Audio Output Stages

Loudspeaker Connections

Before delving into the amplifier, it is worth mentioning that many sets have some means of disconnecting the internal speaker when using an external speaker. This could be a switch, plug or screw, and is often the cause of silent sets. I have been asked to repair a couple of supposedly dead sets, where the problem turned out to be nothing more than a loose speaker disconnection screw!

In this diagram, from the Bush AC11, the internal speaker switching screw is S23. The output transformer is mounted on the speaker, and the connections to the chassis are indicated on the diagram as "1" and "3". If this is a plug and socket connection it could come detached, again resulting in an apparently dead set.

If the chassis has been removed several times, the speaker or output transformer connections could have been disturbed. It is not unknown for a wire to drop off, and then be reconnected in the wrong place.

Speakers do occasionally fail, but this is not common. If in doubt, an external speaker can be connected to confirm. A more common problem is warping of the cone or speech coil, resulting in distortion as the coil scrapes against the magnet. There is of course the possibility of physical damage to the speaker cone, but this would normally result in the grill cloth being damaged too, so would be easy to see.


Audio Output Stages

The vast majority of sets will use a Class A output stage, with a single valve driving the output transformer. The output valve is normally a pentode or beam tetrode. Power triodes were used in early battery sets and high quality mains sets of the 1930s and early post-war period.

The output stage of the GEC BC402 set is shown here. V5 is the output valve, driving output transformer T1. The quiescent anode current is generally between 25 and 60mA (50mA in this case). This current passes through the primary of the output transformer, which will drop typically 15 to 25V. If this voltage drop is present, it confirms that the valve is drawing anode current.

A low value resistor (a few hundred ohms) is connected in series with the cathode, and carries a similar current to the anode. The voltage drop across this resistor (R34) holds the cathode a few volts above the ground level. The cathode resistor is sometimes bypassed with a parallel capacitor (C57), typically 25uF or 50uF.

The control grid is held at ground potential by a high value resistor (R30), and is thus negative relative to the cathode, giving the correct biasing conditions. The biasing voltage is dependant on the valve type used, and can be obtained from the service sheet or valve data book.

Sometimes a resistor (R32) will be positioned in series with the control grid of the output valve. This combines with the input capacitance of the valve, giving a low pass filter that attenuates the high frequencies above the audio range and ensuring stability.


Output Stage Faults

A very common fault is a leaky grid coupling capacitor (C54). This would cause the control grid to be at a potential above 0V, which alters the biasing of the valve causing it to draw excessive current. This problem is so common and familiar that Radiophile magazine and many restorers refer to it as "THAT Capacitor". This is one capacitor I nearly always replace, often before I apply power to the set for the first time.

If the problem is ignored, it will get progressively worse until the valve fails, often destroying the output transformer in the process. Be sure to check this point even if the set appears to be working correctly. It is a simple matter to measure the voltage across the resistor with a digital meter. If there is any voltage here, the capacitor must be replaced.

Internal leakage within the valve can cause a similar situation to occur. This less common than a leaky capacitor, but it does occur particularly in later AC/DC sets due to the high operating temperatures. The UL41 valve seems particularly prone to this - and unfortunately replacements are quite rare and expensive (around £12 - £15).


Shortcut Test

There is an alternative method of checking for both these problems. It is quick and effective, but NOT recommended to beginners because of the risk of serious error!

Connect a voltmeter to the HT supply and note the reading. Now carefully short the anode of the triode amplifier valve (NOT the output valve) to the chassis with a screwdriver blade or whatever (the anode is fed from a high value resistor so no harm will be caused). In the circuit above, one would short the anode of V4 (pin 9) to the chassis. If the HT voltage rises, the control grid coupling capacitor (C54) is indeed leaky and must be replaced. A loud click will emanate from the speaker unless the output valve or output transformer is dead (be ready for the click so it doesn't make you jump!).

If the capacitor is OK, short out the grid leak resistor (R30). If the HT voltage rises there is internal leakage in the valve, which must be replaced.

I must emphasise again that this technique is offered as a suggestion to experienced engineers who know what they are doing. If one accidentally shorts the wrong point, there is a high risk of causing serious damage.


Screen Grid

The screen grid is often held at a similar potential to the anode. Any ripple here would appear as hum on the output, so the screen grid is connected after the next stage of decoupling (R36, C60). This is also at a slightly lower voltage, which compensates for the drop across the output transformer (the screen grid and anode should normally be at approximately the same voltage).

