Site Logo
Telephone Number

Features

Classic Audio is busy building the second batch of Spartan 10 phonostages. Available directly from Classic Audio Ltd. and combining over a decade of pre-amplifier design experience for moving magnet cartridges, and high quality components with a proven track record to deliver consistently excellent performance for many years to come, the Spartan 10 is a completely original phonostage design that offers quite a stunning performance at a very modest price point.


Spartan 10 front view

Spartan 10, first batch, serial number 002

As a consequence of the careful and balanced design choices made during development, along with several new ideas, the Spartan 10 delivers the following features:

At the time of writing (10.01.2022) Classic Audio is putting together the second batch of units, having sold the first batch to great ovation. Waiting on metalwork, these will be available from the end of January 2022 for a base price of £350, coming to £380 total including P&P and a suitable power supply. For the time being, the Spartan 10 is available in the UK only directly from Classic Audio Ltd., although this will change at the soonest possible opportunity!

This is going to be a long read! All the features above deserve an in-depth explanation below. Classic Audio Ltd holds a policy of no magic, no secrets, and a thorough design process that leaves no stone unturned.


Specification

Firstly and fore-mostly, the specification of the Spartan 10 is proudly presented. A stark contrast to many other products that make vague allusions to sound quality using non-definable terminology, all the while hiding a fairly terrible spec behind a cascade of tabs featuring subjective statements that extoll the virtue of the products 'insight', 'expressiveness', etc. The Spartan 10 spec is shown up front, with the tacit challenge to the reader of finding something better! A great deal of care and attention, along with a complete lack of 'marketing' choices, has been paid to component selection, topology, layout and manufacture. After over a year of development, this effort has delivered a superb specification from the first PCB version. See the result compared to a highly regarded phonostage that Spartan 10 has already purged from more than one customer's setup:


Parameter Spartan 10 Era Gold V
RIAA accuracy 0.1dB, 35Hz to 22kHz 0.5dB, not specified
Channel balance 0.1dB, 35Hz to 22kHz 0.2dB, not specified
Channel separation 80dB, 20Hz to 22kHz 64dB, not specified
Signal to noise ratio 78dB, at output ref 5mV cartridge load 65dB, at output, not specified
THD + Noise <0.0007%, 35Hz to 22kHz, 9V RMS output 0.02%, not specified
Maximum output >9V RMS, 18Hz to 100kHz 4.9V RMS, not specified
Maximum input at 1kHz 75mV RMS 39mV RMS
Maximum input at 10kHz 350mV RMS Not specified
Overload margin, ref 5mV 23.5dB 17.8dB
Gain at 1kHz 41.6dB, 120x 42dB, 126x
Output impedance 100Ω 910Ω
Minimum load impedance 2kΩ 10kΩ
Input impedance 50kΩ, ±1% 47kΩ, not specified
Input capacitance 120pF Not stated
Mono switch Toggle, front panel Toggle, rear panel
On switch, muting relay Toggle, front panel None
Subsonic filter 22Hz, 3rd order None
Power supply Split linear ±15V SMPS 24V DC, single 18V rail
Advertised price £350 £430

As can be seen from the table above, the Spartan 10 offers an exceptional performance to cost ratio compared to one of many of its contemporaries! Thanks to intensive investigation during component selection and Classic Audio's 'performance through topology' approach, this performance can be obtained with an excellent economy, delivering fantastic value. The specification shown beats all of the big name brands by a considerable margin, particularly in regards to RIAA accuracy. THD is quoted at 7ppm throughout the audio band, although it is quite likely far lower than this, as this approaches the limits of Classic Audio's test equipment using unbalanced connections. While the spec is excellent, the Spartan 10 has been designed primarily for real-world performance, and not just to generate a series of small numbers on the test bench.

Noise is shown measured with a real high impedance cartridge on the input, not a short which hides the often disastrous effects of amplifier noise current that some manufacturers have to preclude from the spec sheet to avoid embarrassment's. No A-weighting is shown to confuse things by adding an illusory 2-3dB of 'performance' either. The input amplifiers were selected specifically for their low input current noise characteristics and excellent HF linearity at high impedance, so as not to generate a correspondingly large input noise through the high 25kΩ impedance of an MM cartridge at high frequencies. Instead of having to strike a unhappy compromise between MM and MC cartridges, the Spartan 10 is no jack of all cartridges, but an MM master!

