Symetrix SX202 Modifications

updated 1-3-2012

The Output Amplifier

The output circuit is a very simple one. It uses a follower and a unity gain inverter to provide buffered in-phase and out-of-phase outputs from the SSM 2015 amplifier stage. Although the gain of the output stage is low, it creates the bulk of the stock preamp's distortion. While better balanced output topologies could be used, I have decided to stick with the stock arrangement because of its relative simplicity and good performance into balanced inputs, as well as its insensitivity to long cables and reactive loads.

As stock, the SX202 output stage uses a variant of the 4560 dual bipolar op amp, usually either the Exar XR4560 on older units or the Rohm BA4560 on later units, the Rohm version being especially repugnant. At best, these amplifiers are somewhat slow, and this adds excess distortion above 3KHz or thereabouts. The Rohm devices seem to be marginally unstable in addition to being too slow to pass audio cleanly, and this causes the distortion spectrum to be littered with many high order harmonics. The end result is an overly bright and hard sound, solely due to the 4560.

When I last worked on these mods, almost ten years ago, the best possible replacement chip was the LT1469CN8 from Linear Technology. It is a dual, low noise, bipolar input op amp that is remarkable in that it has extremely low distortion, excellent output drive, and a very high gain-bandwidth product. It is unity gain stable and can operate from high supply voltages, so it is a simple drop in replacement. The only possible drawbacks are cost and availability, which in my opinion are quite reasonable.

It is possible that a newer chip has been released that supercedes it, but the state of analog ICs today is to focus on amplifiers for low voltage, surface mount, single supply applications, and not DIP ICs that can be used with +-15V or higher supplies. If anyone has any suggestions, please forward them to me, and I can investigate them.

Some varieties of NE5532 will also perform very well, but not quite as well as the LT1469, which is essentially the long awaited upgrade to the 5532/5534, a chip that took the IC industry almost 30 years to significantly surpass. While the 5532 and 5534 are considered commodity items these days, in my tests and in this circuit, the JRC 5532D from New Japan Radio seems to perform consistently better than most 5532 flavors, and the Fairchild units were fairly close. The Philips 5532 varieties did not do so well distortion wise. The main advantage to the 5532 is price, so I would recommend that chip only if funds are tight and it's hard to find the LT1469. Yes, a 5532 is better than the stock 4560, but it will not produce the best possible performance.

The LT1364CN8 is also a reasonable amplifier, and if ultrasonic distortion is of prime importance, this might be the best choice. However, due to its high idling dissipation, it should be heatsinked, especially when driving low impedance loads at high levels. Its midband distortion performance is not quite as good as the LT1469, so it is not my first recommendation.

I have recommended some JFET amplifiers from Burr Brown (now part of Texas Instruments) in the past, e.g. the OPA2132 and the OPA2604. While these amplifiers do sound very good, they are not quite as clean at very high frequencies as the 1469 or even the 5532, and I cannot recommend them anymore if cleanliness is your goal. My opinion about these amplifiers changed after finding ways to clean up the SSM 2015 stage, which I will discuss in the next section. After that modification, it became clear that the dominant source of HF distortion was the Burr Brown chip, and the differences between the 1469 and 5532 became more noticeable too.

There are three 4560 chips per SX202 to replace. Install high quality machined pin, low profile sockets after you remove the old chips, as soldering the chips directly could damage them. Sockets also make it easier to modify the circuit later on if a better chip or hybrid becomes available.

When removing the old chips, clip the pins off near the epoxy chip body and discard the old op amp. The PC board is far more valuable than the old op amp - sacrifice the op amp to help minimize stress on the PC board. Once the bodies of the 4560s are clipped out, remove each pin of the old IC with a soldering iron and needle-nose pliers and the excess solder with a desoldering tool, solder sucker, solder wick, braid or what have you. Again, if you don't know what you're doing or you don't have the right equipment, have a technician do these modifications! The PC board is fairly simple to rework since it is only a two sided board, but it can be easily damaged by carelessness.