Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Quality aspects of ADC (Read 20068 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Quality aspects of ADC

Hi all!

Background information (safe to skip  )
Currently I'm preparing a large archival operation: my musical society has a collection of recordings on compact cassettes from 1980 to 1990, about 140, which I think have to be digitalized to protect them from deteriorating. I have found a nice tape deck (Nakamichi DR-3) but it has to stay at the office of the executive committee. A friend of mine has a nice ADC for this job, a quality portable mixing console with USB out, but it cannot stay at the office as it is more or less public and he doesn't trust everyone there. However, it is too much hassle to bring it every time, as I think this operation will cost about a year: 3 recordings a week. So, I guess I'm stuck with my own ADC: a Creative X-Fi Surround 5.1 USB-card, which is not really made for the job, but I can at least leave it at the office, as it isn't that valuable.

On-topic
That friend of mine that offered me his portable mixing-console with ADC argued that with a cheap ADC all little nuances and details would vanish, so I should look for another solution than using my Creative ADC. That made me think: i'm much more a theoretician than he is, so I was wondering, what makes a high quality ADC? I know the working principle of several ADCs, I can't imagine which link in the chain would be capable of 'erasing details' anyway. How does a non-linearity in an ADC sounds like in practice? And jitter? I guess uncorrelated, random jitter could sound like 'more noise', as it is random. To me it seems a low-quality ADC doesn't add anything but noise and a non-linearity, let alone it could remove things from the signal.

In short: what could a low-quality ADC do with the signal except adding noise? Probably I'm far too short sighted, can anyone enlighten me?
Music: sounds arranged such that they construct feelings.

Quality aspects of ADC

Reply #1
The only thing I'd even consider is noise.  And, unless the ADC/soundcard is defective the noise will be very low compared to the background noise on the cassette tape.  Most soundcards are superior to cassette in frequency response & distortion.   

Probably the most important thing is to get your recording levels right.  You want a strong enough signal so that the ADC noise remains insignificant, but you don't want to clip (distort) the ADC either.

If altering the sound doesn't violate any archiving principals, you might be able to use some digital noise reduction and/or EQ to make the digital copy sound better than the analog cassette!    Or, you can keep both an unaltered archive and an enhanced version (something I've done).  Just beware that noise reduction can introduce artifacts (depending on the source material) and of course any EQ "improvements" involve of taste & judgment. 

I have no idea what jitter sounds like...  I've never heard it.  From what I understand, the amount of jitter in any normally operating device is inaudible. ...  You have to intentionally design something with gross jitter before it becomes an issue.

If someone starts giving you a bunch of nonsense about jitter, or "high resolution" audio", or "nuance",  Ethan Winer's website[/u] has a lot of useful information to bring you back to reality.

P.S.
When it comes to timing-related issues, I be more concerned about the analog side...  Tape speed, wow & flutter, stretched tape, etc.  These are likely to be at least 10x worse than digital jitter.  But again, I've never heard any speed/timing problems from any properly working turntable or tape machine.  (I have heard these kinds of defects when the unit had a worn belt, or something was bent, etc.)

Quality aspects of ADC

Reply #2
The only thing I'd even consider is noise. And, unless the ADC/soundcard is defective the noise will be very low compared to the background noise on the cassette tape. Most soundcards are superior to cassette in frequency response & distortion.


Digitalizing those cassettes is the case now, but I was wondering about ADCs in general. So, the THD+N (for distortion and non-linearity) and SNR are the only specifications that matter for an ADC?


If altering the sound doesn't violate any archiving principals, you might be able to use some digital noise reduction and/or EQ to make the digital copy sound better than the analog cassette!

I will archive the sound unaltered, but of course, I could improve the recordings if necessary. I'll do that upon request to seperate files, not for the archive as a whole.
Music: sounds arranged such that they construct feelings.

Quality aspects of ADC

Reply #3
One thing that's been an issue in the past is sound cards that converted at 48 khz and the on-the-fly conversion to 44.1 didn't work so well.

I'm not sure if it was an artifact of that conversion, but I once had a Soundblaster LIve card that inserted impulses that could be seen and heard on track gaps etc on recordings from the line-in.  I know it wasn't the normal vinyl clicks/noise because they were only one sample wide.

Edit/add:  I have been perfectly happy with recordings made on the built-in sound chips of more recent computers.

Quality aspects of ADC

Reply #4
Quote
So, the THD+N (for distortion and non-linearity) and SNR are the only specifications that matter for an ADC?

With any audio equipment there are 3 sound-quality specs/factors that concern me.*
1. Noise
2. Distortion
3. Frequency Response

Ethan Winer adds "time-based errors" (and room acoustics).  But as I said,  I've never had any time-based issues with properly-working equipment.  In a "scientific" discussion time-bases issues must be included because you can obviously muck-up the sound by changing/varying the playback speed, or the clock in a digital system.   

