It is a common belief that valves are unreliable and that transistors last forever. In fact valves are nowhere near as unreliable as people think (repairers would often claim a valve had failed and charge for it to cover the cost of the time spent finding and replacing a couple of faulty capacitors, because customers expected valves to fail).
Modern silicon transistors are very reliable if operated within their specifications and recommended operating conditions (temperature, humidity etc). They will easily outlast the expected life of the equipment. However if they are pushed beyond their limits their life will be limited. This may be done to save money or to deliberately shorten the life of the equipment (why does so much cheap modern equipment fail a few months after the warranty expires?).
Early germanium transistors are less reliable, although this should not be regarded as a criticism since the sets were probably only expected to last five years or so, and nobody expected collectors to be repairing them forty years later.
AF117 series transistors
By far the most unreliable transistors are the AF117-series used in the RF and IF stages of many transistor radios produced in the 1960s. These have metal cans and four leads - the fourth lead being connected to the can for screening. The type numbers include AF114, AF115, AF116, AF117, OC170 and OC171. They tend to develop internal short-circuits between the can and either the collector or emitter.
The easiest way to detect this fault is to tap each transistor with the handle of a small screwdriver with the set switched on. When the offending device is tapped, the set will either crackle or burst into life briefly. The set may appear to come back to life permanently, but you can be certain it won't work if you try it again next week!
The following diagram and paragraph are taken from the book Electronic Classics written by Andy Emmerson.
"The cause of death or fault mechanism is rather interesting. Where you to open it up, you would find that the encapsulation surrounding the transistor contains silicone grease and an air space inside the metal can. From the inside wall of the can in the air space grow microscopic hairs (0.008mm across) of an unidentified medium which is tough, springy and electrically conductive. After some twenty-five to thirty years these conductive hairs reach the internal lead construction, giving the fault symptoms described. It is not clear if the air space is part of the design intent or is in fact a process error."
There is no ideal solution to this problem. The traditional engineers solution was to cut the screen lead, which will often get the set working again. However it is only a matter of time before another hair reaches another internal connection.
Any replacement transistor of the same type, whether used or new-old-stock, will have the same hairs growing and will almost certainly have the same problem sooner or later.
The short-circuit can be blown away by discharging a 50uF capacitor charged to 50V between the screen lead and the other three leads joined together. If a 50V supply is not available, a larger capacitor charged to a lower voltage (e.g. 1000uF to 9V) would probably have the same effect, although this has not been tested. Again with this solution we are on borrowed time because sooner or later another hair will reach an internal connection. If the set will remain in your own collection this is probably a good solution because it retains originality, and an occasional repair is not a major problem. This is my normal solution to the problem.
Until fairly recently a newer range of germanium transistors were available - the AF121 and AF125 being useful general-purpose devices that will replace the whole AF117 series. They have smaller cans and the leads are in a different order. This is therefore not an invisible repair, but they are reliable. However the AF121 and AF125 are no longer available, nor are any other high frequency germanium transistors.
Another alternative is to fit modern silicon transistors. Since the bias voltage is different (0.2V to 0.3V for germanium and 0.6V to 0.7V for silicon), the biasing arrangements need to be changed to suit. I plan to investigate the changes needed over the next few months and will add details here in due course. This is a repair method used by dealers selling working sets with warrantees.
Audio output and driver transistors can also give problems. They sometimes become intermittent, and the same screwdriver handle tap-test will often reveal the problem.
This seems to affect many audio transistors including OC71, OC78, OC81, AC127, AC128, AC176, AC187 and AC188. The problem is not as universal as with the AF117 series however, and replacement with a similar transistor from a scrap set will often be a permanent solution.
The AC1xx series occasionally develop intermittent short-circuits between the case and an internal connection. This only causes a problem if the transistor is mounted on a heatsink plate that is connected to the 0V rail of the set. I do not know whether the cause is the same as the AF117-type transistors, or whether the same cure will work. On the rare occasion I have encountered it, I have just replaced the transistor.
At the time of writing a limited range of germanium audio transistors are still available, but it is probably only a matter of time before they are discontinued. The AC128 is a good general-purpose replacement for OC71, OC78 and OC81. Note that AC127, AC176 and AC187 are NPN and the others are PNP.
Replacement with modern silicon devices is also a possibility, but it is currently not such a pressing issue as the AF117 series.
Other germanium transistors and diodes
The OC44 and OC45 used in early transistor sets tend to be reliable. There are enough spares from scrap sets etc. to provide replacements for the occasional faulty device.
Germanium diodes tend to deteriorate rather than completely die; often showing increased reverse leakage and increased forward resistance. A few germanium diodes (such as the OA47) are still available, but as with germanium transistors, their days must be numbered.
Modern small signal Schottky diodes such as the Philips BAT83 and BAT85 have a similar forward drop to germanium diodes at low currents, and could probably be used as direct replacements in many sets. Again I will be investigating this.
A suspect transistor can be checked with a multimeter on resistance range fairly easily. Since a transistor consists of two PN junctions (between base and emitter and between base and collector) the resistance of these in both directions can be checked. Also the resistance between collector and emitter should be checked - it should be open-circuit both ways around.
