Welcome electron tube amplifier constructors - this page is presented for your information and guidance.

It is intended to provide helpful hints to save you inevitable pain and suffering in your quest for audio excellence and to help you find joy in your chosen pathway to audiophilic pleasure!!

Please note it is not intended for the novice constructor. Basic circuit theory and construction techniques are not attempted herein because it is assumed you already know that and are competent in both.

In any event, you will soon see I am not a theorist, so whatever is said is based upon a combination of the reported experiences of others, helpful guidance from my electronic guru friends, reference to published texts from people who are considered to be reliable technical experts, electron tube manufacturers' manuals and data sheets, and my own experiments, research and experiences.

I hope it is of help to you in designing and constructing the very best audio amplifier for your needs.


Copyright in all quoted works remains with their original owner, author and publisher, as applicable.

Please note that no warranty is expressed or implied - see footnote notice.

Intellectual property in the applied engineering concepts expressed in this paper remains exclusively with the author Dennis R. Grimwood.

The whole or part thereof of this paper and/or the designs and design concepts expressed therein may be reproduced for personal use - but not for commercial gain or reward without the express written permission of the author.

All rights reserved.



Do not attempt to design and/or construct a vacuum tube audio amplifier unless you suitably skilled, qualified and/or experienced.

The Author makes no claim whatsoever as to the validity or accuracy or otherwise of any statement, information or opinion contained in these pages and no liability will be accepted for any error or omission of any kind whatsoever.

Proceed only at your own risk!!

No warranty of any kind is expressed or implied as to the workability or performance of designs, concepts or equipment described herein.

I took an IQ test and the results were negative!!


Now that you have been suitably warned, let us proceed together to explore the world of vacuum tube audio.








5:    POWER










The power supply is one of the least understood components of an amplifier, yet may well be the most important.

The primary purposes of the POWER SUPPLY are to:

a)    isolate the amplifier from the AC mains - an essential safety requirement.

b)    convert (rectify) the power available from the AC mains supply to the particular forms of AC and DC power required by the tube amplifier.

c)    shield the amplifier from spurious mains voltages.

d)    immediately provide rectified uni-directional DC current in response to instantaneous demands from the power amplifier - ie under considerably varying levels of load current ranging from zero to full peak current requirements - at more or less CONSTANT DC voltage to discreet sections of the amplifier.

e)    ground the amplifier chassis and metallic components to a reliable earth.

The following commentary attempts to delineate the above main features in more detail:

a)    isolate the amplifier from the AC mains.

Isolation from the mains supply means that the amplifier is electrically separated from the AC mains power supply - an essential safety requirement.

This ensures that whichever way the two AC input wires are connected - ie "active" and "neutral" - the output is always connected in the same intended polarity. The intention is that the active wire is the one switched on and off to control current flow in the circuit. If this is not so then the chassis could be connected to the mains active wire and if contact is made between the chassis and ground - even when the device is switched off - then whammo - goodbye audiophile!!

This is because electric current always tries to return to ground by the shortest possible pathway, so if the user presents that pathway then the user becomes the conductor for the return circuit to ground.

Isolation transformers are essential.

Important Note: "Auto-transformers" must not be used to couple the mains supply to the rectifier circuit, because they do not isolate.
Auto-transformers may be used to vary the secondary voltage up or down before the rectifier however in that configuration, the primary feed to the auto-transformer must be from an isolating transformer.

A useful user safety device is the earth leakage core-balance circuit breaker - otherwise known as a "safety switch". This device senses any variation between its in and out (line and load) circuits - and is designed to detect small leakage currents to earth - such as when a person inserts oneself in the active or neutral circuit, or between them and ground. This can be a painful, if not terminal experience. The safety switch can be installed as a separate discreet component between the mains and the power supply input.

Important Note: Conventional fuses and circuit breakers will not detect small variances in current and do not provide user safety protection.

b)    convert (rectify) the power available from the AC mains supply to the particular forms of AC and DC power required by the tube amplifier.

Direct current must be as pure as practicable - ie free from AC ripple, spurious mains fluctuations or surges, Radio Frequency (RF) interference spikes, or noise.

Direct current is required at particular positive or negative polarity voltages for the tube Plates, Screen Grids, Control Grids and, where applicable, tube filaments or heaters

In the case of large transmitting tubes, the high voltage B+ supply to the Plates may be in the order of 2 to 4,000 volts DC.

As well as Direct Current requirements, it is usually the case that a low voltage AC supply, or supplies, is required to supply the filaments or heaters of tubes used in the amplifier.

The AC supply is usually a standard voltage designed to match the tube type characteristics to be installed in the amplifier.

Standard heater/filament voltages are 1.5, 2.5, 5.0, 6.3, 7.5, 10.0, 12.6 - and a range of higher voltages.

Filament/heater supplies may be AC or DC.

Where the AC is rectified to DC, to allow for rectification the transformer LV voltages must be higher when a choke filter is used, or lower when a capacitor filter is used.

The AC supply is sometimes centre-tapped to provide a cathode path for indirectly heated tubes, or to minimise hum in indirectly heated tubes.

In some tube amplifier designs, particularly pre-amplifiers where low signal voltages are typical, it is common practice to terminate the centre-tap of the LV heater supply to a  DC source fixed a positive potential relative to ground. This is intended to ensure that any electrons that might conduct between the heater and cathode do so from cathode to heater (DC) rather than heater to cathode (AC). Typical voltages for terminating thecentre-tap in this manner are 40VDC to 100 VDC. Voltages higher than this could cause a breakdown in the heater insulating coating - thereby rendering the tube unserviceable.

Where a tube rectifier is use, even when it is an indirectly heated type, it is essential to provide a separate LV supply to it because the heater circuit and transformer winding will be at B+ potential above the chassis.

c)    shield the amplifier from spurious mains voltages

Spurious asymmetrical voltage waves - eg high amplitude spikes - present in the DC supply to the amplifier can modulate the output, resulting in distortion and instability.

Microwave ovens, computers, electric motors, refrigerators, electric drills and saws etc can be an annoying source of interference through the mains circuit.

It is good audio design engineering practice to instal a Mains Filter in the incoming line.

Professionally engineered mains filters are available as modular units from electronic components suppliers or can be recycled from computers.

Electrostatic Shielding

It is also a standard audio amplifier design convention to instal an electrostatic shielding winding between the mains primary and the HV secondary and any other windings. One end of the winding is normally grounded to the chassis, so only one terminal or lead is provided. This winding will protect the amplifier from radio interference and any other spurious signals injected into the power supply from the mains. It will also protect the amplifier from unintended high voltage spikes resulting from switching surges, lightning strikes or power utility transformer inter-winding failures between high and low voltage circuits, such being diverted to ground via this winding in the isolating transformer.

d)    provide rectified uni-directional DC current - under considerably varying levels of load current ranging from zero to full-current requirements - at more or less CONSTANT DC voltage to discreet sections of the amplifier.


In the early days of radio it was common for radios and audio amplifiers to be battery operated. Consequently, the standard nomenclature for battery identification by their circuit application has traditionally been:
  • DC power to the tube Filaments or Heaters (A Supply)
  • DC power to the tube Plates (B Supply)
  • DC power to the tube Screen Grids (B Supply)
  • DC power to the tube Control Grids (C Supply)



    The A, B and C supplies are usually expressed as + or - polarity

    Hence B+ has become the convention for identifying the primary DC supply to the Plate circuit - and any other high-voltage circuits in the amplifier.

    Discreet supplies to different sections of the amplifier - at the same or different voltages - can be identified by B1+, B2+, B3+ etc.

    For linear high-fidelity  amplification - ie where an exactly constant ratio of signal power output to signal input at all signal levels and frequencies is required - the Power Supply must supply as much POWER - ie current at the design voltage- as the amplifier requires - as and when it is required

    For high-fidelity linear amplification the supply voltages to all stages of the amplifier should be practically constant under all operating conditions.

