For some unknown reason, the electrolytic capacitor voltage
ratings were not shown on the circuit diagrams.
The capacitors affected are:
All other capacitors should be rated at 500V, with the exception of C2 which only needs to be a 63V type.C5 220uF 25V C7 47uF 400V C8 47uF 400V C9 16uF 400V
In my March 2001 column I featured a 3W valve amplifier.
The circuit was very similar to the first amplifier I ever built
back in 1969. My original amplifier used an EL84 output
valve and an ECC83/12AX7 for the voltage amplifier stages.
The design in PW kept the EL84 in the output stage but I substituted
an ECC81/12AT7 for the voltage amplifier.
I included a suitable p.s.u. which used an EZ80 rectifier.
I gave the 6BQ5 and N709 as direct equivalents to the EL84. For Octal enthusiasts the 6V6G/GT is a good choice. Electrical equivalents to the 6V6G are the 6BW6 (B9A base) and the 6AQ5/EL90/N727 (all B7G base). An Octal, electrical near-equivalent to the ECC81 is the 6SL7GT. If you don't need so much gain, use an ECC82/12AU7. Or the 6SN7GT, if you're into Octals.
Substituting the rectifier is a little more tricky. In this design the rectifier heater shares the same earthed heater supply as the rest of the valves. Therefore, any substitute must have a high heater-cathode insulation rating. Preferably 450V or higher. Remember, the rectifier cathode is at full h.t. voltage while the heater is effectively at earth potential. Acceptable substitutes are the EZ81 (just a beefier EZ80), the B7G-based 6X4/EZ90 and the Octal equivalent of the 6X4, the 6X5/EZ35.
Most valve rectifiers have their heater internally connected to their cathode (or are directly heated, in which case there's no separate heater and cathode, just a filament) which solves the problem of high heater-cathode voltages. The problem now becomes one of providing a separate heater winding on the mains transformer for the rectifier heater. And this is how it used to be, most of the time. If you have a separate (5V) heater winding on your mains transformer then there are lots of rectifiers to choose from. If you wish to use any alternatives, including those I've mentioned, check the base connections in your valve data books. Alternatively, do a Web search on the valve you want to use. That'll probably get you all the data you want.
By the way, that's not my PW amplifier in the photograph. It's the audio and power supply stages of an a.m. radio I'm building. The frequency changer and i.f. amplifier will be built onto an identical chassis and then bolted to the side of the amplifier/p.s.u. chassis. This 'building block' technique is ideal for experimenters as it allows the rebuilding of part of a design, or even complete replacement, without disturbing the rest of the set. My PW amplifier could be built in the same way, and on a similar sized chassis.
Well, I always found it difficult to remember the equivalent American type numbers for the ECC81, ECC82 and ECC83. Actually, I had no trouble with ECC83 = 12AX7, but I could never remember which way round the other two were. Then came Tux.
Writing the equivalents side by side:
ECC81 = 12AT7
ECC82 = 12AU7
ECC83 = 12AX7
It's easy to remember the 12A-something-7 bit,
then the sequence is: 1 = T, 2 = U and 3 = X.
Which spells the name of our favourite penguin.
Output transformers are not as readily available as they once were. Neither are they as cheap. Those made for the Hi-Fi market are very expensive and are a total overkill for a project like this. And forget trying to use a mains transformer. Mains transformers are not designed to take a d.c. magnetising current and will cause loads of distortion. There are (relatively) inexpensive output transformers being made, but don't ask me where. Try a Web search.
All may not be lost if you have a old output transformer lurking around. Since we're not too bothered about frequency response, there's only two things to work out: the power rating of the transformer and the ratio. Power rating is a bit tricky, you've simply got to guess! For this design you're looking at something at least as big as a 15VA mains transformer. To work out the ratio, connect the primary to a 100V or so, 50 or 60 cycle (no Hz here - this Valve and Vintage) supply and measure the secondary voltage. Divide the secondary voltage into the primary voltage. That'll give you the ratio.
So, if the primary voltage was 110V and the secondary voltage was 2.75V, then the ratio would be 110/2.75 = 40:1. Which is quite a common ratio. Since most output valves (running single ended) like to see an impedance between 3000ohm and 7000ohm, and the most common loudspeaker drive unit impedances in the old days were 3ohm and 15ohm, we can guess what kind of impedances the transformer was originally designed for.
