Vacuum Tube (Valve) Radio and Audio Repair Guide: Finding faults on output stages

Home | Audio Magazine | Stereo Review magazine | Good Sound | Troubleshooting

Since there is little difference in AC and AC/DC techniques other than in the power supplies, in this section and those following the notes will apply equally to either type. The only point to remember is that AC sets can be expected to have higher HT voltages, perhaps even twice as much, than AC! DC models and replacement components should be rated accordingly.

FIG. 1 shows a typical output stage having the triode section of a double-diode-triode driving a pentode, with conventional matching transformer to a permanent-magnet loudspeaker. Capacitor C1 across the transformer primary is included to reduce the high frequency response. If it should happen to become a dead short it would effectively silence the set. The test meter should show 200 ohm to 500 ohm across the primary, and a low reading indicates a fault, hopefully in the capacitor, because it is a lot cheaper to replace than the transformer itself!

FIG. 1 Basic AF amplifier and single-ended output stage. Typical component values.

It is rare, but by no means unknown, for the primary winding to short internally with a drastic reduction in volume. If, on the other hand, C1 should go open circuit it can sometimes result in high pitched whistles in the loudspeaker, giving the impression of RF or IF instability. If a replacement has to be fitted, use one rated at 1000V DC, because very high peak voltages are built up on the transformer. This is particularly so if the set is run with the speaker disconnected for any reason. Sets have been seen in which the voltage has been so great as to break down the insulation of the output tubes (valves) holder and to burn part of it away.

Replacing output transformers

The new transformer (whether genuinely new or second hand) must match (have the correct primary to secondary ratio) the output valve, or tubes (valves) in the case of push-pull stages and it must also be capable of carrying the same amount of anode current. The service data for the set usually will give the anode current for the output tubes (valves) but otherwise it will be found in the Tube (valve)Data Appendix (2) to this guide.

The ratio is determined by taking the square root of the figure obtained by dividing the optimum load for the tube (valve) by the impedance of the loudspeaker voice coil. Optimum ratios for all the common output tubes (valves) will be found in Appendix 2 to this guide. The loudspeaker impedance may be quoted in service data for a particular set, but if not an average figure for vintage tubes (valves) receivers was 3 ohm.

If you are lucky enough to have a new replacement the specifications may be shown on the box or on a label fastened to the transformer itself; if not you will have to use judgment and a simple test. Generally speaking, the physical size of the transformer will reflect the rated anode current, so as long as the new transformer is at least as big as the old, all should be well. In fact, you will find that there was a surprising amount of standardization of size and fixing centers for fastening bolts which often helps a good deal when replacements are made.

As regards ratios, it is no good trying to compare the resistances of the primary and secondary windings as they will be of completely different gauges of wire. What you do is to connect the secondary to a source of low voltage AC, say about 2V, and measure the voltage that appears on the primary. Divide the latter by the former and you have the ratio. Don’t take too long about this in case of possible overheating of the output transformer.

Watch out for negative feedback

Many sets had tone control or tone compensation systems using negative feedback derived from the secondary of the output transformer. A replacement transformer must be correctly ‘phased’ to give negative feedback and you will soon be made aware of a mistake in this respect. When the secondary is connected the ‘wrong way round’ the result will be positive feedback which will provoke instability and maybe ear-splitting oscillation. The remedy is, of course, to reverse the connections to either primary or secondary, whichever is the more convenient.

Transformers with three windings

Philips and Mullard receivers in particular used negative feedback derived from a tertiary winding on the output transformer. In the absence of an exact replacement the following expedient usually will give complete satisfaction.

Experience shows that it is always the primary winding of these transformers which fails by going open circuit. Fit a replacement transformer as detailed above but leave the old one in situ, with the leads to the tertiary winding undisturbed. Connect the secondary of the new transformer to the loudspeaker as usual, and also to the secondary of the old transformer. This will then act simply as a low impedance transformer delivering negative feedback as before; the same precautions regarding phasing must, of course, be observed.

Some causes of low or distorted reproduction

An all too common cause of distortion, but with the level of volume little affected, is the coupling capacitor between the triode anode and the control grid of the output valve. If it starts to ‘leak’ even a little positive voltage will get onto the grid and cause it to draw excess anode current. This is so potentially harmful, not only to the tube (valve) itself but to the output transformer and HT supply as well, that a check on the coupling capacitor should be regarded as top priority in any set. The tests are a little different for sets with conventional cathode bias and older ones having negative bias.

With cathode bias, the control grid voltage should be zero. If a positive reading is obtained on the meter, it can usually be put down to a faulty coupling capacitor, with an internal short in the tube (valve) as an outside chance. There is a very simple test to determine which actually is the culprit. With the meter still connected to the grid, short the anode of the triode to chassis with the blade of an insulated-handle screwdriver (it is perfectly safe). This removes HT from C3, and if the meter reading drops sharply the capacitor clearly has been leaking. If no difference is recorded, the tube (valve) itself probably is at fault, the most common failing being a leakage between grid and cathode.

