Continuing with PC monitor fed with high frequency experimenting, it seems to revolve around power supply issues. With a low level feed in and max vol. on the speakers, I get the same effect as with max input signal, and reduced volume - implying that the output stages of the monitor inbuilt amps are misbehaving, or causing the power supply lines to incur too high a level of modulation.
So, is it really just those particular speakers? Luckily, I also have "real man" monitors,
, Gateway Altec Lansing 2-ways - oops, made in Chi... - haven't been fired up in years. Hmmm, noisy with hiss, fairly crappy volume pot, crackling and farting - but, same problem! Provided I drive with sufficient final treble volume I get the spurious noise coming in, and out, at a certain level. Speakers are more "solid" than the h/k's, but same basic configuration of units, running off wall wart ...
So, more to investigate! ... Any thoughts ... ?
So, is it really just those particular speakers? Luckily, I also have "real man" monitors,
, Gateway Altec Lansing 2-ways - oops, made in Chi... - haven't been fired up in years. Hmmm, noisy with hiss, fairly crappy volume pot, crackling and farting - but, same problem! Provided I drive with sufficient final treble volume I get the spurious noise coming in, and out, at a certain level. Speakers are more "solid" than the h/k's, but same basic configuration of units, running off wall wart ...So, more to investigate! ... Any thoughts ... ?
Last edited:
So what started as a discovery about digital audio evidenced in a PC's internal sound system (not withstanding the PC's sample rate converter) is now localized in a pair of PC monitors?
Depending on the age they may have analog or digital amps. They may have marginal or really marginal power supplies. And most are very sensitive to EMI. The BOM on many did not exceed $5.00. Labtech, responsible for a lot of the PC speaker abuse, was bought by Logitech. I have heard that the current audio guru's at Apple are ex Logitech. Which may explain why the sound quality of the iPod/iPhone/iPad has degraded every generation (the current being really poor).
The PC's native sound system (AC'97 and later) resamples everything to 48K and usually 16 bits with a marginal resampler (low processor overhead being important). Its necessary so the system noises, music and movies can all be played at once. The chipset for the audio rarely cost over $1.00 for a PC. That is a brutal business all about meeting the specs for less and fractions of a cent are the difference between success and bankruptcy.
At a minimum can you bring the audio out to an external DAC and see if you still get the artifacts?
Depending on the age they may have analog or digital amps. They may have marginal or really marginal power supplies. And most are very sensitive to EMI. The BOM on many did not exceed $5.00. Labtech, responsible for a lot of the PC speaker abuse, was bought by Logitech. I have heard that the current audio guru's at Apple are ex Logitech. Which may explain why the sound quality of the iPod/iPhone/iPad has degraded every generation (the current being really poor).
The PC's native sound system (AC'97 and later) resamples everything to 48K and usually 16 bits with a marginal resampler (low processor overhead being important). Its necessary so the system noises, music and movies can all be played at once. The chipset for the audio rarely cost over $1.00 for a PC. That is a brutal business all about meeting the specs for less and fractions of a cent are the difference between success and bankruptcy.
At a minimum can you bring the audio out to an external DAC and see if you still get the artifacts?
I wish you could be transported easily to my environment to hear the Duet system. Another feature I didn't mention is a compressor that I still take a while to understand when I see the schematic. And <blush> I designed the electronics. I discussed it in here briefly a long time ago, and Thorsten recommended FETs. Nope, too expensive and/or too variable in characteristics for production.Continuing with PC monitor fed with high frequency experimenting, it seems to revolve around power supply issues. With a low level feed in and max vol. on the speakers, I get the same effect as with max input signal, and reduced volume - implying that the output stages of the monitor inbuilt amps are misbehaving, or causing the power supply lines to incur too high a level of modulation.
So, is it really just those particular speakers? Luckily, I also have "real man" monitors,
So, more to investigate! ... Any thoughts ... ?
I think the migration to "digital" amps in PC speakers is due to the absence of anyone left who knows analog EQ. Therefore any company selling power amps with bundled DSP EQ will hold his or her hand and do the specified (as if they knew what to specify!) EQ. And the last I heard, symbol domain compressor/limiters were simply lousy --- and with the drastically power-limited systems with teeny transducers to boot, you really need a good, frequency-dependent limiter.
But this too shall pass.
This is part of a journey to discover what influences sound, from the highest to the lowest "quality". It's no good just saying that people should make sure they buy well-engineered components that won't have artifacts intruding into the sound, they don't have stickers on them that guarantee such!
Just had a look at the JBL Duet, the current version, that is. Yes, at face value that's about the quality of what I'm playing with at the moment. Incidentally, the h/k is quite superior to the Altec - which is perhaps not surprising considering how long the latter has been hibernating - determined by running a very slow frequency sweep through them: the Altec has lots of "false" bass, and myriads of distortion resonances at various points; the h/k is relatively clean.
I'm looking at the PC, plus monitors, as a single entity - what's influencing what is not as important as the fact that the combination does not work properly, for whatever reasons ...
