I've considered the total DIY route as well...... but AFAIK there aren't any proven designs that do everything that I want, and designing it is way beyond me.
As far as the 4552 goes it's recommended at 1k because it looks like there's a distortion spike on the factory graphs at about 8-900. We will see how it measures. I can tell you that the eq needed for my DIY Paralines will be atrocious regardless of the driver used 
What are you using for midrange and bass?
My lower end is handled by four peerless nomex cone 8-inch woofers in a two by two array just as close together as we could get them
http://www.parts-express.com/pe/showdetl.cfm?partnumber=264-1098. The goal of the array was to create a polar pattern that resembles the horn's polar pattern in the crossover region, and to gain some driver efficiency (they're wired in series and parallel, so they present electrically as a single 8-ohm driver that's about 6dB higher efficiency than the stock woofer).
A lot of people use a 15-inch driver with 511s, but when we were doing design we couldn't find an affordable 15" that had attractive-looking frequency response up to 800Hz. The cabinets have a port tuning of 28 Hz, and are lined with acoustic tile and heavily stuffed (I know that's unusual in vented designs, see this Rod Elliot article to see what we were thinking when we did it
http://sound.westhost.com/articles/boxstuff.htm). Bottom line (with appropriate baffle-step dialed in) I get more or less flat bass response to 28 Hz, and I'm only three or four dB down at 20 Hz . I've got infra-sonics high-passed out for driver protection.
I use a frankensub for home theater (a cheap commercial sub with the driver replaced), but the peerlesses put out low enough and loud enough for my 2 channel audio listening (and they sound much better than my sub!).
In the amp news front, I demoed another low-priced headphone amp, and was so impressed I immediately picked one up to drive the horns as a stop gap solution until I can save enough for something more serious: NwAvGuy's Objective 2 amplifier (O2) (lots and lots of measurements here http://nwavguy.blogspot.com/2011/07/o2-headphone-amp.html). It's critically limited in one key way, but so far it sounds better than almost everything else I've heard (short of the First Watt J-2).
The good: It's a DIY with completely open source design, and companies are building and selling finished O2s for about $150. The specs are astonishingly good (THD+N in the low thousandths of a percent at 10 milliwatts) , the gain is low (stock gain is adjustable between 2.5x and 6x), and it's smaller than a paperback book. It's also completely and totally silent into the horns with the inputs muted, and (on the low gain setting) is completely silent even with silence playing. But there's a big "but"
The Bad: Its maximum rated power into a 15 Ohm load is around 350 mW (but THD+N is still in the low-thousandths of a percent until 300 mW).
The worse news: I think (but am not certain) that that 350 mW rating is not per channel but total power output (because that's the convention in headphone amp ratings). So realistically, the amp will put out 150 milliwatts a channel of low noise output. [EDIT: additional reading suggests that his 350 mW figure was per channel not aggregate] He actively current limited the design because it's a headphone amp, and he wanted to offer some protection for folks with very low impedance headphones. So while it will cheerfully put out 600 milliwatts into 32 Ohms, it's down to 350 by the time you get to 15 Ohms.
This isn't that big of a loss of power for me: I'm was using a 1-watt headphone amp. Going to a 350 mW headphone amp is a loss of 4.5 dB of overhead in my specific application, and, for me, the better output impedance, better noise specs, and generally better sound are more than worth the loss of overhead and theoretical maximum volume. My maximum volume in JRiver is set to the point where -20 dB pink noise gives me 83 dB at one meter. My estimated power consumption at 83 dB at 1 meter is one milliwatt, so a hundred milliwatts should give me the necessary 20 dB of overhead to play any signal that JRiver would be outputting at my maximum volume setting, and 300 milliwatts gives me 4.5 dB of overhead above that.
I cheerfully admit that I'm cutting things a little fine here (and probably getting some headshakes/eyerolls from the eminently sensible people reading this thread), but the amp sounds amazing, and so far I haven't managed to drive it into clipping (that I could hear). Bottom line, to quote someone upthread, if my ears are the test, this little amp is passing with flying colors. It'll certainly do until I can save up a few thousand for something more serious. I recognize this may not be a good solution for anyone but me, but I thought I should mention it as it's affordable and very low noise (albeit also very low power).
I'll report back later on once I've managed to do some "torture testing" to see what the practical limits are with my set up.
Author
Topic: Low Wattage, Low Noise Amplifiers (Read 34824 times)