of the amplitudes of output to input sinewave of frequency f, and its argument (angle) is the phase difference between output and input sinewave of frequency f.

  In a linear amplifier the magnitude of the magnitude of the transfer function is the frequency response (magnitude of output over magnitude of input at any frequency) and the arqument of the transfer function is the phase response (phase shift between input and output at any frequency). If the transfer function at any frequency we can predict what the output will be for any input. Say an arbitrary input x(t) is applied and the output y(t) needs to be predicted. X(t) can be writen as a sum of sinewaves according to the Fourier integral. To the amplifier it would not make any difference whether these added sinewaves are applied or x(t) because the input is the same. Then the transfer function can predict what would happen to each of these sinusoids if it was input sepparately. The output will be the sum of all responses according to the superposition principle.

  Of course this applies only to linear systems where the superposition principle holds by definition.

  A distortionsless linear amplifier must preserve the Fourier soectrum. Therefore [][] all frequencies must be amplified by the same amount and the phase of each must be preserved or increased linearly with frequency (same time delay for all frequencies).

  In audio amplifiers it has not yet been desided whether the bandwidth must be greater than 20Hz to 20KHz or not. Similarly the phase response once believed to be unimportant on the sound quality now it has been found that the ear can detect phase changes and that further investigation is needed.

 

 

 

 

 

 

 

 

                         CHAPTER II

 

 

 

 

Design of the amplifier              Used               

 

 

  The chosen type of amplifier to be designed was a class A single ended using a triode power tube. One reason for that was that the amplifier should be as linear as possible without any feedback so that phase and feedback (if negative feedback is applied) delay effects could closer be studied. Another was curiosity whether the sonic advantages of the Ongacu could be observed in smaller scale. A important part of motivation was that if the amplifier performed any close to the high quality amplifiers like those described in chapter II it would be nice using it to listening to music and experiment with it trying to improve on the sound quality.

 

1. SOUND REPRODUCTION

 

  The objective of sound reproducing equipment ,ie turntables, amlifiers, loudspeakers etc is to reproduce music as faithfully as possible. But how do we know how faithfull is a sound reproducing system ?. This is not always easy but fortunately there is a final judge and an absolute reference. The final judge is the human ear and the absolute reference is live music. No sound system is perfect but the good ones are those that bring the listener close to the beauty of live music.

  As an example some systems are capable when someone is having two loudspeakers in front of him/her and sits in the midlle to have the impression of the size of the recording hall and the orchestra. It is possible to hear instruments sounds coming from the middle if the recording is arranged so that both loudspeakers produce the same sound intensity at these particular instruments. Another sound will appear to be coming from a bit more right if the right loudspeaker plays a bit louder that particular sound. In this way it is posible to find out the place of an instrument in the orchestra. Soundstage depth and also height are possible.

Even having such equipment able to do this and of course other, going to a live performance one see how far we are from that. This is written as a deffence for those who believe that we know almost all there is to know about music reproduction.

 

2. ELECTRONS AND MUSIC

 

  Electronics can be defined as the subset of science, engineering and art that deals with the control of the movement of electrons to do something useful. This is certainly true in the music reproduction area. In an audio amplifier for example the music signal is used by an active device to produce corresponding variations in the movement of electrons (electric current) in the device. Amplification is achieved by the fact that the small power signal from say a cartridge is used in the amplifier to control the much greater power used to make the cones of the loudspeaker move. This excess energy is given by the power supply of the amplifier. The amplifier must be able to inlarge the input signal without changing its characteristics.

  To go in a bit more detail the reproduction of a simple sound will be described. Probably the simplest sound is that coming from a tuning fork. The sound of it is a pure tone containing one frequency. When it is striked its two metal parts vibrate sinusoidaly at its frequency. Air molecules nearby coliding with it are forced to vibrate sinusoidaly at the same frequency. Then these molecules collide with others and pass on the vibration and so on. This is how sound propagates as a longtudinal pressure wave. Now imagine a microphone is situated nearby. A microphone consists of a membrane that can vibrate atouched to coil from which ends is the voltage output. This coil is situated near a magnet. When the molecules carring the information of the sound of the tuning fork collide with the diaphragm it is made to vibrate. Since now the is magnetic flux changing sinusoidaly through the coil there is an emf induced at it by Faradays law. There is therefore a sinusoidal voltage at the output of the micriphone. The signal can travel now at a cable as a corresponding electromagntic wave. Effectively the electrons in the cable now have a sinusoidal drift velocity superimposed on their thermal movement in the same way that the air molecules had a sinusoidal drift velocity superimposed on their thermal movement. One difference being now that the infomation is passed by forces exerted between each eletron because of their charge, traveling at a speed equal to the speed of light in the dielectic of the coaxial cable used.

  At the end of the cable this sinusoidaly varying voltage with the same frequency as the tuning fork in input to the amplifier. At the output of the amplifier (more of this latter) this sinusoidal volatge appears with greater amplitude. then this siganl is passed to the loudspeaker which works on the same prinsiple as the microphone. Here the voltage is applied to the louspeaker coil attached to its membrane. The is a sinusoidal current in the coil and therefore a Lorentz force sine there is a magnetic field nearby produced by the permanent magnet og the speaker. Then the memprane moves sinusoidaly. Simillarly then with molecules and the sound wave produced until some collide with the ear drum drum of somebody situated with the loudspeaker maby in a different room. The sinusoidal motion of the ear drum is then translated through the mechanism of ear and brain to the sansation of the sound reproduced.

  The same priciple applies to mpre complicated sounds. The link can be made by the siniwave cincept from the point of view of fourier transforms where a complex even non periodic sound can be though of as sinusoidal sounds of infinite frequencies existing together. Throughout this example it was assumed that linearity holds. This is only an approximation.  

  More detail about what happens in the amplifier will be given next. To do this since the amplifier to be described is a valve one a few things will be said about how valves work.

 

3. A FEW THINGS ABOUT VALVES

 

  The first device invented able to control the movement of electrons effectively was the triode valve. It consists of an evacuated usualy glass enveloppe with three electrodes in it (hence triode from the greek word for three "tria").

  Its construction and circuit symbol can be seen in Fig.1 and 2 respectively. These electrodes are arranged as follows. First comes the cathode which can be a fillament. When it is heated some of the electrons in it can gain enough energy to escape. This is what is needed because without free charges in a vacuum a current cannot exist.

  In order to make these electrons move another electrode called anode or plate is inserted which is charged positive with respect to the cathode by a voltage supply Fig.3. (This is quite the same as charging a capacitor. This capacitor exists because we have two metal elecrodes the anode and cathode, close to each other. It is called Cak.) The electrons are now attracted by the anode and move towards it then collide with it and continue their journey  in the cable connected to the anode through the power supply and back to the cathode. Thus there is an electric current. The near vacuum in the valve is important to achieve because the electron movement would be impeded by collisions with gas molecules.