Have moved on, and am now trying a combination of 19k/20k sine waves, which should generate some nice 1k IMD. Yes, at a reasonable output volume there it is, but it's about 50dB down - which is not too bad, considering. How do I know it's 50dB down? Because I generated a separate, pure 1kHz track at that level, by trial and error - and switching between the two tracks there is a good match, by ear ...
Now, what's really interesting is to resample the 19k/20k track to 352k, from 48k - big, big jump in distortion artifacts, quite dramatic! So, what's causing the problem here ...?
The real purpose of this exercise is to point out, contrary to what some people believe, that precisely how certain audio tracks with the same intrinsic information are played back can make a significant, or huge, difference to the sound quality. Just saying, "get a better system!", is not a useful answer ...
Just had a look at the JBL Duet, the current version, that is. Yes, at face value that's about the quality of what I'm playing with at the moment. Incidentally, the h/k is quite superior to the Altec - which is perhaps not surprising considering how long the latter has been hibernating - determined by running a very slow frequency sweep through them: the Altec has lots of "false" bass, and myriads of distortion resonances at various points; the h/k is relatively clean.
I'm looking at the PC, plus monitors, as a single entity - what's influencing what is not as important as the fact that the combination does not work properly, for whatever reasons ...
Have moved on, and am now trying a combination of 19k/20k sine waves, which should generate some nice 1k IMD. Yes, at a reasonable output volume there it is, but it's about 50dB down - which is not too bad, considering. How do I know it's 50dB down? Because I generated a separate, pure 1kHz track at that level, by trial and error - and switching between the two tracks there is a good match, by ear ...
Now, what's really interesting is to resample the 19k/20k track to 352k, from 48k - big, big jump in distortion artifacts, quite dramatic! So, what's causing the problem here ...?
The real purpose of this exercise is to point out, contrary to what some people believe, that precisely how certain audio tracks with the same intrinsic information are played back can make a significant, or huge, difference to the sound quality. Just saying, "get a better system!", is not a useful answer ...
Not really a lack of knowledge on analog eq. Its more a lack of knowledge on real eq period. The current generation of digital systems for the small speaker dock etc. are pretty remarkable. The eq. can work quite well, the crossovers can be tuned to what you really need and the limiters are very sophisticated. The usual Asian design team has no idea how to use them to deal with the lousy box or driver limitations but that's another story. They can make a measured flat system real easy however. The flat system highlights every resonance in the system.
For a BOM of about $25 you can make a 20W X3 digital system that's pretty decent including Bluetooth (required it seems to sell the stuff). There is no way you can make an analog system competitive any more. And well done cheap digital will sound better than cheap analog. Poorly done either is a tragic waste of resources.
Please provide more details so we know what you are working with. There are very few hardware options that can play 352K (and why resample to a non even multiple of the original track?). Resampling well is difficult. keith Johnson still does it the hard way- Digital to analog at the input sample rate and analog to digital and the output sample rate. He has not found an algorithm he likes. If you go here you can see many of the problems illustrated SRC Comparisons
At the hardware level, a HP Compaq dc7700, with, from HP's website, "Integrated High Definition audio with Realtek ALC262 High Definition audio codec". Software is Audacity 2.0.3, which with the latest release has a new resampling engine, libsoxr, which is definitely superior to the previous code module used - I have tested this through a number of rate conversions, and it is "perfect" in its behaviour, no artifacts are introduced when progressing through conversion stages. That is, start at a particular rate, go through a number of conversions up and down, and end up at the rate you started with: do a difference between original and multiple converted, and you get a null result, certainly at the accuracy of the format resolution.
The 352k was a mistake on my part, it was intended to be the even multiple.
I have found, irrespective of what the DAC's 'natural' resolution really is, that feeding it with audio at different file sample rates does alter the sound - my experiemnts here hopefully will shed some more light on things, at least for me ...
The 352k was a mistake on my part, it was intended to be the even multiple.
I have found, irrespective of what the DAC's 'natural' resolution really is, that feeding it with audio at different file sample rates does alter the sound - my experiemnts here hopefully will shed some more light on things, at least for me ...
Well I shall have to take your word for it Demian. I do know what happened pretty much the moment I left Harman in late 2004, and it did have much to do with skill sets lacking in the folks left there. Not too long after that the brilliant new CEO dismantled the entire Multimedia group and assumed everything would be executed (a good word for it) offshore. That group by the way was the only one turning a profit within the Consumer division.
Of course by some measure that was already a long time ago.
I agree wireless has become de rigueur for speaker systems and is mostly done badly. I suppose by now someone would have done something about compressor/limiters too and gotten more sophisticated about other issues; but it wasn't that long ago that TI (for example) was rather confused about multichannel, presenting an absurd five speaker array in a single package as a demo to a client of mine at one point (when I pointed out that there was no power supply the guy pointed to a 3.3V regulator and contradicted me --- complete idiocy). Now I understand that they apply significant resources to making sure certain large customers are properly cared for (to the extent that one particular engineer feels he just rubber-stamps their efforts).