The anode and screen grid voltages are given in the service sheet. Normally the anode voltage is about 15 or 20V lower than the HT due to the drop across the output transformer primary. The screen grid is normally at a similar voltage, but is some sets it is significantly lower. If the screen grid voltage is incorrect, the anode current will also be incorrect, which will be shown by an incorrect voltage reading across the cathode resistor.

Varying the voltage on the screen grid will affect the anode current. On some sets, the screen grid is deliberately held at a lower voltage to give a lower anode current.


Output Transformer

The purpose of the output transformer is to match the high voltage high impedance of the output valve anode circuit to the low impedance of the loudspeaker.

The inductive nature of the windings gives a variable impedance load on the output valve anode, giving a frequency response, which accentuates the higher frequencies. The impedance (and thus the frequency response) is corrected largely by one or more capacitors (and possibly resistors) in parallel with the primary. In this circuit, C56 and R33 are the tone correction components. The capacitors may alternatively be connected between the valve anode and 0V, the effect being the same.

In some cases a variable resistor or switched capacitors are included here as a tone control (R22 and C30 in the Bush diagram at the top of the page), but in most cases the tone control is on the control grid of the output valve (C55 and R31 in the GEC circuit) or across the volume control.

The output transformer is prone to failure, in the form of the primary going open-circuit. This will result in a silent set, with a high HT voltage. Since the output valve has no load on the anode, the screen grid will act as an anode and will become very hot. This can sometimes be seen as a hot spiral of wire glowing inside the valve.

The output valve sometimes fails at the same time as the transformer (it is probably valve failure that destroys the transformer). If in doubt it is worth getting the valve tested, as a repeat performance is hardly desirable! Failure of the valve or output transformer often results in damage to the cathode resistor (R34 in this case), so this should be checked with a test meter. If the resistor is open circuit and is bypassed by a capacitor (C57 in this case), the capacitor will almost certainly be damaged and should be replaced.


Replacement Output Transformers

Because the output transformer carries a standing DC current, a conventional transformer cannot be used. Any replacement transformer must be designed specifically for this use.

RS Output Transformer - Click for larger imageRS Components sell a suitable transformer (stock number 210-6475, price £7.02 + VAT in the September 2001 catalogue), which has several tappings on the primary and secondary. Normally terminals 1 and 4 on the primary, and D and B on the secondary, gives a suitable ratio for mains sets. For a battery set try terminals 1 and 3 on the primary, and D and C on the secondary. These connections are a suggested starting point and may have to be changed in some cases.

The catalogue gives a table showing the connections required for various primary and speaker impedances, which is reproduced here. If the image is not clear enough, click here for a larger copy. The primary impedance is the Ra figure for the valve (taken from a valve data book), and the speaker impedance is generally 3R.

Obviously this transformer is a modern component and it will probably look quite out of place on a vintage chassis. However, it is fairly small so it may be possible to hide it below the chassis, and leave the defective original transformer visible above. Also the sound quality using this replacement will probably not be as good as that with the original transformer - but it is obviously much better than a dead set!

Do NOT use 100V line transformers or mains transformers. These are not suitable for use as valve output transformers, because the standing DC current will saturate the core. They will therefore run very hot, give a weak distorted output, and not last for very long. They may also cause damage to the valve.

Valve radio dealers may stock suitable replacement transformers, or may have a stock of used components from scrap sets. I have used the output transformers from most of my scrap chassis to repair other sets.

If all else fails, or you want to keep the original transformer (particularly if you are repairing a high quality or valuable set), you may be able to get it rewound professionally. This is likely to be more expensive than the replacement options, but sometimes it is worth spending the money.


Alternative Output Biasing

Some sets (notably many made by Ekco) uses an alternative method of biasing the output valve. The circuit of the Ekco U245 is shown here. R13 and R14 are low value resistors (39R and 47R respectively) and are placed in the 0V connection. Thus the total HT current passes them, and they drop a few volts. The cathode of the output valve is connected directly to 0V, and the control grid is biased to the lower end of the resistors, thereby obtaining the negative bias.

From the service sheet, 3.8V is dropped across the resistors. This is less than the figure of 10.4V given in the valve data book for the UL41 valve used. To compensate for this the screen grid is connected to a lower than usual voltage rail (due to R15 and R16). This reduces the anode current, compensating for the incorrect biasing level on the control grid.

