The Phono Preamp

This article on phono preamplifiers is written from my personal experiences with gramophone (phonograph) records which started in July 1969 (at the age of 14), and my subsequent work with AV equipment during the 1970s - some of it quite old and containing valves. It is also based on my personal research and undying interest in record reproduction, and the electronic experimentation which goes with it, which began in 1975. I make no apology for this article not being in keeping with much of what is currently understood about record equalisation - much of it false.
Graham Slee


History and the Need for Equalisation

Phono preamplifiers have been around in one form or another since the earliest electric phonographs.

Cutter Head Response

The cutter head response (top) is "equalised" (bottom) to produce a nearly flat output - the limitation being the cutter head amplifier dynamic range.

The first mass-market electric pick-up cartridges were crystal and ceramic - they worked on the piezo-electric principle, and are still found today on cheap "nostalgia" record players. Their output was high enough to drive the input grid of a record player amplifier valve directly. They had a constant-amplitude output, so did not need rising frequency compensation as do the (constant velocity) magnetic cartridges of today.

However, the output from a record is not flat. If reproduced by a perfect constant-amplitude pick up cartridge, the output from a modern (RIAA) vinyl record, plotted graphically, would show bass boost below 50Hz, mid-range cut at 500Hz, returning to flat above 2122Hz - the rate of boost and cut being 6dB/octave. This uneven output is due to the "equalisation" required by the inductive cutting head of the record cutting machine - without which its output would simply fall-off at a constant rate of 6dB/octave - a 45 degree downward slope if plotted graphically.

It was not possible to provide exact linearization because of the amount of gain required to equalise the then high-fidelity frequency range of 30Hz - 15kHz. That would require a 54dB gain difference all of its own. Today's 20Hz - 20kHz requires 60dB gain difference.

The cutting head amplifier also needed something in the order of at least 40dB signal to noise ratio, as well as 14dB headroom for transients - another 54dB, taking the total dynamic range required to 108dB (114dB by today's standards). Such dynamic ranges were difficult or impossible to achieve with valve amplifiers, but even if they could, distortion levels were quite high too (and solid-state had not yet been perfected).

The answer was to "fold" the response and that produced the boost and cut described above. It shaved off 14dB from the dynamic range requirement (20dB by today's standards). An amplifier with a dynamic range of 94dB was more in keeping with the times.

Why those particular frequencies were chosen was more to do with the mass-market and the use of cheap piezo cartridges with low-cost record players than any other concern. The companies making records were, more often than not, also making and selling record players (HMV, Philips, RCA Victor, Decca... to name a few), and it was in their interest to make those records sound good to the record buyer.

Cheap valve output transformers (and loudspeakers) didn't have much bass extension, so a bit of sub 50Hz boost was welcome. Piezo pick up cartridges easily resonate at mid to high frequencies, so the shelved-down upper frequency response helped make the sound less tinny. Tone controls were added so that records from other manufacturers would play acceptably. However, if you were to play those records using an accurate (and more expensive) piezo cartridge with a high fidelity amplifier and speaker, the upper midrange and treble would sound muffled, plus the lower bass would have sounded over-blown.

Valve Amp Input

An early 1950s high-fidelity valve amplifier input stage shows inputs for both piezo (crystal) and early (more expensive) magnetic cartridges.
(image taken from Hi-Fi, P.A., Guitar & Discotheque Amplifier Design Handbook; Babani Press, 1972)

This is where the phono preamplifier makes its debut proper, around 1948, in the input stage of high-fidelity valve amplifiers. These often had one valve which catered for all the inputs, and had a number of potential divider networks arranged around it, which could be selected by a rotary switch to provide the right gain for a particular input. For the disc input, the potential divider network included resistor-capacitor (RC) filter circuits which applied opposing boost and cut to that of the record, to give a flat output to subsequent stages within the amplifier.

Take away the components required for the other inputs and the rest of the amplifier, and you'd be left with a phono preamp. Put it in a separate box; add a power supply and it would be very similar to today's phono preamps.

