Listening at the level that the mastering engineer used in monitoring his mastering effort is preferable, but almost no publisher ever discloses that level, so we listen at a level that sounds subjectively correct. We can gain some useful guidance from Dolby's current recommendations for setting reference levels for monitors in highest quality mastering, which is Lref = 85_dBA at -20_dBFS signal from both channels summed uncorrelated at the listening position, which is 85+10*log(0.5) dBA = 82_dBA from each separately.

Think of this is in terms of stressing your system, where you can consider this as being highest stress level, worst case stress on maximum crests. Less well mastered music suffering more dynamic compression will have lower crest levels and so will be less stressful to the system.

The manufacturer claims a nominal impedance of 8_Ohms and a voltage sensitivity of 85_dBA at 2.83_Vrms at 1_m, which is 1_W into 8_Ohm, and a maximum output of 108_dB. Sound propagates from a point source as -6_dB with doubling of distance, -10*log(X)_dB where X= propagation distance divided by reference distance, and reference distance in this case is 1_m.

108_dB - 102_dB = 6_dB

Propagation is -6_dB at 10^(6/20) = 2_m

So your speaker playing at 108_dB at 1_m will be 102_dB at 2_m. If you listen at 2_m (or closer), the loudspeaker should be sufficiently loud on maximum crests. Most recorded music suffers much more compression, and you will be listening at well below Lref. If you listen at -3_dB below that Lref, then that distance extends to 3_m, 10_ft.

As for power needed to get 108_dB at 1_m, as noted above the speaker produces 85_dB at 1_W, so...

10^((108_W-85_W)/10) = 200_W @ 0_dBFS

200_W*10^((-20_dB)/10) =
= 2_W @ -20_dBFS @ Lref @ 2_m

Not included in the above info, Dolby recommends linear headroom of +3_dB above 0_dBFS for the loudspeakers, and +6_dB above 0_dBFS for the power amplifiers. That headroom is for very brief crests in the top octave. In theory, at a frequency of half of the sample rate, a peak centered between 0_dBFS samples is +4_dB above 0_dBFS, but with practicable filters +3_dB is a more realistic limit. An amplifier that is clean below clipping can be characterized as clean at half power below significant clipping indicated by 5%_THD or 10%_THD, with no linearity rapidly increasing at outputs above that. T
I think that 5%_THD or 10%_THD would well correlate to the +6_dB amplifier headroom, to be clean at +3_dB and below, with signal never exceeding +3_dB, but manufacturers of consumer grade amplifiers rarely provide that information.

So for your interests, to avoid nonlinearities associated with clipping, you want an amplifier capable of clean crests at +3_dB, which is doubled power above 0_dBFS, 400_W into 8_Ohm. Most manufacturers do not provide ratings for power output for brief crests.

I suggest shopping for two nice used properly working NAD 2200 stereo amplifiers, and operating those as bridged mono blocks. Since these are stereo amplifiers, you can start with using just one, and later add the second one. Using two as monoblocks gives better stereo separation and provides more headroom. You should be able to find these for several hundred Dollars each, leaving more than half of your stated budget for a nice used preamplifier. If you want to add some coloration with tubes, do that in the preamp, or maybe do that in a bypassable line level device (unit gain tube buffer).

Here is a link to a test review of the NAD 2200:
http://www.hifi-classic.net/review/nad-2200-546.html

Fekser wrote:

Hi JRT. First of all thank you for your reply and answer. My first language is not english so maybe I misunderstand your recommodations.

Are you suggestibg the NAD2200 mostly for its ability to deliver power? In this case the Roksan delivers 150W in 8 ohms. Again I apologize about my lack of knowledge regarding the amplification engineering.

My suggestion is to get two NAD 2200, and configure those as mono blocks.

In your application, noise is low enough that you will not hear any noise from the amplifiers through your speakers under any circumstances in your listening room.

The amplifiers exhibit low distortion below clipping levels. In your application, the amplifiers will not clip. So you will not hear any distortion in your application.

In your application, these amplifiers would deliver clean gain, not adding anything audible to the sound.

One more suggestion: Avoid the expensive unusual loudspeaker cables that some people seem to like to play with. In some cases those can introduce excessive capacitance to the load impedance which can create problems. You can read more about this at the following link.
https://sound-au.com/cable-z.htm

Fekser wrote:

In this case the Roksan delivers 150W in 8 ohms.

Other than reading Roksan's specification claims, I am not familiar with the Roksan K3. The manufacturer claims 140_W into 8_Ohms, 220_W into 4_Ohms. If that is not exaggerated, power output would be adequate on most recorded material. Not sure how clean that is in crests at higher frequencies.

The noise specification on the NAD 2200 is -113_dBA below signal at 100_W output, -93_dBA at 1.0_W output. The Roksan K3 is rated at -100_dBA below signal at 140_W output, with no mention of noise on small signal level, so without more information that could be -80_dBA below signal at 1.4_W output, -77_dBA at 1.0_W. While that is not very good noise performance, it is better than most phono preamps. You can be sure that the NAD 2200 is much more quiet than the Roksan K3, but the Roksan might be quiet enough. You can test this with your ears in a quiet room with the amplifier turned on, but with input muted (no input signal), place your ear near the tweeter and listen for noise. If you cannot hear any noise, than the Roksan is quiet enough.

The Roksan power ratings do not mention frequency spectrum, so might not perform well at low frequencies (stresses power supply) or at high frequencies (amp might not be fast enough to perform well at high frequencies). The NAD is clean across full spectrum, 20_Hz to 20_kHz.

The Roksan K3 might be adequate. Improvement will be small, but some improvement can be had.

The NAD 2200 is a better amplifier, is much quieter, exhibits low distortion over full bandwidth, can swing voltages unclipped across the load on all maximum signal crests in your application.

Looking at eBay, pricing on a used Roksan K3 is approximately double the used NAD 2200, so you can buy two NAD 2200 and sell the Roksan K3 to pay for those. But then you will need a preamplifier. I suspect that the NAD 2200 will hold its value better over longer time, because it is a much better amplifier. The Roksan is nothing special and will probably fade into obscurity in the used market.