And with modern electronics, noise is the only defect I ever hear.  It's easy to get low distortion (below the threshold of human detectability) and flat frequency response, although if you want to get "ruler flat" frequency response beyond 20kHz from a real-world DAC or ADC, you might need to go to a higher sample rate (higher than 44.1 or 48kHz).

With transducers (speakers, headphones, and microphones), frequency response becomes a BIG deal!


* I guess I should say power (watts) is a major spec that I consider when it comes to power amps, and of course there are other features & functions to consider that don't directly relate directlyto sound quality.

Quality aspects of ADC

Reply #5
With any audio equipment there are 3 sound-quality specs/factors that concern me.*


Well, I could name another important one for speakers: their radiation patern. For some speakers, there's quite a difference between sitting exactly in front of it or at an angle of about 10 degrees, for some speakers, this difference is only slight. Like you probably overlooked this, I started this topic to see whether I overlooked something on considering ADCs too. When I posted this topic, I was concerned about DAC-slew for example, as would occur in sample-and-hold DACs, and something similar in ADCs. You can think of so much things that can go wrong like improper shielding, aliasing, channel crosstalk, but I guess these can all be measured under those two measures: THD+N and SNR.

However, to really convince myself, I just did some tests: play some music with a high dynamic range on the soundcard and record it through line-in (direct loopback), play that recording and record it, and so on. I 'stacked' the effect of DAC and ADC in this way 12 times, and I really can't hear a difference, so I'm really confident now that my ADC can do the job  It also revealed the soundcard of my laptop is really, really badly build: when I plug-in my USB-soundcard it start to hum at quite high levels, (-50dB or something similar) not really you would expect of a laptop that is priced over 3000 dollar (luckily I got it for much less) but I guess sound isn't a priority on those things.
Music: sounds arranged such that they construct feelings.

Quality aspects of ADC

Reply #6
That friend of mine that offered me his portable mixing-console with ADC argued that with a cheap ADC all little nuances and details would vanish,


Your friend is drinking the High End Audio Kool Aid. ;-)

Quote
so I should look for another solution than using my Creative ADC.


Many of us have been down that road.

We've all learned that once audio gear meets certain technical standards, it all sounds the same. What has changed over the years is that the cost and size of audio gear that meets certain standards has decreased dramatically.

In 1970 the university I attended bought a 200KHz sample rate 16 bit ADC/DAC. I seem to recall that it cost about a half million dollars and filled a full height 19 inch rack cabinet.  Last year I bought a USB audio interface that outperformed it, and had twice as many channels. I got change from $100. I could fit it in my hand but if I closed my hand the corners of the box showed. But, the box it was in was mostly empty, whereas that 1970 DAC was packed full of circuits.

Quote
That made me think: I'm much more a theoretician than he is, so I was wondering, what makes a high quality ADC?


A quality DAC provides basic clean signal processing, IOW low noise and distortion. It is also durable and convenient to use. All failures to provide clean signal processing can be broken down into noise, linear distortion and nonlinear distortion. You may be unfamiliar with the oxymoron-sounding phrase "Linear distortion" but that is just more familiar things like flat frequency response and appropriate phase response.

Quote
I know the working principle of several ADCs, I can't imagine which link in the chain would be capable of 'erasing details' anyway.


Elsewhere on HA we've been discussing just that. I don't know how to erase details with equipment that meets fairly ordinary technical standards. About as close as we can get is to bury small signals in noise.

Quote
How does a non-linearity in an ADC sounds like in practice?


Nonlinearity adds tones to sounds that weren't originally there. The added tones are both harmonics and also sum and difference tones related to the tones in the original recording.  The harmonics tend to fit in with the harmonics that are already in the music. The sum+difference tones don't fit in nearly as well, and there's no way to avoid them if the equipment is nonlinear.  In small quantities nonlinear distortion tends to alter the balance between fundamental tones and the natural harmonics that are in the music. If small enough, added harmonics can be easily ignored or overlooked. They may even make things sound better to some people.  The sum and difference tones don't fit into the natural harmonic structure of the music, and tend to give a sort of sour or gritty sound. 

For example, if you want people to overlook modest amounts of nonlinear distortion in a demo, you might use recordings of a solo instruments or a certain kind of voice. There won't be a lot of different concurrent tones in the music, and most of the distortion will come out as harmonics which might be overlooked or even preferred.  If you want a listening test that is very critical of nonlinear distortion, you might pick a complex sound like a choir. Choirs are particularly good because their signal is dense in the midrange, but leave a lot of less-used frequencies at the extremes. Any sum and different tones that are created will be more audible when they naturally fall into the frequencies where the basic music is less dense.