  How do we now control the movement of elecrons that is the electric current ?. This is done by another electrode called the control grid, which is placed between the cathode and anode nearer to the cathode. The input voltage is applied at the control grid. By changing the voltage accross it with respect to the cathode this means that electrons have come to it or left from it. This makes it charged negative or positive accordingly. If it is charged positively it will attract more electrons towards the anode and the anode current will increase. If it is charged more negatively it will repel more the electrons and therefore the current will decrease. In this way we have a voltage controled current source. Since the grid is placed closer to the cathode than the anode in order to produce the same change in anode current a smaller change in grid voltage is required than in anode voltage. This means that the grid voltage can be amplified.

  The existance of interelectrode capacitances (e.g. between grid and cathode restrict very quick changes in elecrode voltages and hence gives the high frequency limit at which the valve can still

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

amplify. Fortunately this is at radio frequencies.

  In normal audio usethe grid potential  with respect to the cathode is kept always negative. This has the advantage that electrons are not attracted to the grid ,ie grid current is negligible. Thus negligible current and therefore power is taken from the voltage source whose power must be amplified.

  When refering to electrode voltages they are all by convention

referenced to the cathode. Thus for example by anode voltage we mean the potential difference between anode and cathode.

  Information about how a valve behaves to the ouside world can be measured or read in a valve or electron tube manual. A useful family of characteristics is the anode current versus anode voltage with grid voltage as parameter. In Fig.4 this kind of characteristics is shown  for perhaps the most  popular triode used now extensively in most valve hi-fi and electric guitar amplifier circuits, the ECC83. This triode was used for the voltage amplification part of the amplifier designed.

 

4. A SIMPLE VALVE AMPLIFIER

 

  The function of a voltage amplifier is to produce a magnified version of the input signal as a voltage. Such a circuit is shown in Fig.5.

  The input voltage applied across the grid produces coresponding variations in the anode current as explained previously. The function of the load resistance R_L is to convert the anode current to a voltage according to Ohm's law. Since the voltage at the upper end of R_L is kept constant at V by the power supply the voltage at the other end must vary. This is the output voltage. A coupling capacitor is required to get rid of the dc component of this voltage and be left with the varying component.  

  The function of R_K is to provide the grid bias. The grid bias is the grid voltage when zero signal is applied. This must be negative as is explained in 3. Since there is a current through R_K this produces a voltage across it. This makes the cathode positive w.r.t. ground. Since there is negligible grid current flowing the volage across R_G is nearly zero. Hence the grid is at ground potential. Therefore the grid is negative w.r.t. cathode. The function of C_K is to keep the voltage across R_K constant.

 

5. THE LOAD LINE

 

  Since a load resistor is connected in the anode circuit this resticts the possible values of anode voltages as the anode current varies. This relationship must clearly be linear due to ohms law. At the same time the valve must behave as Fig.4 shows.

  The locus of points who satisfy both conditions is called the load line, each point corresponding to some anode voltage, anode current and grid voltage. This is a straight line because of Ohms law (the voltage and current in an ideal resistor vary in phase and proportionally), Fig.6. If two points are known the line can be completely specified. For convinience two extreme cases will be

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

chosen.

  Firsty imagine that there is so much anode current and therefore so much drop at RL that the anode voltage becomes zero. This means the whole supply voltage appears across R_L. The voltage across R_L is known and therefore the current. This gives as the point A in Fig.6.

  In order to find the second point we can think as follows.

Imagine that the grid voltage is made so negative that no anode current can flow. This means that the voltage across R_L is zero and therefore the whole suply voltage appears at the anode. This voltage and zero current defines point B in Fig. .

  Now the load line is known and therefore the anode voltage and current is known for any given grid voltage. For example for -1.5V grid voltage, they are defined by the intersecion of the valve curve at 1.5V and the load line. It can be seen that by varying the grid voltage the operating point moves along the load line.

  The quiescent operating point (the point defined by the grid bias) is usualy somewhere in the middle of the output characteristics. Then the applied signal makes the operating point move about it on the load line. For the particular case of applying a sinusoidal voltage to the amplifier Fig.7 shows how we can predict what the corresponding anode voltage and anode current will be.

 

6. HARMONIC DISTORTION

 

  In Fig.7 we can see that allthough the input voltage (grid voltage) is sinusoidal, the output voltage (anode voltage) is not. The amount of distortion decreases as we can see by decreasing the amplitude of the input signal. This harmonic distortion is characteristic of all active devices. It arises from the fact that the characteristics do not consist of parallel and equidistant lines. Valves espetialy triode ones tend to be better in this respect than tetrode and pentodes or solid state devices. The characteristics of ECC83 for example as can be seen from Fig.4 are not far from parallel and equidistant. This makes triodes in general more suitable if not the only suitable devices for low or zero negative feedback designs. A drawback for many people is that they are more inefficiant compered to other devices.

  Another observation that we can be made from Fig.6 is that the amount of harmonic distortion depends on the slope of the load line. (The slope depends on the value of R_L that is used). By decreasing R_L the load line tends to become perpendicular to the X axis and the points of intersection with the characteristics become less equidistant and harmonic distorion increases. Thus a relatively large value of R_L  is used. For an ECC83 it can be typically 100KOhms. Note that the output of such a voltage amplifier must be connected to high input impedance load so that the overall load resistance is not reduced.

 

7. THE OUTPUT TRANSFORMER

 

  In an output stage of a power amplifier it may be said that the  load which is driven must be of the same resistance as the anode resistance of the valve in order to have maximum power transfer according to the maximum power transfer theorem. (The anode resistance of a valve is defined as the first derivative of the Va w.r.t. Ia keeping Vg constant). This is a conflicting requirment since by making the load impedance larger we can reduce harmonic distortion. A good rule of thumb is to chose a load resistance two or three times the anode resistance. A typical value of anode impedance is 1000 Ohms. Therefore a load of 3000 Ohms will do.

  Now how about the typical 8 Ohms magnitude of impedance of a loudspeaker that we want to drive. The valve wants to see 3000 Ohms and we want to connect it to 8 Ohms!. This difficulty is overcomed by the use of a transformer called the output transformer.

  A transformer can for example step up a voltage. At the same time the current will have to be steped down to keep the product of voltage and current constant according to the principle of conservation of energy. Therefore a transformer also transforms impedance.

  In the tranformer of Fig.? a resistor R can be seen connected to the secondary. The presence of R makes the primary (at mid frequencies) behave as if it were a resistance of value say R_R. Using the relations

1

and

2

the reflected resistance can be found as follows

3                             4

 

8. CLASSES OF OPERATION []

 

  The degree of nonlinearity occuring in the operation of the valve serves as one basis for classification of amplifiers. Definitions of some of the standard classes are.

  A Class A amplifier is an amplifier in which the grid-bias and alternating grid voltages are such that anode current in a specific tube flows at all times.

  A Class B amplifier is an amplifier in which the grid-bias is approximately equal to the cut-off value so that the anode current in a specific tube flows foe approximately one-half of each cycle when an alternating grid voltage is applied.