A transformer reflects impedances by a factor equal to the square of the ratio. Given our example has a 40:1 ratio, it'll reflect impedances by a factor of 40 x 40 = 1600. Trying a 3ohm loudspeaker, that'll 'look' like a 1600 x 3ohm = 4800ohm load to the valve. Not bad considering my PW amplifier likes to see about 5000ohm. If you get a low-ish ratio, say 18:1, then you can assume the transformer is wound for a 15ohm loudspeaker. 18 x 18 = 324. And 15ohm x 324 = 4860ohm. Some transformers have tapped primaries and/or secondaries. Just measure all the voltages and work out all the likely ratios and impedances. Alternatively, read the data sheet. If there is one. :-)
Ideally, a loudspeaker drive unit needs to be driven from a perfect voltage source. That's another way of saying that the amplifier should have zero - or at least a very low - output impedance. The reasoning behind this is two-fold: first, the loudspeaker manufacturer can design his loudspeaker without knowing the output characteristics of your particular amplifier. Secondly, as the loudspeaker cone moves in and out, the loudspeaker drive unit actually generates a voltage - a back e.m.f. After all, it's a coil moving in a magnetic field, isn't it. If this back e.m.f. 'sees' a low resistance, a high current will flow in the unit's voice coil. This current will oppose the movement of the coil, and so oppose movement of the loudspeaker cone.
To demonstrate this, try winding a hand generator. First into an open circuit. Then into a short circuit. See which is easier. :-)
Of course, this back e.m.f. is all mixed up with the driving voltage so don't expect to see it on an oscilloscope. But it is real. If it doesn't see a low resistance, the cone will flap about somewhat awful and the loudspeaker will sound 'boomy'.
Another way of looking at the situation is in terms of the bass resonant frequency of the loudspeaker drive unit. The physical mass of the cone will resonate at some (low) frequency. At this frequency the impedance of the drive unit will increase quite considerably and the loudspeaker will become much more efficient. The loudspeaker designer will try to tame this resonance but it'll still be quite pronounced whatever he tries to do. Still, this isn't a problem if the driving amplifier has a low output impedance and tries to output the same voltage regardless of the load impedance. At resonance the impedance rises and the drive unit takes less current. But that's compensated for by the increase in efficiency at this frequency, so the output of the loudspeaker remains fairly constant.
If the amplifier has a high output impedance, let's say it's almost a perfect current source, then nasty things happen. As the loudspeaker's impedance rises at the bass resonant frequency, the amplifier will try to push the same current through the drive unit's voice coil. But hang on, the efficiency has increased and the same current is flowing as before - the sound level will increase.
The upshot of all this is the sound output from the loudspeaker will be very dependant on the loudspeaker's impedance at any particular frequency. The most noticeable characteristic will be a pronounced bass 'boom' around the bass resonant frequency, although there may be minor resonances and other variations all over the place. Not good!
Back to our output stage, which is a pentode.
The same applies to a beam tetrode, by the way.
If you look at the graph, you'll see that (for a constant screen voltage) the current taken by a pentode is almost wholly dependant on the control grid (g1) voltage, providing the anode voltage is above a certain minimum value. Now, without going into detail, this means that whatever the load in the anode circuit of a pentode, the valve will try to push the same current through it. As we've seen above, the output transformer simply reflects the impedance of the loudspeaker by a factor of its ratio squared.
As the impedance of the loudspeaker varies, so will the reflected impedance as seen by the valve's anode. But the valve will always try to push the same current through its load regardless of the impedance of that load. We've been here before with our current source.
As normally applied, global negative feedback reduces the output impedance of an amplifier, making it appear more like a (perfect) voltage source. But a pentode - or beam tetrode - output stage without feedback will have an inherent high output impedance. This is why old valved radio sets, which seldom had any form of feedback, sounded boomy. Or warm, if you're a lover of old radio sets. :-)
I've included this photograph to show you a wiring technique
that you might like to try.
Fix a tagboard underneath the mains and output transformers.
Where appropriate, wire the components which are mounted
on the chassis to this tag board.
You can then mount most small to medium size components
on the tag board. But, please leave sufficient wire on each
component so it's easy to remove or replace it if you need to.
If you make a rough drawing of the location of the components on the tag board and label them as in the circuit diagram, you'll find that checking voltages and fault-finding is so much easier. One word of warning though, this technique usually means that the interconnecting wiring is longer that it could be. Where this might cause a problem - very high impedance circuits, stray capacitance/inductance, etc - use point-to-point wiring.
Return to the contents page