Quote
And jitter?


Jitter is just another kind of nonlinear distortion. In large amounts low frequency jitter is just vibrato or flutter or wow. Which makes a point - while jitter is commonly obsessed over as applying to digital equipment, we've had tons of it for years in LP recording and playback, as well as analog tape. High frequency jitter can be more like an increase in background noise. High frequency jitter has been around for years as "scrape flutter" in analog tape recorders.

The real kicker is that the best analog playback system has tons more jitter than even a mediocre digital system.

Quote
I guess uncorrelated, random jitter could sound like 'more noise', as it is random.


Yes, I just covered that, but much jitter is correlated. It is due to interference caused by the signal itself or outside influences like the power line or frame size in digital media. Making jitter go away in a digital system is very simple. Every CD player's optical pickup delivers a massively jittered signal to its decoding electronics. All you do is buffer it and reclock it and its as pure as the clock you use to reclock it. BTW, adequately clean digital clocks are now about a dime a dozen.

Quote
To me it seems a low-quality ADC doesn't add anything but noise and a non-linearity, let alone it could remove things from the signal.


The noise and the nonlinearities could add sounds that mask the detail in the signal. Thing is that even inexpensive DACs such as the one in cheap digital music players such as the Sansa Clip, or the motherboard audio interface in your computer, are now really very good. They are capable of sonic transparency, which is the ability to reproduce sound without adding anything that makes it sound different, even in terms of the smallest things that we can reliably hear, or sense in any way.


Quality aspects of ADC

Reply #7
We've all learned that once audio gear meets certain technical standards, it all sounds the same. What has changed over the years is that the cost and size of audio gear that meets certain standards has decreased dramatically.
(...)
Thing is that even inexpensive DACs such as the one in cheap digital music players such as the Sansa Clip, or the motherboard audio interface in your computer, are now really very good. They are capable of sonic transparency, which is the ability to reproduce sound without adding anything that makes it sound different, even in terms of the smallest things that we can reliably hear, or sense in any way.

I really don't understand how that is possible actually: I study mechanical engineering, in which I'm confronted with the aspects of mass-production. The idea is simple: when something costs more at no benefit, don't use it. I guess a motherboard-manufacturer would choose the cheapest audio-gear it can get: most costumers really don't care, they won't complain about sub-par audio. If they complain, it will be about the DAC, not the ADC, as they do not expect much of those cheap built-in microphones. Placement of these chips won't get a high priority either, as placing and shielding an audio-chip as it should would cost them several dollars I guess, at a budget of ~40 dollar for the whole motherboard, that is not an option. As opposed to what I would expect, computer-audio is certainly not bad, is making audio sound good that easy? I was thinking in a similar way about my Creative X-fi, as it major selling point is surround output, not quality input. Why would they spend a few bucks per unit in getting the ADC right, if (nearly) nobody cares? Why would they meet those minimum quality standards?

In 'professional' equipment, things are different. Specs are read, and if the device doesn't meet the expectations, it will be returned, named and shamed. I can understand why equipment made for the job is well built. I wonder with consumer equipment however, as no one seems to care.

Your friend is drinking the High End Audio Kool Aid. ;-)

I'm aware of that. I mentioned him just because he made me think.

In 1970 the university I attended bought a 200KHz sample rate 16 bit ADC/DAC. I seem to recall that it cost about a half million dollars and filled a full height 19 inch rack cabinet.  Last year I bought a USB audio interface that outperformed it, and had twice as many channels. I got change from $100. I could fit it in my hand but if I closed my hand the corners of the box showed. But, the box it was in was mostly empty, whereas that 1970 DAC was packed full of circuits.

I guess that's the blessing of IC's and their miniaturization.

A quality DAC provides basic clean signal processing, IOW low noise and distortion. It is also durable and convenient to use. All failures to provide clean signal processing can be broken down into noise, linear distortion and nonlinear distortion. You may be unfamiliar with the oxymoron-sounding phrase "Linear distortion" but that is just more familiar things like flat frequency response and appropriate phase response.

Right, that was what I was after. Thanks!

Elsewhere on HA we've been discussing just that. I don't know how to erase details with equipment that meets fairly ordinary technical standards. About as close as we can get is to bury small signals in noise.

I'll take some time to read it

For example, if you want people to overlook modest amounts of nonlinear distortion in a demo, you might use recordings of a solo instruments or a certain kind of voice. There won't be a lot of different concurrent tones in the music, and most of the distortion will come out as harmonics which might be overlooked or even preferred.  If you want a listening test that is very critical of nonlinear distortion, you might pick a complex sound like a choir. Choirs are particularly good because their signal is dense in the midrange, but leave a lot of less-used frequencies at the extremes. Any sum and different tones that are created will be more audible when they naturally fall into the frequencies where the basic music is less dense.