  A Class AB amplifier is an amplifier in which the grid-bias and alternating grid voltages are such that plate current in a specific tube flows for appreciably more than half but less than the entire cycle.

  A Class A amplifier has much less power efficiancy than the other classes but it has the advantage of being much more linear since it operates in the linear region of the valve characteristics.

 

9. PHILOSOPHY OF DESIGN

 

  From the introduction to this chapter there are some reasons given about why the design chosen was a single ended triode no negative feedback. Let us now look at it from a different viewpoint. This may give a clue to one of the questions that we want to answer, why this kind of amplifiers turn out to sound so impresive.

  The requirement for a good quality amplifier can be that it must interfier as less as possible with the music. Provided that the source gives a good quality signal the amplifier must amplify it as well as degrade it as less as posible.

  Let as digress a bit to say that the most important element in a sound reproducing system is now cosidered to be the sourse. This was probably first shown by Linn Products Ltd. which designed and produced in the seventies the legendary Linn Sondek turntable still now considered a reference turntable. According to this company the sequence of importance in the audio chain is turntable, tonearm, cartridge, amplifier, loudspeakers. (Note this has the same direction as signal flow). A reason why this seems reasonable is that if the source gives a bad quality signal then the hypothetical perfect amplifier will just amplify it without changing it to give an amplified bad quality signal.

  Concerning music signals now some people feel that the less prossecing is carried on a music signal the worst it would be for it []. A classic example is the grafic equaliser. By inserting one in an audio chain one can really see how the quality is degrated. (This includes unnaturalness, hiss, painfull and agresive midrange and not only etc). What to do is obvius. Most serius manufacturers nowdays take a step further and elliminate bass and treble controls.

  The same is felt about the quantity of active devices and other components. Imagine the poor signal having to pass through miriads of transistors, swuthes, filters, converters, digital signal processing, compresors, connectors etc. This is unfortunately done in most modern studio recording consoles (few feet long full of op amps) and the results are as one might expect not that good.         Turning back to listening to analogue recordings pressed on records made in the sixties one cannot believe how brilliant they sound. The sound (I have also personal experience to this through records from an older person) is full of life, dynamic and there is space between instruments. Another reason is the quality of the equipment used. Examples are Westrex cutters, tubed Neumann condenser microphones etc.

  Conserning the design it was felt that the circuit should use as few components as possible. The least amount of active devices, resistors, capacitors etc. since each would anavoidably degrade the signal. For the same reasons good quality components would have to be used. For example pollypropyline coupling capacitors, suffisiently linear active devices (this sugests triodes), good quality bypass and supply smoothing capacitors since they supply the signal current etc. In fact all components must be of good quality. Some people say that even a better caoacitor can make a lot of difference, but even if this is not so, all the components being better quality will obviusly do make a lot of difference.

  Going to other deteils, no needless switches must be used, same with potentiometers and of course no tone controls. The cabling must be as short as possible ,sensibly laid following sorter distanses and trying to avoid electromagnetic coupling, may introduce hum, frequency restriction due to creating unwanted capacitances and inductances. These aspects will be further discused later.

  Conserning the circuit itself it must be as simple as possible. Thinking about the type of operation there are the following allternatives. Class A single ended, class A push-pull [], class AB (push-pull), class B (push-pull). Class B is most efficient in terms of power input output, but the active device is driven to highly non linear regions (since the device is driven to cut-off). Therefore class B is excluded, some for AB. Class A is left.

  Using class A there will be less output power but this would not matter if the main concern is the sound quality. Apparent loudness can be greater if valves are used because of their gentle overload characteristics (types of harmonics produced) [][]. This will be discused later.

  Class A produces a lot of heat (low efficiency) and transistors are not vary happy with this. Also their characteristics change with temperature. Valve characteristics are not dependent on temperature of course. Also valves are linear enough without feedback unlike solid state devices.

  The desision is now between valves in class A push pull or single ended. From types of valves pentodes, tetrodes we choose triodes which are more linear and produce predominantly second harmonic distortion rather than third etc. Another reason is the rellatively low anode resistance of the triode which makes possible more damping of the loudspeaker self oscilations and hence better control of the loudspeaker.

  Triodes in class A push-pull or single ended ?. Push-pull took the place of single ended design about sixty years ago. The main reason for this seems to be that it almost cancels the d.c. magnetising current in the transformer that can saturate the core in the output transformer see Fig.?. This makes the design of such a transformer less difficult. The amplifier is also less sensitive to produsing hum from the ripple in the power supply due to cancelation again effect in the output transformer. From simillar arguments it can be shown that the even harmonics cansel but odd ones do not.

  This means that that the second harmonic produced by the valve will be reduced but the third will be left untouched. Third harmonic can be heard more easily while second is difficult to identify . Its presence may even partially mask the third harmonic and subjectively the whole effect should sound more natural. Also if we are to have distortion better it be mostly second since it is most musically related.

  Another disadvantage of push-pull is that it needs perfectly matched valves and circuits and is more complicated. It needs more active devices to do the phase spliting and that sugests further signal manipulation.

  The design chosen will be a single ended no negative feedback triode design. It consumes a lot of power, gives out heat and the rest electrical energy to the loudspeaker, needs a very good power supply, difficult to design output transformer, but it has the following sonic advantages which are rellevent to sonic quality. It uses one of the most linear devices ever made, the triode vacuum tube. The power triode operates in its linear region since class A and therefore produces low harmonic distortion. The whole signal is handeled by it the power triode, since single ended. Very few components are used, say a voltage amplifying triode driving the power triode, and the latter the loudspeaker. The power drawn from the power supply is constant now matter what the output signal level []. The harmonic distortion produced is low order and predoninantly second which is musically related. No negative feedback needs to be used since harmonic distortion is low and if a good wideband output transformer is used. Therefore no transient intermodulation distortion and time delay effects will be present. Also the non linearity will come gradually at musical fortissimos avoiding headaches and rushing to turn down the volume. The subjective impression because of the production of small order harmonics will be of a big sound (imitation of the non linearity of the ear at loud sounds) and therefore the apparent loudness and dynamic range will be greater.

 

10. THE DESIGN

 

  As discused the design will be a triode single ended no negative feedback one. The structure of the circuit will be a follows. The source will be driving a voltage amplifing section based on triodes or if possible just one triode of sufficient amplification factor. Then the output of this stage will be connected to the grid of the power triode. Finally the output at the anode circuit of the power triode will be coupled to the loudspeaker via the output transformer.

  Every component in the amplifier will have to be of good quality.

One of the most important ones in valve amplifiers determining the sound quality, most expensive and difficult to design is the output transformer. This is because the music signal will pass trough it. As a first comment it must be mentioned that it will not have to pass fifty or sixty Hertz as a power tranformer. It must do the same with the same attenuation in voltage for any frequancy from 20Hz to 20KHz and possibly more. The turns ratio formula apllies approximately only to mid frequencies. At low and high frequencies inductive and capacitive reactances that exist in a tranformer become important.

