I posted the technique some time back, it would be good if someone dug that up. Basically there is a trick that you can do with some measurement software that is called sychronous averaging. You need to generate a "near" 1 kHz tone that exactly fits a 1024 window. Then you send this into the system and average the time signals at the output on exactly the 1024 (or 2048 etc.) samples. The signal will average to a sationary signal but the noise will average down. This allows you to find the harmonics of this signal well below the noise floor of anything in the system. You then drop the signals input level until it gets lost in the remaining noise. You are now looking at the actual nonlinearities of the system at very low signal levels and NOT the signal + noise. An amps with crossover distortion will show a rapid rise in the higher harmonics as the signal level falls. This is very very bad!
Chip amps have a serious advantage in this test since being all on one substraight the components match and track much better for a smoother transfer at the zero crossing.
I bought mine off Ebay... Here's a similar set.
http://cgi.ebay.com/Image-DynamicsC...m3a53425cfa&_trksid=p3286.c0.m14#ht_500wt_958
If I'm not mistaken, Bruce Edgar of Edgarhorn fame contributed to improvements in the designs from Image Dynamics.
The ones that I am using are from a competing company... They're an older design.
Here's their current offering:
http://www.usdaudio.com/products/wg-a700.php
Personaly I'd like to know the answer to that question too. Does anyone know if current Pioneer receivers also use the same chip amps as the older ones? The older receivers use a chip that said "Pioneer" on it. Don't have the part number handy.
I'm almost sure you only measured with a constant resistor load, which is what most people do. Distortion will vary quite a bit with speaker loading due to back EMF. There are also many other aspects that effect audible performance not viewable in distortion measurements. So it's a good idea to keep an open mind in the process of system improvement.
I also realize that you do make great effort to back you optinion with data; however, it's not possible to do 100% of the time.
Last edited:
This is sort of a quibble and not intended to start an argument. I think this is a bit closer to what Dr. Geddes means. If not, it's what I'm saying from my own experience: Actually I think he's saying that the speakers and rooms matter the most and that there is very little to be gained from an amp after a certain point in design. I more than tend to agree. Whenever I go to Tokyo I also go and listen to all the latest HiFi and otherwise gadgets I can in the Akiba district. A few years ago my wife and I did some listening to many amplifiers--roughly a dozen--all in the same room through the same speakers and same song. The difference is incredibly small and indiscernible even during honest (as can be) sighted listening at the same SPL. The only ones we heard that sounded audibly different with a normalized volume were Rotel models--a bit smoother, and it was only slight. All these were SS designs. There were no Sony, Pioneer, etc... receivers in the room so I can't comment on them. I just had a similar experience at Burning Amp. Of course the set up there wasn't that good for evaluating amplifiers.
Seemed that speakers got switched with nearly every amplifier switch. In any case I didn't hear any amplifier that sounded bad there. The tube amp definitely had a unique sound, but it was playing through speakers that were not "tube friendly" at all. Even in that case the difference was not as large as a speaker change.Dan
Last edited:
Dan
Mostly agreed as long as you realize that "different" does not imply "better". Thats a whole other level of complication.
That's interesting. Your test method led to the use of a chipamp, where there's a whole, cultish following of these things. In thinking about it, it does make sense that crossover distortion is such an issue. The amps I've liked (And indeed, every amp in my house) are all models that would feature low crossover distortion, from single ended triodes to class D to the chipamp I just pulled into the bedroom rig (courtesy of the market place here, $60 for $150 in parts
)There's just less listener fatigue in this type of amp. Makes the efforts some people go to to match transistors carefully seem that much more worthwhile.
Chip amps have an inherent advantage for crossover distortion because of better parts matching and temperature compensation. Its very difficult for a discrete amp to do these things. I should also mention that the worst amp that I tested was also a chip amp.
Then that is exactly the point - most "good" amps do not sound significantly different, if different at all, and if they do sound "different", that doesn't mean "better". Further, if they do sound "different" then you can bet that it's because they are "less accurate". Finally, if the "different" sounding amp is made "more accurate", then you can also bet that it will not sound significantly different than any other "good" amp.
That pretty much sums up my position.
That pretty much sums up my position.