Sounds interesting, as the recordings that have to be digitalized are fairly complex, a large symphony orchestra plus a large choir. I have been wondering, why some microphones do very, very well on small chamber music settings and fail on large orchestra's, while others do both. Could that be non-linear distortion of the pre-amp, or is that related to distortions in the microphone membrane?

The noise and the nonlinearities could add sounds that mask the detail in the signal.

So it comes down to noise for a large part. I will take a look at those sum+difference tones
Music: sounds arranged such that they construct feelings.

Quality aspects of ADC

Reply #8
We've all learned that once audio gear meets certain technical standards, it all sounds the same. What has changed over the years is that the cost and size of audio gear that meets certain standards has decreased dramatically.
(...)
Thing is that even inexpensive DACs such as the one in cheap digital music players such as the Sansa Clip, or the motherboard audio interface in your computer, are now really very good. They are capable of sonic transparency, which is the ability to reproduce sound without adding anything that makes it sound different, even in terms of the smallest things that we can reliably hear, or sense in any way.


I really don't understand how that is possible actually: I study mechanical engineering, in which I'm confronted with the aspects of mass-production. The idea is simple: when something costs more at no benefit, don't use it. I guess a motherboard-manufacturer would choose the cheapest audio-gear it can get: most costumers really don't care, they won't complain about sub-par audio. If they complain, it will be about the DAC, not the ADC, as they do not expect much of those cheap built-in microphones.


There is a spectrum of costs from mechanical engineering to electronic engineering to software engineering. At one extreme (mechanical) materials cost and production costs are relatively high. For electronics, particularly microelectronics, the materials and production costs are small and decreasing. For software, materials and production costs are about nil. IP is an increasing portion of the sales costs of all 3. Value is increasingly perceived value as opposed to being inherent in the cost of materials. Modern audio gear is at the intersection of microelectronics and software.

It takes few if any more materials or processing to make a good DAC as opposed to a poor one. The difference is almost all IP and the IP is becoming increasingly well known (and therefore less costly).

The market for DAC chips is very competitive, and if someone can add a zero in front of his chip's THD spec, he's going to win most design competitions. This goes on until there's only one guy selling DACs which seems a long ways away.

Quote
Placement of these chips won't get a high priority either, as placing and shielding an audio-chip as it should would cost them several dollars I guess,


The sound chips that they put on motherboards generally get no add on shielding at all. A lot of their noise resistance comes from the fact that they are tiny so they act like poor antennas.

Quote
As opposed to what I would expect, computer-audio is certainly not bad, is making audio sound good that easy?


Certainly the vendors are making it look easy. The finesse the board design IP problem by providing "reference designs" of all of the necessary circuit board traces.

Quote
I was thinking in a similar way about my Creative X-fi, as it major selling point is surround output, not quality input. Why would they spend a few bucks per unit in getting the ADC right, if (nearly) nobody cares? Why would they meet those minimum quality standards?


Creative got spanked in the market place some years back when they tried to compromise sound quality

Quote
.
In 'professional' equipment, things are different. Specs are read, and if the device doesn't meet the expectations, it will be returned, named and shamed. I can understand why equipment made for the job is well built. I wonder with consumer equipment however, as no one seems to care.


Things are evolving predictably. Once quality gets so high that nobody hears any faults that are due to the chips, things will probably level off.  In many cases (like the Clip) the DAC becomes part of the same chip that carries the CPU.


For example, if you want people to overlook modest amounts of nonlinear distortion in a demo, you might use recordings of a solo instruments or a certain kind of voice. There won't be a lot of different concurrent tones in the music, and most of the distortion will come out as harmonics which might be overlooked or even preferred.  If you want a listening test that is very critical of nonlinear distortion, you might pick a complex sound like a choir. Choirs are particularly good because their signal is dense in the midrange, but leave a lot of less-used frequencies at the extremes. Any sum and different tones that are created will be more audible when they naturally fall into the frequencies where the basic music is less dense.


Quote
Sounds interesting, as the recordings that have to be digitalized are fairly complex, a large symphony orchestra plus a large choir. I have been wondering, why some microphones do very, very well on small chamber music settings and fail on large orchestra's, while others do both. Could that be non-linear distortion of the pre-amp, or is that related to distortions in the microphone membrane?


I do a fair amount of recording of bands, orchestras and choirs at State festivals. On site people ask me how I get my recordings to sound so good, so I must be doing something right! ;-)

One setup I use is composed of just a single fairly inexpensive stereo mic, the Rode NT4 (about $400), an inexpensive mic preamp (< $200), and a relatively inexpensive digital recorder. Much of "The trick"  is getting the mic in the right place and pointing the right direction. Once I know a room I just walk in, set up, make a few checks and roll.  I can often size new rooms up so that I'm close on my first try and can fine tune between the first few groups.

