Quad 44 and the IEC Amendment

Submitted by dada on Mon, 06/28/2021 - 16:20

The 1976 'IEC Amendment' to the 1954 RIAA phono equalization curve is a controversial extra zero with a time constant of 7950uS, which causes a loss of -3dB at 20Hz, increasing at lower frequencies at 6dB/octave. The reasons behind its introduction are mysterious, nobody apparently having asked for it, and many manufacturers have never implemented it at all. As an LF or rumble filter it is rather useless, being only first order, and it cuts into the music as high as -1dB at 40Hz, or the bottom note of a double bass, and starts to cause phase errors from this frequency down. The Amendment was withdrawn in 2009.

Quad implemented it temporarily (and rather inaccurately) in the 44 MM module M.12515.4, but dropped it for the subsequent revision M.12515.5. The 44 MC modules were not shipped with the IEC Amendment, but instructions for implementing it are provided in the service manual.

To remove the Amendment from module M.12515.4 and thus correct the bass response to be flat to 20Hz, merely increase C304/4 from 10uF to 100uF or even 220uF (or more if you can find a capacitor that will fit: voltage rating of 6.3V is more than sufficient if you choose this path). The ultimate purpose of this capacitor is to reduce the gain to unity at DC, by ensuring that none of the NFB is shunted to ground at LF. Reducing its capacitance increases the frequency at which this starts to occur, and is clearly the implementation mechanism envisaged for the IEC Amendment. Douglas Self gives a better one in 'Electronics for Vinyl', Routledge 2018, p.110.

44 MC modules should have the values of R17/19/21 checked to see whether the IEC amendment is present, and if so they can be restored to their original values. Similarly in the other channel, of course.

On the 44 I tried this on, using 220uF, the bass response went from +9.5dB to +20dB at 20Hz referenced to 1KHz, that is to say from very poor to exactly correct RIAA The original capacitors measured 12uF, so the original measurement wasn't due to bad component tolerances.