Before 1953, there was no official standard for where the "fold" frequencies should be. Each record label which also made and sold record players had its own idea. By the time of the commercial debut of Columbia's 3313rpm LP record in 1948, there were two competing EQs: American and British. They differed only by their "shelving" frequencies, which were 1592Hz and 3183Hz (respectively).

In 1953 the Recording Industries Association of America (RIAA) met, and adopted what was believed to be RCA's preference for a "half way house" between the American and British shelving points - the frequency being 2122Hz. From then on record companies and record reproduction equipment manufacturers were urged to adopt what we now know as the RIAA standard.

Good high-fidelity amplifiers produced in the 1950s and 1960s often catered for all three EQs by additional RC filter circuits, so that the different equalisation could be selected. However, by the 1970s RIAA had been adopted universally by amplifier manufacturers, and today's phono preamplifiers are often referred to as RIAA preamps because of this equalisation standard.

The adoption of the silicon transistor in preference to valves in the 1960s made it difficult to use piezo cartridges. This was because the input impedance of a transistor amplifier (around 50 kilohms) was much lower than the output impedance of a piezo cartridge (above 1 megohm) causing gross bass distortion. However, hi-fi amplifiers continued to cater for piezo disc inputs by using one of a number of methods.

A rather crude but effective method was to add a single low-noise transistor stage ahead of the proper solid-state amplifier input. If, like me, you wanted to make an add-on DIY unit, it could be run on a simple 9 volt E-cell battery, using the transistor as a common emitter virtual-earth amplifier. Its series input resistor would be in the order of 1 megohm or above to suit the piezo cartridge, and to compensate for the shelving down of the record EQ, it would be bypassed by something like a 220pf capacitor.

Often these were tuned by ear to produce a particular sound, and weren't very accurate. Different values were offered for different piezo cartridges, and I remember a DIY Disco article by A C Ainslie and Clive Toms (I knew them by their real names) in Practical Wireless in 1975, which illustrated a number of these add-on "phono preamps".

The Magnetic Cartridge

RIAA Output from a Magnetic Cartridge

The output from a magnetic cartridge is its rising frequency response plus the record's own frequency response superimposed upon it.
(image taken from Audio/Radio Handbook, National Semiconductor, 1980)

By the late 1960s the stereo moving magnet cartridge was becoming affordable and therefore popular. These suited the input impedances of solid-state electronics, and resulted in the almost universally accepted 47 kilohms loading you see today. However, the output of a magnetic cartridge is constant velocity, which means its output rises with frequency - the exact opposite to a linearized cutter head. When an electrical signal is fed to a constant velocity transducer its frequency response "falls off", but when generating an electrical signal a constant velocity transducer's frequency response rises.

At this point we could stop and think that if it were not for the intervening history, we would not need any EQ at all, as the magnetic cartridge "reverses" the effects of the magnetic cutter head. However, one thing gets in the way, that being surface noise (not forgetting surface damage): the cutter head would have cut such low amplitude at high frequencies that on replay, all sorts of noise would dominate.

Instead, we are faced with an output signal from a magnetic cartridge that has tilted the record's frequency response upward by 45 degrees, if presented graphically. However, if we are not too fussy about the resulting sound, we can easily equalise to a flat response using only two transistors (per channel), by using some frequency selective negative feedback. If we are fussy, we can play about with this arrangement by fine tuning circuit impedances, and produce quite a good sound.

Going back to the piezo cartridge, it was found that if they were used into a preamp stage as mentioned above, by careful adjustment of values, they could be used to reasonable effect with modern solid-state electronics. However, they could still sound a little distorted. The piezo cartridge was soon relegated to the cheapest of cheap record players, and the moving magnet cartridge became the de-facto transducer for high fidelity listening.

The End For Vinyl - the Beginning For The Phono Preamp

Solid state stereo amplifiers from the late 1960s until the mid 1980s came equipped with magnetic cartridge inputs, and inside they included the equalisation stage and amplification needed to match the other catered-for inputs, such as tape and radio tuner - and eventually the compact disc.