Mics are generally very free of nonlinear distortion because distortion in transducers is dominated by issues related to diaphragm movement and mic diaphragms don't move much. Far more important is frequency response as a function of acceptance angle. Distances from sources and the room boundaries  to the mic (and what the room boundaries are) together with mic response versus acceptance angle pretty much control how the recording will sound.

Differences among electronics, particularly nonlinear distortion other than clipping, are decreasing as strong influences in system sound quality.


Quality aspects of ADC

Reply #9
For electronics, particularly microelectronics, the materials and production costs are small and decreasing.

While small, there is no reason not to use even cheaper components. Quality of audio is generally not a selling point of motherboard or portable audio devices. (just being capable of outputting sound is a selling point) For example, I own a Sansa Fuze, which is perfectly capable of driving my headphone to unhealthy high levels. While I accepted that as normal, I recently borrowed a (there the name is again) Creative Zen in the same price class: it distorted the sound because it was not able to drive the headphone at levels above average. That again contrasts with the theory: why would Sandisk use this probably more expensive DAC that is able to drive higher loads? I don't think of Sandisk of a higher quality brand than Creative regarding audio, I guess the few people pleased by the performance of that Fuze wouldn't compensate for the expenses on the better chips. Probably theory is just theory, and probably some people do care about making quality products, no matter whether it pays back or not.

It takes few if any more materials or processing to make a good DAC as opposed to a poor one. The difference is almost all IP and the IP is becoming increasingly well known (and therefore less costly). The market for DAC chips is very competitive, and if someone can add a zero in front of his chip's THD spec, he's going to win most design competitions. This goes on until there's only one guy selling DACs which seems a long ways away.

Then again, a better chip can be sold for a better price. Why would an ordinary manufacturer choose a better, more expensive chip? I presume better is more expensive, as that is just plain logic: if you manufacture something superior in specs to the others, you'll sell it for a better price as one should have more to spare for higher quality. Or am I just wrong? At the other hand indeed, in a competitive market, one wants to gain market share by providing higher quality products in the same price range, but probably also by dumping even cheaper stuff.

The sound chips that they put on motherboards generally get no add on shielding at all. A lot of their noise resistance comes from the fact that they are tiny so they act like poor antennas.

When I plugin my external soundcard, the quality of my internal card drops: it starts to hum. I guess that could be prevented by better shielding or decoupling. Again, that's something manufacturers don't want, as it costs them money and most people do not care.

Creative got spanked in the market place some years back when they tried to compromise sound quality

Definitely good to hear that.

Things are evolving predictably. Once quality gets so high that nobody hears any faults that are due to the chips, things will probably level off.  In many cases (like the Clip) the DAC becomes part of the same chip that carries the CPU.

Then again, integrating it, the cost is even more important to the manufacturer producing that CPU, as not every implementation uses this function extensively and this market is probably even more competitive than the DAC-market.

Much of "The trick"  is getting the mic in the right place and pointing the right direction.

Of course, the art of recording itself is the most important. However, we usually use about 8 microphones and pick one or two pairs that sound best. Recordings are so unpredictable that we really can't know which mic to choose beforehand. However, it is surprising that some mics do chamber music (which is not complex) very well, however, when the complexity rises, they miserably fail when compared to the other mics that are just next to it, so it should not depend on the way of recording I guess. I really don't get which specification can describe this ability to resolve complexity.
Music: sounds arranged such that they construct feelings.

Quality aspects of ADC

Reply #10
For electronics, particularly microelectronics, the materials and production costs are small and decreasing.

While small, there is no reason not to use even cheaper components. Quality of audio is generally not a selling point of motherboard or portable audio devices. (just being capable of outputting sound is a selling point)

For example, every Asus motherboard I've ever bought bragged about its audio on the outside of its box. Some of their product is obviously being sold to the HTPC market.

Quote
For example, I own a Sansa Fuze, which is perfectly capable of driving my headphone to unhealthy high levels. While I accepted that as normal, I recently borrowed a (there the name is again) Creative Zen in the same price class: it distorted the sound because it was not able to drive the headphone at levels above average. That again contrasts with the theory: why would Sandisk use this probably more expensive DAC that is able to drive higher loads?


Sansa may have had much choice about the DAC. Its DAC is on the same Australian Microsystems AS3525 chip along with the CPU and DSP.  Please see Rockbox web site Clip+ hardware information 

The output voltage of a digital player is set by its headphone amp, not its DAC.  The Sansa headphone amp is a very sophisticated design with no oubput coupling capacitors and a very low source impedance. AMS idd a good job on teir AS3525 chip. BTW its CPU can potentially run Windows 7.

