Ayre K-5xeMP line preamplifier
The name Ayre Acoustics is always associated in my mind with preamplifiers; the Colorado company's second product was the K-1 preamplifier, which was reviewed for Stereophile by Wes Phillips in March 1997, and it followed the Ayre pattern in being fully balanced and using no overall loop negative feedback. Ayre's cofounder and chief engineer, Charlie Hansen, hates feedback with a passion.
The K-1 cost $5250 when it was introduced at the end of 1996, which was not inexpensive. By the time Paul Bolin reviewed the K-1x in 2002, the price had risen to $6750; and in 2007, when WP wrote about the K-1xe, that version cost $7000. This is only a small jump in price in five years, but it's still a bunch of change for a line preamplifier. It was therefore a no-brainer for Ayre to introduce a less expensive preamp, the K-5xe, which replaced the K-1's elaborate, mechanically coupled volume control and input switching with FET switches and metal-film resistors, while retaining the big model's fully balanced, zero-loop-feedback, all-discrete topology. Sam Tellig reviewed the K-5xe line preamplifier in May 2006 (Vol.29 No.5), when it cost $2950, and enthused about its sound: "The Ayre K-5xe . . . just got out of the way. It's solid-state, as good as it gets. The sound is open, airy, and sweet."
I asked for a sample of the K-5xe so I could do a Follow-Up to Sam's review, but other review commitments kept getting in the way. When I finally spent some time with it (S/N 10J002), I found the sound a little on the robust, forward side, which made system matching problematic. Then, as I was about to spill some ink on the K-5xe, I got an e-mail from Charlie Hansen letting me know that the development of the QB-9 USB DAC had led them to rethink the K-5xe's design, and that Ayre would be sending a sample of what would be called the K-5xeMP. After a longer delay than I had anticipated, the K-5xeMP, priced at $3500, arrived for review.
The new version of the K-5xe can be distinguished from the previous version by the "wave" logo that appears on the front panels of Ayre's digital products, below the Ayre logo, though the "MP" now stands for "Maximum Performance" rather than "Minimum Phase."
The primary difference between the MP and its predecessor concerns the amplifying devices. I'll let Charlie Hansen explain, as the story is a fascinating glimpse into how a high-end audio engineer develops a product:
"The K-5xe was the only non-power amp in our entire product range that did not use JFETs for the output (buffer) stage. The reason was that when we designed the K-5, we had just switched from FET outputs to bipolar transistors (BJTs) in our power amps and were extremely happy with the results. We figured that, since they worked so well in the power amps, we would also use them in the preamp.
When we were designing the analog circuitry for the QB-9, I was determined to make a circuit with bipolar transistors that sounded as good as our JFET-based circuits, the reason being that Toshiba had discontinued the low-noise P-channel JFETs that had been a staple of our designs. We made a lifetime buy of 500,000 pieces, which should last us somewhere between 50 and 100 years, but I still only want to use these parts when absolutely necessary.
When it came time to design the output (buffer) stage for the QB-9, we decided that the easiest thing for experimentation would be to start with the circuit used in the K-5xe rather than build something from scratch. We soon found (for reasons I still don't understand) that the simple complementary BJT emitter follower used in the K-5xe was holding back the performance of that whole unit. The easiest solution was to replace it with a pair of complementary JFET source followers, as found in, for example, the C-5xeMP, the K-1xe, or the KX-R.
Rather than use up the Toshiba JFETs, we found some US-made parts that sounded nearly identical. That is when I originally notified JA of the upgraded unit. We were just getting ready to ship it out to him when we ran it through the battery of measurements to make sure that all was well with the unit. That is when we found that the US-made JFETs caused the THD of the unit to double compared to either the Toshiba JFETs or the previous bipolar output stage. Their transconductance is only about one-third that of the Japanese parts, and this is apparently the source of the problem.