...and really crappy filtering.

Yes, the DAC is 16b, but the output is noisy as hell and most of the high end is simply missing. Even though it's not technically 12b, those upper 4b of information are simply filtered away, making it little better than if it had been 12b. Still not correct to call it 12b, but I can see where that's coming from. That's not anti-Creative angst, just how the stuff sounds. Creative won the commercial war, but that doesn't mean it was technically superior or even at parity. They were 'good enough' though and combined with attractive pricing and successful marketing they managed to become the default that everyone supported. But that still doesn't make the SB16, least of all the first CT1740, a really good card.

Context: I recently did a multi sound card build. GUS, PAS16, SB16, SBPro clone (Aztech Sound Galaxy NX Pro) and MQ. Haven't been able to get it all working at same time (the PAS16 and NXPro really don't like each other), but the SB16 works in every combination so have been able to listen to it head-to-head with the others. I have a dodgy CT2950, but perfectly good CT1740, CT1750 and CT2910. Even the CT2910, hailed as one of the best SB16 cards, is completely outclassed by the PAS16 and in terms of sound (if not noise) fails to surpass the NX Pro convincingly.

dionb wrote:

...and really crappy filtering.

Rich Heimlich has mentioned a number of times that the early SB16s have no filtering at all. I don't know what he's specifically referring to, since the CODEC itself has filtering affecting PCM operations. In any event, the original SB16s have a measured frequency response of 20Hz - 20kHz, compared to a measured 20Hz - 17kHz response in later designs, which presumes that additional filtering was added at some point after-the-fact.

Yes, the DAC is 16b, but the output is noisy as hell and most of the high end is simply missing. Even though it's not technically 12b, those upper 4b of information are simply filtered away, making it little better than if it had been 12b.

Do you have any objective information or measurements to support these statements? Anecdotally speaking, your experience hardly matches my own, especially as concerns the output being "noisy as hell." How do you have the mixer configured? Is the output amplifier enabled or disabled? These things matter, not just regarding the SB16, but any soundcard.

But that still doesn't make the SB16, least of all the first CT1740, a really good card.

That isn't the point, or the intent. Rather, I'm objectively refuting some of the misinformation regarding the SB16.

It's just 16bit with at best +-1LSD precision (but in reality one digit less), due to the reasons @dionb already mentioned.
Good sound cards (i.e. that are of mastering quality) cost several hundred dollars and use at least 20bit DACs. For this reasons RME along with a few other companies is practically unknown in the consumer market. I am quite sure that any RME sound card declasses every Creative sound card ever made in regards of sound quality.

retardware wrote:

I am quite sure that any RME sound card declasses every Creative sound card ever made in regards of sound quality.

Again, not the point; several contemporaries of the SB16 outclass it in sound quality.

The initial driver disks installing with default sound parameters to treble down and up bass probably doesn't help either.

leileilol wrote:

The initial driver disks installing with default sound parameters to treble down and up bass probably doesn't help either.

This wasn't done with the earliest driver set, but one released shortly thereafter. Where Creative's answer to the noise/popping during single-cycle DMA playback was to have the customer turn the treble completely down, it stands to reason that this was a (really poor) attempt to prevent/address those complaints as part of a default installation.

Are there any better codecs that could be used in place of the original?

Cloudschatze wrote:

dionb wrote:

But that still doesn't make the SB16, least of all the first CT1740, a really good card.

That isn't the point, or the intent. Rather, I'm objectively refuting some of the misinformation regarding the SB16.

I strongly suspect quite a lot of the sound card myths that live on would be easy to dispel if people actually read documentation and bothered using some professional equipment (such as a scope) to check what is being output instead of relying on their ears only.

I'm also pretty sure there are plenty of musicians who would disagree with claims that only equipment with relatively recent high-end components can produce decent sound... 😎

You can play a succession of tones that each use one extra bit and see that even the ones using just one or two bits produce detectable output on a spectrogram, despite all the filtering and massive amount of noise many SB16 cards tend to have.

Tiido wrote:

...massive amount of noise many SB16 cards tend to have.