Also featured is excellent high frequency headroom, making sure that the phonostage has plenty of room to breathe when assaulted with otherwise innocuous HF transients from 10kHz upwards, where the output of MM cartridges rises rapidly. Surface clicks, mainly confined to the high end of the audio spectrum consequently do not cause overload, intermodulation distortion, or any intrusive low frequency artifacts that cause an otherwise benign little tick from popping out at the listener, causing fatigue and veiling the music. The Spartan 10s input capability further extends as frequency increases, all the way up to 750mV at 22kHz where most cartridges start to resonate and roll off.


Spartan 10 size

Measuring 170*58*130mm, the Spartan 10 fits well into almost all setups

All of the production units made so far have exceeded the specification shown above. So far RIAA accuracy has been within ±0.06dB, with distortion dropping down to 0.00032% at 1KHz, full output. The specification table is not 'sampled', meaning that the best measuring units have been picked to generate it, but the very bare minimum of performance Spartan 10 users can expect.


Build quality

The Spartan 10 sits inside a simple small 3 piece extruded aluminium enclosure that easily fits into most setups and visually plays down the calibre of the electronics contained within. The thick aluminium of the enclosure provides excellent shielding and RF immunity to the electronics. High quality parts are used throughout with the following construction benefits:

Spartan 10 board rear

Rear of the Spartan 10 board showing, connectors

The Spartan 10 is hand-assembled in Kent, England, using the most suitable parts available for the job, with absolutely no marketing oriented component selection. All components were painstakingly selected for their individual performance relative to the whole product.


RIAA accuracy

RIAA accuracy is very important for uncoloured sound. When records are cut, RIAA equalisation is applied to match the frequency content of recorded music to the groove displacement and acceleration limitations of the format. In order to reproduce accurately across the audible frequency, the equalisation curve must be faithfully reproduced. If it isn't reproduced accurately then the effect will be much like having leaving the tone control on a pre-amplifier off-centre, resulting in a coloured, sub-optimal sound.

Most phonostages on the market use cheap 5% tolerance capacitors, rounded to the nearest available standard component value to roughly equalise the recording curve. You can instantly spot this on the spec sheet of these products as this approach yields a playback accuracy of ±0.5dB (5%) or worse, potentially causing a total deviation of 1dB across the mid-band; quite possibly audible in itself, especially when stacked with loudspeakers and headphones that err in the same often random direction. This then makes the phonostage the weakest link in the analogue signal path in terms of frequency response anomalies.

In all but the most esoteric (and poorly performing!) designs, RIAA deviation from the curve occurs in the critical middle of the audio band, where the 500Hz and 2122Hz turnover points of the RIAA curve occur, and the ear is most sensitive. While it may not at first seem like much, these little deviations can quickly add up in a sound system and cause an audible effect to be apparent resulting in an overly bright, dull, boomy, thin or mid-heavy 'shouty' characteristic to be added to the sound. This same effect occurred in many tone control circuits, and is the reason why many electrically asymmetrical tone controls that exhibited similar frequency deviations were condemned back in the day; excluded entirely from many preamplifier designs towards the end of the 20th century.

So as to avoid distorting the RIAA curve, the Spartan 10 therefore employs a very carefully tuned precision RIAA equalisation network utilising 1% tolerance polypropylene film audio capacitors and metal film resistors to attain a superb guaranteed RIAA playback accuracy of ±0.1dB all the way down from 35Hz to 22kHz. This ensures that you hear only the sound of the record as intended when it was cut, and not that of the phonostage.


S10 RIAA response

Spartan 10 RIAA accuracy before component tolerances

The components in the precision active RIAA network arranged in such a manner as to ensure that the tolerance effects of the capacitors contained within are mitigated to the highest level possible. Extensive investigation was carried out to find the very best arrangement using real-world values, resulting in the response in the graph, deviating by less than 0.01dB in total from the RIAA response, making certain that if tolerances cause straying, they're not going to stray very far from the get-go. All production units that have been tested so far have shown this happy marriage of theory and practice, and long-term production may demonstrate that the already stellar specification is a little pessimistic.


