I'm thinking along the lines of an amplifier that is stable, has flat frequency and phase response across the required range, and keeps the THD and IMD below a certain level at all times. The "straight wire with gain" kind of thing. (I'm not persuaded that it needs ultrasonic capabilities).
The contentious part, I presume, is the allowable level (and nature of) the distortion. For all I know, the distortion levels of the LM3886 (with ordinary power supply) are way, way below audibility, and it's only anecdotes from sighted listening tests that muddy the waters. If someone could show objectively that LM3886 levels of distortion were audible, I would of course take notice. But as things stand, I have no reason to believe that the amplifier is the weak link in the chain, particularly given the easy signals and loads encountered in active speaker applications.
I am open to alternative viewpoints, though!
what I'm undecided about is how one defines "competently designed"
look up the Stereophile measurements for the Channel Islands amp, which is based on UCD180 modules (like my used to be amp). my measurements pretty much mirror those performed by Stereophile. my current amp is the Audio Refinement Complete, also measured and reviewed by Stereophile. go compare the measurements. neither are exceptional but the AR seems to be horrible and actually I didn't expect much from it. until I powered it up. the difference is obvious in 5 secs of listening. I even asked non audiophile friends to describe the difference and they described exactly what I heard. and it's not only a difference, the class D amp simply took away an important part of the music and that was at the extremes of the spectrum. I really wish I could remeasure the UCD amp so that you see that there's nothing indicating problems there, at least not compared to the AR.
The best thing about class A is it teaches the need for good power supplies .
Class H is so close to D as to make it a caprice for most of us to go class D . Why deliberately have an amp which has partial connection to it's power supply ?
My friend John builds gigantic high current amp that drive coils . The have a retail cost of about $80 000 and a waiting list of customers . They are class AB as D was not able to stand the rigors of the back EMF .
John uses a switch mode PSU simply as he will not have to get certification for that aspect of the design . Much money saved . A conventional PSU would have to be carefully designed to meet regulations if not . That wouldn't be a problem as the way the amplifier is used never changes . The strange part of this story is I used an industrial Amcron to do this years ago ( gradients ) . I wonder if his customers realize they could go that route ? The Amcron had 3 phase mains ( 3 KVA ? ) .
My feeling is, if PSU , DAC and output all shared the same clock and basics it might be worth the effort . It seems weird having 3 different clocks as is common . If the clock frequency is too high use a division . Then at least the synchronization is good . If aircraft were designed as most hi fi is, no way Jose would I fly .
All volume controls lead to signals falling into noise floor of listening space, and into noise floor of hearing.
The quietest passages in a recording can't be heard unless the volume it turned up.
Crappy argument of bit loss implies that all volume control should be done by attenuating signal after power amplifier, or that all volume adjustments should be done by adjusting gain of power amplifiers.
DSP as used my most is primarily providing a bunch of sliders that implement all the same old IIR type filters. This is barest tip of DSP.
Even with 32k sliders, I couldn't take this:
An externally hosted image should be here but it was not working when we last tested it.
to this:
An externally hosted image should be here but it was not working when we last tested it.
But the above becomes easy with inversion of measured transfer function of drivers.
http://www.ti.com/lit/ds/symlink/lm3886.pdf
See figure 5. Yes, it can be argued that you get that when, among other things, the heatsink is too small, and that's not a competent design. But I can assure you that it sounds terrible when pushed into that region.
Otherwise they are o.k., just big opamps really.
One of the rarely-appreciated behaviors of amplifiers are short-term overloads. Since most people don't realize what a small increment in perceived loudness requires, and since amplifier companies rarely provide clipping indication, and since some program material has a very high peak-to-average ratio sometimes, this can be the source for a lot of "amplifiers sound different" opinions. I don't think it is the only reason, but it is very important.
The above are independent measurements of woofer and tweeter.
This is how speaker measures with both drivers running:
An externally hosted image should be here but it was not working when we last tested it.
And this is zoomed in view of reverse polarity null of the speaker response when one of the drivers has leads reversed:
An externally hosted image should be here but it was not working when we last tested it.
Response recording of speaker playing 1kHz square wave:
An externally hosted image should be here but it was not working when we last tested it.
And response recording of speaker playing 66Hz square wave:
An externally hosted image should be here but it was not working when we last tested it.
Percussion of drum being struck. It is about 94ms of recording from CD. Top track is recording on CD, lower track is recording of speaker playing CD:
An externally hosted image should be here but it was not working when we last tested it.
This is 10ms zoom from above:
An externally hosted image should be here but it was not working when we last tested it.
My objective is to get direct sound to match input signal.