    Unfortunately except for relatively constant current power amplifiers, this design standard is very difficult to establish - c'est la vie!!

    e)    ground (earth) the amplifier chassis and metallic components

    Noting that small children are attracted by lights and shiny knobs, the mains connections and high-voltage components of the amplifier and power supply must always be adequately mechanically shielded from physical contact by people or animals, to prevent deliberate or accidental contact.

    All the above principles are particularly important for portable amplifiers such as public address and guitar amplifiers, where the grounded chassis is relied upon for both user safety and for shielding of input leads to prevent hum.

    As a self-preserving safety precaution, users of portable ampifiers should always ensure their equipment is reliably grounded before use.

    During normal use and/or servicing, audio amplifiers require direct physical contact between the user and metal components at some point in the system. All exposed metal components must be effectively earthed (grounded). There must be no measurable resistance between metal components and earth.

    It is standard convention to, as far as is practicable, use the chassis as the negative pole/bus of amplifier circuits, including rectified circuits from the Power Supply, and to separately ground the chassis to earth with a reliable connection.

    This design principle is also essential to grounding shielded input leads and speaker leads, to protect signal carrying conductors from picking up external interference.

    Where positive and negative polarity, or floating DC supplies are used then other arrangements are required - but always ensure the amplifier can be safely used by people.

    Important Note:

    Most modern home electronics equipment, such as the TV, VCR, DSTB, DVD etc, operate in a mode isolated from the mains and ground by capacitors - ie "floating".

    However connection of devices to an audio amplifier via shielded lead and RCA or DIN connectors always requires a grounded terminal to enable the shielding to be effective.

    Physical contact between a floating chassis and any metal components connected to it, which may be at a significant potential between it and ground, and actual ground via a shielded lead or terminal connector may result in electric shock.

    Exposed metal parts include handles, control knobs, pick-up arms/cartridges, and tape deck drive parts.


    All tubes are "rectifiers" insofaras they are intended to conduct electrons in one direction only - ie current goes in one end and comes out the other.

    However, tubes which are used primarily to control the direction of current flow are known as "rectifier tubes".

    Tube types designed for exlusive or discreet application in rectifier circuits are called "Diodes" or "Rectifier Tubes" and in schematics look like this:

    2.1    Half-wave Rectifier tube with directly heated Filament - ie one Plate and one Filament

    2.1    Full-wave Rectifier tube with indirectly heated Cathode - ie Two Plates and one Cathode

    The Plate and Filament, or Plate and indirectly heated Cathode - may be arranged in either positive or negative polarity in a circuit, depending upon the application.

    Normally, in tube rectification circuits, the AC input is connected to the Plate and the output is taken from the Cathode or Filament. Because there is a voltage drop across the tube, the Cathode will always be negative to the Plate.

    For negative voltage bias supplies (C Supplies) the Plate is normally connected to Ground or Earth to create the positive pole, and the output is taken from the Cathode or Filament - ie the negative pole of the circuit.


    This graph shows the relationship between the sine-wave AC input voltage "ERMS" (the horizontal axis is zero, or 0 volts), the Peak Input AC Voltage "EPEAK", the rectified output AC voltage "ERMS" and the rectified Peak Output AC Voltage "EPEAK", measured over "Time" - being the horizontal axis reading from left to right.

    It will be thus observed that the output voltage is the same as the input voltage.

    The theoretical equivalent rectified and polarised DC output voltage available to the load is shown as "EAVGE (FULL WAVE)"

    The Peak Inverse Voltage "PIV" is expressed as 2.828 x "ERMS" (the actual AC RMS input Voltage)

    In the case of a symmetrical sine-wave, the mid-point or central axis of the AC input voltage is designated 0 volts, however that is in relation to the positive and negative alternation swings of the AC input voltage about that central axis. It is very important to understand that the central axis may be either "floating" - thus inductively coupled - or fixed by design, at a potential of up to many hundreds of volts + or - AC or DC to ground or chassis.

    eg Some early tube domestic radio receivers used the 2.5V or 5V AC supply to the Rectifier Tube filament for the dial indicator lamp. One terminal of this low AC voltage was in fact the DC output from the rectifier and was therefore at a potential of several hundreds of volts + DC to ground, thus presenting a shock hazard to the unwary serviceman.


    3.1    How Rectification Works:

    The input AC voltage and current supplied from the mains transformer at 50 or 60 Hz are assumed to be a sine-wave.

    The above graph shows voltage over time - ie the horizontal axis is "time", reading from left to right.

    The horizontal axis also represents voltage and current zero - ie the mid-point of an AC cycle.

    For convenience, the alternations shown below the horizontal axis are normally described as being "negative" polarity and those above "positive" polarity.

    In a full-wave rectifier supply, each 50 or 60 Hz AC half-cycle is rectified in the forward direction to produce the output shown in the above graph - ie a 100 or 120 Hz half-wave AC voltage of positive polarity - ie the voltage shown above the horizontal axis.

    For negative polarity circuits, such as in C Supplies for Control Grid bias, the graph could equally show the output as being only below the horizontal axis, in which case the output would be deemed to be of "negative" polarity relative to ground.

    The current is now uni-directional, also shown in the graph as left to right, thus the graph shows one and one-half alternating current cycles.

    3.2    Ripple

    The load in a tube amplifier requires Direct Current (DC), which is represented graphically as a straight line.

    However as the above graph shows, the output from the rectifier is not a straight line - ie a constant uniform voltage - but a series of crests of the rectified sine-wave, with the theoretical DC axis being the line "EAVGE (FULL WAVE)".

    So the object of the Rectification process is to convert the modified sine-wave, including its ripple component, to a straight line constant DC voltage for delivery to the load.

    Thus the challenge is to eliminate ripple.

    This is usually done by means of the process known as "Filtering".

    3.3    Filters

    The primary object of filtering is to convert the uni-directional Alternating Current rectified sine-wave form power input, shown as EPEAK in the following graph, into Direct Current output - shown as the straight line axis EAVGE :

    In simple power supplies, two main forms of filters are used:

    These are more correctly described as: In more complex Power Supplies voltage and/or current regulation may be incorporated to compensate for variations in mains input voltage, transient signal requirements, or losses in the power supply itself.

    Modern solid state equipment utilises the high-frequency "Switch-mode Power Supply" design to both filter and regulate the output, however the extra complexity, cost and potential unreliability of a Switch-mode Power Supply may not be warranted for a tube hi-fi amplifier when a holistic view of Power Supply requirements is considered.

    It has been well demonstrated over the past 50 years or so by the performance specifications of actual commercial equipment and elsewhere in this web-site that triode and ultra-linear output stage configurations may perform satisfactorily with conventional rectification an filtering systems.

    Rectifiers operate on the principle that for current to flow in a circuit there must be a difference in voltage across the load.

    Hence for any filter system to work - at all - there must be a voltage drop across the filter.

    Now straight away we have a problem - which is called "Regulation".

    3.4 Regulation

    The term "resistor" is derived from the verb "resist".

    In the case of DC circuits, "resistance" in the power supply circuit will "resist" the flow of current from the power supply to the amplifier and will cause voltage drop under load.

    The term "impedance" is derived from the verb "impede".

    In the case of AC circuits, "impedance" in the power supply circuit will "impede" the flow of current from the power supply to the amplifier and will cause voltage drop under load.

    Either of these conditions will be detrimental to audio performance.

    "Regulation" is a measure of the difference in voltage drop across the load between zero-signal and maximum signal conditions, usually expressed as a percentage of the zero-signal condition.

    In a 100 Watt RMS amplifier this can be in the order of 0.25A or more.

    Thus for a stereo power supply, a current range of 0.5A is reasonable.