  In an advertisment in Hi-Fi News and Record Review it was found that that Audio Note was selling output transformers for single ended triodes, namely the 300B and the 211. These are the directly heated triodes that have been forgoten for decades and are now used in new designs throuhout the world. Their manufacture has started again by PM components with Shuguang in China.

  For this design the 300B was chosen allthough it gives less output power in class A than the 211. The reason was that it is a triode specially designed for audio use in class A and that it does not need the lethal 1.25KV the 211 needs allthough the 500V it needs is still lethal.

  The 300B was designed at the Bell Telephone Laboratories before 1939 and made by the Western Electric company and others. The 300B obtained is one made by Centron in U.S..

  Bearing in mind that the output tranformer that could be obtained would transform 8 Ohms to 2.5 KOhms the operating conditions shown by the arrow in Fig.? were the ones chosen. Allthough other 

operating conditions can be seen in Fig.? for load resistance of

2.5 KOhms ,these we chosen because of the reasonable amount of maximum output power, namely 12.5 Watts.

 These operating conditions are :

 

(V_b,V_g,I_a,R_L)=(400V,-84V,80mA,2500Ω)

 

 By examining the data sheets of the WE 300B given in appendix A it can be seen that a peak value of sinusoidal voltage equal to the grid bias is needed to get the indicated power output and harmonic levels. In our case this is 84 Volts or 80Volts if d.c current is used in the fillament cathode. Therefore the gain of the voltage amplifier for a sensitivity of 1 V r.m.s. must be

 

3                  [1]

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

looking at the data sheets of the ECC83 reccomended operating conditions for a gain of 59  and 59 V maximum output voltage with a prescribed amount of distortion can be found. Thus only  two triodes may be used, the ECC83 and the 300B, and the circuit can be as shown in Fig.?. The input signal is applied to the grid of the ECC83. As can be seen then the amplified voltage is applied to the

grid of the 300B. The loudspeaker is then connected to the secondary of the output transformer.

  Another function of this transformer apart from impedance matching is that it blocks completely any d.c. voltage which would destroy the loudspeaker. This is because a steady anode current produces a steady magnetic field in the transformer core and therefore zero induced emf in the secondary. On the other hand when non zero signal exists the changing anode current produces a changing magnetic field. This changing magnetic field produces a changing electric field (Maxwell's equation 3) and electrons are able to aquire a drift velocity in the transformer secondary ,loudspeaker cable and loudspeaker coil. Those in the louspeaker coil will expereance a force becaouse of the magnetic field due to the loudspeaker permanent magnet nearby. The sum of all forces on the electrons in this coil will be the force that displaces the loudspeaker membrane attached to the coil thus creating sound.

  Coming back to design in Fig.? are shown the voltages and currents that must exist in the circuit according to the requirements and the manufacturers reccomended operating conditions. It is now a matter of appling Ohm's law to calculate the values of the resistors neeeded.

  Begining with R_K3 the current trhough it is known. This is 80mA since the grid current is negligible. The voltage across it is 80V. Therefore its value must be

 

4

 

  The power it must be able to dissipate is

 

5

 

  Thus a 1KΩ of 8W or more resistor (so that it does get very hot) can be used. Using similar arguments R_L2 and R_K3 can be calculated. R_G2 is equal to the input impedance of the amplifier. This is because the resistance between grid and ground is extremely high since the grid is not to become positive and therefore does not attract any electrons.

  R_G2 can be adjusted according to preference. A value of 100KΩ will be used here for an input impedance of 100KΩ. Of course if the amplifier is overloaded by driving the grid positive w.r.t. cathode the input impedance will increasingly decrease.

  R_G3 is recommended by the manufacturer of the ECC83 to be 330KΩ.

  Therefore the resistors must be are as follows.

 

R_L2=100KΩ              0.5W

R_K2=1.5KΩ              0.5W

R_G2=100KΩ              0.5W

R_K3=1KΩ                 >8W

R_G3=330KΩ              0.5W

 

  Next the values of capacitors will be calculated. The quiescent value grid voltage must be kept steady in order to avoid negative feedback and reduction of gain. For example the 2.1V across R_K2 must be kept constant. This is why C_K2 is used since capacitors tend to keep the voltage across them constant.

  Another way of looking at this is to say that is order that the

voltage across R_K2 be constant at the presence of signal, the current trough it has to be constant. Therefore since the anode

current is equal to the cathode current the varying component of

the anode current must pass through C_K2 and not R_K2.

  A simple rule to ensure this happening at all frequencies of interest, otherwise the gain will be reduced, is to say that at the lowest frequency of interest the reactance of the bypass capacitor must be say ten times less than the cathode resistor []. Therefore

 

6

7                             8

 

and therefore

 

9

 

The nearest coomercial value available is 330μF. The voltage rating must be as can be seen from the circuit diagram grater than 80V. The other bypass capacitor, C_K2 can be calculated similatly by evaluating the above function at (R_K2,5Hz).

  C_C2 will also limit the low frequency response of the amplifier At the extreme case of 0Hz it wont less pass anything and this is what it is used for. In general it is charged will 260V and this cansels will the 260V plate voltage to provide zero grid voltage at the power valve. At the existance of signal the plate voltage will change so will the other end of this capacitor since it will tend to keep the voltage across it constant. At lower frequencies the capacitor voltage will begin to change more since more time will be given.

  The 3db frequency f₁ contributed by this capacitor can be shown [?] to be given by

 

10

 

where R_eq=r_p║R_L+R_g and

 

  Substituting the numbers in this equation C_C2 can be found to be near 0.1μF. The type of capacitor that will be used will be a polypropylene one becauase it can stand high voltages 1KV or so and has very good high frequency properties. Note that at amplifier turn on the voltage across C_C2 will be greater because of lack of anode current.

  Therefore the capacitors with their voltage ratings determined from the circuit diagram Fig.? must be.

 

C_K2=220μF                25V     axial elctr.

C_K3=340μF               >100V    axial elctr.

C_C2=0.1μF               1000V    polypropylene

 

11. THE POWER SUPPLY

 

  What is left now is to supply the right voltages to right parts of the circuit. For the heaters 6.3V a.c or d.c. at 150mA for the ECC83 and 5V a.c or d.c. at  1.2A as recommended by the manufacturers. Then 495V and 400V d.c. must be supplied at the top end of the primary of the output transformer and the top end of R_L2 as shown in Fig.?.

  A 350V 0 350V power tranformer with two solid state diodes and a large electrolytic capacitor to keep the voltage near the peak value of a 350V r.m.s. sinusoidal voltage will provide nearly 495V since

 

11

 

  Such a tranformer could be available together will the right heater voltages.

  The power supply circuit then takes the form of Fig.?. The capacitors are used in series since electrolytics will voltage rating greater than 350V could not be found. R₃ is used to drop the voltage to 400V and then the combination of C₃ and C₄ to keep it constant. These componets together also have the advantage to form a R-C low pass filter which is what is needed since the 100Hz and harmonics must be surpresed. They also prevent variations in the 500V due to current drawn by the power valve to create variations in the 400V rail. This is similar to say that they prevent signal flowing back in the form of positive feedback which may turn the amplifier to an oscilator [?].

