Quad 303 Modification.

The modification to the ' standard class-B amplifier ' is also of interest if you already have an amplifier using the same sort of output stage. There may be many existing designs which could easily be modified in this way, but I will only mention one, the Quad 303, even though it is no longer in production. Properly adjusted, the 303 is an excellent design, and I recall reading an article (Dynamic Testing of Audio Amplifiers, Hi-Fi News and Record Review, Nov. 1970 p.1655-1657.) including comments from the designer, Peter Walker, in which he demonstrated that the distortion was far below the level at which it is audible, even when extracted and heard without the masking effect of the original signal. He claimed that if any other amplifier sounded different to the 303 then that amplifier was wrong.

The amplifier was also part of a challenge to the reviewers in the Hi-Fi press who were then, as now, making claims to hear significant differences in sound quality between what were known to be good designs. Quad were prepared to stake their reputation on the outcome, and declared that there would be no audible difference in direct comparison between an early valve design, the later 303, and what was then their latest design, the 405, provided the frequency responses were adjusted to be equal. The reviewers eventually withdrew after preliminary tests, and later in a full test a substitute panel of experienced listeners failed to reliably identify any differences in sound quality between the three designs. The conclusion is that it is probably small differences in frequency balance which account for most of the claimed differences between amplifier sounds, or lack of care in the testing, or just imagination.

The important phrase in the first paragraph is 'properly adjusted.' The modification suggested would ensure that distortion remained low even if the output stage current drifted significantly. I must stress that I have not carried out the modification myself, and can give no assurance as to the success, or even that it would be safe to carry out. Stability margins may be reduced in the output stage, and although I doubt whether there would be any ill effects, this modification should only be performed by those with the skill and test equipment necessary to identify and solve any problems. Also, I have no information on any different versions or later changes to the circuit by Quad, so be sure the circuit of your 303 is as in the diagram refered to before proceeding.

To see how the circuit can be modified, refer to the circuit diagram, which can be found at

Resistors R120 and R121 are already 100ohms as in my own circuit, and it is only necessary to add two more 100ohm resistors. One goes from TR104 emitter to the junction between R123 and R124. The other goes from TR103 emitter to the junction between R124 and R125.

The quiescent current will need to be readjusted using RV101 after the resistors are added. Unfortunately I do not know the normal value, so I can only suggest that the current should be measured before the modification, and then for safety set as low as possible while the modification is carried out, then set to a value about 50 percent higher than the initial value after modification. Simply measuring the voltage across R124 or R125 with a sensitive voltmeter with no signal applied to the amplifier seems the easiest way to determine output stage current. A current of 50mA for example would give a voltage reading of 15mV.

Current stability is lower after modification, so some care is needed to determine whether thermal drift is a problem. The values of R124 and R125 may need to be increased e.g. to 0.5ohms or more if this is a problem.

This design note is intended only as a suggestion for experimentation, and anyone attempting to modify their amplifier does so at their own risk. I don't have a 303 to experiment with myself, so I hope anyone who successfully carries out the modification will let me know the result, particularly if distortion traces can be produced for before and after.