    Ohm's Law tells us that introducing a resistance and/or impedance of only 100 Ohms into the Power Supply - including resistance and/or impedance in the mains transformer and filtering components - will cause a voltage drop at the Plates of the output tubes in the order of 50 Volts DC at 0.5A - just at the very instant in time when we want maximum voltage to enable us to produce maximum power output. (0.5A is the approx. DC current required for a 100 W RMS per channel stereo or equivalent amplifier)

    In this example, a voltage drop of 50 VDC on a zero-signal condition of 450 VDC would be expressed as a "Regulation" of 11.1 %. Whilst this appears to be a good figure it is really poor, resulting in dramatically reduced performance on peak power signals.

    Since the power output from the amplifier is calculated on the basis of the square of the AC output voltage divided by the load impedance, it follows that power output is affected by the square of any voltage drop caused by poor regulation - so it is worth worrying about.

    3.5    DC Rectification using a Capacitor Input to Filter

    A typical FULL-WAVE RECTIFIER with indirectly heated Filament Tube (ie with a separate Cathode) and single Capacitor Input Filter is shown below:

    The following analysis and explanation of Capacitor Input to Filter Rectification assumes:

    a)    Voltage

    The voltage at the first filter capacitor follows the line A, B, A1, B1 in the upper graph

    The capacitor charges between A and B

    The capacitor discharges between B and A1

    Thus the mean level of DC output voltage is A, B, A1, B1

    In a full-wave rectifier, this is normally 1.414 x the AC input voltage LESS circuit losses, including internal losses within the power transformer (electro-magnetic core and copper losses).

    Tube Rectifiers introduce significant losses in voltage within themselves (Forward Voltage Drop), increasing with load current.

    Solid state rectifiers generally produce negligible losses within themselves at any load current.

    As the value of the first capacitor before the Filter Choke increases, the measured DC output voltage will fall.
    For large values of capacitor (eg in the range 200-300 uF) the output voltage may be in the region of 1.25 times the AC input voltage. Very high values of capacitor (eg in the range 2000 uF or more) will deliver even lower values of DC output voltage.
    It is prudent to do some research before investing in an expensive power transformer.

    When using a Tube Rectifier, refer to the Tube Manufacturer's Data Sheets for limiting values of Plate Voltage and Capacitor Values.
    Excessive values of :

    are likely to damage the tube and/or shorten its life dramatically.

    The Rectifier's Maximum Rated Peak Inverse Voltage (PIV) must not be exceeded.

    b)    Current

    As stated above, for current to flow there must be a voltage differential betweent the poles of the circuit, so the current through the tube rectifier is similar in form to the difference in voltage between the curves APB and AQB.

    In this case, the shaded area between the curve APB and the curve AQB in the lower graph represents the voltage by which the transformer voltage exceeds the value of the charged capacitor.

    Thus the current available to the load through the Plate circuit of the rectifier tube only flows for the interval of time between A and B and between A1 and B1 - because at other parts of the AC cycle the transformer voltage is below the voltage of the charged capacitor.

    c)    Ripple Voltage

    The Ripple Voltage may be determined from the shape of the graph A, B, A1 and B1

    As the load resistance is increased, B A1 becomes more nearly horizontal and the area AOPB becomes smaller until, in the extreme case where the load resistance is infinite, the DC output voltage is equal to the peak voltage "EPEAK"

    It follows that because small, low-power tube amplifiers will present a higher load resistance to the Power Supply than large, high-power tube amplifiers, eliminition of Ripple Voltage will be more difficult for high-power applications.

    It also follows that because Ripple Voltage is expressed as a percentage of output voltage and high-power tube amplifiers operate at high amplitude voltage, eliminition of Ripple Voltage will be more difficult for high-power applications.

    Parallel-push pull amplifiers will present a lower load resistance to the Power Supply than single-pair types hence it will be more difficult to eliminate Ripple in those designs.

    The same principle applies to single-pair push-pull versus single-ended output stages - however in the latter case there is no hum cancellation in the output stage so hum prevention is more difficult than for push-pull configurations.

    For those theoretically minded, the Ripple Frequency fundamental and harmonic components may be calculated by Fourier analysis.

    On the assumption that the Ripple consists of regular symmetrical triangular waves, the RMS Values of the Ripple components will be:

    Fundamental Ripple Frequency Voltage = 0.575 ER

    Third Harmonic Ripple Frequency Voltage = 0.064 ER

    Fifth Harmonic Ripple Frequency Voltage = 0.023 ER

    Seventh harmonic Ripple Frequency Voltage = 0.0117 ER

    where ER is the Peak Amplitude of the Ripple Voltage.

    c) Voltage Doubling

    A typical FULL-WAVE VOLTAGE-DOUBLER Rectifier with Solid-state Rectifiers and Single Stage Capacitor Input Filter is shown below:

    This illustration also shows a negative polarity DC supply for Control Grid bias - note that in the case of the negative bias supply the + and - filter capacitor polarities are incorrectly reversed - ie the + terminal should be earthed.

    In the case of this configuration, the input voltage is approximately doubled, enabling the use of a transformer having half the secondary voltage. However to maintain the required power input the current must be doubled.

    This still enables a smaller and less costly transformer to be used, offering important benefits to guitar amplifier users, such as less weight and more compact equipment.

    3.6    DC Rectification using a Choke Input to Filter

    A typical directly heated filament Full-Wave Rectifier with ONE STAGE (ie one choke) Choke Input Filter is shown below:

    A typical directly heated filament Full-Wave Rectifier with TWO STAGE (ie two chokes in series) Choke Input Filter is shown below:

    The general explanation above is relevant also to Choke Input Filters  provided the choke has a high-impedance.

    DC Output Voltage "EAVGE" is typically 0.9 x the AC input voltage LESS circuit losses, but may be lower with high inductance values of choke

    WARNING: When using silicon diode rectifiers in a half or full-wave configuration, or a full-wave silicon bridge rectifier, peak DC voltages at switch-on may be as high as 1.414 x the RMS input voltage - ie the same as for a capacitor input filter.

    In the case of voltage doubler circuits, the peak voltage may be as high as 2.818 x the RMS input voltage.

    Even though this peak voltage may only be applied for the first half-cycle after switch-on, unless ALL capacitors in the power supply rectifier and B+ rail circuit are adequately rated, failure is likely. (Until the tubes heat up and draw current there will be no significant load on the power supply).

    The DC voltage will threafter progressively reduce as the capacitors charge to normal operating conditions.

    3.7    Combination Power Supplies - ie use of different methods of rectification and filtering for different purposes

    The following schematic of the Acrosound 100W 6146 Tube Amplifier Power Supply shows one example of this.

    The B+ supply is 725 VDC - note the full-wave rectification with dual tubes for higher current capability and improved regulation

    The B1+ supply to the Screen-Grids is 220 VDC - note the variable load used to adjust the applied voltage

    The B2+ supply to the drivers is 400 VDC - note the voltage divider network

    The A supply to the tube heaters is 6.3 VAC - note the 100 Ohms potentiometer used to minimise hum

    The C supply to the 6146 tube Control Grids is 115 VDC negative polarity - note the half-wave silicon diode rectifier

    This Power Supply is as about as complex as they become for tube amplifiers

    In this case because all of the transformer windings are simultaneously similarly affected by a sudden or peak increase in load on the primary winding and core, improved performance can be obtained by using separate power transformers for each discrete amplifier function and full-wave silicon rectifiers throughout (preferably full-wave bridges).

    Running the Bias supply from the main power transformer is also not desirable, because loss of bias voltage on peak power draw signals will reduce power output dramatically - in addition to any loss in the Plate supply.

    In a large and heavy Power Supply such as this, where the transormer secondary winding is 1650 volts rms, substantial weight reduction can be obtained by using full-wave voltage doubler silicon diode rectification. The same B+ voltage (725 VDC) could be supplied from a 300 VAC transformer having a single winding without a centre-tap as per the above schematic.


    The A and C supplies are usually determined by choice of tube, so that leaves us with the B+ supply to consider.

    One of the most challenging aspects of an amplifier design is to select the most suitable operating voltage for the power tubes - ie the B+ supply to the Plates.

    Most tube handbooks publish a range of optional operating conditions for power tubes and the designer is free to choose from a wide variety of tube types and voltages.