The AMS 3525 DAC's specs are:

DAC: 18bit with 94dB SNR (‘A’ weighted)
ADC: 14bit with 82dB SNR (‘A’ weighted)
Sampling Frequency: 8-48kHz
32 gain steps @ 1.5dB and MUTE

Which it meets or beats according to my measurements and listening tests. Basically it preforms as well if not better than a good stereo receiver being driven by a good CD player.

Quote
The sound chips that they put on motherboards generally get no add on shielding at all. A lot of their noise resistance comes from the fact that they are tiny so they act like poor antennas.


When I plug in my external soundcard, the quality of my internal card drops: it starts to hum. I guess that could be prevented by better shielding or decoupling. Again, that's something manufacturers don't want, as it costs them money and most people do not care.


The hum is no doubt due to a ground loop, and is very likely lowering your estimate of its sound quality. The ground loop is your doing so it is up to you to address it.  I'll bet it does not hum when driving just headphones.



Quality aspects of ADC

Reply #11
I'll bet it does not hum when driving just headphones.

...nor when the laptop's power supply is unplugged and it's just running on its battery - @ktf, try it.
"Not sure what the question is, but the answer is probably no."

Quality aspects of ADC

Reply #12
The hum is no doubt due to a ground loop, and is very likely lowering your estimate of its sound quality. The ground loop is your doing so it is up to you to address it.  I'll bet it does not hum when driving just headphones.

That would be strange, as that soundcard is just plugged into my notebook and nothing else, how could there be ground loops? It happens when I plug the line-out of the USB-card into the mic-in of my notebook, with no other cables going to my notebook except the power supply. I came across this when I was trying to test the ADC of my notebook with a signal from the DAC of my USB-card.

I'll bet it does not hum when driving just headphones.

...nor when the laptop's power supply is unplugged and it's just running on its battery - @ktf, try it.

Currently I can't test it.

We finally started digitalizing tapes, the first one went pretty well. Thanks for all responses.
Music: sounds arranged such that they construct feelings.

Quality aspects of ADC

Reply #13
I do a fair amount of recording of bands, orchestras and choirs at State festivals. On site people ask me how I get my recordings to sound so good, so I must be doing something right! ;-)
That sounds like a quote from an audiophile magazine 
You do use special power cables, don't you ?

Quality aspects of ADC

Reply #14
Today I made some recordings which made me wonder again on something else, but pretty closely related.

I usually record at the highest setting of the recording device (96kHz, 24bit in this case) because I thought 96kHz brings in some oversampling-like improvements when resampled, and an off-line resampler would do a better job at brick-walling the frequency response, as when directly recording at 44kHz, the filter has to be causal. However, my harddrive is filling up too quickly to my taste (and SD-cards do too), and I don't like deleting original files when I resample them at my computer. It's just the feeling you're throwing away information...

That made me think, would setting the ADC at 96kHz be better anyway? I guess recording 24-bit is nice for some extra headroom (it the other parts have a SNR of over 96dB, which I'm not really sure of) but is there anything like that in the sample rate? Not that long ago, I thought 96kHz would be some kind of nice oversampling, but all ADC do oversample, so would it bring any improvement?
Music: sounds arranged such that they construct feelings.

Quality aspects of ADC

Reply #15
I do a fair amount of recording of bands, orchestras and choirs at State festivals. On site people ask me how I get my recordings to sound so good, so I must be doing something right! ;-)


That sounds like a quote from an audiophile magazine 


Most audiophile magazine writers would be totally horrified if they knew how most of the recordings they use to judge equipment were made.

Quote
You do use special power cables, don't you ?


Yes, I typically use this one or something very much like it:

25' roll up power cord

Quality aspects of ADC

Reply #16
I usually record at the highest setting of the recording device (96kHz, 24bit in this case) because I thought 96kHz brings in some oversampling-like improvements when resampled, and an off-line resampler would do a better job at brick-walling the frequency response, as when directly recording at 44kHz, the filter has to be causal.


I suppose that it might be possible for a software downsampler to in theory do a better job, but back in the real world even average 44/16 converters pass ABX tests related to sonic transparency.

Quote
However, my harddrive is filling up too quickly to my taste (and SD-cards do too), and I don't like deleting original files when I resample them at my computer. It's just the feeling you're throwing away information...


I think you need to distinguish between information and data. There is no doubt that transcribing audio at higher sample rates increases the amount of data. However gathering as much data as is possible is not  the essence of recording. Gathering useful information is the essence of recording. If the data makes no sense and adds nothing, then it is clearly not relevant information.

