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: Benefits of oversampling (Read 7052 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Benefits of oversampling

Hi !
Thanx for all of you to help me to choose a sound card !! 

Just another question :
Do all sound cards have oversampling technique (to avoid aliasing)?
I think it could be finaly good to choose a 16/48 or 20/48 with 128 oversampling rather a 24/96 which is not necessary better in terms of quality !
I mean it should be better to use16/20b with 64/128x oversampl. than a "standard" 24 b converter without oversampling at all :
  what do you think about it ?

thanks
best regards,
Nick

Benefits of oversampling

Reply #1 –
I think all cards use oversampling today, no matter if they are 16 or 24 bit. Well, maybe some very crappy ones don't, it's a possibilty.

Benefits of oversampling

Reply #2 –
There's really no reason not to oversample anymore, including cost. Oversampling is a means to relaxing final anti-aliasing filtering in the analog domain. The latter is more difficult, and more expensive. Oversampling in the digital domain reduces the final analog filter order drastically, improving upper frequency phase response in the process. The days of brick-wall filtering are over, and this applies to sound cards at all price points.

Benefits of oversampling

Reply #3 –
Yes, just one thing, oversampling filters are supposed to be still brickwall filters, but of digital type instead.

Benefits of oversampling

Reply #4 –
Of course - the brickwall I was referring to was the post-DAC analog smoothing filter. The digital side of the equation involves a high-order decimated FIR filter pair to move Nyquist (half the oversampling rate) out to a more manageable frequency band for the analog back-end.

I remember the good old days when the difference between a base-sampling player and a 4X oversampling player was night and day as far as audible artifacts were concerned. I once bought a Philips player advertized as 4X oversampling and returned it the same evening insisting it was in fact a 1X player. After a lot of amateurish posturing by the store regarding my lack of audio expertise and listening experience, and a call to the service depot to verify the chipset in the player, there were a few red faces on the salestaff. Life goes on...

Benefits of oversampling

Reply #5 –
Thanks for yours answers !!
but just one thing:
After reading technical specifications (terratec for example), it's not mentionned "128x oversampling" for all sound cards ,just for the EWS88...
Compare EWS88 and EWX24/96.
I guess the second don't use this kind of technique...
If i'm right the sound quality should be not the same between the two cards

Benefits of oversampling

Reply #6 –
It's quite possible, even probable, that both cards use the same converters (codecs), and they're playing a marketing angle on the two cards. Even if the second card only oversamples at 64x, the filter designs for either are elementary from a theoretical electronics perspective. What will set the two cards apart are power supply noise rejection and component & layout issues affecting the transfer function of the analog filters. What sets high-end cards apart is usually not the degree of oversampling, but rather how well the front/back end filtering is implemented both electronically (component selection) and physically (layout). The higher priced card is not necessarily the more transparent card. Unfortunately, this kind of information is never part of a marketing effort (and for good reason)...

Benefits of oversampling

Reply #7 –
FYI:

"The AudioSystem EWX 24/96 is based on the EWS88 audiocard, featured in CAW back in Summer 1999, but boasts the latest 24bit, 96kHz oversampling chipset for..."

http://www.computeraudio.co.uk/caw/cawrevi...tecewx2496.html

Benefits of oversampling

Reply #8 –
Be a little careful about the specs here:  typcially 128x "oversampling" means that the card uses a 1-bit ("Sigma Delta") converter.  Said another way, the DAC (Digital to Analog Converter) has 1 bit of resolution running at 128x oversampling rate.  This inherently bad resolution (1 bit) can perform quite well in the audio band with some advanced techniques like high-order noise shaping to push the quantization noise (a 1 bit = 6 dB singnal to noise ratio).  Philips called this technique "bitstream conversion" and other manufacturers have marketing names for the same technique.

Technically this can work well in the audio band, but it does have lower resolution and much higher noise floor in the highest octaves.  Should not be an audible problem though, and the inherent amplitude linearity of a 1-bit converter reduces distortion at low amplitudes (below -60dB).

The other common DAC type is a "ladder" or "R2" type of converter.  This typically runs at 4-16x oversampling, but since it is just a straightforward conversion of every sample it can achieve higher S/N ratios (>110 dB) without noise shaping.  This is a more expensive DAC to build since the tolerances of the resistors in the "ladder" become extremely difficult to maintain in production.  Many high-end pieces of equipment use ladder DACs due to the lower noise levels at higher frequencies and the belief in "better sound."

There are a large number of factors that will go into the final analog output signal's performance, this being only one (albeit an important one).  Analog circuits, board layout, shielding, etc. can all play equally important roles. 

