The Analog Addicts Phono Preamp

By Thorsten Loesch

In recent months I have spend a lot of time playing around with various RIAA preamp circuits. The final configuration I arrived at may seem extremely familiar and common, but great care and much work has ensured that the resulting sound is rather better than usual. Please read the following carefully and do take my recommendations seriously.

If you use substantially different parts or values, or otherwise change the design, it's your design. By all means tell the world and me what you think about the sound of the design IF YOU BUILD IT AS INSTRUCTED. If you change things, don't say it's my design that sounds so and so: it's now your design.

First, let me set out my main considerations for a phono-headamp of the 90's:

1. The phono-headamp must have enough gain to have real-world output voltage similar to that of CD player. That is about 45 - 50db for a high output moving magnet like the Goldring G1042 that I use. To use low-output moving coil cartridges a gain of 70-80db is necessary. This is a much higher gain figure than usual, where 35db for an MM and 60db for a 0.1mV MC is more common.
2. Due to the relatively high gain, we MUST use fully active RIAA equalisation as we will otherwise run into severe headroom problems (if using monolithic integrated circuits). In addition, very few monolithic integrated amplifier circuits with low enough noise have a wide enough bandwidth to satisfy my personal axiom #1: Any amplification circuit MUST have a OPEN-LOOP bandwidth which exceed the bandwidth of the signal to be amplified by at least one octave into each direction. Interestingly, a active RIAA Equaliser could also be seen as a 50Hz 1st order lowpass, so monolithic op-amps can fulfill this condition, while in a flat amplification stage they fall woefully short. At the same time, the high gain used will reduce the amount of NFB needed, opening up the sound. For an application like this, it is essential that the device used has exceptionally low distortion levels.
3. The RIAA equalisation shall have the best accuracy that can be obtained using of-the-shelf parts and MUST implement a replay curve which mirrors the curve used when cutting a record, NOT THE documented RIAA Curve, which is incorrect. The Neumann Cutting Amp Manual states that the boost in high frequencies is being rolled off at about 50kHz. Neumann cutting lathes and amps are pretty much the Industry Standard, we can assume this as being a de-facto standard. Any kind of warp-filtering is bound to introduce low frequency phase shifts. Most decent record players use clamps to minimise warp related effects, and no-one should need attenuation of any rumble, so the misguided IEC amendment to the RIAA curve must be avoided, and the absolute lower cutoff point of the phono headamp should be as low as possible. The lower cutoff should be a single dominant pole at 5Hz or less. With current off-the-shelf parts of moderate cost but high quality, an RIAA accuracy of +/-0.1db or better can be achieved in the midband. Due to the additional high frequency breakpoint we expect an error (with respect to the standard RIAA curve) of about +0.25db @ 20kHz.
4. For a simple realisation of the circuit, the main RIAA EQ should use a readily available, high quality integrated circuit. The RIAA network shall be optimised around this integrated circuit. The gain of the RIAA equaliser should be set in such a manner that it in itself is sufficient to amplify and equalize the signal from a moving magnet cartridge.
5. The bandwidth of the circuit shall substantially exceed 100kHz and distortion shall be minimal. The power supply needs to be implemented in such a manner and fashion as to not degrade the performance of the circuit.

So, what did I do to fulfill this brief. Well an awful lot really, though in the end (as always) KISS proved to be the best method.

(KISS stand for Keep It Simple Stupid)

Below you can see the circuit I managed to come up with as so far the best MM equaliser. The circuit (currently implemented inside my integrated Amplifier) has a excellent technical performance AND sounds exceptionally good.

The Noise-levels are very low, though not as low as the best, but close.

The Circuit is designed around a Burr-Brown precision FET-input op-amp. At the moment I use the OPA604 (actualy a dual one), but simulations in PSpice have shown that the OPA637 would be the best choice, with lower distortion levels as well as a advantage of about 4-5db with regards to the Signal-Noise Ratio.

Both the OPA 604 (~USD 5.00) and the OPA627 (~ USD 21.00) are readily available from

A key feature of the circuit is the fact that throughout this circuit, direct coupling is employed as much as possible. This is possible due to the exceptional DC performance and low offset of the Burr-Brown parts.

The output offset does reach almost -1Volt in one channel though (the other is +150mV). This is relatively harmless when blocked out, but could cause serious problems with certain Amplifiers/Preamplifiers when direct-coupled. I consider the loss of headroom due to the offset inconsequential when compared with the possibility of inducing noise through the offset adjustment.

After careful tests, I chose an output capacitor realised as film capacitor (or Film/Electrolytic Compound Capacitor) as the best compromise. I found that using even a much oversized and bypassed electrolytic cap in the usual DC blocker position did audibly degrade the Sound. I have also tried servos and found them unsatisfactory.

The capacitors in the RIAA network are Philips non-inductive 1% tolerance polystyrene capacitors (again, Maplin stock).

In my case, I took the time to hand-match the capacitors between the two channels to a difference of less than 1pF for each cap (measurement limit of the RC meter). This is HIGHLY RECOMMENDED, as stereo imaging relies on accurate channel balance.

Please note, that most cheap meters are quite inaccurate with regards to absolute values, but even the cheapest unit is very useful to match the relative values of capacitors.

Even with the maximum specified tolerances and assuming the worst-case scenario for accumulation of tolerances into suitable directions, the RIAA error should remain within +/- 0.15 to 0.2 db of the specification. Typically the RIAA accuracy should be better than +/- 0.1db between 100Hz and 10kHz.

All resistors used where 0.5Watt 0.1% Holcos, though more mundane Metal-Film resistors (1% though) should work well too. If you really want to have a go, try Vishay bulk foil resistors or Vishay EE10 Series resistors.

Compared to other (also highly accurate) networks that I tried the one shown below sounds much more musical. I think this is due to the use of higher impedances in the circuit, which will load the op-amps output less at higher frequencies than the other networks tried. This will result in an improved distortion behavior.

The MC pre-preamp is a very simple implementation based on a circuit published in the French l'Audiophile Magazine. The FET can be the 2SK147, 2SK389 or the American 2N5566 (NPD5566). It should be noted that the MC step-up inverts absolute phase, a compromise I accepted as otherwise much more electronics would be needed without much sonic benefit. The coupling capacitor should be as good as possible, a WIMA MKP4 (or Siemens stacked film) bypassed with a 10nF polystyrene will likely do the trick.

Finally, the layout of the circuit is dual mono and uses full star grounding. My current implementation feeds off the amp's PSU, which is external and uses batteries. The voltage is taken from the + and -12-Volt "taps" on the Battery stack.

I cannot recommend any type of regulated supplies. You MUST use Batteries to have the full performance possible with this circuit. May I humbly recommend the Yusea NP Series 12Volt 2.1 Ah sealed lead acid batteries (or equivalents). These should be charged in parallel.

The electrolytic capacitors I used in the PSU are specifically designed for low distortion and good frequency response. I personally use the ELNA Starget Series capacitors in any audio application (if possible - voltage limitations apply). If these are unavailable please use Panasonic HFQ or HFZ Series electrolytic capacitors.

The 2.2uF bypass caps are ideally WIMA MKP4 Type metalised polypropylene. These have become hard to obtain and I have found a capacitor series that is sonically as good, but the mechanic stability is compromised. These are Siemens "stacked film" type capacitors without the protective plastic casing used by ERO and WIMA.

The 10nF Philips polystyrene film and tin-foil bypass capacitors further extend that wide frequency response achieved (again Maplin stock). The makes and types of capacitors (both 'lytic and film) are critical. Using different parts may result not in an extended flat impedance for the PSU but in a very reactive PSU line, which will degrade the sound.

Notes:

If only the 2.2uF film capacitor is used at the output (nothing larger or a electrolytic bypass), the load impedance should be preferably kept above 22kOhm. This is to avoid any audible phase problem effects, which can subjectively lead to slower bass transients.

Any parts substitution should be taken very carefully.

Specifically the Op-Amp used MUST be a FET input type. Very few of these have suitably low noise. The AD OP275 is NOT SUITED, neither is the AD797, NE5534 or OP27/37 Series as all these MUST use DC blocking on the input, which to avoid was part of my design goal.

Ideally the phono pre should be mounted in a non-metallic enclosure (together with the batteries) and be kept close to the turntable. It may be beneficial to hardwire the output cables (and maybe inputs) to eliminate one RCA connection. The input switch is best a high quality rotary (Elma) and second best a silver contact miniature toggle of the type using one piece contacts.

So, here is the circuit:

* Note, a 1% polystyrene capacitor of suitable value should be connected across the MM input to achieve the correct loading of the cartridge. The 100 Ohm Resistor at the input of the MC pre-preamp should be set so that the value is optimised for your cartridge. For neither of these are DIP switches, jumpers or the like acceptable. I use a second RCA jack to connect the loading resistor/capacitor which I mount inside decent RCA plugs.

(refer to the manual for your cartridge for the recommended loads - note that the capacitance of the cable connecting the cartridge to the phono headamp is part of the total capacitance) .

The simulated frequency response of the circuit is shown below:

If you want to build this from scratch, have fun. If you want to modify an existing circuit, enjoy, it's easy. For a unit build from scratch, using parts at full retail, I'd expect to spend about 200 dollars (US) including batteries, charger and case.

But most of all, enjoy the music afterwards.

As for a few listening impressions, the phono pre is very detailed with glorious soundstaging and a very natural and transparent treble. It can be a bit "analytical", "cold", "matter of fact" with certain recordings, but it still makes rather bad recordings enjoyable.