LFC

It has long been known that switching to mono brings an audible reduction to the low frequency noise, caused by surface imperfections in the record and bearing rumble of the turntable. This low frequency 'road noise' can be particularly distracting when listening with headphones, as it sits across the stereo field, pulling the listener's attention to the sides of the stereo-field. Due to inherent limitations in the stereo groove system, all bass content has to panned to the centre of the stereo field to prevent the stylus from jumping up and out of the groove, so there is no stereo separation at low frequencies on any well mastered record.

Wouldn't it be nice if there was a way to correspondingly merge the two stereo channels down to mono at the low frequencies where all the low frequency content is already merged during mastering and therefore cancel out the low frequency noise, while still maintaining separation at higher frequencies? The Spartan 10 does just this, being the first commercially available phonostage to implement 'low frequency crossfeed', or 'LFC'. It really works!


LFC waveform

Vinyl stereo noise waveforms, top with LFC in, and bottom with LFC out

The waveform above clearly shows the beneficial effects on the noise floor of the average vinyl pressing that activating Spartan 10's LFC switch can bring about, with both waveforms showing the improvement after 22Hz subsonic filtering - all the low frequency content is audible. The RMS power of the noise waveform has now decreased by almost 6dB (a factor of two)! All of the attenuated noise energy is under 200Hz which means that previously unheard low-end detail will be apparent to the listener while also reducing the distracting cross-field effect of the noise. Customers have already commented on the difference that this has made to their listening, dramatically improving the clarity of bass instruments during quiet sections.


Noise floor with LFC in

Noise floor with LFC out

The samples of amplified groove noise above demonstrate the real-world effect of LFC, a significant reduction to audible rumble and the 'throbbing' noise that occurs with every revolution of the record. This disc was in fairly good condition, but for warped discs or discs with visible surface imperfections, that noticeably produce a shimmering effect on reflected light as the records spins, the effect is even more pronounced. The greatest difference occurs when listening with headphones as they realise almost 100% stereo separation.

While blending low frequencies the LFC function maintains 18dB of channel separation at 500Hz and 24dB of channel separation at 1kHz, better than most cartridges, extending to 30dB and beyond past 2kHz where the direction-finding ability of human hearing is most sensitive; the baby isn't thrown out with the bathwater. Unlike previous attempts at making an LFC circuit for vinyl playback featured in electronics magazines over the years that used complicated summing and differential amplifier arrangements, all pass filters, and didn't even work in the first place, LFC is implemented using an elegant bridging arrangement that excludes any additional series components from the signal path. The function is also fully defeatable; completely switched out of the circuit by a toggle switch on the front panel, satisfying purist sensibilities.


Mono switch

If the best results are going to be had from mono discs, then a switch that puts the phonostage into mono mode is a must. Due switching to mono, all stereo information is cancelled, which on a mono only disc will consist only of extra noise and distortion, the latter being particularly unpleasant as like the low frequency noise described in the previous paragraph, it sits distractingly across the stereo field. The mono switch can also be used on very worn or distorted stereo pressings, reducing distortion and surface noise enough to make otherwise un-listenable discs enjoyable again.


Simulated stereo label

Switching to mono undoes the effects of 'simulated stereo'

Perhaps one of the most useful functions of the mono switch is the ability to instantly cancel out the jarring effects of 'stereo reprocessed', or 'simulated stereo' re-releases of mono recordings. These pressings can be picked up today for modest prices, mainly because the use of reverb and phase effects across the channels are fatiguing and not pleasant for modern listeners, who unlike their original audiences of the 1960s, have quite gotten over the novelty of stereo effects. Switching to mono not only reduces noise and distortion on these discs, quite high if they were played back on the heavy tracking equipment of the day, but also undoes the processing effects that exist in the stereo difference, allowing for enjoyment of the original mono source once again. Customers have been surprised at the night-and-day difference this has brought about on many of these discs, uncovering new detail under the clanging reverb effects.