For example I have an audio interface that records at 24/192 and has 107 dB dynamic range, and I've verified this in lab tests.  However, if use it to record at 24/192 the resulting recording has only about 80 dB SNR if I measure noise over the full 96 KHz bandpass of the device. To obtain the 107 dB number I have to follow good practice, which is to measure noise over a reasonable bandwidth which is 20-20 KHz.  If I look at the spectral content of the actual recordings I make at the 192 KHz sample rate, most of what I record above 20 KHz is noise, both random and also spurious coherent signals such as harmonics from switchmode power supplies elsewhere in the room.

In short, if I used this audio interface at 24/192 to transcribe LPs or cassette tapes, I'm spinning my wheels by wasting time and disk space with noise that contributes nothing to my enjoyment of the music. Even if I use it to record live music, I'm still spinning my wheels because I know for sure that I can brick wall my recordings @ 20 KHz with no audible change whatsoever.

Quote
That made me think, would setting the ADC at 96kHz be better anyway? I guess recording 24-bit is nice for some extra headroom (it the other parts have a SNR of over 96dB, which I'm not really sure of) but is there anything like that in the sample rate? Not that long ago, I thought 96kHz would be some kind of nice oversampling, but all ADC do oversample, so would it bring any improvement?


You have to ask yourself - are you working for your enjoyment or are you a slave to the excess performance of your equipment?  The current the SOTA of electronics is that we can use reasonably priced electronics to overkill just about anything in audio except for the actual limitations to the sound quality of our recordings which have nothing to do with 24 bits or very high sample rates. We still can't overcome the basic limitations of recording music in rooms with microphones and we still can't overcome the limitations of listening via speaker or headphones.

Quality aspects of ADC

Reply #17
Quote
You do use special power cables, don't you ?


Yes, I typically use this one or something very much like it:

25' roll up power cord


Luckily, the high end listener can make up for that deficiency in recording by using extra-extra special power cables on playback. 

Quality aspects of ADC

Reply #18
Woops, I forgot about this topic. I have set tracking for this topic, but for some reason I don't get notifications.

If the data makes no sense and adds nothing, then it is clearly not relevant information.

(...)

We still can't overcome the basic limitations of recording music in rooms with microphones and we still can't overcome the limitations of listening via speaker or headphones.


Indeed, I guess I should just go back to work then  Last week I did over 3 hours of recording in 4 different places, and it has (again) become clear that placement, acoustics and microphone choice is far, far more important.
Music: sounds arranged such that they construct feelings.

Quality aspects of ADC

Reply #19
It also revealed the soundcard of my laptop is really, really badly build: when I plug-in my USB-soundcard it start to hum at quite high levels, (-50dB or something similar) not really you would expect of a laptop that is priced over 3000 dollar (luckily I got it for much less) but I guess sound isn't a priority on those things.

That's more likely caused by a ground loop than a defect.

Apologies if someone else has already mentioned this.

Quality aspects of ADC

Reply #20
Apologies if someone else has already mentioned this.


I even replied to that.

The hum is no doubt due to a ground loop (...)

That would be strange, as that soundcard is just plugged into my notebook and nothing else, how could there be ground loops? It happens when I plug the line-out of the USB-card into the mic-in of my notebook, with no other cables going to my notebook except the power supply. I came across this when I was trying to test the ADC of my notebook with a signal from the DAC of my USB-card.


I really would like to know how this could be a ground loop (I don't understand how ground loops work ) I still haven't got a change to test it with the laptop on batteries.

Music: sounds arranged such that they construct feelings.

Quality aspects of ADC

Reply #21
In your setup, there are two ground paths between the computer and the "USB-card" - via USB and via the audio connection. The root problem is that you will now have some power supply return current flowing through the audio cable. Running the laptop on battery will not help you with this kind of ground loop. You might get a minor improvement by using different USB and/or audio cables. If the USB thing has the option to power it from a wallwart instead of USB, that could fix it. Otherwise, you'll need to insert an isolator in the audio path.

Quality aspects of ADC

Reply #22
In your setup, there are two ground paths between the computer and the "USB-card" - via USB and via the audio connection. The root problem is that you will now have some power supply return current flowing through the audio cable. Running the laptop on battery will not help you with this kind of ground loop. You might get a minor improvement by using different USB and/or audio cables. If the USB thing has the option to power it from a wallwart instead of USB, that could fix it. Otherwise, you'll need to insert an isolator in the audio path.


All good points. Two others only a little less troubling - a mic in is generally way too sensitive for recording a line-out, and most laptop mic ins are mono.

Quality aspects of ADC

Reply #23
Cleaning and demagnetising the heads, matching the azimuth to the tapes, using the correct Dolby setting, and setting the recording levels correctly are so much more important than all of this I can't begin to comment

ground loops, and interference to the sensitive and maybe not adequately screened* electronics in the DR3 are also killers!