In the rush for the new compact disc (CD) the LP record soon became so rare that most amplifier manufacturers simply stopped providing a disc input altogether.

By 1990 the LP record was seen as a joke by some. However, a small number of vinyl devotees hung on to vinyl as their prime listening source. It created a niche market for all things vinyl, and a few of us decided to make the piece of electronics missing from most modern amplifiers, as an add-on box. From that moment on, the phono preamp became a piece of equipment in its own right.

Moving Coil - the Final Solution?

Blumlein's MC cartridge

(image taken from New Scientist magazine
- Jan 12, 1978)

Prior to the launch of CD in 1983, I remember the rising popularity of moving coil cartridges marketed as the "final solution to high-fidelity". However, moving coil was nothing new: the moving coil stereo cartridge had been invented by Blumlein and used by EMI in 1933 (New Scientist, Jan 12, 1978). It was patented in 1935. In 1962 Joseph F Grado successfully patented improvements to "the prior art"

Whatever your views are on moving coil, it didn't stop the hysteria over, and mass exodus to compact disc.

Although excellent results could be obtained (but mostly weren't) from much cheaper moving magnet cartridges, by the mid to late 1990s most vinyl devotees had been convinced of the superiority of moving coil. Unfortunately, because of the expense involved in both the cost of buying a moving coil cartridge, and that of buying a phono preamp to suit them, which few people could afford, playing records became something of a rich man's hobby.

The Common Man's Phono Preamp

One or two entrepreneurs stepped in to "help" those on a more meagre income and a new breed of low cost (cheap) phono preamps entered the market in the late 1990s. Some of these were based on the two transistor stage mentioned earlier in this text, and some were based around quite cheap and limited performance operational amplifiers.

In 1979 I had spent quite a long time trying to develop a preamp and power amp, which I intended to try to market. Line level inputs were in the order of 250 - 500mV being from radio tuners and tape (or compact cassette), and power amplifiers could be made sensitive enough such that the preamp didn't need any amplification for them - it could be a "passive preamp".

The real effort was needed for the disc input, so most of the circuit was all about the phono preamp.

Audio-quality operational amplifiers (op-amps) were on the horizon but were not quite in "circulation", so I did lots of experimentation using discrete transistors. By far the best sounding design (to my ears) was the simple two transistor "DC coupled" configuration, which used negative feedback RIAA equalisation. One of the things I noticed was that its sound staging was very convincing.

The First Phono Stage?

If I had succeeded in my venture, that preamp could have been the first proper phono stage! It would have had the additional feature of a "passive preamp" (for line inputs). However, life took a different route which took me into the printed circuit industry, then the broadcast audio industry, and threw me back out just in time to see the return of vinyl after its devastation by CD.

Whilst in broadcast audio I was called upon to design a phono preamp to go with a range of radio studio equipment. Although it worked, the design brief was quite peculiar, requiring audio to be routed along with its DC supply, and through its remote power supply enclosure! Needless to say the BBC rejected it. Sound quality was not good at all - although measurably it was pretty good indeed.

I decided to refashion it at home. It had used passive EQ which was very much in vogue in hi-fi, and even after stripping out the weird power supply arrangement, and after numerous rehashes of the circuit, I still had the urge to fiddle (a similar symptom to "upgraditis" which so often troubles hi-fi customers).

In the end I threw the thing in the trash bin. My next DIY phono stage design was an experimental balanced input design which used active EQ, and it sounded much better. Swapping back to the standard arm wiring it became a "typical" active phono preamp, and I had no inclination to tamper with it, or even use anything else until I started this company.

MM and MC Don't Mix!

Moving magnet and moving coil cartridges have two things in common: they are both magnetic giving rise to their constant velocity outputs (rising with frequency), and they can both be used to play records. After that all similarities end.

There are a number of phono preamps available which claim to do both, and also claim to be high-fidelity. To the layman that may look OK, but to an experienced electronics circuit designer of advanced years (like me), claiming high-fidelity for both is questionable. If the phono preamp were to use separate circuits for each cartridge, then I could be in agreement. The type of circuit that simply alters its voltage gain by a switch cannot do both justice.