I do recall some very high priced brands also use chipamps. The issue is how the whole design is integrated.That's interesting. Your test method led to the use of a chipamp, where there's a whole, cultish following of these things. In thinking about it, it does make sense that crossover distortion is such an issue. The amps I've liked (And indeed, every amp in my house) are all models that would feature low crossover distortion, from single ended triodes to class D to the chipamp I just pulled into the bedroom rig (courtesy of the market place here, $60 for $150 in parts
There's just less listener fatigue in this type of amp. Makes the efforts some people go to to match transistors carefully seem that much more worthwhile.
I have to agree with you Dr. Geddes. Seems to pretty much mirror my experience though I've done only elementary testing. It doesn't seem worth it to do much more. It's fine if you want to learn and enjoy a hobby, but don't think the improvements you get from tweaking a good amp, if any, are going to equal your time and monetary investment.
In any case, this is a great thread Patrick! I've learned a lot from it. It's laid the foundation for my horn/WG comprehension. Most of what Dr. Geddes and you horny guys speak is Greek to me, but now I'm starting to get it. Who knows, in a few years I may roll my own.
Thanks!
Dan
In any case, this is a great thread Patrick! I've learned a lot from it. It's laid the foundation for my horn/WG comprehension. Most of what Dr. Geddes and you horny guys speak is Greek to me, but now I'm starting to get it. Who knows, in a few years I may roll my own.
Thanks!
Dan
Well put, mine too.
Higher power is a valid reason to look elsewhere, and I like Hypex class D
Cheers.
I've heard a few people report that they didn't prefer the sound of their horns with the foam, and I had some thoughts on why this may be:
- There's a significant hit on the efficiency with the foam. The foam is acoustically filtering out high frequencies, and EQ must be used to bring it back. If you do not adjust the response, it's going to sound muted - because it is.
With the roundover alone, there is virtually no hit in the efficiency. There IS a modest improvement in polar response. The improvement is quite audible, because the location of the horn is more difficult to pinpoint, because there are no longer strong reflections at the mouth creating secondary imaging cues. (see Haas Effect.) (This is a polar measurement of the USD horn in my car, adjusted by 10db for clarity.)
With roundover AND foam, we're taking a big hit in efficiency, particularly at the top end. Keep in mind, we have efficiency to burn and the areas that need additional gain are at the frequencies with the lowest distortion. (Distortion is closely related to excursion, and there's less excursion at high frequencies.)
At 10khz the treated horn needs 10db of gain to bring it back up, at 5khz it needs 6db, at 2500hz it needs 4 and at 1250 it needs 3.
It's interesting to consider that at many frequencies, an untreated horn has more energy from reflected energy than from the primary wavefront. No wonder they can sound so poor! - The roundover is a no-brainer - it has no effect on the efficiency, is cheap, makes an audible improvement, and requires no EQ to implement. I literally can't think of a single downside, except that it looks weird. If the cosmetics are a problem, put a big fat roundover UNDERNEATH. In my measurements, a roundover UNDERNEATH was nearly as effective as one that extends the mouth.
- The foam is a bit trickier to implement. Ideally, use a microphone to do gated measurements, and fix the response via EQ. If that is not an option, the measurements that I have performed can provide some guidelines to apply EQ. If you don't have the ability to re-EQ the horn, go get an EQ before implementing this enhancement
- There's some psychoacoustics to consider too. If you've grown accustomed to a certain "sound", your brain will take some time to adjust to the new paradigm. That's one of the reasons I love to do a lot of measurements, your brain will often fool you into a preference for things that are novel and ALSO for things that are comforting and familiar. I spend a lot of time trying to correlate why speakers measure one way and sound another. It may take a few days to get used to a horn without the horn sound.
Last edited:
Either I don't understand or your numbers can't be correct. The most loss that I have seen from foam is about 3 dB and its usually more like 2 dB. The HF fall in a CD device because of the CD and this adds another 6 dB or so resulting in about a 10 dB boost required at 10 kHz to yield a response that is flat to 1 kHz. But only a few dB of this is foam, the rest is the fact that the energy is going over a much larger polar angle at HF than it is a typical system.
It depends on the driver: many drivers use diaphragm breakups purposely to increase HF output (think diamond surrouds).
Plus, this additional HF power will also heat the voice coil.
How can you conclude that??
Last edited: