Solid State Preamp Reviews

Following this mechanical switch array, we get to the heart of the Iris, what Hafler has cutely (and somewhat annoyingly) termed the "Cyber-Optic" volume control. One problem with relatively inexpensive preamplifiers is that the lower financial resources available to the designer forces him (or her) to choose an inexpensive, logarithmic-taper, volume-control potentiometer. As well as having a limited lifetime, these often feature variable tracking between the two channels, particularly at low levels, which leads to drastic changes in channel balance dependent on the listening level chosen. In such high-end preamplifiers as the Krell and Mark Levinson, the designer solves this problem by using a highly specified component from, for instance, Penny & Giles, but as these pots can cost as much as the raw parts that go to make a mid-fi cassette deck, this is only applicable when the customer is willing to bear that cost. Other alternatives are to use a voltage-controlled amplifier, but these have had generally bad press—the Meitner PA6 being an honorable exception—and are also not inexpensive.

At a more affordable level, the use of solid-state switching, using, for example, a resistor ladder switched by something like a 4051 chip, can ensure good channel balance at the expense of limiting the volume control to a number of preset levels. Such a solid-state volume control, designed for Hi-Fi News & Record Review magazine in 1980 by my then-colleague Ivor Humphreys (now Technical Editor at Gramophone), featured 32 steps, which was definitely not enough. Such switched volume controls also often give an audible click at every step. And compared with a good, old-fashioned knob, solid-state switching using up/down buttons lacks any kind of visual feedback to let the user instantaneously know how loud the music will play before it starts.

The problem facing a designer who wants to dispense with a conventional volume pot is therefore twofold: to arrange for a large enough number of steps that the volume can be set to an effectively infinite number of preferred levels, and to include some kind of level indication system. So, what do you see on the Iris's front panel? An old-fashioned volume knob! But the fact that the remote control also sports good old analog knobs for volume and balance should indicate that all is not what it seems.

The heart of Hafler's "Cyber-Optic" control is a very familiar device, a Cadmium Sulfide (CdS) photosensitive resistor, as used in countless photographic light meters and also in automatic-nulling distortion meters. The principle is simple. Use the CdS resistor as the shunt leg to ground in a potential divider and shine a light on it. Increase the brightness to lower both its resistance and the volume; dim the light to increase the volume. But to actually make use of a CdS cell in a practical volume control is not that simple. Inside the Iris, the output from the switch array is split into two, each leg fed to just such a voltage divider, with the shunt resistance to ground being half of a CdS cell. The illumination is provided by an LED mounted in a sealed enclosure with the CdS cell, and the outputs from each leg are recombined after passing through an RC series network. But the trick used by Jim Strickland to a) get sufficient dynamic range from the dividers, b) arrange for a familiar logarithmic characteristic, and c) ensure good tracking between channels, is both complex and ingenious. (A patent has been applied for.)

Basically, both CdS elements and the LED driver are included in a servo circuit that monitors the resistance of the elements and ensures that the LED illumination and photoresistor value correspond exactly to either a DC voltage set by the position of the front-panel knob, which operates an inexpensive linear pot operating as a voltage divider, or by a DC voltage supplied by the input signal from the remote control. This may sound simple, but to implement it requires a number of ICs—dual op-amps to act as buffers, error amps, etc., and a transistor array to act as a log converter. Purists noting all these ICs under the Iris's hood should be aware that none are in the signal path; the volume-control element is the purely passive CdS photoresistor, the rest is housekeeping circuitry.

In addition, to avoid unnecessary operation of the servo—being feedback devices, servos can sometimes misbehave—the whole thing is arranged only to work when you turn the balance or volume knobs on the front panel or on the remote. Assume that you want to turn up the volume with the front-panel knob. The instant you start to rotate the knob, the Iris detects that you are doing so by the rate of change of voltage from the pot's wiper and transfers control of the preamp functions from the remote to the main unit. The rate-of-change signal also turns on the servo operation so that the "Cyber-Optic" volume control will instantaneously follow your rotation of the knob. When you release the knob, the servo relaxes, holding the volume-control parameters constant at your last setting.

At your listening seat, you want to adjust the volume from the remote. Touching either balance or volume knob on the remote transmits a digital code representing a DC voltage to the preamp. When the Iris receives this code, it transfers control of volume to the remote (selecting a source with the remote without touching the volume or balance controls leaves the main unit in charge of volume and balance). Once checked for errors, this code is applied to an 8-bit DAC per channel to provide the volume-control servo with one of 256 control voltages. Whereas the front-panel control offers effectively infinite resolution, the remote will offer 256 discrete steps. This, however, is pretty close to continuous operation of the volume control. One useful aspect of the volume control is that when set to its 12 o'clock position, it offers approximately 20dB of attenuation, which almost exactly cancels the gain of the line stage. For some comparison purposes, it is useful to be able to tell at a glance whether you are overall attenuating or boosting the input signal.

The balance control operates in an identical manner. Unusually, it attenuates the opposite channel by up to 13dB but boosts the selected channel by up to 3dB to maintain constant music-power operation. The signal can be muted by 20dB again with either a front-panel switch or via the remote. This muting function is achieved by the "Cyber-Optic" control and is canceled by a Normal button.

The line stage, again, is a complementary-symmetry, all–J-FET circuit. The input is a long-tailed–pair differential amp with a constant-current source load. The volume control feeds the non-inverting input while negative feedback taken from the output is taken to the inverting input, along with a fraction of the pre-volume-control signal. The output is AC-coupled to the single pair of sockets by a 2µF plastic-film capacitor. A final FET crowbars the output to ground when the preamp is first powered to avoid turn-on thumps.

As ingenious as the Iris's main circuit is, it is apparent that as much attention has been paid to the remote. With the transmitter and receiver chips used in TV sets widely and cheaply available, you may wonder why Jim Strickland chose a more complicated route. Again, it comes back to the volume control. Jim didn't want to reduce the number of steps below 256, and that was just too many to be implemented with the common chips. In addition, with so many components using the same chip sets, data security has become a major problem. The last thing you want to happen is to operate a CD player or VCR and find that you have inadvertently set the preamp volume to maximum!

The Iris therefore uses a dedicated code, comprising 32 bits per frame at an 8fps rate. Each frame consists of an 8-bit word describing one of the 256 possible values of volume, this determined by an 8-bit ADC that measures the DC voltage set by the remote's volume control, then a second 8-bit word for one of the 256 values of balance; these 16 bits are then repeated in inverted form to provide 100% parity checking—after receiving the first 16 bits, the Iris then stores them temporarily to compare with the inverted version in the second half of the frame. If these differ, it takes no action. Gratifyingly, Hafler chose to use two high-intensity IR LEDs in the remote. It would appear that as long as the remote is in the same room as the preamp, it will operate: no more carefully aiming the remote at the target.