Shellac 78

A typical American recorded 78

Mono switching, combined with the Spartan 10's strong subsonic filtering and excellent high frequency headroom, also allows for the satisfactory playback of a great many 78 RPM discs, including many British recordings from the post-war period and most American pressings from the mid 1930s onwards that typically equalise to within a couple of decibels or less with the modern RIAA curve. All 78s from 1955 onwards use the RIAA curve so can be played back through the Spartan 10 with the correct stylus on a swap-able head shell, made even easier by the full input decoupling that makes for hot swapping without any loud bangs or thumps caused by bias currents running through the cartridge resistance.


Subsonic filter

Even with a perfect setup, subsonic disturbances (erroneously referred to as 'rumble' - it's not audible on its own), can seriously upset loudspeakers, headphones, and even some amplifiers. With the worst levels of subsonic content being in the 8-12Hz region they can easily push the unloaded drivers of bass-reflex loudspeakers past their precious few millimetres of linear excursion at the most moderate listening levels, promising severe 'shimmering', and 'fluttering' intermodulation distortion to be occur when audio is present creating a vague and unstable sound with an ill-defined stereo centre. A sure sign of this is movement of the loudspeaker cones during quiet sections of the record; if you can see it, you can be sure that it is going to be causing distortion, burying low level detail and directional cues under a sea of insidious amplitude-modulated artifacts.

Undesirable excursion effects can also occur in delicate headphone drivers, especially open-back types that don't have the 'springiness' of the air behind the driver to prevent excessive excursion and distortion when faced with subsonic disturbances. Contrary to popular belief, on a well-adjusted setup the worst source of disturbance is in the surface of the record itself and can be easily verified by observing the changing pattern of reflected light from the surface of the record as it spins on the turntable.

A subsonic filter that rejects the troublesome 8-12Hz disturbance region by a factor of 10 or so while not affecting the response at by more than 0.1dB at 40Hz, the lowest reasonable expected recorded frequency on the record, should therefore be considered mandatory for high quality reproduction. Unfortunately these filters don't have a particularly good reputation as component tolerances can cause significant deviations in the frequency response above the cutoff in the lower bass region, especially if compromises are made to pick standard component values 'close enough' to the optimum calculated ones.


Subsonic filter deviation

Effects of 5% capacitor tolerances in an otherwise ideal subsonic filter

As the graph above shows, capacitor tolerances alone can warp and fray the low-end response in conventional subsonic filters by a fair degree, that could stack up with other errors to cause a noticeable change in some systems. The plot doesn't account for anything except capacitor tolerance, and could be a good deal worse for a filter built with wide tolerance resistors and already truncated component values. The deviation extends quite a distance past the 22Hz cut-off towards more audible bass frequencies as well and starts to add a fairly high group delay on the red trace around 30Hz. If 10% capacitors are used, as they sometimes sadly are, then things deteriorate by a factor of 2. This sort of thing has lead to otherwise highly useful subsonic filters receiving a bad name for 'doing weird things to the bass', spurred on even further by HiFi marketeers looking to sell their customers less electronics for an ostensibly purist premium.

The Spartan 10 side-steps this problem by firstly optimising the subsonic filter for real component values, not truncating ideal values to standard component values and hoping for the best. Once the network is tuned for real world values, 5% capacitors hand matched to within 1% of each other are used in each phonostage so that the filter response is ultra-flat above the cutoff frequency with perfect balance between both channels. All that changes is the absolute cutoff frequency by about 1Hz in either direction and, importantly, not the flatness of the filter; not a big deal if both channels are using identically matched capacitors. The best, and most expensive, capacitors available for the job without matching are typically ±1%, meaning that they might deviate by a total of 2% from one capacitor to the next, causing a potential 0.2dB of response variation, potentially losing the 0.1dB precision RIAA tracking instantly.


Spartan 10 subsonic filter

Spartan 10 subsonic filter

The Spartan 10 implements a 18dB/octave subsonic filter with a -3dB turnover at 22Hz, attenuating subsonic disturbances by over 22.5dB (a factor of 12) at 10Hz while ensuring response flatness within 0.1dB all right down to 35Hz, preserving the bass response while stopping insidious subsonic content from causing problems further down the signal path. Even better, and quite uniquely, the subsonic filter is implemented before the final 7dB of the total 41.6dB of gain is realised, so the subsonic content can't eat into the final overload margin of the phonostage, like it does in all other designs that implement the filter after all the gain has been applied. Splitting the gain this way also helpfully shares the burden of amplifying the signal between the Spartan 10's two amplifier stages, improving the overall linearity.


Linear PSU

Many small phonostages today use an external switching power supply that can inject nasty audio frequency currents straight from the mains into the audio ground path between the phonostage and power amplifier resulting in unpleasant buzzes, hums, and whines adding a most unwelcome accompaniment to the music. With these supplies it is not possible, without the further addition of more noisy switching electronics inside the enclosure, to derive a proper split power supply for the audio electronics to work off, and a single supply arrangement where the sensitive audio ground has to be precariously derived from half the power supply voltage, sharing the same ground path as the power supply rails. With the standard 24V DC single supply this means that only 7V RMS of headroom is available, or even less than 5V RMS in some cases where the 24V supply is regulated down to 18V in an attempt to reject some of the switching noise, seriously increasing the risk of obtrusive overload when attacked with subsonic content and surface scratches.


Spartan 10 Board

Linear split power supply, left side of Spartan 10 board

Unlike the majority of small phonostages available, the Spartan 10 uses a tried-and-true ±15V split linear power supply that affords far better isolation from the mains. Alongside the front panel power switch, a switch-on muting relay is employed so there are no loud thumps through the speakers when the Spartan 10 is switched on, only a satisfying 'click' from inside the enclosure after the second it takes for the power supply to stabilise. Having a true split supply of ±15V means that the Spartan 10 can handle an internal level and line output level of over 9V RMS, realising not only a very comfortable overload margin, but also permitting the gain to be increased so that the output with a standard moving magnet cartridge can subjectively match that of a modern DAC without the smallest risk of overload. Using a split supply also keeps nonlinear amplifier power currents out of the signal ground ensuring no injection of extra distortion into the signal path.


Linear PSU

External transformer PSU available from Classic Audio Ltd.

By using an external transformer based power supply, the Spartan 10 keeps the power transformer and its troublesome magnetic field well outside the enclosure, while also keeping mains voltages outside the enclosure. As the external PSU is double insulated, the risk of ground loops between the amplifier and Spartan 10 is also entirely removed. External transformer based power supplies, while not as cheap or convenient as switching ones for a low parts count and corresponding low cost, have an excellent track record for longevity and are easily replaced in the highly unlikely event of failure. Cheap switching supplies, however, are notoriously unreliable due to the proximity of low cost electronics and heat, noisy, and have a very undesirable habit of allowing the ground to 'float' at 80V RMS or so, damaging sensitive line inputs when hot swapping equipment.


Topology

Designing a high performance phonostage is a considerable technical challenge. A high quality MM phonostage must take a high impedance input with a rapidly rising high frequency response and amplify it by more than 1000 times at 50Hz, remove subsonic content to prevent intermodulation distortion further down the signal path, and be immune to very strong high frequency transients that can be well over 10 times higher than the nominal cartridge level of 5mV. It must also equalise the RIAA curve to within a fraction of a decibel to prevent colouration of the sound, and have a very low input noise so as not to eat into the dynamic range of the vinyl medium. All of this has to be done with very low distortion and plenty of headroom, to keep those high level frequency transients crisp and clean, without any overload artifacts in the midrange.

The Spartan 10 uses an original new phonostage topology, consisting of several unique features that push the practical performance through the stratosphere. Particular attention has been paid to the crucial first amplifier stage. The first stage not only has to form a well-behaved, low noise MM input, but also must never have work so hard that it's linearity is degraded in the mid-band. Single stage designs that ask for gains of 40dB or so from unity compensated op-amps fall down here, most often copied out of the service manuals of cheap 1980s audio equipment by marketeers masquerading as designers. In most of these designs a flashy op-amp with impressive seeming characteristics on the datasheet is shoehorned in for marketing purposes, that either doesn't make a difference in practice, or degrades the compromised performance even further.

Classic Audio firmly believes in performance from topology, circuit architecture and layout, more than any other design choices. It is better to have a palace made out of solid, reliable, and ordinary bricks, than a dilapidated pigsty built out of Italian marble covered with gold leaf! The paragraphs following the schematic diagram below will cover this approach in detail and explain how the Spartan 10 generates a specification and subjective impressions, using non-exotic parts, that outdo virtually all phonostages available below the triple figure mark.