    Is it better to run a tube well below its rated maximum limits for long life and cool operation, or to run a tube at or near its maximum ratings to increase power output and efficiency?

    Do tubes sound better when pushed?

    These are questions only the user/designer can answer, because like "sound" itself, assessment of results is very subjective.

    It is well to remember that tube performance is primarily controlled by the Screen Grid - so plate voltage is not as important as it
    might seem. Low Screen Grid voltage translates into less Cathode current, lower operating temperature, longer tube life, reduced
    distortion and more faithful reproduction.

    However, because a tube is a voltage amplifier and AC POWER output from the amplifier is calculated as the square of the output voltage divided by the load impedance, it follows that for power tubes when driving into a given load impedance, a small increase in voltage will increase power output dramatically.

    The limiting factor to applied DC voltage will be the maximum tube ratings and/or flashover across internal electrodes, base pins or  the socket.

    Important Note: See my paper Screen-Grids to understand why the applied Plate voltage must not exceed 2.5 times the applied Screen-Grid voltage.

    However a practical solution is to select the B+ voltage upon available working voltage ratings of the electrolytic capacitors used in the power supply filter circuit - because these are the most difficult voltage dependent components to acquire.

    Rated working voltages for electrolytic caps tend to be available in 50V increments, however there are some popular ratings that are more readily available - particularly in the larger capacitance sizes - such as 350 VDCW, 400 VDCW, 450 VDCW and 500 VDCW.

    Not much is available at affordable cost with ratings over 500 VDC.

    Hence, for B+ voltages over 450 VDC, it is usually the case that electrolytic capacitors need to be connected in series to withstand the applied voltage. Unfortunately this has the effect of halving the effective capacitance (assuming both caps are identical).

    Consequently, if the designer chooses to use a B+ voltage greater than about 450 VDC then considerably more chassis space and cost is required.

    It may be more practicable to consider parallel-push-pull output.

    On the other hand, if a full-wave voltage doubler power supply is used, two series caps are required in any case, so no extra componentry is needed - at least in the rectifying section. Furthermore, the power transformer size will be smaller than a full-wave centre tapped or full-wave bridge style, so the final space and cost may well be less.

    It is well to remember that tube performance is primarily controlled by the Screen Grid - so plate voltage is not as important as it might seem. Lower Screen Grid voltage translates into less Cathode current, lower operating temperature, longer tube life, reduced distortion and more faithful reproduction.

    IMPORTANT: Short-time Start-up Surge

    Directly heated tube rectifiers - ie those having a directly heated Filament - produce a surge voltage upon switch-on.

    This surge will be significant and typically up to 1.414 x the RMS input voltage, and will definitely be substantially higher than the full-load output voltage. Hence filter capacitors MUST be rated to withstand the DC surge voltage - or else they will fail (a somewhat spectacular and messy process).

    The surge voltage is caused by the Rectifier tube heating quickly and commencing conducting current very quickly after switch-on, compared with the load upon it (usually the power tubes) which heat relatively slowly. Of course if the power tubes are also directly heated types then the surge voltage will be of shorter time duration but there is always the possibility of the surge being present for a short period of time - long enough to damage the filter capacitors.

    As the indirectly heated power tubes warm up and commence drawing current, the tube rectifier output voltage will fall dramatically due to the output voltage falling from the crest of the input sine-wave to an average RMS level. To this must be added internal losses in the power supply circuit resulting from passive devices such as resistors, transformer windings and choke windings.

    Indirectly heated rectifiers - ie those having a separate Heater and Cathode - do not usually produce a surge upon switch-on because they are deigned to warm-up slowly - slightly slower than the tubes they supply. Voltage drop under load is also lower than with directly heated rectifiers. Hence power supply regulation is better with indirectly heated tubes than with directly heated tubes.

    For example,  Rectifier Tube types 5Y3-GT and 5V4-G have identical heater/filament voltages and current and are of similar physical size.

    For an AC voltage input of 500 VAC per Plate (1,000 VAC CT) and choke input to filter, the 5Y3-GT produces an output voltage of 380 VDC @ 125 mA

    For an AC voltage input of 500 VAC per Plate (1,000 VAC CT) and choke input to filter, the 5V4-G produces an output voltage of 410 VDC @ 175 mA

    Furthermore, the 5Y3-GT offers a maximum DC current output of 80mA with Capacitor Input to Filter, whereas the 5V4-G offers a maximum DC current output of 175 mA under similar conditions.

    Indirectly heated rectifier tubes are more expensive to manufacture, thus have been consequently limited to better class equipment.

    Solid state full-wave or bridge rectifiers do not produce this surge at switch-on because the full-load continuous output voltage is essentially the same as the switch-on voltage. They also offer substantially better regulation under load due to their very low forward voltage drop. Normal current ratings are available in the several amperes range at very low cost.

    IMPORTANT: See note above re surge voltages when using choke input filters.

    5.    HOW MUCH POWER ?

    For an excellent technical explanation of how power supplies work and options available see http://www.tpub.com/neets/book7/27a.htm and follow the "next" links to scroll pages

    See also paper by Williamson and Walker

    5.1    Source Power

    As noted above, the power supply is one of the least understood components of an amplifier, yet may well be the most important.

    The purpose of the power supply is to provide approximately CONSTANT DC voltage to the amplifier under considerably varying levels of load current.

    Volts X Amperes = Watts = "POWER". Hence the term "POWER SUPPLY".

    That is to say, a "power supply" is a supply that delivers electrical "POWER".

    However, in a power supply that derives its source power from a 50 Hz or 60 Hz mains, the source energy switches COMPLETELY OFF 100 or 120 times per second, as the case may be.

    Closer examination of the sine-wave input shows that the bulk of the useable electrical energy (calculated by the area under the curve) is available only about one half of the real time input.

    Hence in an AC mains derived power supply converting that energy to DIRECT CURRENT there are very clearly defined periods of time when there is no useful power input at all.

    The challenge then is to convert the AC energy into DC but at the same time making the DC available continuously in such a manner as to instantaneously enable transient peak currents of not less than 100 x zero signal DC plate current (=20 db dynamic range) up to ? x zero signal plate current (= 100 db dynamic range) - see plate current characteristics of most tubes.

    The only practical way this can be done is to introduce a storage device, or storage system, into the process.

    This is why large filter capacitors are essential for high-fidelity.

    Furthermore, the type of capacitor should be high-ripple industrial can type chassis mounted.

    The basic principle is that the energy needed for the power tubes in the amplifier can be supplied either from the mains or a storage device - or both.

    Since filter capacitors are usually required to remove ripple from the B+ supply, a minimum value will be included in any event - with or without a filter choke.

    By substantially increasing the value of the filter capacitors, they can be converted from being solely a filtering component, to a power storing component.

    When the capacitors become large enough, the power rectifier becomes simply a "topping up" device and is no longer relied upon to deliver primary energy to the power tube plates.

    Important Note: There are practical limitions for the size of storage capacitance incorporated into a tube audio amplifier and consideration needs to be given to the risks of current inrush at switch-on, transformer or capacitor component failure, fire and death by electrocution in the event of inadvertent contact with them.

    5.2     "How much capacitance is needed?"

    First. let's review a few traditional vintage designs.

    The original and popular Quad II 12W amplifer had a total of 32 uF capacitance in the B+ supply - 16 uF for the plates and 16 uf for the driver stages.

    The original McIntosh Model A116 30W amplifier had a total of 100 uF for the complete amplifier, of which 80 uF was for the plate supply.

    The Acrosound 1955 "Williamson" ultra-linear ampliers had a total of 20 uF for the B+ supply.

    The GEC 1957 "Williamson" triode amplifier using KT66's had a total of 16 uF for the plate supply.

    The original Hafler and Keroes 1951 Acrosound ultra-linear amplifier had a total of 80 uF for the plate supply.

    It is reasonably obvious that these values of capacitor were universally considered at the time as being adequate for the intended purpose. That thinking is underscored in vintage textbooks, which explain power supplies in terms of reducing "ripple" to an acceptable audible hum level.