The above setup makes for highly objective ideal. Subjectively it sounds great.
Reasonable start - how is the IMD going to be measured? The traditional way (with un-music-like stimulus) or using some kind of multitone waveform?
Its based on sighted listening (goes without saying) but my experience with LM3886 has been that power supply noise gets injected into the signal stages via decoupling caps. Then then when this common fault is fixed ISTM the PSRR becomes the limiting factor. If you check out the DS you'll see the -ve rail PSRR is appalling at higher frequencies. KSTR on DIYA has a circuit configuration to mitigate this, by re-referencing the signal grounds to the negative rail. He's said there's an improvement subjectively (as far as I'm aware) but I doubt this is from a rigorous DBT.
What measurements do we currently have which are sensitive to PSU-induced noise?
This is why I say I am developing a 100WRMS amp to have good quality 50 Watts at my disposal.
You're right, the really odd thing is that everyone's thinking in absolute terms, how much power at full output. This is ridiculous as it assumes pure sine waves only, room for transients = 0. First transient to come along clips the amp.
To me, a nominally 100W amp is good 50W amp and an excellent 25W amp. Bring it on!
Not to even mention of how few people realize what power demands are made once you turn that bass pot to say +6 dB. And, as luck would have it, they choose to do so during parties, when the amp is already stressed out. They want "bass slam".
Frankly, CopperTop, I never investigated that route. No idea.
My feeling is that for an active speaker system, one would be best off to develop (or use if he already has it) a really well made 100WRMS amp dedicated to the bass driver, a separate 50W amp for the midrange and a say 30W in pure class A for the tweeter. Number quoted off hand, may need adjusting to specific drivers.
While the power figures may appear to be on the low side, one shouldn't forget that in a fully active system, there is no more power robbing passive crossover, which can be a veritable power robber.
In general, designing a high quality 25W amp is easier than a say 100W amp. Everything is scaled down, for example, you can safely use capacitors rated at 40V rather than 63V or more, etc. It tends to add up, but going full class A will kick the price up because you need a hefty power supply for it to be what it can be.
Too many variables to have a clear cut answer, also much depends on what you end up wanting.
My feeling is that for an active speaker system, one would be best off to develop (or use if he already has it) a really well made 100WRMS amp dedicated to the bass driver, a separate 50W amp for the midrange and a say 30W in pure class A for the tweeter. Number quoted off hand, may need adjusting to specific drivers.
While the power figures may appear to be on the low side, one shouldn't forget that in a fully active system, there is no more power robbing passive crossover, which can be a veritable power robber.
In general, designing a high quality 25W amp is easier than a say 100W amp. Everything is scaled down, for example, you can safely use capacitors rated at 40V rather than 63V or more, etc. It tends to add up, but going full class A will kick the price up because you need a hefty power supply for it to be what it can be.
Too many variables to have a clear cut answer, also much depends on what you end up wanting.
My feeling is anything up to 100 watts is easy . Anything above less so .
Point taken about many drive units and small amps . Even a chance to stagger things a bit if it helps .
For a sub woofer one can use tough transistors without any thought as to FT . Sometimes even bias isn't required . Funny thing is although measurements say unimportant a bit of bias sounds better even for a sub .
Class H or G against D anyone ? The maths seem to say H is a good idea . I dare say it can be AB if switched ?
My friend in PA says class D is half the weight of H which is half the weight of AB . He thinks in Audio it would be less dramatic as we don't run high sustained power . I notice some class D data sheets say crest factor of 10 . In PA it is typically 3 . I always say minimum 6 for hi fi . My friend loves the sound of class H amps . D he thinks are OK and save his back .
Point taken about many drive units and small amps . Even a chance to stagger things a bit if it helps .
For a sub woofer one can use tough transistors without any thought as to FT . Sometimes even bias isn't required . Funny thing is although measurements say unimportant a bit of bias sounds better even for a sub .
Class H or G against D anyone ? The maths seem to say H is a good idea . I dare say it can be AB if switched ?
My friend in PA says class D is half the weight of H which is half the weight of AB . He thinks in Audio it would be less dramatic as we don't run high sustained power . I notice some class D data sheets say crest factor of 10 . In PA it is typically 3 . I always say minimum 6 for hi fi . My friend loves the sound of class H amps . D he thinks are OK and save his back .
Hi dvv
The impression I have is that active makes it all much easier, despite the need for more individual amplifiers. If the differences between amps show up most on the extreme loads and signals, then the active speaker scores on both the easy load (some nominally 8 ohm passives drop to 4 at times apparently) and the easy signal, with the required bandwidth for each amp being so restricted. It looks to me that even though the passive crossover itself may be 50% efficient, say, you can get away with much less than half of the total power previously required of the amps, when going active, simply because you don't need as much 'in reserve'.
