In the world of audio, one kind of format is often juxtaposed to another and they are treated as two opposing parties. MC cartridge against MM cartridge, tube amps against solid-states, horn speakers against cones etc. etc.
     I myself do not care whatever format or medium it is as long as they deliver the music the way I want. There are many good tube amps and bad ones, as are solid states. The good ones remain in people's memory long after the end of their production. Some of them are even reproduced later on, like Marantz 7 and 8B. There is a natural selection of history going on here.

What is an Op-amp?

     I would like to start this article with a brief explanation of op-amps and its take in general, because, unlike transistor amps, there are not many people yet who consider op-amps as a serious audio device today.
     In the case of tubes, there is a recognition of the character of each different kind of tube, as in KT-88 or 300B, and we can start our topics based on that understanding. In contrast, I never heard of anybody mentioning any famous power transistor in their discussion. If tubes are compared to the games of individual players, transistors' must be that of a team play.
     You won't find a clear definition of op-amps in the guide books, but their characteristics are clearly stated in them with standard circuit applications as in the case of tubes. This will make it possible for DIYers to challenge designing their own amplifiers using op-amps as its gain stage.
     The whole variety of op-amps are available in today's market, from headphone use to portable radio use, and of course, there are some for hi-fi applications with very good sound quality. Using those high quality op-amps, you can create your own amplifier that rivals well designed tube amplifiers.
     Today, op-amps are widely used in boom-boxes and game machines and this situation seems to make audio designers distance themselves from a serious application of op-amps in hi-fi components they design. This is another unfortunate prejudice toward a medium and its use.
     The amplifier I designed with an op-amp as its only gain stage (Model 4706 GAINCARD) was most welcomed among tube amplifier fans. It is all up to its application and, for all I can see, op-amps have a great potential as an amplifying stage no less than that of tubes.

Beginning of an affair

     When I first started designing a CD transport, I was hesitant to use a tube amp as its reference amplifier. I wanted to bring the sound of CD up to challenge that of LP, which, in my mind, was always associated with tube amplifiers. But to design a new solid-state from ground up requires totally dedicated attention and will divert my energy from designing the transport. Op-amps came in handy for my purpose at that moment.
     By then, after 30 years of experience in audio, I was determined to solely depend on my own set of ears when it comes to finally decide on the sound of the equipment I was designing.
     My first prototype contained an op-amp and a little more than ten parts(resistors and capacitors) attached on a circuit board about the size of two post cards and, honestly speaking, it didn't look promising at all. When the sound came out the first time, I was surprised by its dynamic presentation of music and my opinion toward op-amps were inherently changed at that moment.
     That was also a beginning of new struggles.
     The guide book suggests to insert a film capacitor of 0.1µ F right after the power supply to prevent an oscillation. It also suggests to connect the input to the ground using a 100pF capacitor. I followed those suggestions at the beginning and was somewhat content with the result for a while.
     Then, as usual, you get more greedy. If the standard application gives you this much result, you can improve it further ahead by tweaking with your own creativity. Right?
     My first idea was to give more uniform conditions for each left and right channel to get a better stereo performance. You can't handle this overwhelmingly tiny component with the same sense of size as you do with tube amplifiers. It actually felt as if 1cm of the op-amp circuitry was equivalent of 10cm of that of the tube amplifier, a small difference of signal pass length making a great effect on the outcome. It was quite an eye opening experience. When I set the conditions of each channel in the order of millimeters, the whole stereo performance improved dramatically.

Oscillation and the sound

     Next, I took off the capacitor which the book suggests to install parallel to prevent the oscillation. I have used the same method in the past to cut off the oscillation related to NFB when I was designing a tube amplifier, and this bothered me quite a bit. If you have to insert such, that means the amplifier itself is constantly under a danger of the oscillation.
     In case of op-amps, the oscillation happens at really high frequencies, like in 100MHz or 200MHz, and you can't check them with normal oscilloscopes which are usually capable of up to 20MHz range.
     Then I realized something else. The temperature!
     To solve this oscillation problem, I hit upon an idea to solder the resistor for NFB directly to the feet of the op-amp instead of on the circuit board. In my case, the op-amp has eight feet and the NF resistor comes between the foot No.3 and No.8. This is really the shortest length imaginable.
     On the circuit board, there are patterns of 5mm each, plus the length of feet themselves and the length of the feet of the resistor. These add up to a big difference in the scale of op-amps. The effect of this approach appeared in the difference of the operating temperature of the amplifier.
     Before this treatment, right after you connect the DAC and power it on, you could feel the temperature rising without playing any music. But once you directly connect the NF resistor and eliminate the condenser, this temperature rise was immediately gone.
     This lead me to a conclusion that the oscillation of the op-amp, which escapes a normal testbench measurement, and the noise leak from the DAC, also in high frequencies, are forcing unnecessary extra work to the amplifier.
     Unfortunately, no automatic insertion machine can perform in such a small scale. You need some experienced hands to do it properly (in my case I need to use a jewelers loupe). The guide book's suggestion, allowing certain amount of oscillation and eliminating it by a condenser, may be based on this observation, but it also allows the demerit of NFB obviously present. It may sound good at the beginning but you'll notice listening fatigue setting in after a while which is a typical side effect of NFB.

Op-amp and NFB

     For tube amplifiers, we use NFB to improve the overall characteristics, but with solid-states, we need NFB to control its gain.
     There is a strong belief among audiophiles that NFB is harmful to the sound and they hang their hopes on variations of non-NFB circuitry. I see potentials in that approach too, but through my experience with op-amps which require an attention in order of millimeter scale, I also realized a possibility to solve the problems associated with NFB by shortening the length of its loop.
     The solid-state devices' high gain is a necessary condition to ensure the wide frequency response and you need NFB to control that gain. Engineers' concern is how to apply a large amount of NFB while taming its side effects.
     One of a very interesting aspect of op-amps lie in this relationship between the gain and NFB. The op-amp I'm using has 90dB gain while I only need somewhere around 30, so I need more than 50dB of NFB to cut that down. Like some companies, the problem is their having too much profit! In this respect, op-amps are a quite ironical component as a gain stage in the audio design. However, it also presents for us a great advantage in terms of its physical dimension to make everything compact.
     I remember seeing an ad of an amplifier featuring its short NFB loop length, claiming that it's worthy of the Guinness book record. They were talking about 30 mm length and it wasn't even the major loop. GAINCARD's loop length is 9 mm including the length of the resistor and I wonder what they are going to say about this one!
     Solid-state amplifiers which have to depend on a large amount of NFB are always facing the danger of oscillation, and in the worst case, hauling. It is quite a symbolical fact that the temperature rises when the amplifier is operating near its limit toward the oscillation, beside its strained and uncomfortable sound.

Tubes vs. op-amps

     Tubes and solid-states are in contrast in almost every respect except one aspect. In either case, NFB plays an important key role.
     I find the recent trend of non-NFB circuit somewhat misleading. Unlike non-NFB of a tube circuitry, which releases the circuit from the use of NFB by pursuing its original characteristics, non-NFB of a solid-state circuitry requires a clearer definition. In some cases, manufacturers claim it's non-NFB just because it doesn't have NFB on its final gain stage. I find it inappropriate and misleading
     Op-amps have a possibility to solve the problem of NFB associated with the delay or smear in the time domain by minimizing the loop length. It also opens up a wide window for many adventurous DIYers to experiment with their own unique ideas. There is still a prejudice about their sound quality, especially in relationship with the use of NFB, but it's not anything fatal and you can experiment with your own ideas to steer around it. In addition to that, manufacturers are actively developing new models, giving us high hopes toward the future of its applications. DIYers who's projects were almost strictly limited to tube amps can find a new play ground with op-amps.
     You will also find its sound quality similar to that of very well designed tube amplifier. I myself was surprised by this and deepened my belief against those prejudices about different types and formats of audio design.
     The output transformer usually incorporated in the tube amplifier design also works as a bandpass filter, automatically limiting the necessary audio frequency. In contrast, op-amps are DC amplifiers, and the gain and the frequency response are left untouched, making it more difficult to control. It feels as if you are forced to struggle directly with nature, instead of working in a limited space of a garden. If you could find this worthy and interesting, op-amps can be more fun than tubes.