    However, in my experience, verified by extensive listening tests, values up to 10,000 uF are preferable in a hi-fi tube amplifier to supply adequate transient power over and above normal signal requirements.

    The question is - "How much capacitance is needed?"

    This can be answered in part by the formula:

    W = C x E squared


    W = energy in Joules (Watts per second)
    C = Capacitance in Farads (1,000,000 uF)
    E = DC voltage applied across the capacitor


    C = 2W
    E squared


    The amplifier is rated at 100 W RMS per channel (stereo) from a 400 VDC common power supply.

    If we assume we need the stored energy in the Capacitor to sustain Plate Current at design Plate Voltage for one full cycle of power mains input to the power supply at 50 Hz mains frequency, and the amplifier is 50% efficient (ie output power in Watts RMS divided by input power in Watts RMS), then we will need the following size capacitor:

    C = 2W
    E squared

    In this instance, W = (100 W x 2) x 2 (= 50% efficiency) x 1/50 second

    or, W = 400

    = 8 Watts


    C = 2 x 8
    400 x 400

    = 16

    = 0.0001 F

    = 100 uF

    Now anyone with experience in hi-fi amplifiers knows that 100 uF is not adequate for this purpose.

    This difference between theoretical and practical may be explained by the fact that 100 uF is not sufficient to maintain Plate Voltage at a reasonably constant level during the peak transient power output period.

    Consequently it must be assumed that either:

    Practical experiments demonstrate that at least 1,000 uF Capacitance is required at 400 VDC to supply the requirements of the above example.

    So, as a practical solution, we can assume the following empirical formula:

    C = W
           E squared


    C = Capacitance in Farads
    W = Rated power of the amplifier (x 2 for stereo)
    E = DC voltage across the capacitor

    Thus in the above example, the formula becomes:

    C = 200
          400 x 400

    = 200

    = 0.00125 F

    = 1,250 uF

    This is the minimum capacitance needed.

    Obviously more is better.

    WARNING: Large capacitors will store energy for a long time - often measured in days. The risk of electrocution is high unless measures are incorporated to discharge capacitors after the amplifier is switched off. A 15k-20k Ohm 20W wire wound resistor permanently connected across the B+ is one method available.

    If preferred, this can be connected in series with a normally-closed relay so that the resistor is in circuit when the Plate Supply power is off - and out of circuit when the Plate Supply is on. In this case the value can be reduced to a 1500 ohms 20W wire wound resistor for fast discharge.

    WARNING: When large capacitors are used, there is a risk of voltage doubling or spiking if the amplifier is switched off then on again after a short period, resulting in damage to capacitors, tubes or transformers.

    WARNING: Large values of capacitance create large values of transformer inrush current and switching surge current in the power transformer, when the amplifier is first switched on. This can result in loose or deformed windings, hum and premature failure of the insulation protecting the winding wire - ie shorted turns or shorted windings. The rectifier must also be capable of handling the surge current without damage.

    WARNING: Large capacitors should be supplied through a protective fuse to ensure that each capacitor is isolated from the supply in the event of internal failure (not a common event but still a possibility).

    5.3 Tube Rectifier Applications

    WARNING: Tube Rectifiers are not suited to supporting large values of input capacitor.

    To avoid some of these problems, a practical approach is to use relative small values of capacitance in the rectifying stage - ie 32 uF for tube rectifiers and 200 uF for silicon diode rectifiers - then a large value after the filter choke(s).

    In practice, my experience suggests that regardless of the size of B+ capacitance at the centre tap of the output transformer some level of audible instability in the power supply will be experienced if the value of the first capacitor after the choke is too large.

    A value of about 200-300 uF after the choke is sufficient and seems to be stable.  It is likely that a smaller value of capacitor at this point assists the rectifier to top up the capacitor more quickly and effectively than with a larger size, allowing time for the filter choke to discharge into the second stage of the rectifier/filter system..

    Note however that a simple rectifier/capacitor filter system does not seem to be affected in this way.

    Generally speaking, the higher the value of the first capacitor after the rectifier, the lower the DC output voltage, the better the regulation and better amplifier performance ("sound") BUT some form of current limiting device is required before the rectifier to limit inrush current at first switch-on. A 20 ohm 20W heavy-duty WW resistor should suffice. (This method will cause further voltage drop in the B+ supply when under load)

    For guidance on optimising capacitor values see my paper on OPTIMISED ELECTRON STREAM © TECHNOLOGY

    Choke input to filter rectification - with a sufficiently large value of inductance - helps in this regard, because the choke will limit the inrush current to a safe value.

    For further information on capacitor safety see http://www.repairfaq.org/sam/captest.htm

    WARNING: NEVER exceed the surge voltage rating of a filter capacitor. In the case of choke input to filter power supplies this means the capacitor must have a rated surge voltage greater than the unloaded rectifier output voltage - ie 1.414 x the rms input voltage


    6.1    Filtering and Regulation

    In any event, the power supply must have good regulation, and effective filtering, to supply reasonably constant voltage DC to the output tubes.

    It has been established that the B+ DC supply should have minimal AC ripple and good regulation, so a two stage filter choke input to filter rectification system is recommended - see typical design below:

    If the first stage of a two stage filter is the preferred choke input to filter style, then the first choke may be installed between the negative DC terminal of the rectifier and ground. To improve regulation - ie voltage drop as load current increases - the second choke may be a swinging choke - ie a choke that increases its inductance as the through current decreases.

    Typical one, two and three-stage Capacitor-input to Filter systems are shown in the following diagramme:

    A full-wave voltage doubler or bridge circuit is recommended for superior DC voltage regulation. However to ensure stability in the rectification/filtering system it is recommended that the following extra components be installed:

    1.    A silicon diode connected across the full-wave bridge rectifier output, using the same type and polarity as for the diodes in the bridge

    2.    A 0.1 uF capacitor connected across the bridge output and before the first filter choke - having a DC voltage rating of at least three times the DC B+ voltage. If necessary connect capacitors in series to provide adequate voltage rating - if not the caps might just short-circuit and do some nasty damage

    3.    A 0.47uF capacitor across each and every electrolytic capacitor in the filtering system (Note though that this will affect the audio signal and may produce too much top end frequency boost)

    4.    Most importantly, to prevent spikes from the rectifiers carrying through the system and modulating the audio output, instal a shunt capacitor having a value within the range 0.022 uF to 0.47uF 630VDCW across (in parallel with) the first filter choke - but not the second.

    The value of this capacitor should be fine-tuned to create a tuned shunt filter. The values of the choke and capacitor will determine the tuned frequency, ideally being the fundamental or first or second harmonic of the rectified mains frequency. This simple design modification will also improve the power factor of the first choke stage, thereby delivering more usable power to the output tubes. The optimum size of capacitor should restore the power factor to 1

    5.    Instal a commercially manufactured mains line filter between the mains input and the power transformer. These are readily available from computer supplies stores

    The DC voltage output from a full-wave or bridge rectifier into a choke input filter will be about 90% of the AC input voltage, so a higher AC input voltage is required than for capacitor input to filter rectification. The output voltage from a choke filter will be further reduced by DC resistance and AC impedance in the choke(s) and power transformer.

    The DC voltage output from a full-wave or bridge rectifier into a capacitor input to filter will be about 140% of the AC input voltage, so a lesser AC voltage is required than for choke input to filter rectification. The output voltage from a capacitor filter will be further reduced by DC resistance and AC impedance in the choke(s) and power transformer.

    However we get nothing for nothing in electricity, thus the AC current in a full-wave or bridge capacitor input filter will be about 1.4 x DC output current, compared with only about 0.9 for a choke input to filter rectifier. It can easily be seen that either way, V x A = Watts

    Note that if the first filter capacitor after the choke is too large it may cause instability in the power supply and excessive current inrush at switch-on.

