Back to english start-page
Amplification techniques:

Conventional semiconductor power amplifiers:
An input stage drives a driver stage which in turn drives the output transistors. The total amplification can reach 100 000 or more, but with a distortion of 50% or so.
To reduce the distortion a feedback is made output-input. If the feedback factor is 1000 both the amplification and distorsion are reduced the same amount. The amplification becomes 100 and the distorsion 0.05%, a rather typical figure for modern semiconductor amplifiers.
It seems easy, but there is a catch. The three stages will at some frequency (let´s assume 200 kHz) produce a total of 180° phase shift, making the amplifier oscillating at that frequency.
Now, the NORMAL approach to that is to limit the drive stage slew rate (speed at which the voltage can change, dV/dt) by adding a capacitor between collector and base of the transistor (or source and gate if it is a FET or MOS) so that the total amplification goes below the limit set by the feedback (in the case above, 100) before the phaseshift reaches 180°. The amplification to keep the oscillation running will not be enough. The amplifier will stabilize.
Now, everything is fine. Or is it? Well, there is still a catch. Limiting the driver stage slew rate will make it saturize when an input signal makes it reach its maximum slewrate. And as this happens 1000 times below the non-limited oscillation frequency, it happens at 2000 Hz, well within the audio spectrum. A transient (like a hit on a drum) can easily make the driver stage reach its maximum slewrate, silencing all other sounds as long as the transient lasts. After that, the other sounds appear again.
This is called transient intermodulation distortion (TIM). Your brain will fill in the empty gaps, so TIM is not conciously perceived. But it puts a strain on your brain, after a short while you get listening fatigue and just want to turn down or even turn off the music.

Tube amplifiers:
An input stage has an internal feedback, limiting the amplification to, lets say 10. There can be no 180° phase shift in a single stage, so this causes no oscillating. The feedback makes the distortion low, let´s say 3 %. Three such stages are put after one another, the last one with a higher power capability. The total amplification will be 1000, and the total distortion no more than 9%. Using a feedback factor of only ten will give an amplification och 100, and a distortion of less than 1%. There will still be instability, causing instability, this time also at 200 kHz.
Now, for some reason many tube amplifier designers make a smarter move. They don´t limit the slewrate, but instead give the input stage a frequency feedback, so that it has an amplification of less than one at 200 kHz, and a -3dB limit at 20 kHz. The result? Stability without TIM, as there is no stage that saturates because of dV/dt limiting. And a power frequency response all the way up to 200 kHz. Of course, the distortion factor will not be an amazingly low 0.05%, but more like 0.5-0,8% but so what? You can´t hear that anyways, and the benefits are so much greater. The sound is not limited in any way within the audio spectrum, and so you get an amplifier which has a very clean, transparent sound.
Unfortunately there are drawbacks of other kinds. Tube amplifiers work with high voltages/low currents, so both for power and safety reasons you need an output transformer (and input transformer). High power transformers with high linearity are extremely expensive, and still introduce both distortion and phaseshifting at low frequencies. That is one reason that many tube amplifiers, even very expensive ones, produce a softer bass than what really was recorded.

CM-130, the best of two worlds:
We wanted the transparent, real-to-life sound of the tube amplifier, so we simply (well, it wasn´t that simple in real life, but anyhow) used the tube amplifier design principle.
But we also refused to use output transformers, so we had to use semiconductors. This is how we made it:
An input stage is connected to a driver stage. There, between the driver stage and input stage, we put frequency feedback limiting the amplification in these two stages to 2.25 at about 320 kHz. The driver stage then feeds an output stage consisting of two amplifying stages, internally limited by linear feedback to an amplification of 21, and giving an extreme small signal bandwidth of several MHz.
The total amplification after feedback output-input is limited to 47. This design makes the whole amplifier totally stable, as the input stage due to frequency feedback (not dV/dt-limiting) dives below 0.4 well before the output stage reaches its -3dB limit. You get an extreme bandwidth and stability and freedom from any trace of TIM.
Someone might object to this, claiming that the internal bandwidth of the input stage (its -3dB limit) must be low to make amplification as low as 2.25 at 320 kHz. And that is true. BUT... it is a frequency feedback, not dV/dt limiting. So, the only drawback is that the distortion factor starts to raise at high frequencies, there is still no TIM. And as the output stage is a dual stage with internal feedback, the distorsion in each part of the amplifier is inherently low. Even at higher frequencies the THD is therefore very low, definately far below what can be preceived by the human ear.
We use high-current HEXMOS semiconductors as output devices, and ultra-fast (ft=100-300 MHz) bipolar semiconductors in other parts of the amplifier.
We made it possible to combine rock-stable, high definition bass response typical of high-current semiconductor amplifiers with the overall transparent, totally lovable midrange and treble which is typical for high-quality tube amplifiers.

Are we alone? Frankly, we don´t know. Looking at other manufacturer´s specifications it seems so. The problem with our approach is that it is not suited for mass production. The idle current setup alone takes several minutes because of the complex temperature curve of the output stage combining bipolar & HEXFET semiconductors. At a production plant where the PCB:s can be produced in seconds, it is impossible to wait several minutes just to do the idle current test. In fact, producing one single amplifier takes us about one full workday. That is why you have to stand in line to get a CM-130. It can´t be produced faster without loosing quality. And we will not let that happen.