|
Page 2 of 4
Many are the supposed problems of zero-feedback (overall feedback) triode amps, but like the impedance difficulty I've just talked about, they are either illusory or soluble. The other hoary old chestnut, that of electrical damping and its affect upon bass quality, happens to be in the former category. I have spent weeks listening to this amplifier, swapping between it and various top quality solid state units and in all cases the triode amplifier had tighter and deeper bass. In reality, the series bass inductor within the crossover of a loudspeaker has a resistance of many ohms, high enough to cancel out any electrical damping effect. Add to this the predominance of magnetic and acoustic damping on a speaker cone and you realise that in truth amplifiers do not control loudspeakers to the degree commonly supposed. So the high output impedance of a zero-feedback triode amplifier is not a problem in theory or practice, as experience proves.
This amplifier was designed to be easy to build, possessing very few parts. That's why Tim decided to use high performance phase splitter/driver transformers. They eliminate the need for an entire valve stage, plus the imbalance problems intrinsic to such stages. Also, 300Bs are directly heated (there's no cathode) and need 5volts a.c., so a dedicated mains transformer is required. The circuit is pure Class A and self-bias ensures that if anything fails, the valves don't burn themselves out. Tim feels the potential safety and reliability of self-bias more than compensates for its limiting effect on power output.
In the power supply two GZ34 fullwave valve rectifiers have been used, for slow thump-free turn-on, to extend valve life by applying H.T. only after the filaments have lit, and to make experimenting easier and safer. Removing a valve kills one channel, effectively disconnecting it from the power supply. Tim also insists, like others, that valve rectifiers don't switch like diodes and therefore give a cleaner supply and better sound.
Similarly, for reasons of quality, we have used a smoothing choke in each of the independent power supplies. It gives less hum and noise, and better regulation. It also allows high performance polypropylene power supply capacitors (C II & 12) to be used - a very unusual feature in any amplifier.
So this amplifier, because it makes no concession to convenience and parts availability, as It could not, being designed for very best sound quality, is unlike modern pentode based amplifiers, but it is not difficult to build. Tim has designed the transformers, since these days such specialised items aren't available off the shelf. But if the 300B valve is to be used, this is a necessity. The transformers will be made available separately, for all those who feel they can do the rest themselves. All other parts are available from Electromail and valve suppliers like P.M. Components, Billingtons, Chelmer, Collomore, Cricklewood Electronics, G.T. Audio. (See Kit Suppliers on Page 8)
Finally, I must warn you that lethal voltages exist in the circuits and that you should not consider building this amplifier unless you are fluent with handling valve circuits and know what precautions to take. This circuit is offered for those who know what they are doing.
A kit will be offered in Part II of the project (October issue), which makes life easier. It will be based on a professional 16 gauge steel chassis fully punched and finished. All parts, including transformers will be included. You may be able to get someone like an electrician to build it, but tell him first that in addition to the lethal mains voltages present, the H.T. runs at 550volts!
THEORY OF OPERATION
One half of an ECC83 and ECC82 triode act as input stage and driver for the transformer stage on each channel, being shared by the two channels.
The input ECC83 acts as a low noise voltage amplifier which feeds the ECC82 as a Class A 0.2 watt anode dissipation driver with the driver transfonmer as its load.
A small amount of feedback is taken from a tertiary winding back to the input stage to provide low distortion (0.1 %). High voltage push-pull drives the two output valves (about 400V peak to peak).
A pair of 300Bs are used in a push-pull output stage with self bias by means of a cathode (filament) resistor bypassed by a parallel capacitor. A potentiometer is used to balance out hum in each valve (a problem with directly heated triodes). The output transformer matches the output to an eight ohm loudspeaker load from a 3,000 0hm plate-to-plate (anode) resistance. The amplifier is run on 550volts d.c. which is about 450volts across the power valves at 80mA (approx 40watt plate dissipation).
The power is derived from a reasonably smoothed H.T. using two 25µF polypropylene capacitors and a 5H choke acting as a pi section filter fed by GZ34 full wave rectifier valves from the 450 volt windings on the mains transformers. The mains transformer also has six 5 volt windings (one for each directly heated 300B and GZ34 rectifiers). Finally, a 6.3volt winding is for the input valves.
Total power consumption is about 220 watts.
Tim de Paravicini
SAFETYWARNING
Lethal voltages exist in this amplifier. We do not suggest you attempt to build it unless you are conversant with valve circuits and safety precautions.
For safety, never hold earthed metalwork when testing. Make sure your body is isolated by rubber soled shoes. You should posses a voltmeter capable of reading up to 1000volts.
The final power amplifier should have a protective underplate and a wire mesh top cover, since the 300B valves run very hot and will burn.
SOUND QUALITY
Valve amps have always had the affection of audiophiles, the 300B is no exception. My first acquaintance with the 300B came from listening through Heybrook Quartet speakers and a Sugden CD player. An Audiolab 8000 pre-amp and monoblocks were at hand to represent the higher end of transistor amps.
What struck me first was a sense of delicate spaciousness, and superb articulation. With Paul Simon's The Boy in the Bubble', bass lines progressed with fluency and speed, totally unhampered by bass boom. Imaging was excellent with great depth as well as a sound stage which truly disguised the speakers positioning and height. With 'Homeless', voices seemed to come from beyond the usual sound stage between the speakers.
Mary Black provided the female vocals.
On 'Katie' a natural subtlety prevailed that allowed her voice to breathe freely. There was a slight reediness to the upper midrange vocals, however, as I later discovered this was due more to the speakers sonic footprint than that of the amplifier. 'Columbus' went on to show further qualities of the 300B; beguiling, smooth, and a lack of sibilance that is denied to lesser transistor amps. Separation was accomplished with the music being presented rather than hurled at the listener. One was able to clearly distinguish between the divergent guitars of Mike Oldfield. It was possible to pick out a particular guitar and follow its rhythm, yet the piece remained coherent throughout.
MEASURED PERFORMANCE
Power output measures 28watts per channel before clipping (1.5% distortion). That's enough to make all but the most insensitive loudspeakers go loud, including our own three-way DIY loudspeaker featured in this issue. We found the two went together surprisingly well.
Frequency response extends right down to 5Hz, and up to 36kHz. The output transformers have very good cores, allowing them to deliver full power from 20Hz up to 22kHz (-I dB), an excellent full power bandwidth figure for any valve amplifier, let alone a specialised no-feedback design like this.
Sensitivity is high at 250mV, meaning the amplifier can be used with a passive pre-amp and will match all sources, even those with a low output. These days the lowest outputs from tuner or cassette are usually no less than 300mV or so.
Hiss was very low at -I 00dB. However, direct heated triodes have a little hum, although if well suppressed it will be just about inaudible, as it is on this amplifier. Measured hum level was around -72dB on our prototype, but this low level can be reduced still further by careful balancing of the hum bucking potentiometers on the 300Bs, so hum is not a problem.
Distortion rises Iinearly with output level on a no-feedback amplifier such as this. Below I volt output distortion is lower than 0.1 %. At 3V (I watt approx.) it measures 0.2% at I kHz and 6kHz, and 0.4% at 60Hz; the analysis (I kHz) shows second harmonic dominates. At I 0 volts output (12watts) levels hover around 0.6% and at full output of 15volts, or 28watts, as defined by visual clipping on a 'scope, it measures 1.5%. There's no great increase in distortion at low or high frequencies, such are the abilities of the output transformers. NK
Power 28watts
Frequency response 5Hz-36kHz
Separation 54dB
Hiss -IOOdB
Hum -72dB
Distortion 0.2%
Sensitivity 250mV
|
