Hi Peng, the info I was looking at was output impedance graphs (which is inversely proportional to the DF) for the 3805, 3806, 3808....and all these are weaker than the larger 4306 by a factor of 10. BTW, my observation I was discussing was just slight. Hence the uncertainty. We would also need corresponding impedance graphs for our speakers to really draw any [theoretical] conclusions. Ironically, the link I was just given discusses the importance of system coordinated impedance as well, but for a headphone application. I had used the 3805 for years too. If it wasn't for that darn HDMI connection, I would still be using it.
I hear what you are saying. I went from 3805+/11amp to 4308+/-amps to AV7005+amp(s) and like fuzz said, without doing a proper AB, I, and none at home can tell any improvements in SQ but then we typically have the volume at around -20 to -28. On rare occasions I would crank it up to -10 to -15. So 99% of the time, the amps basically just sit there wait for those intermittent 15 to 20 dB peaks and then stretch just a little.
Fuzz seems to enjoy reference level so during some peaks typically found in movies I can see why in his application a Cinepro or big Outlaw would be put to use, and as they say YMMV too..
Back to the DF thing, if you read one of the links provided by fuzz you will see that it is only inversely proportional to the output impedance of an amp if the other signal path including the speaker's impedances are ignored. So in real world it is hard to prove with facts that even a DF of 20 to 30 would sound noticeably different than say 300 and above. I prefer to stay well clear of 50 but that's just for peace of mind, same reason why I have more than one amps in my system. You really don't need the speaker's impedance graph to draw conclusions if you are comparing DF >30 unless your speaker has dips below 2 ohms, just go over the math yourself (fuzz;s link) you would likely agree with me/us.
Lastly, don't assume the 3800 series Denon are that different from the 4000 series. I have read pretty much every available googleable free lab measurements for the 3808, 4310, 4311, 4308, 4810 and found no evidence whatsoever that their amp sections are all that different. In fact HTM did both 3808 and 4308 and the 3808 yielded better measurements in the power output department even though it was rated slighly lower. Base on those numbers I believe it is likely (just an educated guess) they use the same amp section for the sake of standardization and cost effectiveness.
HTM measurements on the 3805:
'Into 4 ohms, the amplifier reaches 0.1% distortion at 218.4 watts and 1% distortion at 243.0 watts.'
That's quite similar to that for the 430X's.
Now if you are talking about Dennon's lower models such as the 2000 series then all bets are off and you may get to see lower DF dips down to the 20 (again it may still be not noticable) range for 4 ohm rated speakers. Even then I am just taking a wild guess and I doubt too many people would use the 2000 and below series for 4 ohm speakers except for those minute HTIAB types.
Edit: let me add one more link:
http://www.trueaudio.com/post_013.htm
Seems extreme but I guess he wanted to make a point..