  Resistors R₁,...,R₄ are used for two purposes. One is to provide a path for the capacitors to discharge when the amplifier is switched off. The other is to ensure that the voltage is equaly shared between the two capacitors. This is in case the capacitors are not of the same value which can be the case with electrolytics because of their large value tollerance. These resistors were chosen so that the current through them is about ten times less than the average current the capacitors supply to the circuit.

  R₃ can be found since the voltage across it must be 495V-100V=95V and the current through it the sum of the quiescent anode current of the ECC83 and the maximum bleeding current, i.e

1.4mA+0.13mA=1.53mA.

  By examining the circuit of the power supply it can be seen that the current to the circuit is supplied by the smoothing capacitors for most part of a 50Hz cycle. The diodes are conducting only for a very short time. This happens because the capacitors keep their anode near less than 495V. They conduct only when their cathode becomes greater than this (forward biased). At the negative going cycle the voltage accross them will go up to nearly 1KV which is the voltage across the capacitors plus the peak value of the sinusoidal voltage across the secondary of the transformer. Therefore they must be able to withstand this reverse voltage and be able to provide high pulses of current so that they can put back charge to the capacitors in the short time they are conducting. The types that were chosen are the 1N4007 used in series.

  The smoothing capacitors C₁,C₂ supply for most of the time the anode current to the power triode. This is the quiescent 80mA plus the varying component which is the music signal. This shows that these capacitors must be of good quality. They must be able to supply large currents very quickly, in case for example of reproduction of percusive instruments. Therefore they must have low series inductance and good high frequency properties. This is not the case with elecrolytic capacitors which also become quite nonlinear at high frequencies but the ones chosen were specified for low series inductance. What can also be done if elecrolytic capacitors must be used because of cost constraints is to bypass them with a smaller value of polypropylene capacitor which would supply the high frequency components of the current needed.

 

12. THE CONSTRUCTION [][]

 

  Before deciding on the layout of the components the following had to be taken into account. The two transformers would have to be as far as possible from each other to avoid electromagnetic coupling. the order of the components would have to follow the signal path. The input should be far from the output. The components should neither be very close together to avoid them interfier with each other and to avoid overheating, nor to far to avoid long connections with greater inductance and greater capacitance to ground. Signal carrying conductors should not be to close to the chassis to reduce this capacitance which tends to limit the high frequency response (one plate of this unwanted capacitance is the conductor and the other the chassis). The ECC83 voltage amplifier because of its high output impedance ahould be as close to the 300B to avoid high frequency loss due to the higher capacitance of a longer cable as explained below.

  When an output of a circuit which can be represented by a ideal voltage source in series with an output impedance is connected with a coaxial cable to the input of an other stage the high frequency components are attenuated. This is because what has been made is a low pass filter. The resistance is the output impedance and the capacitance is the cable itself since it consists of two conductors close together. In order to look at it in a bit more depth it must be said that the voltage across the coaxial cable can not change quikly because capacitors tend to keep the voltage across them constant. The existance of an output impedance further slows down the process since electrons find it more difficult to flow and charge the cable. The smaller the capacitance and output impedance the higher the cut off frequency. The effect of long cable can be heard by connecting a turntable cartridge to a cartridge amplifier using a five or so meter of coaxial cable. This has a great effect in the reduction of the upper harmonics of instruments making the sound dull.

  Back to the layout it was chosen to be as follows. The chassis would be of dimensions 350mm x 265mm x 55mm. The big components (valves, smoothing capacitors, transformers) would be placed on the top. The componnents are layed out in the same order as energy flow. On left and top the mains, mains switch, mains transformer, under it in the chassis the diodes, next to the transformer the smoothing capacitors, then the connection of the h.t. to the output transformer primary, the output transformer, the other connection of the primary to the anode of the power tube etc i.e. the same order the componnents are connected together on the circuit diagram. The above requirements are simultaniously satisfied.

  The chassis was made by bending a sheet of 2mm thickness aluminium. The corners were welded and then filed. Then it was desided on paper where the holes should be made and cut. After the holes were made the chasis was processed with wet and dry, and finaly was sprayed black which made it look more proffesional. After the big components were mounted the various connections were performed on the inside of the chassis.

  The amplifier outside and inside is shown in Fig. and Fig.  respectively.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

                          CHAPTER III

 

 

Subjective Evaluation

 

  The power amplifier was used to drive a KEF Carlton III loudspeaker at home and a smaller Sony from a midi system in the laboratory. The sourses used in the lab were a small Sony CD player, and a relatively expensive Sony Walkman. At home the Linn Sondek LP12 turntable was used with a Linn Basik tonearm ,Linn K9 cartridge. As peamplifier the Naim Nait 2 integrated amplifier was used connected to a Metz tape recorder used as a voltage amplifier to increase the output. It has to be realised that none of these compinations is acceptable for a quality source. Exception is the turntable and preamp combination but the signal was found to be degrated as expected passing through the tape recorder. It should be much better using directly the Naim preamplifier out but this needed opening it and trying to find where this output is etc ,when time was not available. Also a good quality electric quitar was used.

 

1. SOME FIRST COMMENTS ABOUT THE SOUND QUALITY

 

  Unfortunately for time reasons only one amplifier was constructed so all listening tests were performed in mono and therefore no information about stereo imaging can yet be available.

  First for safety reasons (avoiding blowing out expensive equipment) the amplifier was tryed with a tape played from the Metz tape player and the Sony loudspeaker. The result was neither good nor bad.

  Almost immedeatly the KEF loudspeaker having a better and more extended low frequency response and good subjective mid performance was connected. The improvement was as expected. Then the turntable combination was connected. There was a vast improvement in the sound quality. The best area was the mid range especially at vocals were people commented that it is close to having the singer in the room. The bass on the other hand was not very deep compared to listening to the Naim transistor amp but this may be because the speaker the Naim was driving at direct comparisons was the one in the corner of the room.

  The amplifier sounded dynamic and quite loud. Very high volumes could not be obtained partly because the amplifier was not driven properly, the maximum output of the tape recorder used as voltage amplifier was less than 1V and it most certainly started clipping before the amplifier.

  Compared to the less expensive Naim amplifier the vocals were found more natural, but the sound of the Naim appeared more clean. This was again found to be because of the tape recorder in the valve amp chain. When both amplifiers were fed by it the valve one souded less distorted. Listening to records the high frequency responce of the valve amp did not seem very extended. Here again it may be because of bandwidth restriction due to the tape recorder voltage amplifier so it cannot be desided yet whether it is the power amplifiers fault or not.

  Going to CD the sound was less natural and tyring. The same and even wosrst at earlier times was observed using the Naim amp with a CD sourse. This agrees with the reputation of CD players for harsh sound especialy in the high frequencies. Apart from that the sound was impresive at times, dynamics of piano notes and other percusive instruments were apparent and at some times frigtening. Although on mono in a good recording a listener commented that he could fell that the piece was played in a big hall. The presence of instruments could be some times felt in the lab. But at long term listening the sound was uninteresting, unnatural and fatiguing.

  The Walkman used allthough sounding more natural than the CD player has a dissadvantage that whenever it has been connnected to amplifiers a very tyring form of distortion exists. This distortion is not so apparent when connected to headphones. It seems to be either crossover or intermodulation distortion. Connecting the valve aplifier was no exception and allthough the sound on good recorded tapes was quite natural and enjoing on strings, it was imposible to listen for a long time due to headache. 

   Probably the most impresive results came from the electric guitar. Here it was almost directly connected to the power amp. In the begining the mic input of the tape recorder was used. The sound was getting distorted very easily and it was not very impressive. Then an high gain ECC83 voltage amplifier substituted the tape recorder and the improvement was great. The proffesional guitarist playing his guitar commented that it was the best amplifier he has ever connected his guitar to. Playing some complicated Jazz chords every single note could be distinguised when all were sounded together. He also commented that each note sounded decaying much slower than usual.

  Equaly impressive results were observed recently when the amplifier was connected at the headphone output of a good quality Roberts radio. The vocals had a great pressence, the sound was open and had a certain body.

  Summerising it may be said that the amplifier was exposing both strong and week points of the sourses used which is what a good quality amplifier should do. But it can not be yet decided in many cases whether the bad critisisms are due to the sources or the amplifier itself. In order to do this a different source combination will be used. The power supply was not proved to be a success due to some hum apparent and probably degrating the quality of the sound of the amplifier. Solutions to these problems together with other further work will be discused in chapter V.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

                         CHAPTER IV

 

 

Objective Evaluation

 

1. THE OUTPUT TRANSFORMER

 

  Before the amplifier was built it was found worth doing some measurments to the output transformer that was obtained since much of the final sound quality in an amplifier depends on it. This was also an attempt to understand how an output transformer limits the low and high frequency response of an amplifier.

  For small signals a valve can be shown [] to behave at a first approximation as an ideal voltage generator in series with its plate resistance. The transformer was then driven by a signal generator of a small output impedance typicaly 60Ω in series with a resistor to increase it. Voltage waveforms were observed by an oscilloscope across the primary and across R_L in the secondary.

  First a bad quality transformer was tested. Using this transformer in an existing amplifier and listening through a loudspeaker gave the conclusion  that the bass was lost.

  By applying sinusoidal voltages of different frequencies the voltage across R_L was found to decrease at bellow 300Hz as was expected and above 20KHz. What was found interesting was that at low frequencies by decreasing the frequency the voltage across the primary was decreasing. The voltage across the genetator was the same. This gives the clue that the signal is lost as voltage drop in R. The explanation for this effect is that the primary of the transformer behaves also as an inductor which is what would one expect of a coil wound on a core. At mid frequencies this primary inductance L_p reacts more and a small current is drawn from the sourse. By decreasing the frequency ,the lower it is the lower the reactance of L_p and the more current it draws. This leads to more voltage drop in R and therefore less voltage across the primary. The consequence of that is that less voltage appears in the secondary. Therefore for good low frequency responce the primary inductance must be high and the valve plate resistance low.

  Unfortunately this is a conflicting requirement with wanting to extend the high frequency response. A large primary inductance may require more turns on the primary and this will increase the flux leakage between primary and secondary. Flux leaking implies that those turns whose flux does not arrive in the secondary are as if there are wound on a different core ouside the transformer. This implies that a so called leakage inductance exists in series with the primary. This inductance restricts the high frequency response since since its reactance and therefore voltage drop across it increase with frequency. One way of overcoming this is to wind some primary turns then some secondary then again some primary and so on to reduce the leaking flux. This is one of the reasons why good transformers are difficult to make.

  The tranformer to be used in the amplifier (much bigger and heavier) was then under a simillar test and the frequency response

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

measured is shown in Fig.2. Note the extended low frequency response.

 

2. CHECKING VOLTAGES

 

  Since inside the amplifier can be found voltages that can be lethal, great care plus common sense is needed when operating the amplifier with the back cover removed or making neasurments. The obvius thing to do is be not to touch anything when it is connected to the mains. Even when disconected from the main one must wait a few minutes for the capacitors to be discharged before touching the circuit, but it is best to check with a voltmeter since the bleeder resistors might have been disconected.

  Measurments connecting probes for example can be done before suppling the amplifier with the mains. If for good reason a connection must be made while voltages are present it must be done with one hand and the other hand must be isolated say in the back pocket. If for example the other hand touches a conductor at 0 potential say the chassis and the hand connecting the probe a 500V rail the circuit will be closed ,the person involved will experience 500V and worst there is path for the current trough the heart. For further precausions and first aids reffer to [ ].

  Using the precausions described above the fist measurment were checking the anode voltages, bias voltages, heater voltages etc. They were found in good agreement with what was expected. One of the most close agreements was the 300B bias voltage measured to be -79V and varying a little when music was played at high volumes as was expected.

  Input sensitivity and power output at begining of overload were found as expected. The power output was found by measuring the voltage across an 8Ω resistor.

 

3. MEASURED FREQUENCY AND PHASE RESPONSE OF THE AMPLIFIER

 

  The frequency response or output over input as a function of input frequency (magnitude of the Fourier transfer function) of the amplifier was measured by connecting the input of the amplifier to the output of the siganal generator and one channel of the oscilloscope. The output was terminated to a non inductive resistor of 8Ω and the other channel of the oscilloscope. In Fig.3 the frequency response can be seen. The Y axis is the ratio of output to input expresed in db referenced at 1KHz.

  The phase response i.e. phase difference between output and input as a function of input frequency (argument of the Fourier transfer function) is shown in Fig.4. It was measured by setting the oscilloscope to x-y and measuring the two parameters of the parens created [ ]. It can be shown that the input and output signals do not have to be of the same amplitude and this made the measurment quite easy since the input signal and x or y sensitivity could be varried to convinience.

 

4. TIME DELAY

 

  The opposite of the derivative of the phase response is equal to the group delay. In Fig.4 it can be seen that the smallest group delay exists at the mid frequencies and the highest at low

frequencies, since the frequency scale is logarithmic. By evaluating the slope of the phase response near 20Hz and 20KHz group delays of 2ms and 4μs are roughly obtained.

  Gated sinewaves of different frequencies as test signals were produced by means of a mechanical switch. Their time delay through the amplifier did not seem to be rellated to the group delay though. The time delay seemed to be much small to be measured with the equipment available. Phase nonlinearities especially at the low frequency end seemed to affect the way the output decayed. Also the first cycle at the output was found to have less amplitude than the rest Fig. . This is the transient before the phase of the output leads the input.

  By appling real music from a tape the input an output waveforms were freezed on a storage oscilloscope. They looked like Fig. . The output being a little different from the input, the edges were found to be a little rounded. Δt was measured to be near 0.02ms.