Speaking of hyperbolic, often misrepresented statements regarding the SB16... 😀

I am curious about the now second mention of excessive filtering though. This is not a common complaint. What cutoff or rolloff has been observed during 44.1kHz PCM playback, and with which specific SB16 models and configurations?

I have that very same SB16 in the photo in first post and have had many others and only ones that I have found not to be very noisy are Vibra16 and some AWE64 based cards.
Once bass and highs sliders are set to mid positions the noise levels become awful, it is like a white noise generator is turned on and you can hear the hiss across to room (but I am using a large HiFi for sound rather than typical tiny desktop speakers) while most other cards I have access to are much queieter or actually silent.
Main source of the noise seems to be the external analog circuits, replacing the opamps in the less integrated cards will bring down noise levels considerably, in more integrated cards like Vibra and AWE64 cards there's nothing much to do but they don't really have a problem anyway.

As far as muffledness goes, with most sources (games) it is like someone turned the last few bands on my EQ all the way down and then some more. This is done by the brickwall digital filter on DAC path, some love it, some not, I'm in the latter camp. External filtering doesn't seem to play a very big role. I don't use SB16 for anything but games so this is a big drawback for my personal tastes. For general ISA based sound purposes there's WSS based stuff that is much better on all regards for that, including 48KHz sources support which none of the Creative's ISA cards seem to support. PCI cards tend to be locked to 44 or 48KHz, rarely having capability to do both (VIA Envy24 based stuff is one notable exception, having clock sources for both types).

The cards I have at hand right now : CT1740, CT2290, CT4180 and CT4520. I used to have a handful of other models but they have been given to friends over the years. Last time I did any serious measurements were over 10 years ago, the results have been lost. I only remember the cards all being worse on all regards compared to a YMF71x card that was used as a reference. I used CT2290 as a secondary card for a long time, partially because it had an IDE interface I used for a CD drive and its integrated OPL3 uses a Philips I2S DAC rather than the noisier YAC512.

The Best Soundblaster Card ever in terms of noise are the CT4170.

CT4170 and a good AT/ATX Supply and you get a nice Headphone enjoyable Sound.

Tiido wrote:

Once bass and highs sliders are set to mid positions the noise levels become awful, it is like a white noise generator is turned on and you can hear the hiss across to room...

Conversely, with the output amplifier disabled (or use of the line-out otherwise), and with optimized mixer settings, I find the noise floor of the SB16 and AWE32 cards to best that of several, similarly-configured contemporaries. That's not to say that there is, or should be, some wall-of-noise without those changes though - Keyboard Magazine ran a battery of tests against a CT1750 using an out-of-box configuration, with comparably acceptable results.

Even when focusing on just the CT17xx line, I'm not sure what explains the discrepancy between our relative experiences.

As far as muffledness goes, with most sources (games) it is like someone turned the last few bands on my EQ all the way down and then some more.

I can definitely tell a difference in the playback of "CD-quality" PCM between some earlier and later SB16s, where the low-pass filtering of ~17kHz is assumed to apply to the latter. Beyond that though, I don't know that I've ever done a gaming comparison. For most 8-bit material, I would think the additional filtering might be welcome.

matze79 wrote:

The Best Soundblaster Card ever in terms of noise are the CT4170.

Noise qualities notwithstanding, the CT4170 uses 8-bit DMA transfers for 16-bit PCM playback, causing issues for some DOS software titles expecting to find and use a 16-bit DMA channel for 16-bit playback.

I always have the speaker amplifiers disabled, either via jumpers or via hardware mods if that option isn't given. The most common TEA2025 is quite noisy on its own and the PSRR is poor aswell. SB cards don't seem to have any consistency in the opamps used and a variety of them are used which could explain some of the performance differences. All of the cards use various low end parts, some less bad than others.

It could also be that the absolute noise level may not be significantly different from other cards, but noise spectrum makes it more offensive. But this is something I haven't investigated nor really plan to.

Filteringwise I never compared the cards at 44KHz sources but games which generally do not do it. 44KHz sources would mainly be music etc. in Windows environment where I no longer use SB16 for anything especially due to lack of 48KHz support. It is also something I no longer plan to investigate, sooner or later I'll have my own SB16 (among other standards) compatible thing made where there are none of the problems.