The impression I have is that active makes it all much easier, despite the need for more individual amplifiers. If the differences between amps show up most on the extreme loads and signals, then the active speaker scores on both the easy load (some nominally 8 ohm passives drop to 4 at times apparently) and the easy signal, with the required bandwidth for each amp being so restricted. It looks to me that even though the passive crossover itself may be 50% efficient, say, you can get away with much less than half of the total power previously required of the amps, when going active, simply because you don't need as much 'in reserve'.
I really have no experience of Class H. Is the idea that the power supply is basically a crude amplifier whose imperfections the class AB amp then rejects? (same PSRR problems that abraxalito mentioned earlier?). Does the system rely on digital delay in order to allow the PSU to adjust in advance of the signal?
Last edited:
I also feel that active is by far the best way to go. In fact, that's where I'm headed.
When my friend and I developed my loudspeakers, we started with the idea that they should end up as fully active, but if required, in steps or stages. Hence the three pairs of binding posts at the back of each loudspeaker.
Step 1 could be simple biamping. Just add another power amp, take out the connectors joining the lowest pair of posts to the middle one, and presto!, you're in business. True, you are still relying on the passive XO inside for frequency division, but the mids and highs are powered by a different amp, which doesn't give a damn what the bass amp is doing. One more amp and you could be triamping.
The last thing you need is an electronic XO. Then, you need to open the boxes and rewire some wire, disconnecting some, and adding new ones (inside the speaker, attached to a side) for a direct link to the drivers. Stack amps beneath the speakers.
Not elegant, but that's the price step-by-step upgrading.
I beg to differ. Once you are out of the SEPP region, it's no longer easy. Not deadly, but not easy.
One thing seldom seen with active is basic one or two pole filters passive filters . You might not class that as active . To me it is a free lunch and offers protection of the simplest sort . Problem is most people who would do active almost embrace it's complexity as a virtue . To me it seems a missed opportunity . The bass amplifier seems best done as an active filter where the amplifier is considered an op amp ( it works very well ) . The drive unit would have to be modeled to suit . It is even possible to have 4 poles whilst keeping it simple . Two active and two passive . That's as far as is know all the bases covered ? One day I will do a blind test of capacitor coupling . I very much doubt that anyone can hear it whatever they say to the contrary . Especially if using a 22 uF polyester ( 100 V type ) . Measurements support this . No op amp has such low distortion and we might be saying by 50 dB . Now if someone can find a good reason to use extra circuitry I can not see what it might be . DSP , OK I will accept that . If it is being used no reason not to extend it a bit . As an analogue lover I doubt it would float my boat anyway . If building a CD/ digital system for a friend then maybe .
I have friends in PA who would not agree . Difference is a sound check might be 20 minutes max . Johns ears and 20 minutes will do the job . He dare not risk it to something that I would take weeks to adjust . Fair enough .
Dvv , 100 watts is falling off a log if music surely ? For a motor not so easy .
I have friends in PA who would not agree . Difference is a sound check might be 20 minutes max . Johns ears and 20 minutes will do the job . He dare not risk it to something that I would take weeks to adjust . Fair enough .
Dvv , 100 watts is falling off a log if music surely ? For a motor not so easy .
It all depends on what you want it to do and be capable of doing, Nige. The more you want, the more you ask it to do, the more complex it becomes.
Ditto for "accoutrements", it seems these days even basic protection circuits are considered a luxury.
I refuse to play that game. Either complete, or none, thank you.
Ditto for "accoutrements", it seems these days even basic protection circuits are considered a luxury.
I refuse to play that game. Either complete, or none, thank you.
For any active or multi-amped system, surely it would be better to use a line-level crossover and have the amps drive the speakers directly. That seems like the way some of the main benefits to sound quality would accrue; nothing between amps and speakers; not even cables!
And then you also get to use tiny line-level crossover components, which makes the highest quality components probably cheaper than even the standard-quality high power ones that would have been needed if the crossovers were after the amps.
Then maybe also eliminate the effects (and cost and inconvenience) of the line-level cables and use wireless (digital of course)...
And then you also get to use tiny line-level crossover components, which makes the highest quality components probably cheaper than even the standard-quality high power ones that would have been needed if the crossovers were after the amps.
Then maybe also eliminate the effects (and cost and inconvenience) of the line-level cables and use wireless (digital of course)...
Last edited:
- Status
- Not open for further replies.