There are expensive and cheap versions of both converters, but my main point is that "oversampling" by itself does not mean an apples to apples comparison.  Different "orders" of noise shapers can affect performance - all else being equal a higher order noise shaper in a 1-bit converter will result in better audio band noise performance at the same "oversampling" rate, so even between 1-bit converters a higher oversampling rate is not necessarily better.

I recommend that you look at the final analog audio performance specifications more than the chipset "features" like oversampling.  And be careful about the specs too - there are a lot of unsubstiantiated bogus claims on paper out there, especially with the cheaper, consumer cards.  For example, if a card really had 24 bits of resolution then it would have a signal to noise ratio of 144dB.  I haven't seen anything even close to that, so typically "24 bits" means the card can process 24 bit data and the DAC will accept and try to convert it, not that it really has 24 bits of resolution.  Heck, 18 bits should be 108dB which would be an amazing sound card.  I would take a "true" 18 bit card over a bogus 24 bit card any day.
Was that a 1 or a 0?

Benefits of oversampling

Reply #9 –
Holly Cow !!
What an explanation !
thank you very much DigitalMan !!


Philips called this technique "bitstream conversion" and other manufacturers have marketing names for the same technique.

Is there a sound card that use this technique ?
Does this has something to do with "I2S" D/A chip ?


I would take a "true" 18 bit card over a bogus 24 bit card any day.

Do you mean turtle beach and creative ? 

Benefits of oversampling

Reply #10 –
Most sigma-delta converters now are multi-bit...

Benefits of oversampling

Reply #11 –
DigitalMan, you are discussing an issue which you are not well-versed in, and you have made a few clearly misleading and blatantly false statements. For instance:

"Many high-end pieces of equipment use ladder DACs due to the lower noise levels at higher frequencies and the belief in "better sound"."

"typcially 128x "oversampling" means that the card uses a 1-bit ("Sigma Delta") converter."

You are somewhere between 5 and 10 years behind the current technology...

Benefits of oversampling

Reply #12 –
Yes, although I guess I'm not as knowedgeable in this area as you, that's what I thought, that nowadays most converters were multibit sigma-delta. I'd say that pure multibit is rarely used today.

Benefits of oversampling

Reply #13 –
I'm certainly not trying to come off as an expert, but 20 years in the industry in a design capacity has taught me a few things about both the technology and associated marketing madness. Most converters these days use multi-bit sigma-delta modulation, and shouldn't even be mentioned in the same sentence as native bit resolution conversion...

Benefits of oversampling

Reply #14 –
Quote
DigitalMan, you are discussing an issue which you are not well-versed in, and you have made a few clearly misleading and blatantly false statements. For instance:

"Many high-end pieces of equipment use ladder DACs due to the lower noise levels at higher frequencies and the belief in "better sound"."

"typcially 128x "oversampling" means that the card uses a 1-bit ("Sigma Delta") converter."

You are somewhere between 5 and 10 years behind the current technology...


@ F1Sushi
Hmmm.  Not well versed?  I've done my share of engineering for high volume audio equipment.  I have over 2 million units in the field from my designs, so I suggest you keep the insults to a minimum.  How about you contribute facts instead of attacking me personally?  Perhaps you missed the point and will be providing a more helpful post for Nick Jr. III soon; I look forward to it.

In trying to keep the information in one post to someone trying to understand the basics of audio technology I believe that the information is essentially correct, not "blatantly false."  The point was not a detailed technical text of all versions/variations of DAC technology but rather an introductory illustration of how the marketing spins on different technologies make comparisons difficult.  Would you like to contribute some useful information on clock jitter so that everyone can better choose their sound cards?

Certainly hybrid, multi-bit sig/delt are very prevalent today.  I don't believe that the fact that many new converters are hybrid changes the description of the tradeoffs in the underlying technologies and the recommendation to focus on the final audio performance.
Was that a 1 or a 0?

Benefits of oversampling

Reply #15 –
I don't attack people...but I do tend to call a spade a spade:

1. "Perhaps you missed the point and will be providing a more helpful post for Nick Jr. III soon; I look forward to it."

My response to the question was low-tech and quite explanatory. Perhaps you should read it again. No need to look forward to a second post...


2. "Would you like to contribute some useful information on clock jitter so that everyone can better choose their sound cards?"

No. That is beyond the scope of this gentleman's query. However, your comparison of a native bit resolution DAC with the technology relevant to this gentleman's query is out of date. I suggest you familiarize yourself with these two technologies, or else provide more insight on your post which states:

"Many high-end pieces of equipment use ladder DACs due to the lower noise levels at higher frequencies and the belief in "better sound"."