Leaving high output moving coil to one side, the conventional moving coil cartridge has very little output. The output is measured in micro-volts, and is approximately one-tenth of the output of a moving magnet (which is measured in milli-volts). The lowest output moving coil I have come across outputs just 100uV (0.1mV). The output is measured at a standardised centre frequency of 1kHz because, due to the constant velocity characteristic common to all magnetic cartridges, its output at other frequencies will differ.

At 20Hz when playing an RIAA record its output will be one-tenth of 100uV which is 10uV (0.01mV). To prevent circuit noise - which is mainly a function of the silicon transistor that will be used to amplify the signal - "coming up" to meet that tiny 10uV signal; the amplifier has to be extremely low-noise. To get that low-noise the amplifier's transconductance needs to be very high indeed. This is the reason why you cannot use valves to amplify a moving coil cartridge directly - their transconductance is too low. In addition, most silicon amplifier 1/f noise tends to rise at frequencies below 100Hz. It presents a serious difficulty which is hard to overcome whether the amplifier uses discrete transistor circuitry or op-amps.

Another problem for low-output moving coil is that high-transconductance equals small slew-rate and reduced bandwidth. In high-fidelity audio, a slew-rate below 4V/uS results in much transient intermodulation distortion - it becomes audible.

The numbers of options open to the audio circuit designer are very few indeed. Add to that the international requirements regarding immunity to spurious noise sources needed (mobile phones being the worst domestic culprits) to comply with EMC (which is mandatory), and there is possibly only one or two op-amps which just "scrape home". Forget discrete circuits - they will be wide-open to today's interference.

Having been successful in implementing solid-state pre-amplification for low-output moving coil by using one of the rather limited methods, can that same input stage be used for moving magnet? The short answer is no.

Whereas the output from moving coil is sufficiently low as not to exceed the linear input region of a silicon bipolar transistor (the type that has the highest transconductance - discrete or inside an op-amp), the output from a moving magnet cartridge can easily exceed it. Remember, all magnetic cartridges are constant velocity, and at 20kHz (percussion and high harmonics) the output of a 5mV @ 1kHz moving magnet cartridge is going to be 50mV. The peak to peak signal the input sees is nearly three times that size, and 150mV is over twice the linear region of the device. The result is incurable distortion, but because distortion simply adds to the percussion sounds and harmonics, some like the sound - but it is hardly high-fidelity.

A moving magnet cartridge therefore requires a different input circuit. Silicon bipolar transistors can still be used, but with emitter degeneration, which reduces transconductance and "stretches" the input linearity. J-FET inputs work in a similar manner. Noise is less of a concern, and the lower transconductance makes it possible for valves to also do the job. In the case of active EQ, where the gain of the stage includes the equalisation, switching gain sends the frequency response off target - another reason why a switched gain "all-in-one-box" MM/MC phono preamp is a poor choice.

The best choice for low-output moving coil is a step-up transformer (S.U.T.) used in combination with a dedicated moving magnet sensitivity phono preamp - either that or use an electronic version of the S.U.T., such as the Elevator EXP step-up amplifier. Another choice would be a dedicated MC phono preamp - which does low output moving coil only.

Conclusion

It has been a long journey since my interest in vinyl was kindled in July 1969. Along the way I have discovered a lot, and it was always my intention to share my findings to help the vinyl devotee understand and get more pleasure out of music from the humble vinyl record.

To some, vinyl is the de-facto pinnacle of "hi-res" music. In my opinion, it simply offers an alternative to digital, and it's good to still have vinyl around as a comparison. Both have limitations - they have good points as well as bad, and some of us are happy to realise that, and get on with enjoying the music.

Whatever your choice of "software" be it the vinyl record (or even shellac 78s), compact disc, or digitised music from a computer or smart phone, I hope you will spare the occasional thought for those who made it all happen.


Union Jack Flag

MADE IN BRITAIN

Our audio components are designed and handmade in Great Britain for your listening pleasure