 

5. OBSERVED LINEARITY

 

  The transfer characteristic was obtained using x-y plots at a sinusoidal input siganl of 700Hz. It is shown in Fig. . Note the gentle overload characterisrics.

  The same was done with music input in the amplifier. This maintained a blurred image of the transfer characteristic.

 

6. SOME COMPARISON WITH AN OPERATIONAL AMPLIFIER

 

  Some simillar tests were performed on a 741 with a gain of 10.  From what was done in the time spent similarities and two main differences were found.

  A similarity was when comparing input and output waveforms of music in the time domain as in 4. both amplifiers maintained the escential characterists of the waveform allthough they were not idectical to each other.

  The transfer characteristic using again x-y plot at 700Hz was as shown in Fig. .

  Also at a music signal the x-y plot maintained a blured image of the transfer caracteristic. The diferemnt overload characteristic to the valve amplifier was apparent at music peaks.

  At steady state the op amp behaved quite differenty. By increasing the input frequency at a reasonable level above about 20KHz the output waveform were triangular instead of sinusoidal. By increasing the frequency the waveform amplitude was reduced. Reducing the input signal could make the output sinusoidal but this was less effective  the higher the frequency. Also the maximum output at overload was 12V (the supply voltage) only at frequencies up to a few KHz. Then the the higher the input frequencie the lower was the output voltage at which clipping begun.

  The valve amplifier on the other hand never produced triagular waveforms no matter how high the frequency and the input amplitude. Before overload the output waveform was sinusoidal at all frequencies that could be produced (up to a few MHz). Of cource the amplitude was starting to decrease beyond 20KHz. The amplifier could produce the same maximum output at all these frequencies, clipping begining at about 12V.

  The clipping of the op amp  was very abrupt. The 12V, -12V rails could not be exceded and clipping started at 12V amplitude of output. Therefore at mid frequencies say 1KHz a bit of overload resulted in a flat top and bottom. The valve amplifier had a gradual overload characteristic. Top and bottom were not becoming flat. Once overload started at about 12V output amplitude, by increasing the input voltage the output could be increased at a slower rate (this can be seem from the transfer characteristic). With the input voltage available (up to 8V) outputs near 16V could be obtained.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

                         CHAPTER V

 

1. PROBLEMS ENCOUNTERED

 

  One of the problems accociated with the construction of the amplifier was finding component suppliers and making sure components arrive on time. In some cases especially in transforms it took a couple of months to get them from the time the order was made.

  It was generaly found that the more thought was given to the design the easier it was to apply these thoughts in practice. Areas that were not much thought (e.g.power supply) proved to be not very sucessful and in the end more effort consuming to improve them. Also it turned out as expected that the areas in the circuit that were more undertood were much easier to design and construct. The same applies to the construction of the chassis. The more it was thought were and how each component would be placed where exactly holes should be drilled the less problems were encountered later.

  The amplifier when connected to the mains and swiched on did not work immedeately and this was not because of delays due to valve warm up time. It seemed that the fuse was blown up due to the transient current to charge the capacitors. After trial and error a 3A fuse was used succesfuly. A surge fuse may be more suitable. Then the high tension across the smoothing capacitors was confirmed to be 500V with the valves unpluged . Then the valves were pluged in and nearly two seconds after swiching on there was sound coming from the loudspeaker. Unfortunately after a few minutes the sound started to get distorted and it was observed that the heater of the 300B was getting less bright. It turned out that this was because the regulator suppling the 5V was overheated because it was not attached to a chassis and the average voltage across it was geting less and less the more hot it became. Also the voltage across the heater was always even a bit after turn on less than 5V. It was the decided to use the 5V a.c. power transformer secondary to heat the cathode. To avoid ecsess hum the cathode return was connected to a 10Ω potensiometer [].

  Hum was apparent in the loudspeaker when the room was quiet and this was with or without the regulator. This may possibly affect the quality of sound.

  Possibly the most annoying problem was that no source was driving the amplifier properly. Not much thought was given to this before the amplifier was made. The turntable the best source of reproduced music from the sources available was amplified by a good preamplifier, but then there had to be used a not so good quality voltage amplifier and it still turned out that apart from degrating the signal it could not supply the maximun 1V needed.

  It also is unfortunate having to supply an amplifier working in class with possibly op amps in class B with crossover distortion, and a lot of feedback (the external voltage amplifier, the CD player and the walkman.

 

 

2. FURTHER WORK CONCERNING THE AMPLIFIER ITS EVALUATION AND IMPROVEMENT

 

  Solution to this problem seems to be the design of a triode cartridge preamplifier (in class A) so that all that will exist between the signal from the cartridge will be no more that three triode voltage amplifiers. The output can easily be a few volts including a passive RIAA filter. No negative feedback will be needed for the additional reason that the signal levels involved will be small (up to a few hundred mV) and therefore the operation of the valves will be in the linear region. The RIAA filter since it is the inverse filter of what is used in LP recording will also correct the changes in phase response that arrise. Then direct to disc records (no tape recorder involved) usualy cut through valve amplifiers can be used. This chain could then make a refference source for studing the effect of making changes in the amplifier and provide at the same time a good system for listening to music or evaluating the effect of these changes.

  As an easier solution which may be tried before the above is to modify an allready exists one stage only ECC83 voltage amplifier to an ECC82 which has a nearly ten times less plate impedance and therefore can drive properly the capacitance of the connecting cable. It can also provive a lower gain of 10 which is what is needed to provide 1.5V or so when driven from the tape output of the Naim Nait 2 cartridge preamplifier.

  Allthough the components were ordered in double quantity only one (monoblock) power amplifier was constructed in the time that was used for this project. The other identical one is hoped to be finised soon so that listening tests can be made is stereo and the stereo image (position of instruments in the soundstage) can be evaluated.

  The money spent to buy the large smoothing capacitors, and the hum present, made the power supply a failure. Allthough larger capacitors tend to make the bass deeper due to the fact that at low frequencies the voltage across them changes less, this should not be overdone since series inductance may increase plus other effects like impedance mismaching with the power transformer and rectifier section may exist.

  The ripple of the supply voltage must be as small as possible which is a dissadvantage of single ended amplifiers in order to avoid hum. Thinking about it now the most suitable filter may well be an inductor capacitor filter or a π filter. This also has the advantage that capacitors of smaller value and better quality can be used, paper in oil ones being considered some of the best.

  Also when electrolyticss are used for example as by pass capacitors since they have to carry the signal they will be bypassed by smaller valued polypropylene or polystyrene capacitors which have very good high frequency properties.

 

3. SOME DISCUTION AND FURTHER WORK

 

  Since this project has started many more triode feedbackless amplifiers have appearded on the market. Reviewers agree that their sound is much different. Some of the differences they find is unbelievable holografic sound stage[][][][], openness, realism etc.