Chas Miller adds:

Negative bias was found in a fair number of mid-1930s receivers, notably those made by EMI, and was used in connection with biasing the output valve(s) and for amplified AVC systems. The more modest type shown in your diagram was used again for biasing the output valve and also for providing both a delay voltage for the AVC and standing bias for the mixer and IF amplifier valves. By this means, the number of bias components was minimised and much space was saved in small sets (to say nothing of cost).


Output Transformer Tapping

In many cases, a small section of the output transformer primary winding is connected in series with the decoupling resistor feeding the remainder of the set. The arrangement in the Bush VHF90C is shown here. The incoming HT arrives at the tapping and the HT to the remainder of the set is decoupled by the smaller section of T1 primary and R33. The purpose of this arrangement is to cancel out hum.

This can cause problems if the output transformer fails, as the suggested RS replacement cannot be used this way. You may be able to obtain a replacement (perhaps from a scrap set) with a suitable tapping, or decide to get the original rewound.

Otherwise, the best approach would be to ignore the tapping and rely on R33 alone for decoupling. This will generally work fine, but if hum is a problem it can generally be overcome by increasing the value of the decoupling capacitor (C55).


Push-Pull Outputs

A few sets use a push-pull output stage. This circuit has two output valves and a centre tapped output transformer. The output valves are biased slightly into conduction, and are driven by a pair of anti-phase signals from a phase splitter circuit.

This arrangement is used on some battery sets to reduce current consumption (particularly at low volume). The two output pentodes are often contained in one valve.

It is also used in some larger mains sets, radiograms and valve hi-fi amplifiers to give an increased power output, higher quality and reduced distortion.

However compared to the normal arrangement, the circuit needs an extra output valve, a phase splitter valve and a more complex output transformer so it was often limited to the more expensive receivers (the Barker 88 being a notable exception). High power output valves are normally pentodes, and occasionally beam-tetrodes or triodes.

The circuit shown here is the output stages of the GEC BC9835 radiogram. V6 is the phase splitter, and V7 and V8 are the output valves. The output transformer has a centre-tapped primary, the centre tap being connected to the HT supply, and the ends to the output valve anodes.

Depending on the biasing the stage may be operating on Class A, Class AB or Class B mode. I do not intend to go into this in great detail because push-pull outputs are not encountered that often, and it is generally somewhat irrelevant to know which biasing mode is used when repairing the set. Very briefly:

For more information on push-pull output stages, please refer to some of the valve amplifier websites listed on the Links page.


Additional Audio Amplification

The output stage is normally preceded by an extra stage of audio amplification. This is generally a triode circuit, and a single valve often contains this triode and the detector diodes (UBC41 or UABC80 for example).

In some later low-cost sets the triode is incorporated in the same envelope as the output pentode (for example a UCL82 valve) and the detector diodes are in the same envelope as the final IF amplifier (e.g. UBF89 valve). This reduces the valve count by one compared to a standard lineup, giving an appropriate reduction in cost.

In the GEC BC402 set shown earlier and repeated here, the triode amplifier is part of V4. This stage normally has no external biasing voltage applied. The biasing is derived by the diode action of the grid and the very high value of R28 (typically 10M) giving a slight negative voltage on the grid. Because of the high impedance, the grid coupling capacitor (C52) should be in reasonable condition, but because it only has to handle a couple of volts this is generally much less of a problem than C54.

This stage is normally reliable, although the anode resistor (R29) does have a tenancy to go open-circuit or high in value. This resistor normally has a value of around 220K.

In a few sets, the anode supply is separately decoupled (typically a 100K resistor and a 0.01uF capacitor), and the decoupling resistor or capacitor may fail.


Short Superhets

Sets which do not have this extra amplification stage are known as "short superhets", and sometimes need a good aerial to give acceptable performance on more distant stations.

The output valve used in these sets is a "high slope" type, which means it has more gain (amplification) than normal output valves. This partly compensates for the missing stage.

Ultra made many short superhets; the 101 featured in this diagram being a typical example. The output valve (V3) is an AC2-Pen-DD which also contains the detector and AGC diodes.

The complete set uses only four valves including the rectifier, and given a reasonable aerial would give perfectly acceptable reception. They were normally sold as lower cost sets intended for good reception of more local stations, and would be absolutely fine for people living in towns and cities served by a nearby transmitter.