    As noted above, the Screen-grid is the effective ANODE in a tetrode, pentode or beam power tube - hence any fluctuation in Screen-Grid voltage under varying conditions of load on the power supply will result in non-linearity in the output signal.

    Manufacturers' Tube Handbooks explain that electron flow (Cathode current) is not much dependent upon Plate Voltage but very dependent upon Screen-grid voltage - because the Screen-grid is a regulating electrode, whereas the Plate is simply a collector of electrons.

    Traditional tube amplifier designs having unregulated Screen-grid supply - or unregulated Plate/Screen-grid supply where the Screen-grid supply is common to the Plate supply, will not be linear - ie not true hi-fidelity under transient signals (because the B+ supply voltage to the Screen-grids and therefore power output, will drop under load), so it is essential to ensure the Screen-Grids are supplied from a generous or regulated power supply separate to the Plate supply.

    The same design rules that apply to the Plate supply also apply to the Screen-Grid supply.

    On paper, Ultra-linear amplifiers do not suffer from this problem, because the Screen-grid voltage rises and falls in a direct proportion to the Plate voltage - at any instant in time - however an Ultra-linear output stage must be supplied from an adequately regulated - read either electronic or huge capacitance - to ensure adequate power is available to supply constant DC voltage under varying conditions of Plate/Screen current - ie Cathode current. This is not an easy challenge to solve but the benefits are there for those willing to make the effort. Silicon bridge rectification, choke input to filter and much capacitance are essential pre-requisites for a reasonable power supply for Ultra-linear operation.

    6.2    Filtering

    The 5Y3GT is a commonly used tube rectifier for audio preamplifier and power amplifier applications.

    The following diagrams, courtesy of RCA, show some important features of this tube, which are typical of tube rectifiers generally when used in Capacitor Input to Filter or Choke Input to Filter power supplies .

    Of interest are the foward voltage drop related to choke value, capacitor value and load current

    These are explained further by RCA in the following way:

    Hence it is important to adhere within the prescribed ratings for each type of tube rectifier.

    These ratings ensure long and reliable tube life and freedom from "Cathode stripping" resultant from excessive current through the Cathode during switch-on and warm-up of the amplifier.

    Unfortunately small values of capacitance cannot deliver the stored energy required to support high levels of instantaneous current during signal transients - which is why full-wave bridge silicon rectifiers is the way to go.

    Large values of capacitance create high values of transformer inrush current and switching surge current in the power transformer, when the amplifier is first switched on. This can result in loose or deformed windings, hum and premature failure of the insulation protecting the winding wire - ie shorted turns or shorted windings. The rectifier must also be capable of handling the surge current without damage.

    However all is not lost. One technique that is useful to minimise ripple voltage (hum) but protect both the transformer and rectifier is to instal high value capacitor(s) in the filter circuit (ie several thousand microfarads) fed via a series resistor to each capacitor. The series resistor will prevent surge currents during switch-on and warm-up but will assist to sustaining B+ voltage during transients. A value of about 500 Ohms per 50 VDC is useful. The resistor must have a rating of about 20W or more. It will be found that the resistor will get extremely hot during warm-up of the amplifier but will quickly cool then remain cool. This is because the capacitor charging current is limited by the peak DC voltage divided by the resistance value, whereas once charged, the capacitor will effectively stay at the DC charged voltage and pass minimal AC current. This configuration reduces the B+ voltage however so some experimentation may be needed to optimise values and voltages.

    A bonus benefit is that under typical music signal conditions, this configuration may serve as a simple voltage regulator - because the resistor prevents significant current flowing out of this capacitor, thereby maintaining DC voltage into the power supply. The voltage drop across the resistor can only be the difference between the charged voltage of this capacitor and the discharged voltage of the next stage - not likely to be more than a relatively low momentary value at full signal power output.

    Other more expensive techniques include using a time-delay relay and/or a Varistor.

    6.3    Choke and Capacitor Values

    Traditional electronic engineering texts have treated each stage of a rectifier filter system as a discrete entity, proving mathematical formulae to calculate optimum values.

    One option so described and widely used is the "swinging choke" system, where one of the filter chokes - either the first or second after the rectifier is conventional. Thus one choke has a fixed value of inductance and the other a swinging choke having a variable value of inductance. The purpose of a swinging choke is to present a high value of inductance at low currents, resulting in improved ripple voltage reduction, and a low value of inductance at maximum current, to minimise voltage drop across the choke. This reduction in inductance also increases ripple voltage, so the device is a compromise between improved regulation and decreased ripple reduction.

    Graph courtesy of "High Fidelity Circuit Design" - Crowhurst and Cooper 1956

    Filter Chokes are a simple "blunt instrument" device that rely upon internal out-of-phase electromagnetic forces to cancel out alternating current passing through the winding. The energy in the current is still there so it appears at the output terminal of the choke as a flatter, or relatively straight-line current. Obviously the higher the "inductance" of the choke the more the AC current will be opposed and converted to direct current.

    However if we examine the characteristic of the current waveform in a choke we can see high spikes or pulses of current appearing at the output of the choke and corresponding in time with each input half-cycle from the rectifier - see graph above. Since the object of the filter is to eliminate these spikes to produce pure direct current, it follows that to prevent the current pulses from being passed through to the amplifier they must be disposed of. That task is assigned to the filter capacitor.

    However since the spikes appear (measured) at the output filter capacitor positive terminal immediately after the choke, it follows that they have NOT been disposed of as intended.

    Therefore, unless further filter stages are added these pulse-current waveforms will pass all the way through to the B+ supply of the amplifier.

    This situation illustrates where theory and practice diverge.

    If the power supply has only a single choke filter then these spikes in current, which are in effect "pulses" of current, form the direct current supply to the amplifier and will modulate the amplifier output stage producing permanent distortion in the system - particularly noticeable in high-frequency audio signals where the spikes cause a "fuzziness" in the sound. Negative loop feedback systems in the amplifier will be working full time to try to eliminate this abnormality.

    If the power supply has a two-stage filter then the spikes in current amplitute are injected into the second stage of the filter.

    It follows that if the second choke is smaller - ie has a lesser value of inductance than the first choke - then the second choke will restrict the flow of current through the entire filter system and produce an irregular current flow - particularly if the amplifier draws more current than the second choke can supply. Not only that, but because the pulse waveforms produced by the second choke will be of a different shape to those of the first, the output current waveform will have two sets of pulses in it each having different and varying waveshape.

    If the second stage choke is a "swinging choke" it follows that its inductance will vary in response to the amplifier signal current and the resultant current pulse waveforms supplied by the filter will also vary with amplifier signal current.

    Very importantly, harmonics from the mains supply are also injected into the rectifier/filter system and appear as current pulses, superimposed upon the waveforms shown in the above graph.

    Since there will be a short time delay between the creation of the current pulse by the choke and its discharge through the adjoining filter capacitor that completes the discrete filter stage circuit, there will be a further complexity in current and voltage waveforms at the various frequencies through the filter.

    Consequently, the output current waveform from the filter will be complex and simultaneously filled with spikes at various frequencies.

    Since Ohm's Law tells us that current times resistance/impedance results in voltage drop, it follows that the output voltage from the rectifier/filter system will also have a complex voltage waveform.

    The consequence of this complexity in voltage and current waveforms results in instability in the amplifier and distortion of the output - perticularly in the higher frequency signal range.

    In my experience, to be sure of stability in the Power Supply:

    1.    it is essential for ALL filter chokes in the rectifier filter system to be identical.

    2.    it is essential that each successive capacitor in the filter be of the same value or higher value than the unit preceding it.

    3.     a small damping resistor of 20 ohms or higher is required to be installed between the output of the power transformer and the input of the FW bridge rectifier - in the case of a centre-tapped transformer one resistor per leg

    For maximised performance, each choke/capacitor combination can be optimised in value such that the capacitor and choke have similar charge/discharge characteristics.

    Note: The above comments apply generally, however in the case of pure Class A amplifiers where the DC current requirement is relatively constant, fewer problems become evident from shortcomings in the power supply.

