The application of both traditional principles and some innovative electronic design work affords this dual‑channel preamp impressive performance. Hugh Robjohns likes what he hears.
It seems as if there are almost as many microphone preamps to choose from as there are microphones these days. The current state of electronic component and circuit design means that pretty much all of the currently available devices are entirely competent — the XDR preamps built into my little Mackie 1402VLZpro are good examples of the excellent quality that can now be achieved at surprisingly little cost.
However, there are still only a few mic preamps which stand out above the crowd as being something really special. Inevitably, most of these are fairly expensive — typically costing over a thousand pounds for a stereo or dual‑channel unit. These high‑end units are mainly solid‑state designs, although many incorporate valve stages in the signal path. It all comes down to whether the designer is going for a purist, character‑free amplifier or something with its own distinct sound signature.
Aphex Thermionics are a division of the well‑known Aphex Systems Inc, based in Sun Valley, California. This new subdivision of the parent company has been formed specifically to develop a number of innovative valve circuit‑design techniques and to apply them to high‑end pro‑audio products. The Model 1100 is the first unit to employ some of these patented techniques and it already sets a very high standard for the company.
The Model 1100 is essentially a two‑channel microphone preamplifier with built‑in high‑resolution A‑D converters. Each channel has only one electronically balanced XLR input which, though nominally for microphone signals, will accept line levels up to +26dBu. Each channel's output is of the earth‑compensated or quasi‑balanced variety and is available paralleled on both an XLR and TRS quarter‑inch jack socket. Each signal path includes a pair of triode valve stages, one as part of the input circuit and a second as part of the output buffer. Adjustable high‑pass filtering is included, as is a switchable peak limiting system — both innovative designs in their own way.
The styling of the 1U rackmounting unit is very attractive, with chromed rack ears and a front panel machined from a substantial chunk of aluminium, the latter polished to a blue mirror finish and using clear and long‑lasting anodised silver legends. The casing measures around 235mm deep and the unit weighs under 4kg, but don't be tempted to slot it into that spare 1U of space in the rack, because it burns 35W of energy and runs surprisingly warm. The handbook recommends leaving space both above and below for ventilation and I think that this is good advice.
The rear panel is clearly labelled and the socketry for each channel is well laid out. In addition to the audio connections, each also sports a second quarter‑inch jack socket which provides a remote mute facility — a switch wired across the tip and sleeve connections mutes the channel's output when open. The internal normalling contact on the socket ensures the remote mute facility is switched off when the socket is unplugged.
To the left of the individual channel connections is a third interface area concerned with the digital output and word‑clocking arrangements. A standard AES‑EBU digital output is presented on an XLR connector with the two preamp outputs encoded in the stereo data stream as left and right. The A‑D can operate from its internal clock at 44.1, 48 or 96kHz sample rates, the latter being output in 'double‑fast' mode over the single connector. A pair of BNC connectors provide an input for an external word clock with either the internal clock output or a relay‑linked loop‑through of the external clock input available on the second connector. No automatic or switchable impedance termination is available on the Model 1100, so a 50 or 75Ω terminating plug will be required if synchronising the unit from an external master clock. The standard IEC power inlet is accompanied by a mains fuse and voltage selector which accommodates supply voltages from 100 to 240V.
The front panel controls are arranged simply and logically, both channels being almost identical. Each channel has a pair of sturdy rotary switches with knurled brushed‑aluminium knobs. The first adjusts the gain in 4dB steps from 21 to 65dB and the second determines the turnover frequency of the 24dB/octave Butterworth high‑pass filter, variable from 195 to 30Hz in eleven steps with a final 1Hz 'off' position.
A recessed section of the panel contains seven buttons, an LED bar‑graph meter, and a host of indicator LEDs. The meter actually displays system headroom and spans the range from 50 to 0dB. The bottom half, up to 20dB headroom, employs green LEDs at 3dB intervals. The mid‑section is yellow with 6, 4 and 2dB steps up to 4dB of headroom, with the last four steps at 1dB intervals in red to 0dB. The final red LED indicates output clipping or the activation of the 'MicLim' switchable peak limiter. A column of three further LEDs provides different information in each channel's display: Channel 1 shows the word clock source for the A‑D converter (Internal, External, and Locked) whilst Channel 2's column shows the sample rate when clocked internally.
The seven buttons all have associated indicators, which is just as well, as it is hard to tell whether the buttons are engaged or not from a quick glance. The first button selects a 20dB input pad, which would normally only be required to accommodate either line‑level signals or those from a microphone in close proximity to something very loud, such as a trumpet. The Model 1100 has 'absolute polarity', meaning that a positive signal on pin 2 of the input XLR is output as a positive voltage on pin 2 of the output XLR, but the second switch reverses this. Both of these facilities are carried out directly on the input signal before it reaches the first stage of amplification.
The third switch has to be held down for over a second before the 700Hz sine‑wave oscillator is engaged. This replaces the microphone signal just before the output muting circuit and is available to both the analogue and digital outputs of the unit. The signal level is internally aligned to ‑20dBFS and 20dB below the analogue clip point. The next button activates phantom power, which has the rather nice feature of ramping up and down over a five‑second period to avoid the audible 'thump' that normally occurs when phantom is switched on or off.