Tone Control

Most sets have just a simple form of top-cut tone control. This is normally in the form of a capacitor and variable resistor in series, connected between the audio signal and 0V at a suitable point. Normal positions are across the volume control on the input of the preamplifier valve, between the control grid of the output valve and ground, or around the output transformer as mentioned previously. This is effectively a treble or top-cut control and does not affect the bass response. The circuits on this page show various examples.

A few sets are fitted with "speech" switch. This normally removes some bass by reducing the value of one of the audio coupling capacitors or altering the negative feedback. Many cheaper sets has no form of variable tone control at all.

At the opposite extreme, some expensive sets have separate Bass and Treble controls, each giving both boost and cut. These are normally incorporated in fairly complex negative feedback circuits, and in a few cases an additional stage of amplification is included to cope with it.


Noisy Controls

Noisy and crackley tone and volume controls can often be fixed with a squirt of contact cleaning fluid (such as Electrolube X2). If there is a suitable gap in the metal cover the fluid can simply be squirted in using the tube supplied with the aerosol can. Suitable gaps can often be found above the connection tags or on the other side where a small section of the cover is pressed inwards to form an internal stopping lug. If there is no suitable gap, the control may have to be removed and dismantled. Do NOT use WD40 for cleaning potentiometers as this may remove the resistive material.

Another alternative if there is no suitable opening for injecting the cleaning fluid is to drill a small hole, about 1/16" (1.6mm) in the side of the case, squirt some fluid in, then reseal the hole with solder. Before drilling the hole, clean the area with abrasive paper to allow the soldering to be successful. If possible, orientate the chassis and drill so that you are drilling in an upwards direction; this reduces the chance of swarf from getting inside the control. A low speed rechargeable drill is ideal for this job. Be very careful as the drill breaks through, so it does not damage anything inside the control.

If the cleaning fluid is not successful, you will probably have to replace the whole control. This is usually straightforward with single controls and switched controls. The small locating lug on the new control will probably not match the appropriate hole in the chassis, and the easiest solution is to simply cut or break the lug off.

Many volume control pots are 500K and include the mains switch. This value is not readily available with a switch now, but a 1M0 log pot can be used if a 1M0 resistor is connected in parallel with the track. Note that volume and tone pots are nearly always log law - so be sure to order the right type! In some sets, additional tone correction components are connected to a midpoint tapping on the pot track. Since this type of pot is no longer available, the tone correction components may have to be omitted, unless a second-hand control is available.

Switched pots will sometimes need to be replaced due to the mains switch failing, rather than the pot section. If the defective switch is removed from the faulty control, the remainder can be used as a spare non-switched pot for another set, or the pieces can be used to repair other controls.

In some later sets, two controls are incorporated into one component. This has a thin shaft passing through the centre of a thicker shaft, and is fitted with concentric knobs. These components are no longer available new, but you may be able to salvage something similar from a scrap set or obtain a second-hand component from a dealer.

Alternatively the control could be dismantled and the faulty pieces replaced with parts from other controls (such as the pots with faulty switches mentioned previously). This is only possible if the spare parts are from controls of the same make and style, but there seems to be four or five common types.


Negative Feedback

Negative feedback is often used in the output stage, and sometimes the preceding amplifier stage, to improve the frequency response and reduce distortion. It can be implemented in a number of different ways, so I will just describe a few typical examples. Obviously as more negative feedback is used the gain will be reduced, so there is always a balance between acceptable frequency response and distortion, and adequate gain to give good volume on weaker stations.

I previously mentioned that the output valve cathode resistor is sometimes bypassed by a capacitor. When the capacitor is omitted, negative feedback is applied to the output valve only. This arrangement will bring about a reduction in distortion, but will not correct the frequency response.

The Philips B3G85U set shown here has a full negative feedback arrangement around the amplifier stage, output stage and output transformer. The output from the transformer is attenuated by R33 and R25, and fed back into the bottom of the volume control.

The method of returning the negative feedback to the bottom of the volume control is fairly common, and has some advantages. When the volume is low the negative feedback is greater, giving improved quality when it most noticeable. When the volume is at maximum there will be no negative feedback, so there will be no reduction in gain on very weak signals, and adequate volume when it is needed. The only slight drawback is that there will be some output, albeit very faint, when the volume control is set to the minimum setting.

In some sets the negative feedback is picked off from the cathode of the output valve as shown here (Bush VHF90C again). This obviously only works when the cathode resistor is not bypassed with a capacitor.

