Phono preamps can be tricky builds due to the need for high gain with low noise. In tube land, linear high gain is not too difficult to achieve even without feedback. Power-supply-based noise can often be brute forced with extra filtering, actively regulated B+, and/or DC-powered tube heaters. High PSRR topologies (eg differential) also have an advantage in the early amplification stages.
The place where most DIY builders are probably tripped up is the mysterious RIAA voodoo. Because the physical limitations of the vinyl medium and cutting process require a limiting of low frequencies and a boosting of high frequencies, we need to reverse this EQ on the playback end in order to get back to ‘flat’ frequency response.
At it’s most basic, the RIAA equalization standard defines three frequencies: 50hz, 500hz, and 2122hz. We should have a 20db boost to 50hz, a -20db/decade transition from 50hz to 500hz, flat playback from 500hz to 2122hz, and a -20db/decade falling response above 2122hz. Note that 20db/decade is equivalent to 6db/octave, so these are not especially steep filters.
Splitting the RIAA requirements between low (<1khz) and high (>1khz), the low frequency manipulation requires at least 20db of gain from whatever device we are using. This type of EQ is commonly referred to as a shelving filter. The high frequency portion is only reducing the response and so it doesn’t require gain (ignoring the overall gain needed to get to line-level signals). This reducing of the high frequencies can be as simple as a first order low pass filter (just a resistor and a cap).
Tubes, with their fairly high output impedance and finite Mu, complicate RIAA frequency-dependent impedance calculations. Operational amplifiers, on the other hand, make filter maths fairly straight forward. Here’s an example:
Starting at the output, the R1 and C1 combination form a simple low pass filter. Because the output impedance of opamps is so low, our equation need only involve the cap and resistor:
f(-3db) = 1,000,000 / (2 * Pi * CuF * R), rearranged as:
R1 = 1,000,000 / (2 * Pi * 2122hz * C1uF)
Begin with a tight tolerance capacitor (say 0.1uF) and you’ll get a resistor value that may come off the shelf or be created with a parallel/series combination (in the case of a 0.1uF C1, the resistor would need to be 750 ohms). The resistor appears in series with the output, so large values may require a high input impedance in the following stage.
The shelving filter created by R2, C2, and R3 appears in the feedback circuit of the opamp. Because we need 20db of gain, we know that the ratio of R2 to R3 should be approximately 10:1 (a 10x voltage gain difference corresponds to 20db). The 50hz point is set by the combination of R2 and C2 and is found with the same kind of capacitor reactance equation as the low pass:
R2 = 1,000,000 / (2 * Pi * 50hz * C2)
Again, start with the cap value because caps have fewer options and are harder to find in a tight tolerance. A 0.047uF cap gives an R2 of about 68k, meaning R3 should be about 6K8. The overall gain of the stage is further set by R-gain (Av = 1 + R3/R-gain).
So that’s a pretty simple way to EQ your vinyl to flat. More gain to bring the signal up to line level could be added by following the EQ/opamp stage with a ‘normal’ tube stage or two. Expect to see some more on this topic in a future project!
2 Replies to “A different way to RIAA”