The truth on low efficiency speakers like B&W?

I can't really comment on 802s. I have 803d and 804s which I have driven with Parasound JC1s and Halo A21 amps.

Its not so much the total power, but the stability of the amp driving 3.5 ohm loads. Both the 803 and 804 (and I assume the 802) have a minimum impedance at low frequencies of 3.5 ohms (latest versions are 3 ohms). The A21 is rated for a minimum of 4 ohm loads, as is probably the 2250. At 3.5 you are stretching it. The JC1, on the other hand, is rated to drive 2 ohm loads.

To my ears, the A21 did not seem as open and relaxed as the JC1 driving the 803d. And I am not talking high SPLs where a more powerful amp would be needed. I am talking 85-90 dB where you are still under 50 Watts peak. While the efficiency of the B&W is not as good as a horn loaded driver, its not that bad either. I think mine are 88 dB at 1W, 1 meter. Newest versions are 90 dB, 1W at 1 meter.

You only need 600 W if you listen to really high SPLs. It has nothing to do with the size of your...

Got an idea what Elizabeth calls small but "normal" is up for grabs, so to speak.

No matter how powerful your amp is you'll never get bass from B&W 802. It's a unique speaker that doesn't benefit much from quality or quantity of amplification and it's a biggest looser in audio industry IMHO(pardon 801's bigger).

Well you have to consider the crest factor of high quality recording.

Excellent orchestral recording have peak to average ratios that can exceed 20:1, which means that the average levels are 20 times lower power than the peaks. If you like to listen at an average of 85dB at your listening position, which might be 4 meters from the speakers, well the power requirements add up pretty quick.

Let's say you have speakers that run 85dB/W. You're 4m from them, so the level drops to 73dB (6dB for each time you double the distance, 1m - 2m - 4m). So, to reach your 85dB, you now need 4W instead of 1W. Now, the crest factor of 20:1 needs to be factored in, which puts the power required on peaks to be at least 80W. If you want your average level to be 88dB instead of 85, then you need 160W peaks. Any amp that puts out less than that will be clipping.

For clean reproduction from speakers that run low to mid 80s and normal listening distances, you should always pick an amp with higher power reserve. 250W would be my minimum choice with speakers like that.

Speakers that run near 90dB and above can get by with amps of quite a bit lesser rating.

09-11-12: Elizabeth
Yes heavy bass played loud taxes amplifier power.
The "45 watt tube amps" are being used to play normal music at normal volumes. period.
The 600 watt amps are for dudes with small dicks...
As compensation. (i guess)
The Parsound seems to be in the middle, or 'reasonable'

Elizabeth, like I told you on our date at the beach, it was just "shrinkage"!

But seriously, I've measured 105db peaks at my listening seat on acoustic jazz played at realistic levels. I've also measured about a 5-6db loss between my listening seat and 1 meter from the speaker. Assuming a 4 ohm, 86db/2.83v/1-meter speaker is 2w at 86db, then 110db output at 1 meter would take 512w for a momentary peak. Some rooms aren't as lossy as mine and don't suffer much loudness difference at all between one meter and the listening seat, but some will suffer more. 105db peaks are, admittedly, very loud, and a lot of speakers can't comfortably get that loud anyway. Nonetheless, in a large room with a 4 ohm, 86db sensitivity speaker, 600w/ch is not entirely compensating for something. Of course, for Maggie owners this is an entirely academic conversation, because even a 20.1 doesn't sound very good at live listening levels. Or if your listening levels peak at, say, 97db at one meter, then a 45w amp will indeed work just fine. But a trumpet is louder on peaks than 97db at one meter, no less a rim shot on a snare drum.

Most of time I'm listening at 85db or less (mostly less) volume levels at my listening seat, and the highest peaks are in the 95-100db range, depending on the material. 45w/ch would be quite marginal, even under normal circumstances.