The Mute switch comes next and this kills both analogue and digital outputs irrespective of whether the programme or tone source is selected. The remote facility allows a 'cough‑key' system to be easily arranged — this enables the 'talent' to mute their microphone and avoid deafening the engineer when clearing their throat. The two mute buttons (local and remote) have sensibly been wired in series, so one cannot override the other — whichever engages the mute must also disengage it.
The penultimate button, and the last pertaining to the preamp channel, selects the MicLim facility. This is unique to Aphex and provides a fast‑acting peak limiter with an intriguing design. Essentially it employs a variable pad at the input to the first preamp stage, the attenuation of which is controlled by a signal peak detector monitoring the output level. This variable pad is based around a custom‑designed opto‑resistor wired across the balanced input, forming an attenuator with the microphone's own output impedance. This design therefore controls the signal level before it reaches the amplifier and so preserves the amplifier headroom.
Normal CdS (cadmium sulphide) opto‑couplers tend to have relatively high resistance, even when switched 'on', which wouldn't work in combination with the very low impedance of a typical microphone at, say, 150Ω. To overcome this Aphex had to develop a special version of the CdS opto‑cell with an 'on' resistance some 100 times lower than a standard device so that the relative impedances form a workable and useful attenuator. In practice, with a 150Ω microphone, the system can introduce up to about 16dB of input attenuation, although the exact figure depends on the source impedance of the microphone — the higher the source, the greater the possible attenuation.
The resistance of the opto‑cell is controlled by the amount of light falling upon it, which is in turn controlled by the drive current generated by the signal‑peak detector circuit. Most opto‑coupled compressors and limiters operate in a feed‑forward configuration, because CdS cells suffer both latency and hysteresis (lag and memory effects) which tend to make feedback topologies unstable. At a simplistic level, when the drive current to the light source is increased nothing happens for a short time so the controlling circuit increases the drive current further, by which time the photocell has reacted and the signal level is reduced far too much! The controller then backs off, allowing the signal level to rise too far and self‑oscillation usually results with the output signal varying in level in an inappropriate and unpleasant way.
However, to employ the opto‑cell at the input to the amplifier, a feedback arrangement is essential and Aphex have had to develop an ingenious control circuit to ensure a sufficiently fast response to control peaks, but without a tendency to self‑oscillate. The resulting MicLim system works extremely well as a protective peak limiter, although it is not designed for creative control of dynamic range. It is particularly effective with the odd peak transient from voices, but is rather too fast and aggressive‑sounding for use on most other sustained musical sources — especially a complete mix. Good as the MicLim is, it certainly doesn't replace a decent general‑purpose studio compressor/limiter system for overall dynamic control, and the handbook warns specifically against driving the MicLim system too hard.
The Model 1100 is quiet, smooth, clear and with the subtle but unmistakable warmth of valve circuitry. In comparing the Aphex against my reference GML mic pre, I felt it fared extremely well indeed. The two units have very different characters — the GML is very clean and neutral with an immensely full but natural bottom end, whereas the Aphex sounded a little more coloured and richer throughout the whole frequency range, and didn't have the same fullness at the bottom. I wouldn't like to say which was the quietest: they were both more than sufficient to justify 24‑bit conversion.
The unit not only looks as classy as it sounds, but also feels good to operate and works in a beautifully understated but professional way. For example, the ramping up and down of phantom power is a total luxury — completely unnecessary, but a nice feature which reminds you of the design effort that has gone into the Model 1100. Likewise with the remote muting facility and the inclusion of a tone generator.
The internal word clock source for the A‑D stage seemed highly stable and jitter free, and it was happy to lock up to external clock references from my Apogee PSX100 at all three rates. Likewise, I had no trouble using the Apogee to act as D‑A for the Aphex in all three modes. The quality of the 24‑bit converter seems to be very good, although I feel the lack of a 16‑bit output mode is a little restricting: although 24‑bit recording is definitely the way to go, there is still a great deal of 16‑bit equipment around which cannot be used directly with the Aphex. The only other omission here is of a termination facility for the word clock loop‑through — a slide switch to introduce a 75Ω termination would have been handy.
The MicLim system works remarkably well in catching the occasional peak transient from spoken or singing voices — the odd plosive pop or kicked mic stand. I found it typically allowed the gain to be set up to 8dB higher than was possible without the MicLim switched in, which equates to more than one bit of extra digital resolution. The handbook recommends its use with mics of more than 150Ω impedance, which includes the vast majority of moving‑coil, condenser and electret mics. Indeed, testing with an historic 30Ω ribbon mic caused barely sufficient attenuation for effective peak control, though this is unlikely to present a problem for most users.
The MicLim was inaudible until 'tickled', and barely then, so I generally left it in circuit without any problems. However, by deliberately overworking it I was able to detect a small tilt in its frequency response — the top end started to become rolled off in proportion to the amount of attenuation being applied. No great surprise, really, given the way the system is engineered, and certainly not noticeable under normal operating conditions where the brief attenuation of a transient is more than sufficient compensation for a momentary and barely audible change in the frequency response.