Spartan 10 topology

Simplified circuit topology of the Spartan 10

Starting with the input network the cartridge hits a DC draining resistor that prevents any charge from building up on the input network when removing cartridge headshells, and then discharging with a loud thump through the speakers when the next headshell is put in. DC decoupling that stops the amplifier input bias currents from running through the cartridge occurs through an electrolytic capacitor, realising a low noise, low impedance coupling between the cartridge and input amplifier. Next the MM signal passes through the RF filter which not only attenuates radio frequency interference, inevitably picked up on the turntables output leads, through a low value series resistor, stopping the phonostage from becoming a mobile 'phone detector. The RF filter also usefully loads the cartridge with the total 100pF it wants to see from the phonostage to attain a flat high frequency response. Most cartridges like to see about 250pF of loading; 50pF through the tone-arm wiring, 100pF through the turntable output cable, and finally 100pF from the phonostage.

There is no option for variable loading, adding additional capacitance only peaks the high frequency response and results in premature roll-off. The input impedance is however raised from the standard 47kΩ to 50kΩ with an aim to slightly extending the high frequency response of the cartridge. Raising the input impedance slightly also allows the capacitance of the cabling, and hence the length of the cabling and associated hum and RF problems, to be slightly reduced without sagging the high end response.

After DC decoupling and RF filtering the cartridge output hits the first amplifier stage. This is done through an NE5534 op-amp selected specially for its input noise voltage and current characteristics, that ideally match the impedance characteristics of MM cartridges, more so than any other readily available device on the market. Bipolar input amplifiers are ideal for MM cartridges as they exhibit very low nonlinear input capacitance compared to FET devices. Some new JFET input devices appear to come close with low voltage noises, but exhibit large non-linear input capacitances that will be further exacerbated by high cartridge impedance at high frequency to degrade distortion performance. Conveniently, for manufacturers who use FET input op-amps, the high frequency distortion does not appear on the test bench with a low impedance distortion analyser. They're also only available in surface-mount package, and based on past experience cannot be relied upon to be available for very long into the future. Readily available with a solid reputation in professional audio, the NE5534 is rather special as it isn't unity gain compensated.

When any amplifier is unity gain compensated that means it's high frequency response and output speed have been curtailed, so that it is stable at gains of one where 100% feedback occurs. Because high frequency response hasn't been curtailed through unity compensation in the NE5534, it can achieve much better high frequency distortion performance, and lower noise as the input stage isn't resistor degenerated to make up for the loss of linearity. It can realise this improved performance in a phonostage, providing that it isn't destabilised by returning more than 30% of the output to the feedback point at high frequency.

Previous designs that use the NE5534 have simply applied compensation with an external capacitor, and then returned 100% feedback at high frequencies. The Spartan 10 takes a different approach; it modifies the active RIAA network so that it can never return enough feedback to destabilise the amplifier, with a maximum of 27% feedback returned at high frequency. The result? Greatly improved feedback factor at audio frequency, more than twice the slew rate, and finally less than half of the mid-band and high-frequency distortion compared to ubiquitous compensated designs. Using this approach, Spartan 10 can be savagely attacked with all kinds of high frequency transients, well outside the capabilities of MM cartridges, and still retain a graceful composure.

A high quality series-feedback RIAA network, built with polypropylene and NP0 capacitors for zero distortion, then induces the amplifier's output to track the RIAA curve with a high level of accuracy and low output impedance. The impedance of the RIAA network was carefully chosen for loading characteristics that bring out the very best distortion performance from the first stage. The total gain of the stage is a modest 35dB at 1kHz, so as to retain a strong feedback factor and reduce the working level on the already linearised first stage. The remaining gain will be made up later down the signal path.