Cloudschatze wrote:

Can we put this to rest? The soundcard wars ended decades ago.

You started this thread asking to put this "myth" to rest, but other than showing that a 16-bit CODEC was present on the card, provide no supporting evidence that this product line actually delivers 16-bit-quality sound. That would be the way to put the claim to rest. If one cares to waste the time to measure the output and show that, for each of the 65536 input values, a unique, stable voltage level is produced in a repeatable manner, then we will have irrefutable proof of the truly 16-bit pedigree of these cards. Anything less is just as debatable as the 12-bit claim itself.

Personally, I was quite disappointed when I switched to a Sound Blaster 16 Basic at the time. The very first time I ran the Creative demo software, I understood that I had compromised on sound quality to gain software support. Even at only 12-bit resolution, the Ad Lib Gold it had replaced produced far less noise and crackle when playing back digital audio. Yes, the card was properly configured, and the amplifier disabled via the on-board jumper intended for that purpose; it just wasn't particularly good. Part of the problem with the SB16 line seems to be the large number of variants and varying quality of parts used through the years.

All of the reviews I read at the time were in agreement: the SB16, like all other consumer sound cards of the time, did not give full, flat 20 Hz to 20 kHz frequency response. To get that, one had to get into RAP-10 or Multisound territory, and those were hardly gaming cards.

640K!enough wrote:

You started this thread asking to put this "myth" to rest, but other than showing that a 16-bit CODEC was present on the card...

Yes, and as far as 16-bit capability is concerned, it's case-closed.

...provide no supporting evidence that this product line actually delivers 16-bit-quality sound. That would be the way to put the claim to rest.

Published evidence already exists. Unlike any "12-bit" sound claim, for which I've yet to see a single review, measurement, or reliable statement in support of, there are at least two sets of mostly-favorable data points available through even the non-optimal testing performed by both Keyboard and PC Magazine. Wouldn't it make more sense for someone making a completely unrepresented claim to prove their point instead? I certainly think so.

If one cares to waste the time to measure the output and show that, for each of the 65536 input values, a unique, stable voltage level is produced in a repeatable manner, then we will have irrefutable proof of the truly 16-bit pedigree of these cards.

Tiido's initial response suggests that he performed such a bit test.

All of the reviews I read at the time were in agreement: the SB16, like all other consumer sound cards of the time, did not give full, flat 20 Hz to 20 kHz frequency response. To get that, one had to get into RAP-10 or Multisound territory, and those were hardly gaming cards.

Are you suggesting that a 20Hz - 20kHz frequency response is required to claim 16-bit resolution?

In any event, the Keyboard Magazine results for the CT1750 show a +1.11/-2.04dB, 20Hz - 20kHz frequency response. The RAP-10 measured in at +0.37/-3.00dB, 20Hz - 17.1kHz, for what it's worth.

Cloudschatze wrote:

there are at least two sets of mostly-favorable data points available through even the non-optimal testing performed by both Keyboard and PC Magazine

did both magazines publish full page (or sometimes even two pagers) of Creative Ads by any chance? this is a rhetorical question 😀

https://books.google.pl/books?id=AzwEAAAAMBAJ … 0varied&f=false
"quality varied wildly ... 44KHz 16bit was very distorted", blamed by Creative engineers on SCSI card 😀

rasz_pl wrote:

https://books.google.pl/books?id=AzwEAAAAMBAJ … 0varied&f=false
"quality varied wildly ... 44KHz 16bit was very distorted", blamed by Creative engineers on SCSI card 😀

They apparently ran into an issue that affected the playback of uncompressed, CD-quality PCM. A storage subsystem unable to keep up with the playback/buffer rate could explain this behavior. The same review also refers to "high quality sound," and rates the audio quality as being "good" despite the technical troubles encountered. Absolutely nothing about 12-bit sound anywhere.

Review said distorted, not pausing/choppy. 200KB/s is not too much to ask from a HDD in 1993. Not to mention same thing happens if you try playing midi at the same time.