In a recent post we covered the fundamental differences between tubes and transistors and one of those we’re interested in is linearity.
Tubes and transistors are only partially linear devices – which means they will not always faithfully reproduce a larger version of the input signal on their outputs.
In our example yesterday we placed a phono cartridge on the base (input) of a transistor and connected a loudspeaker to the output of the transistor. This works as an amplifying device but at the quietest and the loudest extremes of the phono cartridge’s output the transistor (as well as the tube) don’t produce an exact but larger replica of the input. Only signals that fall somewhat in the middle loudness region are handled with perfection. Now, granted this is a major over simplification for those of you amongst our readers that already understand this, so please bear with me. The thrust of this argument is what’s important to our understanding.
Take a look at the following picture which shows a “typical” linearity curve. What this is showing is signal out for a constant rising signal in. Starting left to right, we are seeing an increased output voltage – with the quietest signal on the left and the loudest signal on the right. If I included a picture of the input signal you’d note that would be a simple straight line going from the lower left corner to the upper right corner. A perfect output curve would be the same – whatever I put in, I get out – just larger.
You can see that only a portion of the middle signal is linear – output perfectly matches input. Now, don’t panic because as circuit designers we have many ways to “linearize” this response curve which include different biasing schemes, feedback methods etc.
What’s important to this discussion is that from a high-end perspective we would like a perfect device because the fewer tricks and schemes we employ to achieve a linear response from one extreme to another the better, cleaner and more open our musical presentation will be to a given piece of equipment.
There are no perfect devices – but remember in yesterday’s post we mentioned that tubes run at much higher voltages than most transistors in a typical amplifying circuit? Think about the curve in the graph on this post which is showing the output voltage of a device. It should be obvious that the greater the voltage the greater the linear region.
Let’s imagine that the graph represents 10 volts and the linear region is about 50%. That means that even on a good day you get a 5 volt linear region to play with – not a lot from a designer’s standpoint. Now imagine that the same graph represented 100 volts which is a 10X increase – which gives us 50 volts of linear region! This is huge and more than we would ever need as designers.
From a sonic standpoint we hear image and apparent micro and macro compression as we get near the edges of our linear region and the ear picks these cues up immediately and recognizes something has changed. This is one of the primary reasons why most tube circuits sound so open, effortless and compression free at the quietest and the loudest extremes of the musical presentation. Think of this as a car analogy – if you have an oversized V8 engine in a small car, chances are good you won’t notice any difference in performance from the slowest speeds to the fastest speeds. Change the engine out for a tiny 4-banger and your acceleration and top end speeds are compromised as the tiny engine struggles to do its best.
So while neither tubes nor transistors are true linear devices, the one with the highest voltage wins when it comes to an open and effortless soundstage with perfect micro and macro dynamics.
Notice that I didn’t pick out tubes as the winner in this closing statement – just that used without a lot of knowledge to these effects, a pedestrian solid state audio designer (vs. a pedestrian tube audio designer) will always lose the sonic battle – hence, most tube designs sound more open and effortless than most solid state designs.
However, once armed with this knowledge we can have the freedom to express ourselves with either discipline – so as solid state designers we can choose our working voltages to take advantage of this knowledge. For example, in most every PS Audio amplification device we have always run two to three times the typical voltage used by other designers for this very reason. Others have done the same – but it is NOT common practice.
Paul McGowan / PS AUDIO