With negative bias, there should be an actual negative reading on the control grid; it may not be large, even when a sensitive meter such as the AVO ‘8’ is used, because the voltage probably will come via one or more high value decoupling resistors. Carry out the same test of shorting down the triode anode and see if the negative voltage increases. Alternatively measure the anode voltages of the output tubes (valves) with and without the triode anode shorted down. If the voltage rises when the short is applied you again need to change the coupling capacitor. If no rise is recorded, don’t immediately assume that all is well as another possible fault must now be investigated. It has just been stated that the negative bias arrives via high value resistors. Associated with these are various decoupling capacitors which, if leaky, can reduce severely the negative voltage applied to the grid. The only true test here is to remove one end of each capacitor from circuit and measure across it with the ohmmeter. Even a very slight leakage means that a replacement capacitor must be fitted.

Replace faulty coupling or decoupling capacitors with good quality types of adequate voltage rating. Experience again shows that a very high proportion of receivers coming in for repair suffer from this type of failure and unfortunately, before it has been discovered and put right, damage may have been done to other components, plus the output valve. For instance, the excessively high anode current due to the positive grid bias may have caused the bias resistor R2 to overheat and change its value. In extreme cases it is known for the voltage to rise above the working value for the by-pass capacitor C2, breaking it down completely. If it should not actually read dead short, try the effect of bridging another across it, since an open circuit capacitor can reduce the output considerably in sets not too well endowed with gain. Note, however, that certain models were deliberately deprived of this component in order to introduce negative feedback and thus improve the bass response of the output stage.

The most serious result of a persistently over run output stage is, apart from damage to the tube (valve) itself, an overheated and ultimately shorted output transformer, plus, of course, consequent harm to the HT supply components. It is certainly not rare for a simple coupling capacitor with a ‘leak’ to cause a mains transformer to burn out, so never neglect to apply the simple tests outlined above.

A less common cause of low volume is failure of the triode anode load resistor, R5 in the diagram. When this goes open circuit the effect is usually not to silence the set completely, as might be expected, but rather to reduce the volume to a whisper. Here’s a curious fact: the writer has found over fifty years of repairing radio sets that for some reason or other the value of anode resistor most likely to fail is 220 k-ohm. Why this should be is inexplicable, but it has happened time and time again. Whatever its value, though, the triode anode load resistor should always be a prime suspect in cases of low or distorted sound.

What about the output valve?

It was stated elsewhere in this guide that a radio set could act as its own tubes (valves) tester to a certain extent, and here is a good example. If all the coupling and bias components around the output valve, and its HT supplies, are in good order but there is still low or distorted output, you need to know whether or not the tube (valve) is drawing the correct anode current. You could, of course, disconnect the HT feed to the output transformer and insert a milli-ammeter but a much easier and effective method is to measure its cathode voltage. The correct voltage will usually be given in the service data for the set; otherwise consult the tube (valve) tables in Appendix 2. A low reading, indicating low anode current, suggests that the tube (valve) has lost emission, a likely sequel to persistent overloading. A high reading, indicating excess anode current, suggests an internal short in the valve. In either case replacement is going to be the only answer.

In receivers using negative grid bias there will be no handy cathode voltage to check, as it will be connected directly to chassis. In this case, discover the DC resistance of the primary of the output transformer, either from the service data or by direct measurement with the ohmmeter, then determine how much voltage is being dropped across it. Simple application of Ohm’s law will then tell you how much anode current is flowing through the winding, with the same interpretation being put upon the readings as for cathode voltage.

Heater-to-cathode leaks

It is worth mentioning at this point that even a slight breakdown in the insulation between the heater and the cathode of a tube (valve) can put AC on the latter and bring about a 50 Hz hum in the loudspeaker. If a hum of this type is experienced that does not appear to be due to an HT smoothing problem, suspect an h/k leak. Note: this fault is largely confined to AC! DC tubes (valves) with high heater voltages.

Push-pull output

This feature is found mainly in ‘luxury’ table models and large radiograms. Decca, HMV/Marconiphone and RGD were probably the leading exponents both before and after the Second World War, all producing large (sometimes immense!) record players and radiograms with tremendous outputs, measured it must be stated in good old British watts (RMS). Push-pull was by no means confined to the expensive end of the market; however, one of the cheapest examples being provided by the Barker ‘88’, a mail-order set from the eponymous firm which boasted eight tubes (valves) and eight watts push-pull output for a price of only eight guineas.

Fault finding on push-pull output stages follows the same lines as for conventional (what are called ‘single-ended’) stages. Be warned, though, that generally speaking the larger the set, the higher the HT voltage, and in the real monsters it is frequently around 500V, so they need to be treated with respect.

