What's better 100w class a or 200w class ab?

Interesting point Audiomax. Too bad more companies don't print max current. Creek would advertise peak current of over 30 amps I think but never in their specs. I can subjectively agree with Drrdiamond about Krell as their integrated is punchy.
Musical Fidelity now publishes this info:

A3.2 integrated 115 watts: peak to peak output current 24 amps.
A3.2 power amp 130 watts:72 amps
A308 integrated 150 watts: 48 amps
A308 power amp 250 watts: 72 amps
From listening, the higher current gives a crisper sound. Also there is more detail possibly because the high current fully magnifies every small detail.
I can also subjectively say going from N804 to N802 and ATC active 20 to ATC active 100 the bigger speaker presented more detail. Again, I think because of the higher power and ease of presentation that a larger speaker can do.
This is all a little off track but hopefully interesting.

Sean, what is a high rail voltage? What are the typical ranges? My amp runs at 60 volts and I'm wondering if that's normal, low or high compared to others. I love the sound, so I'm assuming it's high ;-)

Great posts already. Much depends on how you listen to music -- e.g., quiet or loud. Much depends on the efficiency of your speakers. An AB design done right can give the best of both worlds. A pure A design can sound fantastic in a great design with high efficiency speakers.

tank's guys for the input ,i think i will have to listen to both of them in my system before i make a choice
enjoy the music

Ummm that 72 amp "peak current" thing is a bit, shall we say, "optimistic."

Let's take a look at it. In order to get 72 amps drawn by a load it has to fit into the P = I^2 R formula. So, 250w = 72^2 * R - where 250 watts is the max output of the amp, that gives us a load resistance of 0.048ohms. That's about as close as you can get to a dead short... Ok, ok, so you say that the amp can really put out more than 250 watts? Let's look at what the rail voltage is likely to be. It's rated at that power at 8 ohms, so P = E^2 / R... 250 = E^2/8 or E^2 = 2000 or 44.72volts... that's the RMS value of one rail's voltage. The DC rail voltage is about 1.414 times that or 63vdc. Assuming few losses in the output stage the loaded rail voltage at 8 ohms must be at least that, so let's set it a bit higher to take losses into account... about 70vdc. Now we know something of use. *IF* said amp *could* put out 72 amps for *any length of time* at all, then the max wattage required would be 72amps x 70vdc =
5040watts. Ask yourself how many amps have power supplies anywhere near this size. Even if there is a fairly large power supply bank, this sort of current can only be sustained for an incredibly short time by an amp that does not have a huge power supply. I doubt strongly that the MF line has that much supply in it.

BTW, IF it could handle a 2 ohm load, it would have to chunk out 968 watts, and if a 1 ohm load was there, only a mere 1936 watts. :- )

My Symphony No.1 Amp *does* have a huge power supply. It uses 4kva worth of transformers in the original configuration, along with 500,000ufd of filter caps AND 140 amp Hexfred rectifiers. The amp's power section is fed via 10ga SPC wire. The output stage is modest, using only 6 Hitachi metal case Mosfets per RAIL (12 per channel) each of which is rated at 7 amps continuous = 42 amperes per rail, or 100 watts each = 600 watts max continuous. Hitachi rates them at LESS than that for a 10ms 1 shot with rail voltages above 50 volts! BJTs have similar deratings. This amp is a 180 watt per channel rated unit with 60vdc rails.

This amp weighs 128 lbs., none of which is "filler," it's made from aluminum, not steel. Most of it is power supply components.

You can get more power from a smaller, lighter switching supply, but then you are depending upon the *line* stiffness to control the supply and provide the required current on peaks - there are power factor issues there. Halcro has to deal with that sort of thing...

Anyhow... "peak" anything is a very questionable specification, FYI.

I've always preferred the approach of
1) listen
2) try to use the theory to explain the differences

rather than the other way around. If you do it in the other order then it's too easy to overlook something that might render your theory inapplicable in this particular situation (unless you're really a very good audio engineer, with detailed knowledge of the makeup of EVERY component).

Plus at the end of the day you have to listen to the thing !

Thanks all for the interesting thread, though. Since I studied basic transistor amp design in college and built a few it's always nice to read some more discussions on the subject.