  With the amplifier described in this project it was found that the character of the sound was in this direction. Since the listening tests were for the time being in mono, not much can be said about the stereo image. Even in mono one could get an impression of the characteristcs of the place (size etc) where the recording was done. Also every instrument could be well distinguised from each other even if they were all played from one loudspeaker. This gave also the impression that feedback may mix the sounds together (of course further work needs to be done on that). The openess, directness and large dynamic range were apparent, especially in vocals which were very realistic.

  The amplifier gave the impression that its sound quality is very dependent on the quality of the source. This agees with comments from reviewers and users about same type amplifiers.  Considering the limitations due to sourcing problems and restriction to mono a quite large improvement in the sound quality is expected when the changes discribed in 2. are performed.

  The apparent loudness and dynamic range were large as expected. As an example the guitar player could not believe that this is a 10W amplifier. He commented that transistors amplifiers he has used at 10W just sounded very distorted. This can be explained [] due to the overload characterists of the amplifiers. Valves especially with no feedback like biological systems become non linear gradualy [] Fig.?. This approximates the way human hearing becomes non linear when listening to loud sounds. When the amplifier becomes non linear allthough the objective intensity may be low the brain interpretes this as if a loud sound really exists []. This explains also the large apparent dynamic range. Valves act as almost ideal signal compressors. When overload is beggining the ouput is still increased but less effectively. But the dynamics are not lost because the small overload sounds louder ,because it immitates the behaviur of hearing at loud souds, than an intensity meter would indicate. Transistors amplifiers on the other hand produce at once flat top and bottom at the output waveform which sounds very distorted and tyring because of the high harmonic content. One has to consider that in practice when listening to music the maximum level of the recording is not known. The piece may start quitly and then a big orchstra may begin at fortisimo. The effect of that may well be hard to imagine if clipping is abrupt. Even in more common situations the maximum level wil not be known and overloads frequently occur.

  By doing some measurments in the amplifier it was realised why vaves especially triodes overload softly. By looking at the characteristic of the ECC83 in Fig.? this can be seen. By moving along the load line in the possitive direction of increasing signal when the grid voltage is getting possitive grid current will flow and therefore there will be a voltage drop in the previews stage. This overload begins much before the plate voltage becomes zero. Similarly on the negative part of the input signal the operating point moves in the other direction. At a large input siganl the characteristic lines become progresively closer to each other and therefore it is made incrasingly difficult for to the anode voltage to become equal to the supply voltage.

  The above also serves as a good example to illustrate that measurments must be carefully interpreted. This applies certainly to distortion, frequency response and other measurments stady state or not. By studing what was written on the subject it was found that nowadays thre is even more controvercy on matters like whether the frequency response must be wider than 20Hz-20KHz, whether phase nonlinearity can make difference, limitations of steady state measurments since music is transient etc. Especially now with discussions about how different single ended no feedback amplifiers sound and the supperiority of records against cds many people say that measurments are irrelevent or that there do not exist so far many meaningfull ones.

  Such discussions demonstrate how extremely complicated humans are so that models even of perception are difficult to make and must be well critisized because they may be wrong. It also demostrates that different disciplines like physics engineering phycology arts and possibly others must not work independent of each other. After all they can all contribute to knowledge about nature which is endless.

  Coming back to sound reproduction it is now more believed that larger bandwidth is advantegeus. This may be because the bandwidth affects the transint response of the amplifier. A tuned circuit of say 1KHz reasonant frequency will not let pass undistorted a short duration 1KHz wave. This is becaouse the filter will ring, ie keep oscillating like a spring and mass excited for short time. This is equivalent to say that the small duration pulse has a greater bandwidth. The filter makes the bandwidth of the signal narrower and therfore the duration i.e. lengh in the time domain increases (ringing). Different bandwidths may make differnce in sound quality. Direct cut to disk records which have greater bandwidths that 20KHz can be used for experimentation and the bandwidth can be adjusted with a low pass filter. The ear mechanism being non linear sugestets that frequencies above 20Khz can be heard. If two such frequencies are present nonlinearity in the ear will produce difference frequencies which can be in the 20Hz-20KHz band and therfore heard.

  Also the frequency response will affect the phase response. If linear phase response makes a difference then then bandwiddth must be greater so that the phase is linear in the frequency range of interest. Allthough not believed so a few decades ago it has been shown [][] that phase non linearity effects can be heard in certain sounds but not yet in music and that further work is needed.

  Such tests lie on the thought that a distortionles system is one that has constanst frequency response and linear phase response (uniform time delay for all frequency band components). But it must be remembered that this aplies to linear systems. Fourier analysis does not apply to non linear omes because the principle of superposition does not hold by definition. Amplifiers are not strictly speaking linear. Transient intermodulation distortion exists in large negative feedback amplifiers []. Even if transient The steady state tests done on the operational amplifier showed triangular voltage output at high frequencies showing much demarcation from linearity.

 The amplifier designed employing no feedback, Therefor no possible influence to linearity exists because of the time delay feedback takes to arrive at the input. Also giving sinusoidal output at a much greater bandwidth threfore behaving closer to linearity than operational amplifiers may be used for phase distortion tests and may turn out to highlight diferences in phase response. Its phase response can be controled a passive all pass filter. The Quad electrostatic loupspeakers are very suitable since they are close to linear phase.

  Also the amplifier designed is very suitable for studing the sound effect of adding negative feedback. For this purpose another triode voltage amoplifier will be added which has already been designed and overall feeback applied.

  It wil also be interesting to see how the phase response (related to group delay) affects the delays including those in the amplifier that are to be studyied in connection with feedback not arriving in time to correct the input signal. It was found that delays in gated sinwaves were not much related to group delay or there may be a different relation. Further work must also be done on this aspect.

 

4. WHY THIS PROJECT WAS USEFULL

 

  The aim of this project was to design and built a single ended triode amplifier and studying various aspects in its subjective and objective performance including phase response effects and negative feedback effects. Due to time restriction not much work work was done on the last two.

  Nevertheless the project gave the oporunity to read what is known about phase response and negative feedback and think about these matters. By reading papers and articles it was found that what is known is much less than it is appears. For example the problem of dissagreement between measured and auditory performance is unsolved and especially now even more controversial.

  Experience and knowledge was gained while designing and building the amplifier in all the different stages and aspects. This gave the oporunity to compare what was expected (theroretical knowledge) and what happended in the real world. This improved the knowledge of the constructor in both theroretical and practical aspects and showed him that both are equaly important. The need of different discplines working together such as science and art was also seen.

  The performance of this single ended no negative feedback showed agreement with what reviewers of amplifiers say about this kind. This was especially true with the reproduction of the human voice which proved to be so realistic that could deceive .This was even at small levels were the amplifier may be assumed to be linear and therefore harmonics enhancing effects were not present. Bearing in mind limitations of the sourses used and mistakes in the design once they are solved, agreement with even more aspects of performance are expected.

  Now that the intial design and construction is completed it can be seen that such an amplifier is suitable for studying some aspects of steady state and transient measurments versus listening tests, negative feedback and phase linearity effects and other.

  It is true that if the amplifier is improved the pleasure derived from listening to recorded music through it will be grater. After all this is the aim of high quality music reproduction.