Quite often the output tubes (valves) used in the very large receivers were directly heated power triodes, exemplified by the Marconi-Osram types PX4 and PX25. Their low internal impedance facilitated good matching to loudspeakers and extremely high quality sound reproduction could be realised (the term ‘hi-fi’ dates from the 1930s). Even today the PX4 and PX25 are in great demand and fetch large sums of money in specialist radio auctions such as those run by Radiophile magazine. For this reason it is sensible to make quite sure before assuming that one of these tubes (valves) has ceased to work.

Incidentally, don’t trust entirely to a resistance reading to assume that the filament is in good order, for only part of it may be continuous. There is a good visual test which may be applied. Their multi-section filaments are supported and kept taut by four tiny springs at the top of the electrode assembly. If any of these springs should appear to be loose or even missing it is a sure sign that part of the filament has collapsed.

Various types of bias are employed with pairs of directly heated tubes (valves). The least expensive method, with a single LT winding to supply both, is to use the filaments as a virtual cathode by wiring a ‘hum dinger’ across each with the sliders connected down to chassis via bias resistors. A dearer alternative is to employ two separate LT windings, each centre tapped with the taps returned to chassis via bias resistors. In both these cases the actual values of the resistors should be checked carefully to ensure that they are the same within a few ohms, especially after the sort of overloading brought about by leaky coupling capacitors.

A third method is to use grid bias, usually supplied by negative HT smoothing. The same remarks apply regarding possible leakage via decoupling capacitors as for single-ended output stages.

The other favorite output tubes (valves) for high quality push-pull stages was and is the Marconi-Osram KT66 beam tetrode. This tubes (valves) is almost identical electrically to the American 6L6G but such is the power of cachet that it costs five to ten times as much to buy. It is fairly common for the tube (valve) to be triode connected, with the screen grid strapped to the anode. Either cathode or grid bias may be used. A little further down the ladder came most of the popular beam tetrodes and pentodes, which could offer very good quality but did not have the charisma of the types just mentioned. In all cases, when cathode bias is employed, there may be a single common resistor or separate ones for each valve. Negative bias is not quite so likely to be found. Carry out the same checks as already described above.

For a push-pull stage to work properly it is, of course, essential for both tubes (valves) to be ‘matched’ as regards emission. This may be checked in the set itself by using the same techniques as described above for single-ended stages. Don’t forget when you measure the resistance of the output trans former primary, or check the voltage drop across it, to apply the meter from the center tap to each anode in turn. The resistance check also will reveal any serious imbalance in the windings such as might be due to shorted turns on one half of the primary.

Note that push-pull stages driven by phase splitting tubes (valves) usually will provide a reasonable output with one of the output tubes (valves) removed.

This makes it possible to compare the performance of the two tubes (valves) by removing each in turn. It will not work, however, with some of the rather cheap and nasty push-pull stages used in certain ‘high quality’ sets of the late 1950s, in which one of the output tubes (valves) acted as a phase reverser to drive the other. Beware of this feature particularly in sets using EL41 output pentodes.

It is also necessary for both output tubes (valves) to receive the same amount of AF drive. This is one of the few times when an oscilloscope comes in really handy, especially the ‘double beam’ type with which both grid inputs may be viewed and compared simultaneously. Otherwise it is a matter of ensuring that all the resistors in the phase splitter whichever type this may be, are of the specified values as shown in the service data. It is particularly essential for the anode and cathode load resistors in a ‘concertina’ splitter to be matched accurately.

Hot tubes (valves)

Experience has shown that the later all-glass tubes (valves) such as the EL84, ECL82 and ECL86, which all run very hot in service, are all subject to internal leakage rather more than the old-fashioned large tubes (valves), which had wider electrode spacing and dissipated heat better. When these small tubes (valves) are used in push-pull amplifiers it is essential to keep an extra close eye on the bias resistors and capacitors for signs of overloading.

‘Crackly’ tone controls

When a simple variable top cut type of tone control consisting of a fixed capacitor and variable resistor in series is wired between the anode of the output tubes (valves) and cathode or chassis, a leakage on the capacitor will permit DC voltage to flow through the track of the resistor. This will cause a ‘crackling’ sound as the control is turned, the intensity of which will vary with the amount of voltage leaking through the capacitor: if this should be very serious the track of the resistor could burn out.

It may be possible to clean up a control that has suffered from only a minor DC leak. Take off the metal cover by bending out the three or four small tabs which hold it to the molded part of the assembly, which will give access to the track. Gently rub a soft-lead pencil (2B or more) over the track to deposit graphite on it. Then smear it with silicon grease or Vaseline and reassemble.

With any luck this treatment will have done the trick if not there is one more old service engineers’ trick worth trying. Gently pull the spindle of the control outwards as you turn it to see if this removes or reduces the crackling. If it does, with the spindle pulled outwards wrap a turn of fine wire around it between the retaining circlip and the threaded bush. Twist the ends of the wire together to hold it in place and then secure it with a blob of solder.

Note: the same methods are also applicable to crackly volume controls.

Prev. | Next

Top of Page   Related Articles    Home

Updated: Friday, 2020-05-22 19:03 PST