Actually, most everywhere, the paucity of knowledge about acoustics is probably the biggest deficiency, and that's been true for a long time. Besides having the means of backing out nonlinear distortions of all sorts, but not using them, just the absence of understanding of what makes for decent dispersion, box rigidity, quiet porting or compensation for sealed boxes and complementary transducers, etc. seems still mostly lacking. When you have approached these issues intelligently, the resulting needs for equalization (for example) are significantly diminished. I'm amused by some efforts to heroically flatten loudspeaker response on axis while neglecting off-axis response. Another dozen biquads ain't gonna cut it.
There is certainly no good reason why chores done with analog processing should not be accomplished in the digital domain --- indeed, many things are nearly impossible otherwise --- and once one has the now-cheap DSP engine at hand, by all means use it. And it is also quite plausible that the principal problems in systems are ignorance and arrogance on the part of the users, qualities still in prevalence about everywhere. One person who is primarily a transducer designer said he had to fight mightily just to assure that the order of certain signal processing in a system for a smartphone was optimized --- the manufacturer thought the whole thing was unimportant. You probably know him.
Again, this too shall pass, whether in our lifetimes or not is another question.
At some level this can yield to compute power. There is always a common sample rate even though it might be absurdly high. The problem is related to getting ALL tones in exact FFT bins.
I would argue that it is not possible to inject jitter by manipulating the 16/44 source material. Not jitter as what happens between clocks and D/A anyway. One could try to distort the signal in the same way as jitter would but jitter as we know it arises when there is a prolonged (or shortened) duration between generating the "next" level to be output. What really happens inside a D/A during these timing deficiencies is unclear and I doubt that anyone could sit in font of a DAW generating the equivalent.
/
/
Intrigued, I tracked this down: The second jitter test is online. (Page 1) - Reference - AVI HiFi Forum. This is real music, unfortunately, girl and guitar,
"not everything is uderstood" is a quite orignal criticism, can't recall the last time I heard that
maybe you shouldn't be listening to any digital source at all - of course I'd like to see your complete explaination of vacuum tube behavior too
it turns out some things are at least strongly suspected - try reading the ref I gave
maybe you shouldn't be listening to any digital source at all - of course I'd like to see your complete explaination of vacuum tube behavior too
it turns out some things are at least strongly suspected - try reading the ref I gave
And an appropriate low pass filter. However for some, like Keith, there is no drive to find a better solution. He has 4-5 Pacific Microsonics converters that are adequate for now.
We must not forget that the really talented and experienced engineer/recordist knows the limits of his tools and how to get the most from them. Others of us (like me) would just find all the problems and get lost. . .
This brings to mind the episode when Mohamed Nasheed, former (disposed) president of the Maldives, was interviewed by Jon Stewart, who asked the question why the United States woud again have to step up to the plate; wouldn't it make more sense if for example India assumed some role in its own area? Nasheed's lighting quick come back was: "So you are now going to outsource your security to India too?". It is a shame that so much research is taken away from its origin, in Europe as well.
I think it is due to the arrogance of marketing and the inflated role awarded to them by top management. They 'know' what sells, and the consumer is deaf to begin with and subsequently blinded by branding. Great audio engineers like Allan D. are sucked into management to play under cohoots with the sales guys, so all the real talent gets corrupted. This is double bad because the old Harman invested so much in knowledge, with Floyd Toole in the leading edge.
Well I'm somewhat relieved to note --- for the sake of my few friends left there --- that Paliwal seems to consider some research and development worthwhile, including the subjective evaluation facilities. Sean Olive got a call from the man to congratulate him for being the President elect of the AES. After a bit, the CEO said "Now, what is the AES?"
I'll say it again: you can't make this stuff up.
If you can get your hands on a JBL Duet system, and replace the skinny cable to the left sat with something heavier, and blu-tac down each lightweight sat to the desk, you might be quite pleased if you listen at reasonably low levels. It helps to pull the grilles off too. The transducers do have a lot of IM distortion, unfortunately, although when you see the excursion you can see why.
An earlier version was ported quite a bit lower, and when demoed there were people who were looking around for a third channel woofer. However the IM really got bad, so the sane decision was to port somewhat higher. The equalization was quite simple, and the curved boxes are rather stiff for the amount of material. The compressor starts to work early on, sooner than typical but agreed in listening tests to be the best compromise. About 6W per channel from a discontinued class AB amp chip that was running iirc about 19 cents. There is much less diffraction than typical compared to the squarish enclosures typical of most multimedia monitors --- again, the industrial design, which was based on an alien-ET notion echoed in the three-piece system Creature, for once married well with desirable acoustical properties. I wish I could say the same for Creature, whose "grille" should have been made out of mesh rather than solid metal. The tech who measured prototypes evidently mixed up the data and gave me plots for grilles off. Ooops. We did the best we could to compensate, and the iLounge guy liked them
I was told that Harman sold about 800k of the Duet some years ago. The margins were quite healthy and made the company some serious money. However, the cash cow was and remains the OEM automotive business, and other activities exist to support brands adopted by Harman-Becker.
These are the beasts I'm using, Harman Kardon HK 19.5 Reviews - PC Speakers | dooyoo.co.uk, the cloth grille started buzzing some time ago, was removed ...
- Status
- Not open for further replies.