Aleksandar Kitic
RIAA amplifier using ECC88

The voltage drop is very limited, as the initial raw DC is almost 200V (the first cap is oversized for security reasons, while the second cap is appropriate - both are very small as dimensions (thick and short). The remaining caps are obviously poly types. The 10uF cap slows a little the PS and improves ripple rejection some 20 dB, to achieve about 80dB at the regulator output. The type, brand and quality of the first two electrolytic caps are not important, since their influence on the sound cannot be heard (what you could hear is the LM regulator...).

Well, this is heresy, is it not... still, Curcio does similar things. Not that I keep Curcio in high esteem - his designs are somewhat different anyway. Point is, in this case the story is all about ripple rejection: because we are talking about sub milli-volt signals (MC cartridge) and all the ripple present - especially in the first stage will be amplified to become loud hum (hum can be induced in many ways - but in this case all precautions were already taken against hum: what you are left is eventual hum from the power supply - 100Hz ripple...).

Well, with such a regulator, ripple will be gone, for good (along with hum). It is now time to address the second, maybe more important issue: noise. I had an excellent start: using LM217 by SG, there was no added noise, so I actually forgot at all about it, although I read many stories about zener diode and similar source induced noise... One day, I made a new board and put another brand LM317 (if the first one did well, why shouldn't they all?) and got bad, bad, bad noise! Actually, so much noise you could hear no sound: the regulator noise is like buzzzbuuu! Quickly, I turn it off, control everything, turn it on again: same sound... So, take out the bad LM317 by UTC (ban them from commerce!) and in goes another specimen of LM217 by SG - no noise, same result as the first version.

To be frank, there must be some noise, since the voltage reference in the LM chips cannot be of high quality - but with good specimens, you do not get to hear it. The first version of the PS would have been much more to your liking: nice caps (good quality, LCR) although relatively large (500uF), choke, high wattage resistors - and large voltage drop (you need about 150V Udrop in order to achieve similar ripple rejection results). When I changed to regulated, there were no shortcomings: the new supply is better in the hum department and better in the noise department... the sound is cleaner, even more dynamic and details are a little more in evidence: what is missing is a slight bass bloom (the latter most probably caused by low levels of hum interfering with the signal).

Click schematics to enlarge

Questions or comments?
Mail to: kitic@eunet.yu