3. "In trying to keep the information in one post to someone trying to understand the basics of audio technology I believe that the information is essentially correct..."

Yet, you specifically stated:

"Be a little careful about the specs here: typcially 128x "oversampling" means that the card uses a 1-bit ("Sigma Delta") converter."

Do you call that a "correct statement"?


4. "I don't believe that the fact that many new converters are hybrid changes the description of the tradeoffs in the underlying technologies and the recommendation to focus on the final audio performance."

I'm sure deep down you're a nice chap, but you're not up to speed on the technology this person is looking to sink his hard-earned cash into...that's what is relevant here.

Benefits of oversampling

Reply #16 –
Quote
I don't attack people...but I do tend to call a spade a spade:

1. "Perhaps you missed the point and will be providing a more helpful post for Nick Jr. III soon; I look forward to it."

My response to the question was low-tech and quite explanatory. Perhaps you should read it again. No need to look forward to a second post...


2. "Would you like to contribute some useful information on clock jitter so that everyone can better choose their sound cards?"

No. That is beyond the scope of this gentleman's query. However, your comparison of a native bit resolution DAC with the technology relevant to this gentleman's query is out of date. I suggest you familiarize yourself with these two technologies, or else provide more insight on your post which states:

"Many high-end pieces of equipment use ladder DACs due to the lower noise levels at higher frequencies and the belief in "better sound"."


3. "In trying to keep the information in one post to someone trying to understand the basics of audio technology I believe that the information is essentially correct..."

Yet, you specifically stated:

"Be a little careful about the specs here: typcially 128x "oversampling" means that the card uses a 1-bit ("Sigma Delta") converter."

Do you call that a "correct statement"?


4. "I don't believe that the fact that many new converters are hybrid changes the description of the tradeoffs in the underlying technologies and the recommendation to focus on the final audio performance."

I'm sure deep down you're a nice chap, but you're not up to speed on the technology this person is looking to sink his hard-earned cash into...that's what is relevant here.

I'm sure you are a wonderful chap F1Sushi...

1) It wasn't the technical content but the personal tone, not very diplomatic, that.  Perhaps a tradeoff in technical and people skills.

2) I am very familiar with the two technologies.  Because sigma delta DACs use high order noise shaping they typically have a noise floor that rises aggressively in the treble region.  As you know, the noise floor must remain constant in power and the power is usually pushed into the higher frequencies where they are less audible and easier to filter.  This is the "higher noise levels at higher frequencies" that I reference.

3) Uh, okay, so multibit sigma delta converter should be specified as more prevalent.  This is not technically correct, you are correct.  Does that invalidate the post and my ability to discuss this technology?  Perhaps it does for some... 

4) I think the recommendation to look beyond chipset features to end audio performance is still a valid help for this person - do you disagree?  True the multibit sig/delt blend the strenghts of the different DAC topologies, and that is an extension of the underlying technologies of multibit and sig/delt technology that were described.
Was that a 1 or a 0?

Benefits of oversampling

Reply #17 –
...and "the recommendation to look beyond chipset features to end audio performance" was precisely the jist of my earlier post:

"What will set the two cards apart are power supply noise rejection and component & layout issues affecting the transfer function of the analog filters. What sets high-end cards apart is usually not the degree of oversampling, but rather how well the front/back end filtering is implemented both electronically (component selection) and physically (layout). The higher priced card is not necessarily the more transparent card. Unfortunately, this kind of information is never part of a marketing effort (and for good reason)..."

As for my people skills...I still think you're a nice chap.

Best,
f1sushi

Benefits of oversampling

Reply #18 –
Quote
...and "the recommendation to look beyond chipset features to end audio performance" was precisely the jist of my earlier post:

"What will set the two cards apart are power supply noise rejection and component & layout issues affecting the transfer function of the analog filters. What sets high-end cards apart is usually not the degree of oversampling, but rather how well the front/back end filtering is implemented both electronically (component selection) and physically (layout). The higher priced card is not necessarily the more transparent card. Unfortunately, this kind of information is never part of a marketing effort (and for good reason)..."

As for my people skills...I still think you're a nice chap.

Best,
f1sushi

Well done.

Agrees with:
Quote
There are a large number of factors that will go into the final analog output signal's performance, this being only one (albeit an important one). Analog circuits, board layout, shielding, etc. can all play equally important roles.


Cheers!
Was that a 1 or a 0?