    Conversely, the more the amplifier approaches Class B operation, the more important the power supply characteristics become and the more they will affect the amplifier performance.

    6.4 How many Filter Stages?

    In general the more stages in a filter the better the performance - ie the more pure the DC output voltage and current.

    But since each stage introduces losses in voltage due to internal resistance/impedance of the components the choice of stages is a compromise between the quality of regulation and the quality of the DC output.

    Three stages seems to be the ideal - ie three chokes in a choke input filter or capacitor input filter system.

    Voltage drop under transient loads can be offset by using quality chokes having low DC resistance and large values of capacitor, however as noted below, it is desirable that respective choke and capacitor values be equal.

    One of the benefits of this approach is that there will be a reasonably linear voltage drop (gradient) across the stages, resulting in a potential (voltage) difference between each stage under all operating conditions. This will help to ensure DC current can be transferred between stages of the filter. The greater the voltage difference between stages the greater will be the current flow between them.

    For example if the voltage difference between filter stages is 10 VDC and the Plate Supply drops 10V under load, the Plates will still be able to draw current from the previous filter stage capacitor because its voltage is not less than the Plate Voltage.

    The listener with a keen ear will hear the difference between a one, two or three choke filter, since the quality of the sound heard is directly dependent upon the quality of the Direct Current supplied to the amplifying stages.

    I use a three stage capacitor input-filter (3 chokes and 4 capacitors) because the amplifier sound quality is superior, however voltage drop under load becomes a real issue. A simple solution is to use a large value of third and fourth (final) capacitor to store sufficient energy for short transients.

    Note: Since filter chokes release their energy in pulses, it follows that to prevent spurious transient spikes from occuring in the B+ rail series string (PI Filter) as a result of differences in choke characteristics, filter chokes should have the same value of rated inductance and DC resistance.

    For any given power transformer voltage, large capacitors - particularly the last in the chain - will deliver a lower DC voltage to the amplifier B+ supply but the voltage and current are of better quality - ie less ripple. The "PSU Designer II" software described in 6.5 below is an ideal tool to determine optimum values for filter components.

    It will be found that the first capacitor after the rectifier may need to be smaller, in order to maximise output voltage, limit inrush-current at switch-on and deliver a fast recharge time for the filter. The latter is important because modern music typically requires support for heavy bass signals lasting one or two seconds - a long time for a power supply.

    Note re Guitar Amplifiers:

    The most popular solid-state power supply design for guitar amplifiers seems to be a full-wave rectifier or full-wave bridge rectifier followed by a 50 uF capacitor.

    Analysis of this design by critical methods demonstrates the ripple voltage delivered to the output tube plates is very high - in the range of about 50 volts peak to peak for a 450V DC supply - ie the plate voltage fluctuates about + or - 50 volts downwards from 450VDC at 200 mA load.

    The effect of this is to limit the stability of the output stage, limit high-frequency response and increase distortion. Obviously the output from the amplifier is modulated by the power supply ripple.

    However the primary advantage of this design is a very rapid replenishment of power to the capacitor at minimal cost and the result is satisfactory for the purpose.

    Further analysis shows that an increase in capacitance to between 500 and 1,000 uF is need to reduce ripple to a low level.

    However provided the power transformer can handle the peak current at switch-on, then such values are not only practicable, but deliver superior performance in both "sound" and power to the guitarist.

    There is no point to this improvement though if the object is distorted sound, because the improved power supply regulation will reduce distortion and improve sound quality.

    6.5    Optimising the Components in the Power Supply

    Thanks to the design expertise of Duncan Monro, a simple but valuable rectifier/ pi filter design programme is available called "PSU Designer II".

    "PSU Designer II" is a software package configured to help with the design of simple linear (unregulated) power supplies.

    "PSU Designer II" is available as a free download from Duncan Amps and is suitable for use with solid state or tube rectifiers.

    This programme enables the designer to model different combinations of inductors, resistors and capacitors in the rectifier filter system and determine the best fit, or optimum, design.

    Voltages and currents within the power supply are calculated as the power supply starts up, and also when it stabilises.

    These results are displayed graphically, and can be printed out as required.

    One of the fascinating attributes of PSU II is that it graphically demonstrates the behaviour of the various individual components in the filter system.

    Thus the designer can look at the behaviour of each component and combination of components to determine if the filter will do the intended job.

    By inserting values for minimum and maximum amplifier current requirements, the B+ supply voltage can be easily calculated for minimum and maximum signal conditions. These results enable the "regulation" of the power supply to be determined.

    My recommendation is to start with the same values for each of capacitors and inductors throughout, then experiment with values to determine the effect.

    PSUII software modelling produces some very interesting and unexpected outcomes, so take the time to do some serious modelling of planned design arrangements.

    To be sure the calculation allows sufficient time for the output to reach full voltage, set the time to 30,000 milliseconds and reporting delay to "0". Once the time to steady-state full-output is known the time-scale can be reset to a shorter period for closer examination of performance.

    "PSU Designer II" provides full details of voltages and currents at each step of the filter system - a very beneficial feature.

    In my opinion, to ensure stability in the power supply and its output voltage and current, it is essential that current and voltage curves for capacitors and inductors do not cross over each other. Each of these curves can be viewed in "PSU Designer II".

    Although there are other more complex design programmes available, "PSU Designer II" is a great starter programme that should satisfy the Power Supply design requirements of most home constructors.

    I am happy to endorse it as a great design tool.

    6.6 Line Filtering

    Very importantly, over-voltages, switching surges, harmonics and a range of spurious signals from the mains supply are also injected into the rectifier/filter system and may appear as noise, voltage spikes and current pulses.

    In this era of high frequency communications, the mains cabling - in the home and in the street - acts as an efficient antenna for certain kinds of signals, thus bringing unwanted interference into the amplifier system.

    There are some standard ways of eliminating these, such as the use of commercial line filters, bypass capacitors across the mains input and bypass capacitors across the B+ supply.

    One vital improvement is available to assist reducing mains interference is to instal a small capacitor - eg 0.1 to 0.5 uF - across the mains transformer secondary winding BEFORE the rectifier.

    To eliminate spurious voltages and currents from circulating in the rectifier circuit it is essential to ground the centre of the winding. This is easy with a centre-tapped winding such as for a full-wave rectifier - but for a bridge rectifier two series-connected capacitors are required, with their centre-junction being earthed.

    Capacitors should have a DC working voltage rating of at least twice the applied rectifier AC input voltage.

    This arrangement is required for each discrete power supply - particularly Control-Grid bias supplies, where unfiltered spurious signals from the mains supply can modulate the output tube grids.


    7.1    How Many Power Supplies?

    For true high-fidelity reproduction, a separate discrete power supply is required for each of the:

    Some designers prefer DC supply to all the above however, in a well designed amplifier and particularly stages using indirectly heated tubes, AC to heaters/filaments is usually satisfactory.

    AC heater wiring should be spiral twisted to minimise hum. One side, or the centre tap, should be grounded. In some amplifiers it is necessary to use a different power supply to output tube heaters than to driver tube heaters, to eliminate stray inducted signal between the stages - ie to prevent instability. In the case of indirectly heated cathode tubes (ie with heaters, not filaments) if a centre tap is not available then two 50 to 100 ohm resistors in series across the heater winding with their centre junction grounded will suffice.

    In large tubes having directly heated filaments, the mains frequency filament supply voltage will appear in the cathode (filament) to plate circuit, so some precautions against hum injection are essential - usually a grounded centre-tapped filament supply is adequate however some designers prefer rectified and filtered direct current.

    Some constructors prefer to use a power supply separate to the main amplifier. This method has many advantages and few disadvantages, but cost is obviously more.

    It is important to remember that when a power supply is installed in close proximity to the power amplifer or pre-amplifier, the same rules and precautions for inductive coupling between wiring and components apply.

