The Artemis Labs PH-1 is a tube-based, made-in-USA, all-triode, phono stage designed by John Atwood of One Electron™ fame. John has been involved in some high-powered tube design projects. For example, his work on the now defunct Fourier Components Sans Pareil and Triomph OTL amplifiers is well know to me. The PH-1 represents his latest thinking on subject - the result of several years of perfecting a simple yet extremely high quality phono stage, optimized for medium to high-output cartridges, either moving coil or moving magnet. While many affordable phono stages tend to be rather basic in conception and execution, this product features such a veritable cornucopia of interesting design aspects that I feel honor bound to do all of them justice, so please bear with me.
The Artemis Labs PH-1 takes its cue from the 5687 used in the output stage. This is a medium mu dual triode from the tube design team at Tung-Sol that was intended for industrial and military applications. It runs at a much higher filament current relative to other 9-pin miniatures, which is partially responsible for its linear response, and in its mil-spec version, is known for unit-to-unit consistency. Fortunately, it has seen little use in high-end tube gear and thus it is readily available at very reasonable prices. This is a tube that I also like a lot; it just sounds good even when driven hard.
A large (500 Hy) Lundahl choke is used for the 5687 plate load instead of a "Plain-Jane" resistor. It is a high-cost solution to the age-old quest for the ideal plate loading. A large plate resistor makes the tube behave like a current source, provides excellent power supply rejection, and generates the lowest distortion possible. Unfortunately, because the tube and plate resistor act as a voltage divider with respect to the high-voltage bias supply (B+), there is a practical limit to the plate resistance that may be used: the larger the plate resistor value, the higher B+ needs to be. Recall that a choke features a large AC impedance but a low DC resistance. Thus, from the tube's perspective, the choke appears as an ideal "infinite" resistance load to the audio signal without a significant drop in B+ voltage. Choke loading makes it possible to have your proverbial cake and eat it too: current source operation with a moderate plate supply of only 160 VDC.
Another benefit results from the fact that the choke is also an energy storage device, permitting plate voltage to swing to almost 100% above the B+ plate supply, doubling the 5687's headroom and accounting for its huge output voltage of over 40 V rms before clipping. The downsides, in addition to cost, are the need to shield the choke from magnetic fields (hum) and the required core size to avoid low-frequency distortion. The weight of the two chokes is reflected in the unit's weight, being quite substantial at 24 pounds.
The output stage is actually a two tube affair with a 12AX7 high-mu triode feeding the grid of the 5687. A modest amount of feedback (6dB) is used between the plate of the 5687 and cathode of the 12AX7 to stabilize the gain of the output stage and reduce the frequency response sensitivity to loading. In my opinion, this degree of inter-stage feedback is a good thing, and purists need not be concerned.
To quote Atwood, "the first stage is a low-noise "preamp" to bring the signal level up before hitting the passive equalization network. A high-transconductance tube with reasonably high gain is needed. I used to use 6DJ8s for this position in my old designs, but have come to dislike the sound of that tube. The Russian 6N1P sounds much better, and is quite low-noise, too." Being tasked with amplifying microvolt level signals, the input stage's design and execution are critical to the overall success of the phono stage. Low-noise operation is imperative. Passive parts are said to have been chosen carefully for low noise and minimum low-level nonlinearities. To get around the problems of cathode biasing and the typical electrolytic bypass capacitor, a single "N" alkaline cell is employed in the grid circuit to provide fixed bias. This allows the cathode of the 6N1P to be connected straight to ground. Being in the grid path, virtually no current is drawn, resulting in many years of battery life. It is important to note that any battery non-linearities are not manifested by the tube's current flow, which would have been the case had the battery been inserted in series with the cathode.
Atwood's philosophy is to use feedback very carefully. He says that phono stages with feedback-type equalization are especially prone to slew-rate distortion, due to high-energy high-frequency crud coming off the record. Thus, the PH-1 uses fully passive equalization, followed by a two-stage amplifier with limited feedback. This means that little high-frequency energy interacts with the feedback stage. EQ accuracy is said to be within +/-0.2dB. In order to achieve such an EQ tolerance, network capacitors are hand-selected on a precision impedance bridge to better than +/-0.5% accuracy.
A totally cool feature of the PH-1 is its Cool-Swap™ technology that facilitates cooler running tubes and provides a built-in spare. Since many dual triodes, including the 5687 and 12AX7 (but not the 6N1P), have split heaters it is possible to operate only a single triode section. In the Cool-swap configuration, one 5687 and one 12AX7 are used per channel, each with only one half heated. The right channel uses triode #1 and the left channel uses triode #2, the unused triode in each tube being essentially a spare. After a couple of thousand hours of use, you may want to swap the positions of the two 5687s and 12AX7s, thus bringing the unused spare triode into use. In other words, the left channel tube changes places with the right channel tube. You might say that Cool-Swap™ effectively doubles tube life, and since only half the tube is heated, these tubes run cooler, an important factor in the typically hot running 5687. Tube life is further enhanced by a soft heater turn-on and delayed application of high voltage. The B+ plate supply is delayed by about 40 seconds to give the tube heaters ample chance to warm-up without the chance of "cathode stripping." In addition, a muting relay shorts the audio outputs for about 4 seconds after high voltage is applied to allow transients to die away. Note that all time delay and regulation circuits use proven circuitry based on discrete semiconductor devices.
The power supply uses high-speed solid-state rectifiers for both the plate and heater supplies. A MOSFET based voltage regulator is used for each channel B+ rail to maintain a stable operating point, independent of power supply fluctuations.
All critical circuits in the audio path are wired point-to-point on military-style terminal boards which feature silver-plated turrets. According to designer Atwood, fiberglass FR-4 material has odd dielectric behavior, something that Tektronix found out in their oscilloscope designs. By using military-style terminal board construction and point-to-point wiring in the areas where high-impedance audio signals are present, the sonic effect of the PC board is minimized. Conventional fiber-glass PC boards are used for the power supply and for the rear panel jacks and switching. All film capacitors in the signal path are either polypropylene or silvered-mica types.
While moving-magnet phono cartridges are typically designed to work into a 47 kOhm load resistance, most moving coil cartridges sound their best into a much lower resistance and may even benefit from additional load capacitance. With almost all phono stages one is pretty much stuck with a fixed loading. Thus, unless you are handy with a soldering iron and probably willing to violate the product warranty, you have no chance to experiment with cartridge loading to squeeze out the best sound possible. Not so with the PH-1. In fact, it facilitates experimentation. A 47.5 kOhm resistor is hard-wired across the inputs of the PH-1. That's the starting point, but a 3M TexToolฎ ZIF (Zero Insertion Force) Socket, originally designed for the semiconductor industry, makes it very convenient to add shunt resistors or capacitors. Simply take the cover off and flip the lever out to open the socket. The top three positions of the ZIF socket are available for the right channel and the bottom three positions are for the left channel. Be sure to shorten the resistor legs and bend them to fit into the apertures. Closing the socket clamps the resistor/capacitor leads between gold-plated blades for a secure electrical connection. Note that any added resistance is in parallel with the fixed internal resistance of 47.5 kOhm. It was a good old-fashioned analog blast experimenting with my Symphonic Line RG-8 Gold cartridge, eventually settling on 681 Ohm shunt resistors (1/4 W, 1% Vishay/Dale metal-film), which give an effective load resistance of about 670 Ohm.