* - we have mobile phones now, and PCs, and ... - don't get them too close. Most cassette decks don't like them at all.

Cheers,
David.

P.S. and for goodness sake rationalise your work flow - three tapes a week with the equipment dragged in from different locations or left in an unlocked office?!
P.P.S. consider how you process and store the results, and make access easy for others when you're finished, and make sure your digital copies don't vanish any time soon.

EDIT Google finds this about your sound card...
http://ixbtlabs.com/articles3/multimedia/c...ound-51-p2.html
...not the best, but not the worst. Probably more than good enough for almost any cassette. Most important: listen critically to the results.

Quality aspects of ADC

Reply #24
Hi all!

Background information (safe to skip  )
Currently I'm preparing a large archival operation: my musical society has a collection of recordings on compact cassettes from 1980 to 1990, about 140, which I think have to be digitalized to protect them from deteriorating. I have found a nice tape deck (Nakamichi DR-3) but it has to stay at the office of the executive committee. A friend of mine has a nice ADC for this job, a quality portable mixing console with USB out, but it cannot stay at the office as it is more or less public and he doesn't trust everyone there. However, it is too much hassle to bring it every time, as I think this operation will cost about a year: 3 recordings a week. So, I guess I'm stuck with my own ADC: a Creative X-Fi Surround 5.1 USB-card, which is not really made for the job, but I can at least leave it at the office, as it isn't that valuable.


According to this published test from an independent reliable source, your USB audio interface performs at a very adequate level for your intended purpose:

Link to iXBT Labs Audio Rightmark Test Results

"
Frequency response (from 40 Hz to 15 kHz), dB   +0.10, -0.08   Very good
Noise level, dB (A)   -96.1   Excellent
Dynamic range, dB (A)   95.9   Excellent
THD, %   0.0046   Very good
THD + Noise, dB (A)   -82.7   Good
IMD + Noise, %   0.0077   Excellent
Stereo crosstalk, dB   -96.1   Excellent
IMD at 10 kHz, %   0.0075   Excellent
General performance       Excellent
"

Quote
On-topic
That friend of mine that offered me his portable mixing-console with ADC argued that with a cheap ADC all little nuances and details would vanish, so I should look for another solution than using my Creative ADC.


Frankly, he is talking out of the back of his neck. ;-)

Quote
That made me think: i'm much more a theoretician than he is, so I was wondering, what makes a high quality ADC? I know the working principle of several ADCs, I can't imagine which link in the chain would be capable of 'erasing details' anyway. How does a non-linearity in an ADC sounds like in practice? And jitter? I guess uncorrelated, random jitter could sound like 'more noise', as it is random. To me it seems a low-quality ADC doesn't add anything but noise and a non-linearity, let alone it could remove things from the signal.

In short: what could a low-quality ADC do with the signal except adding noise? Probably I'm far too short sighted, can anyone enlighten me?


Depends what you call a low quality ADC.  Back at the turn of the millennium, people were still putting ADCs and DACs on PC motherboards and PCI add-on cards whose performance with 16 bit data was essentially equal to 8 bits.  They had dynamic range of less than 50 dB and clearly added audible noise. I even ran into some motherboard DACs that for all the world sounded like they were not only 8 bits or less, but were also undithered.  The original SoundBlaster 16's ADC would switch into 8 bit mode when run full duplex (recording and playing at the same time).

A low quality DAC or ADC will add noise, and if it is undithered it will also add low-level distoriton.  In addition there have been DACs that had no reconstruction filters, and actually output the stairstep waves that some analog advocates still use to characterize all digital equipment.

The DACs in CDROMs and DVD drives for your PC are still pretty poor to this day. Not only are they noisy, but they have simplified reconstruction filters that are either peaky or prematurely roll off the high end. 

Even today some common consumer-grade audio interface cards will have relatively substandard ADCs that not only have poorer (now about 80 dB) dynamic range and may be 3 dB down at 16 Khz which might actually cause audible dulling or certain sounds. I've also encountered DACs with too-small coupling capacitors that rolled off the low frequencies into a headphone load, starting as high as 100 Hz or more.

To summarize, a poor converter can add:

(1) Audible noise
(2) Audible low level distortion
(3) Audibly poor frequency response, particularly but not limited to the high end of the audio band.
(4) Ultrasonic noise that may spread down into the audible range by poor-quality analog amplification.

That all said, there are sonically blameless converter chips that cost less than a dollar as parts, and are found in products costing a few dozen dollars.  For example the CPU System chip in a Sansa Clip has 2 90 dB dynamic range DACs and 2 slightly lesser ADCs , and 2 output capacitorless headphone amps built into it, and costs less than $10 in quantity.  The finished product incorporating it with 4 Gb flash memory, li ion battery, display and case  sells in stores for as little as the $20 range.