    7.2    Bias Supply

    The negative DC voltage bias supply to output tubes must be full-wave, to eliminate ripple modulation of the output stage (no output stage is perfectly balanced at all signal levels, because even "matched" tubes are not wholly linear across their operating range).

    In a wide-range high-fidelity amplifier, it is essential to ensure complete AC grounding, or low-impedance path, back to the Cathodes.

    To achieve this the bias supply should be AC connected between the junction of the grid (grid-leak) resistors and their respective Cathodes by a sufficiently large non-polarised paper or poly capacitor. This capacitor will be additional to any electrolytic caps used for DC filtering.

    Of course in a fixed bias output stage the Cathodes will normally be grounded, so the AC bypass capacitor may be connected to any convenient grounding point in the bias supply circuit - but preferably directly to the Cathode terminals of the tube sockets.

    The bias supply must also include adequately filtered and regulated circuitry.

    WARNING: Overly large filter capacitors in the bias supply may introduce a time delay of up to 30 seconds between amplifier switch-on and availability of steady-state full bias voltage. Adequate bias voltage must be applied BEFORE applying voltages to other tube electrodes. This is particularly important for quick-heating tube types - ie those not having a delayed heater warm-up period.

    Usually, but not always, tubes having the suffix "A" in their type number have a delayed heater warmup time - check the manufacturer's data sheet.

    Cathode bias and back-bias are not recommended for hi-fi, because both limit transient response, however in budget amplifiers theydo offer an economical and practical alternative to an expensive bias supply and regulated B+ supply (because the Cathode current swing from zero signal to maximum signal is less than with fixed bias operation).

    Cathode bias also introduces undesirable polarised capacitance into the output stage Cathode signal circuit.

    8.    PROTECTION

    Components recommended for protection include cathodes of vacuum tubes - particularly in parallel push-pull output stages, large electrolytic capacitors, power transformer primary and secondary windings.

    8.1     Fuses

    Professional amplifiers have always incorporated fusing to critical components

    To protect the main power transformer and power supply, to minimise the risk of fire, and to prevent the amplifier from self-destructing it is advisable to fuse anything that is likely to fail.

    Do not fuse Plates of Tetrodes and Pentodes because DC voltage will still be applied to the Screen-Grids and the tube will self-destruct (because the screen-grid will try to carry all the signal current) - insert the fuse in the Cathode circuit only.

    8.2    Circuit Breakers

    Where practicable, circuit breakers are far superior to fuses in that they respond very quickly to over-voltages and/or over-currents (as applicable to the design of circuit breaker used) and also provide current limiting properties.

    A good magnetic circuit breaker will respond within the first half-cycle of over-current, whereas a fuse may well take 3 minutes to blow at rated fusing current and several cycles at short-time current.

    8.3    General

    As a general guide, fuses and magnetic circuit breakers are designed to switch the circuit off when the "rated" current is exceeded by 160%.

    In the case of fuses this may take several minutes at the over-current condition, whereas a circuit breaker will respond in less than one-half cycle - resulting in substantially less damage to components.

    Note: In the case of a magnetic circuit-breaking device, breaking a Direct Current circuit when under load is a difficult exercise because of arcing at the contacts - which can cause severe damage and/or welding of the switching contacts. A general rule is to use a circuit breaker having a rated AC breaking capacity at least twice the required DC breaking current capacity. In the case of high voltage - eg B+ circuits, a two-pole circuit-breaker should be used and the circuit breaker contacts connected in series, to halve the voltage across each set of contacts.


    Two very commonly asked questions are:

    1.    Does the power supply affect the sound of an amplifier?

    2.    If I change the power supply design will this change the sound?

    The answer is a definite "YES" to both questions.

    The reasons are simple.

    First, the entire power supply is connected in parallel with the entire amplifier and in series with the output stage - ie between the Plates and Cathodes of the power tubes, so any characteristic of the power supply will be in series with the output and will therefore directly affect it.

    Second, the amplifier power output, and therefore transient response capabilities, are wholly dependent upon the power supply being able to deliver the required power at the required instant in time.

    Another way of expressing it is that the amplifier is actually a simple modulator of the power supply. It is easy to see that if the power supply was connected to the amplifier unconstrained, it would deliver its full power capability. Conversely, if the power supply is completely constrained, then no power will be supplied. What happens in practice is that the power output from the power supply is "modulated" or "regulated" by the audio signal.

    Class B output stages are most affected by this inter-relationship and inter-dependency.

    It follows that:

    a)    a tube rectifier power supply will sound different to a solid state one.

    b)    a full-wave bridge, voltage doubler or full-wave design will each sound different when connected to the same output stage.

    c)    the total value and type of filter capacitor(s) and choke(s) installed will affect the sound.

    d)    a choke input to filter regulation will sound different to a capacitor input to filter design

    e)    a small VA power transformer will not deliver as much useable power as a large VA transformer

    f)     a regulated power supply will sound different to an unregulated power supply

    g)    a power supply common to all stages of an amplifier will not perform as well as a discrete separate power supply for each stage (because all stages and the filaments/heaters will drop voltage simultaneously)

    h)    the better the regulation of the power supply, the better the sound - particularly transient repsonse

    i)    installing a 50k Ohm 20Watt wire-wound resistor across (shunting) the final electrolytic capacitor at the output transformer centre-tap has the effect of stabilising the AC return circuit to ground and will minimise effects of variations in impedance v frequency in the power supply - ie helps to smooth-out frequency response and reduce "resonance". RCA recommend a resistor drawing 10% of B+ total current to improve regulation.

    This resistor will also automatically discharge the filter capacitors in the Plate circuit when the amplifier is switched "off". Note however that with this configuration the Screen Grids should have a similar circuit installed to prevent them from being energised before the Plates when switching on.

    j)    when the tube Plate to Plate Load impedance is low - eg as in parallel-push-pull tubes - a non-polarised polypropylene or oil-filled paper capacitor will sound better than a polarised electrolytic capacitor because electrolytics do not conduct evenly in both positive and negative alternations of the output signal. Being in series with the load impedance - and therefore being a direct component of the load - they will directly affect the sound - particularly at the upper and lower ends of the audio frequency range.

    Note however that adequate direct current (DC) must be stored for transient peak signals.

    Therefore a combination of polarised and non-polarised capacitors will deliver superior performance than either alone.

    k)  where a separate discrete Screen-grid power supply is provided,  it is most important for full-power hi-fi reproduction at very low frequencies to ensure the value of the final Screen-grid B+ bypass filter cap is not less than the value of the Plate circuit B+ bypass filter capacitor situated at the centre-tap of the output transformer.



    Review of past design practice shows us that many designers had no idea of the principles essential to optimising chassis layout.

    In a wide range amplifier, the wiring and componentry act as antennae, to pick up stray and induced signals, such as ultra-sonics,
    RF, hum and noise, from adjacent circuitry.

    The basic rules for component layout and wiring are:


    In some cases it is necessary to instal a mains voltage matching transformer to increase or decrease the mains supply voltage available to that required by the amplifier.

    This is common for top-end or public address amplifiers sourced from foreign countries.

    The user should be aware that the introduction of an extra transformer into the supply chain has a similar effect to the regulation characteristics of the conventional on-board power transformer.

    Consequently the fastidious user may notice some slight reduction in performance on transient signals resulting from voltage drop in the matching transformer.

    One way to minimise the effect is to lightly load the transformer. Just use a matching transformer that has say twice the rating as the amplifier rated full-load power requirement as provided on the rating plate - not its rated power output.

    A convenient rule of thumb is to use a mains matching transformer having a rated VA power of not less than FIVE times the rated audio power output.

    For solid-state amplifiers this could be reduced to THREE times the rated audio power output.

    12.    APPEARANCE

    It is an axiom of engineering practice that a component or complete device must look attractive.

    This rule also applies to audio amplifiers.

    If it does not look good then it is unlikely to work well, so abandon it and start again.



    Contact: "electron"


    Email:      contact

    This page is located at http://www.oestex.com/tubes/power.html

    This page last modified 11 July 2012

    OES Home Page