Its a common theme on the audio forums and you see this perception of "euphonics" very frequently, and often from people who have been in audio for a long time.
"Use a tube buffer to soften the sound" or "use a XXXX type of tube for a lovely lush tone".
This type of perception is very misguided and completely wrong.
Let me explain why.
Firstly the term buffer can be applied to either tubes or solid state. The term is correctly used as a non-amplifying device (no gain) which couples two devices so they can work optimally together. If you have an amplifier with a lowish input impedance and a source with highish output impedance , combined together they will probably have a degraded frequency response.
You need a "buffer" to couple these devices together, something that will present an agreeable load to the source and the amplifier.
The most common devices are opamps, usually used on the outputs of CDP, DACs, etc.
The problem with opamps are they are 100% feedback devices, they do a very good job technically , but they sure leave a lot to be desired in respect to sound quality.
Good designers often revert to discrete designs that require much less levels of feedback.
(There are design methods and types of opamps that sound good, but rare)
Or some people say "use a tube buffer- much better than opamps".
A tube buffer that has no gain is usually a design known as a cathode follower, it has just less than a gain of 1 (unity gain) and has high input impedance and low output impedance which is ideal, but like an an opamp it has 100% degenerative feedback.
It is possible to make a tube buffer sound ok, but all the tube buffers tacked on to the back of dacs , preamps etc are no better than the opamps they are supposed to improve on, and in many cases are inferior.
Its easy to tack a tube on the back of a solid state design, but many solid state designers do not understand tube design, and follow a tired formula to get some tube "credibility".
It is quite common to see many tube buffers used at far lower voltages than they need for optimum operating conditions, and the tube is struggling to do its job.
The cathode follower design is commonly used because it allows a tube that can not normally be used for buffering to have a low enough output impedance to do the job.
But remember, like an opamp it is a 100% feedback device.
What determines how well a tube can drive a following stage is its Plate Resistance (Pr)
If a tube has a high plate resistance, and used in a zero feedback design (nice) it will have a high output resistance (Z).
Tubes like the 12AX7, 6SL7, 12AT7, etc have Pr in the range 50-100Kohm and also have very high gain.
They cant be used as buffers as they have too much gain, and no drive ability. They are high gain devices very much limited to specific use.
These devices when used as amplifying/buffering devices really do have the "warm, lush and romantic" sound of tubes, because they are not operating as they were designed for.
Tubes with medium Pr like the 6SN7, 12AU7, 12BH7, etc have Pr around 5K ohms, and are better at driving loads. You could use one of these tubes to drive a reasonable load of 100Kohms, or higher, but not much lower, they struggle with power amps of lowish input Z . Typically they have medium gain, usually too much to use as non-amplifying buffers, and they are not ideal as buffers anyway.
Depending on the situation the sound can have varying degrees of "euphonics".
An ideal input/output ratio is a minimum of 10 . A preamp with output Z of 10K will be just ok with a power amp's input Z of 100K, although the lower the better.
The higher the ratio the better the performance, and generally speaking, the accuracy will increase and tube "euphonics" will be reduced.
Tubes such as the 12B4A, 6H30, D3a, etc have low Pr around 1-2Kohms and are ideal as interfaces between devices. They have extremely low distortion, provide accurate driving conditions and are ideal used as buffers, although depending on the tube and operating points there will be gain.
Some of the DHT tubes are also ideal in this situation, although a bit more finicky to use.
Some really good tubes were designed at the end of the tube era, high gain and low Pr was a difficult, almost impossible task in the early 19th century but by the 60/70's some excellent tubes were designed- the D3a, E810F, E280F and many others had very high gain, and also low Pr, really the perfect amplifying device.
It should be mentioned that the operating points of all tubes is critical to performance, the level of harmonic distortion plays a major part in any tubes sound, for good and bad, and the designer needs to adjust for the performance required.
Many "lush" circuits are the direct result of improper operating points, sometimes done deliberately, eg guitar amps, SE amplifiers.
It should also be noted that feedback is not necessarily a "bad" thing, feedback that is used to fix design or component faults is always counter-productive, but used sparingly and wisely , it can be a powerful and good sounding tool.
Feedback can be especially useful with high and medium Pr tubes as it lowers the Pr and output Z of these tubes.
Naturally I like tubes in my preamps and as the devices on the output stages of my dacs, I think for dac outputs the low Pr tubes are very much ideal and have very much better sound than any opamp or other discrete device (although I do have a fondness for j-fets).
There are no "euphonics" with the low Pr tubes if used properly- they have very low levels of distortion, are accurate, clean and powerful sounding , very much broadcasting the fact that they are completely zero-feedback, which leads to very good phase performance, and that can be heard in excellent sound-staging and imaging.
But how to get the unity gain that is required in this situation without using a 100% feedback "tube buffer" (also referred to as a goats turd) when I typically use a low Pr tube with a gain of up to 50?
Fortunately the dac often needs some amplification, but the best option is to use a step-down transformer which brings the gain down to useable levels, and lowers the output Z even further.
It is imperative to use a high-bandwidth device that is matched to the tube used, and these devices are expensive. I use Magnequest, Lundahl and Jensen transformers, all imported.
There are different methods of loading the plate resistance of the tube, and many different methods of biasing the tubes.
In fact there are so many variables to tube design, it does take many years of experience to sort the best designs from the less desirable.
But the advantages are very worthwhile and it is very hard to listen to an op-amp buffered dac after living with a good tube output stage that is not a 100% feedback tube buffer.
"Euphonics" has its place , but preferably in a glass of wine, or your choice of ecstacy.
Of course there are many who actually prefer the sound of euphonic systems, actively chasing "warm and lush" sounds.
In todays hi-resolution world, that is perhaps illusionary , but each to his own, and I can quite readily enjoy such a system for a limited time
Here is one of my preamps, used as an example of tube selection and the reasons.
The bigger tubes at the back are 71A Directly Heated Triodes. These are actually very low power tubes, used back in the 30's in radios, but with a low Pr and low gain they are ideal for driving modern power amps. DHT's are finicky to use in terms of noise and minimal components, but there are ways around this. They are used with output transformers so gain is very low, to counter this I use a pair of 6SN7 medium Pr tubes (on the outside) to provide gain. They see an easy load to the 71A's and need a simple non-feedback circuit to drive the 71A's.
The inside tubes are E280F's which are used to provide a separate tube gain stage for one of my dacs. The output from the dac chip is fed straight into the E280F and the output can be switched into the line stage. The E280F is chosen as it is a high gain/low Pr tube. It is transformer coupled and makes an excellent compliment to the digital-to-analouge conversion.