Passive RIAA equalisation is not used as it's grossly inferior, breaking the two most fundamental rules of high quality audio design; don't amplify and attenuate, throwing away valuable headroom, and don't attenuate and amplify, amplifying a greater ratio of noise vs signal. The designer suspects that passive RIAA equalisation is only popular because of its ease of design by charlatan marketeers. Active RIAA equalisation is difficult to design, with a great deal of calculation, and then simulation with real component values required to closely follow the RIAA curve at high frequency. When it's designed and built correctly, series-feedback active equalisation completely blows the passive method out of the water, with superior noise and overload performance. Sadly, many manufacturers have hoodwinked their customers into paying a premium for their own inability to design an active RIAA network. An active network makes the very best of the drive amplifier's gain, never requiring available gain and headroom to be thrown away in a passive network when that gain could be put to good use, reducing distortion at the ever-rising MM cartridge high frequencies.

Up next, the first pole of the 3rd order subsonic filter DC decouples the intermediate level signal from the input amplifier. It is then fed into the the LFC bridge which works through the first pole of the subsonic filter, therefore requiring no additional components in the signal path when disabled. Mono bridging (summing) is obtained across the high frequency correction pole. The correction pole takes over the tracking of the RIAA curve when the first stage reaches maximum feedback and stops attenuating the signal at 32kHz. Because this is a low impedance point in the circuit, crosstalk is very low across the bridging switch, almost by a factor of ten vs other designs that apply mono bridging right at the sensitive high impedance MM input. A dead giveaway that this is the case is if the mono switch is inconveniently placed on the rear panel, right next to the MM input connection. By using the series resistor for averaging with the opposite channel, no additional components are placed in the signal path by the mono switch. All units are tested to a mono summing balance of 40dB of vertical attenuation, with most reaching well past 50dB.

The signal passes through the next two poles of the subsonic filter, a mild Chebyshev filter with a super-flat in band response brought about by hand selecting and matching the series capacitors. DC decoupling of the last amplifier stage from the bridging switches is done through the subsonic filter, preventing bias currents from producing clicks when the switches are operated. The final amplifier stage then applies the last 7dB of gain to bring the total gain up to 41.6dB at 1kHz, taking this burden off the first amplifier stage. Because the gain is applied after subsonic filtering, the headroom of the system is defined by the second amplifier stage, which is free of subsonic disturbance, and not the first amplifier stage. This means that subsonic disturbances cannot eat into the total headroom of the phonostage, further improving output capacity and overload margin in real world use.

The second stage, made up of the legendary NE5532 op-amp, does not need to work hard at high frequency or apply high gain, as it sits downstream of the correction pole and only has to amplify the signal by slightly more than a factor of two. With the input impedances to the op-amp matched at the input and feedback network, combined with the lower common-mode voltage as a result of the gain applied, the second stage's input distortion characteristics are kept as low as possible. The second stage also has no heavy load to drive internally, only the light loading of its feedback network and the subsonic filter, so is quite capable of driving line loads as low as 2kΩ and beyond with excellent linearity.

Unlike single stage designs, the correction pole sits in between the two stages and not across the line output like it does in single stage designs. This isolates the correction pole from the line load, which can vary from 10kΩ to over 200kΩ depending on the equipment downstream, maintaining a flat frequency response regardless of the load on the line.

Lastly, the line output network DC decouples the second stage from the line output, preventing clicking transients from being heard when a source selector switch is used. A series resistor defines the 100Ω of output impedance and makes sure that the phonostage is always stable when connected to high capacitances, inductances or any combination of the two present in the cabling or the next amplifier downstream. The soft start muting relay makes sure, with the help of control circuitry not shown, that the output is always disconnected from the phonostage and grounded when the phonostage is switched off or powering up or down to stop any loud bangs or thumps being heard as the circuitry stabilises.

All of these unique topological improvements have come together to produce a phonostage that is not only economical enough to be a reality for many record collectors with limited financial resources, but also realises a performance that can beat many competitor products priced a factor of 10 times or more higher. Hopefully it will be demonstrated at this point that while high quality components are important, there is no replacement for topology if the performance characteristics of the Spartan 10 are to be had.


Get in touch

If you have any questions or comments relating to this product, then feel free to contact classic audio. You'll receive a quick response and all technical questions will be answered in full, gory detail. At the time of writing, bank transfer is the only means of making payment. Postage is to the UK only for the time being, but this will change in as soon as there are enough spare units to go around the world.