The results obtained depend on whether you are using a digital meter or an analogue meter. For a PNP germanium transistor you should get the following results:
Digital meter on 2k-ohm resistance range:
|B to E and B to C||Negative to B||Few hundred ohms|
|B to E and B to C||Positive to B||Open circuit|
|C to E||Negative to E||Open circuit|
|C to E||Positive to E||Open circuit or several hundred ohms|
|C to E and B||Positive to E||Open circuit|
Analogue meter (AVO-8) on X1 resistance range:
|B to E and B to C||Positive to B||About ten ohms|
|B to E and B to C||Negative to B||Open circuit|
|C to E||Positive to E||Open circuit|
|C to E||Negative to E||Open circuit or about 2 k-ohms|
|C to E and B||Negative to E||Open circuit|
The readings will vary with different meters, but the general pattern should be the same if the transistor is OK.
Note that in theory we would expect to get open-circuit readings between emitter and collector with the meter connected both ways around. However in practice we often get a small forward leakage current. This is more apparent on a digital meter due to the lower test current. To confirm that this is not a problem, link the base to the emitter to bias the transistor fully off and the resistance reading should increase to open-circuit. This is shown in the last line of the tables above.
You may have noticed that the meter polarity for an analogue meter is the opposite of that for a digital meter. This is due to the way the circuits in the meter are constructed and is exactly what we would expect.
When testing NPN germanium transistors the polarity should be reversed for each test, then the results will be the same. Most early transistor radios use PNP transistors. The only NPN transistors you are likely to encounter are in transformerless output stages.
When testing a transistor with a screen lead (such as the AF117 series), check the resistance between the screen and each of the other leads. It should read open-circuit.
Often when testing transistors, you don't really need to think about the polarity. Just check the base-emitter and base-collector junctions, and for each you should expect to see a reading one way round and no reading the other way round. Then check that the collector-emitter junction is not short-circuited.
Diodes can also be tested with a multimeter on resistance range. Since it only has two leads, there are only two tests - forward and reverse.
Digital meter on 2k-ohm resistance range:
|A to K||Positive to A||Few hundred ohms|
|A to K||Negative to A||Open circuit|
Analogue meter (AVO-8) on X1 resistance range:
|A to K||Negative to A||About twenty ohms|
|A to K||Positive to A||Open circuit|
Again readings will vary with different types of meter. If the second check reads anything other than open-circuit the diode is leaky.
Replacing transistors and diodes
If you read period books they will advise that great care should be taken to prevent the heat of soldering reaching the body of transistors and diodes. They will often suggest the use of a heat-shunt (croc-clip or long-nosed pliers on the lead to take away the heat) and will warn of dire consequences if this is ignored.
In practice, this is easier said than done. Transistors are often mounted close to the PCB so that there is no room to get a heat shunt in if you wanted to. If the leads are longer they are normally sleeved to prevent short-circuits, so again a heat shunt cannot be used.
The solution is to forget about the heat shunt, and to take care when soldering so that the component does not become too hot. The period books often suggested the use of a small soldering iron of around 15 Watts when soldering transistors. The idea was that the lower power would prevent that parts becoming too hot. In practice, if you use a low powered soldering iron it takes longer to solder the joint, so the heat spreads further, increasing the risk of damage.
I prefer to use a reasonably powerful iron, around 35 to 50 Watts, which has been allowed to heat up fully. I use a temperature-controlled iron set to about 350°C. With this, the joints can be soldered or desoldered quickly, and the heat doesn't have enough time to spread very far. Try to avoid heating joints for more than about 2 seconds. This should be sufficient time to solder or desolder a joint. Then wait 10 seconds or so to allow everything to cool, before doing the next joint.
For desoldering, use a spring-loaded desoldering pump to remove most of the solder. If the lead is folded over against the track, you will then need to heat the joint a second time and move the transistor body so that the lead lifts away from the track slightly. This should be sufficient to free the lead, and you can repeat the operation on the other leads - remembering to allow time for cooling between operations.
When fitting the new or replacement transistor, you may find it helpful to cut the leads to different lengths so that you can insert them into their holes one at a time. This is particularly helpful if the transistor is in an awkward position. Double-check that you have the leads in the right holes before soldering. I like to apply the minimum amount of solder required to make the connections at this stage, then test the set. If it works, the joints can be soldered with a bit more solder and the leads trimmed. I don't usually fold the leads against the tracks, since this makes any subsequent repairs more difficult.
Transistor and diode connections
The lead connections of the transistors are often included in the service data for the set. If not, it can usually be obtained from transistor data books. Data on many transistors is available on my Valve Data CD-ROM.
This diagram shows the lead connections for the Mullard transistors most often encountered in British transistor radios made in the 1960s (viewed from underneath). Although not indicated, the second and third package types also normally have a coloured dot on the side of the can by the collector lead. Do not assume that other transistors in similar cases have the same connections!
With diodes, the cathode lead is marked with
a stripe or different colour. Some diodes used for temperature stabilisation
of output stages are in the same type of packages as transistors; with these
the cathode lead is marked by a dot on the side of the case. The cathode lead
of a diode is sometimes marked with a "+" symbol, since it is the
positive output when the diode is used as a rectifier.