The vast array of corner frequencies in the high‑pass filter section allowed precise control of the low end of the spectrum and the slope was sufficiently steep to remove unwanted rumbles and subsonic clutter without undue damage to the wanted source. The other facilities — tone source, metering and so on — all worked as expected.
Overall, the Aphex Thermionics Model 1100 is a very nice machine. It does everything you would expect of a mic preamp in this price bracket and has a sonic quality to equal other competing units. It has a discernable character, unlike many high‑end units, but that is no bad thing and will appeal to many — especially as the character can be tweaked by exchanging the valves (see 'Tube Tweaks' box).
Amongst the mic preamps which I have personally tested and rate highly, there is my benchmark GML 8300, the DACS MicAmp, the Drawmer 1962, the TL Audio PA1, the Focusrite ISA110, and the Focusrite Red 8. However, with its sublime sound, the Aphex Model 1100 can also definitely earn itself a place amongst this distinguished company.
The Model 1100 was designed to use Russian‑made 6N1P dual‑triode valves in its amplifer stages, although the more readily available American 6DJ8 version can also be used with almost identical performance — perhaps just the merest hint of increased distortion separates the two valve types. These 6N1P and 6DJ8 valves have not been widely used in audio applications, yet they exhibit very low microphony, have a rugged construction, and have low‑noise characteristics which make them ideal for this application. However, other more familiar dual‑triode valves can also be used, as the RPA circuit topology affords automatic servo‑biasing, which not only means that it is self‑compensating for small variations in valve characteristics, but also that a number of alternative types may be substituted. For example, the common 12AX7 and 12AU7 (ECC82 and ECC83) dual triodes may be used quite happily to deliver the same general performance — identical gain, frequency response and so forth — but with an increase in the distortion (roughly ten times more with the AU7 and AX7).
To facilitate valve substitution, Aphex have included an internal selection switch for either 12V or 6V heater circuits. Aphex Thermionics include in the handbook a list of ten tested types of alternative valves, as well as a list of untried but theoretically suitable designs, and suggest that the user is free to experiment with substitute valves in order to tailor the sound of the unit, changing the amount and characteristic of the valve distortion to suit personal taste. This provides the Model 1100 with high‑end 'tweakability' and the facility to create your own personal sonic signature.
The Model 1100 employs a number of ingenious, innovative and patented circuit topologies which Aphex Thermionics claim provide a superior sound quality. Following the pad relay, MicLim opto‑resistor and phase reverse relay, the audio signal arrives, still balanced, at a discrete transistor differential amplifier. This stage comprises special low‑noise PNP transistors operating in Class A and has a variable gain ranging from ‑3 to +41dB, controlled from the front‑panel rotary switch. The transistor stage's output is then directly coupled to the differential dual‑triode valve amplifier circuit employing Aphex's patented 'Reflected Plate Amplifier' (RPA) topology and providing a fixed gain of 21dB.
Aphex's design operates the valve with relatively low anode voltages (bestowing almost infinite valve life and low operating temperatures) which are effectively held constant by a transistorised current mirror output stage. This circuit detects the anode currents in the triodes and 'reflects' them to the second transistors in the mirror circuit which convert them to the output voltage, thereby acting as output buffer and driver. The power rail voltages for each section of the amplifier are cleverly graduated so that the headroom available in each stage is the same and thus everything clips at the same point, including the A‑D converter. Moreover, the adjustable second‑order high‑pass filter is incorporated in a feedback path around the hybrid RPA stage, which has the advantage that headroom is increased at the frequencies being removed, minimising the risk of LF overload from unwanted subsonic signals.
After the mute facility the signal is split in three to feed the headroom bar‑graph meter, the internal 24‑bit A‑D converter, and the analogue output stage. The last route passes through a multi‑turn potentiometer (accessible from a trimmer on the front panel) which may be used to reduce the analogue output level calibration by up to 12dB. The output buffer employs another RPA hybrid stage, this time with a fixed gain of 3dB and configured to drive a single‑ended (unbalanced) output with an output attenuator controlled from a rear panel switch to operate at +4dBu or ‑10dBV. The unbalanced output signal is used to drive the output XLR and TRS sockets in a ground‑compensated or 'quasi‑balanced' configuration, which has the advantage of working equally well when connected to balanced or unbalanced destinations. This popular and widely used arrangement behaves much like a transformer‑coupled output in that both the correct signal level and the amplifier headroom are maintained whether the output is connected for balanced operation or for unbalanced operation. Obviously, there is no galvanic isolation, but neither is there the inherent distortion characteristic that transformers are well‑known for.
- Excellent sound quality.
- Very low noise and wide bandwidth.
- Effective peak limiting facility.
- High‑quality integrated 24‑bit/96kHz A‑D converter.
- Classy styling.
- Opportunity to 'tweak' the valve stages.
- Digital output fixed at 24‑bit resolution.
- Runs pretty warm.
- Up there with the most expensive dual‑channel mic preamps
A high‑performance preamp which is short on frills but long on clever design features where they count. Top‑notch sonic performance accompanies ease of use and no‑fuss operation.