In some cases, the secondary of the output transformer is connected in series with the output valve cathode resistor or the bottom end of the volume control. This may be the same winding as that used for the speaker, or a separate winding. A point to note here is that some AC/DC sets (notably Ekco) have a separate secondary winding on the output transformer for the negative feedback so that the main speaker secondary can be taken to external speaker sockets. This speaker winding will be isolated from the chassis. If a replacement output transformer (with a single secondary) is fitted the negative feedback must not be connected to the speaker winding as this will result in the external speaker sockets being connected to chassis potential, which could be very dangerous particularly if the mains connections were reversed. The best way around this problem would be to modify the negative feedback circuit by picking the signal from across the output valve cathode resistor, as in the Bush VHF90C above, and removing the cathode bypass capacitor if fitted.

Ranulph Poole added:

On the subject of negative feedback. 'Anode-to-anode' feedback (from the anode of the audio output stage to the anode of the audio amplifier) was a popular technique. However, it has two big disadvantages. Firstly, any hum on the output stage's supply rail gets injected into the amplifier; secondly, the amplifier stage has to work harder for a living, and could be overloaded as a consequence. It is usually better to take the feedback from the secondary of the output transformer. This should not be tried with AC/DC sets where the secondary feeds an 'External LS' socket - someone might get fried.

Negative feedback invariably uses just passive components, and is generally reliable. Capacitors are, of course, prone to leakage and this can cause similar problems to the leaky grid coupling capacitor mentioned previously.


Philips Positive Feedback!

Just to cause confusion, some Philips sets (such as the 310A shown here) have a peculiar arrangement, which uses negative feedback at low volumes and positive feedback at high volume. The tone control is often included in this arrangement too, resulting in a circuit arrangement which is rather complicated and difficult to follow. The positive feedback is insufficient to cause instability, but does give some increase in gain with a corresponding reduction in sound quality. Many such sets use a special type of output transformer with an additional winding or tapping for the feedback. From my experience these sets do not sound that good even when they are working properly (they seem shrill and lacking in tonal quality).

It is clear that the design is more critical, since this area of the circuit uses a number of 5% tolerance resistors, whereas 20% resistors are used elsewhere in the set.

This sort of arrangement may be encountered in sets made by Stella and Cossor too, since sets having these brands were produced by Philips (Cossor from the mid-50s).

If this circuit arrangement is causing problems and cannot be repaired properly (due to a faulty output transformer for example), you may have no choice but to remove the positive feedback components and accept a lower maximum volume. By fitting a bypass capacitor across the output valve cathode resistor, you can recover some of the loss in volume, again at the expense of some quality.


Replacing Awkward Output Transformers

There is another workaround to problems like the Philips 310A above.

Take a look at the Philips 170A circuit section shown to the right. The primary is tapped as described above, and there are two separate secondaries, one for the speaker and the other for feedback. Since the feedback winding has a potential divider across it (R32, R33) and the centre of this is connected to the output valve cathode, it is likely that this set uses both positive and negative feedback in the tone control arrangements.

A similar situation may be found in AC/DC sets, where one winding feeds the speaker and external speaker sockets, while a separate winding is used for negative (and occasionally positive) feedback. In this case it is for safety - the speaker winding (and therefore the external speaker sockets) will not be connected to the remainder of the circuit because the set has a live chassis.

These circuits are a real pain, because the transformers are completely non-standard.

Assume section D-E (the main primary winding) is open-circuit. Unless you can get an identical transformer (unlikely), a possible workaround is to obtain a more standard transformer (such as the RS one described above) and connect it as follows:

Disconnect the primary of the original transformer. Connect wires C and D together (we can live without the hum naturalising bit). Leave the old transformer in place, with the secondaries connected. Connect the RS replacement's primary to D & E, and connect the secondary to F & G. The new transformer will be acting as the output transformer, and the original will be acting as an autotransformer to provide the feedback. If the result is unstable, reverse the connections to the primary OR secondary (not both). The problem is that you have to fit the second transformer somewhere!

This suggestion is based on an idea in Chas's new book, and has not been tried by me. If you decide to try it, or a variation, please let me know how it goes.




This website, including all text and images not otherwise credited, is copyright © 1997 - 2006 Paul Stenning.
No part of this website may be reproduced in any form without prior written permission from Paul Stenning.
All details are believed to be accurate, but no liability can be accepted for any errors.
The types of equipment discussed on this website may contain high voltages and/or operate at high temperatures.
Appropriate precautions must always be taken to minimise the risk of accidents.

Last updated 14th April 2006.