VOGONS


The "12-bit" Sound Blaster 16 Myth

Topic actions

Reply 80 of 129, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

That clarifies things, thanks. Most D/A devices have an output filter for just this reason, unless they're always driving enough of an inductive load that they don't have to worry. Roland leaving one out on a piece of their gear (although admittedly further down in their product range) seemed strange.

All hail the Great Capacitor Brand Finder

Reply 81 of 129, by yawetaG

User metadata
Rank Oldbie
Rank
Oldbie
Tiido wrote:

Some of that is lost in cable capacitance (at line stages) and various parts in a typical amplifier. Most amplifers intentionally limit the bandwidth to some extent and gain at the final stage is much lower for ultrasonics than audible range so that more bandwidth is left for where it matters to work against non linearities present. Luckily energy of the ultrasonic stuff is fairly low and at normal listening levels (with typical speakers it isn't even 1W range) there shouldn't be any danger. I'm not sure how damaging it can be to the high freq elements at louder volumes when the levels reach tens of W or even beyond (I'm in no hurry to test hahahaha).

Apparently sending a 24,000-30,000 Hz tone through a bunch of studio monitors at high volume will burn out the tweeters, sometimes including smoke.
E.g.:
https://www.gearslutz.com/board/high-end/1412 … s-esp-adam.html

For consumer-level speakers there are even more limitations, and they are fairly easy to smoke: https://www.gearslutz.com/board/live-sound/11 … ml#post12655369

SirNickity wrote:
rasz_pl wrote:

can you hear them? 😀

... 😦 touche.

You may be able to, but likely not using consumer-level amplifiers and speakers, which tend to have much more muddy, less detailed output than studio equipment and (depending on who made the equipment) may emphasize certain output frequencies more than others (usually bass).
Disclaimer: this is a general statement, and not meant to be specific for the particular Sound Blaster that this thread is mainly about. It is merely based on what I read on Gearslutz.com, which so happens to corroborate with my own observations using a professional Tascam audio-interface and various sound modules (including the very low-fi Kawai K1) vs. my consumer-level PC speakers, consumer-level Sony amplifier, and various modern laptop sound cards.

Reply 82 of 129, by Tiido

User metadata
Rank l33t
Rank
l33t

Intenional 0db signal at such freqs is definitely gonna destroy most drivers, a tweeter is generally not even 10W capable but an amp can easily make in dissipate 100W and things will fail very shortly.
Energy of the high freqs of just music and any aliasing/quantisation/noise shaping artifacts are tens of db down the 0db point and shouldn't be a problem for most speakers. Now I'm curious if there's any pathological sources out there than purposefully inject such high energy signals at high freqs with intent of destruction...
High freq drivers in my speakers are 6W rated, 16ohms with a rheostat to equalize levels according to user preference and equalize freq range due to them having much higher efficiency than the other drivers in the speakers. In worst case the series resistance is 0 so it takes sqrt(6W * 16ohm) ~ 9.7VRMS / 13.85Vpp to get the tweeter to its limits. The same level on the bass driver will produce ((9.7 ^ 2) / 4ohm) ~ 23.52W, far below its rated 150W power handling. I'm not completely sure how to go about calculating the SPL out of it when sensitivity of entire speaker is rated 87db/1W/1m (not sure what all the individual drivers are) but it'll certainly be neighbor disturbing loud and far beyond normal listening levels which I've measured at not reaching even 1Vpp for the most part.

T-04YBSC, a new YMF71x based sound card & Official VOGONS thread about it
Newly made 4MB 60ns 30pin SIMMs ~
mida sa loed ? nagunii aru ei saa 😜

Reply 83 of 129, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

Overcompressed and clipped music can sometimes generate a lot of high frequency content. Certainly a 10khz square wave PWM source is going to make a real mess if it isn't LPFed.

All hail the Great Capacitor Brand Finder

Reply 84 of 129, by Cloudschatze

User metadata
Rank Oldbie
Rank
Oldbie

So, here's a small preview of things to come. These metrics are being compiled into a master spreadsheet that includes specific differentiators, such as the PCB revision and chip types/versions relevant to the testing. The full RMAA results for each card will also be hosted and available in both HTML and SAV format.

TestComp.PNG

Couple of observerations/points, based on just this sample set:

  • The "A" mixer chip versions, as found on the tested CT1770, CT2740_1, and CT2290 cards, appear to result in the better THD and IMD + Noise metrics than the non-"A" mixer chips.
  • As pertains to the three CT2740 cards tested, differences in filtering components are presumed to account for the discrepancies in the upper-register frequency response.
  • The CT2290, despite being the only 2-layer PCB card tested thus-far, produced "better" results than any of the 4-layer boards.
  • Each card tested received a "general performance" rating of "Good."

Reply 85 of 129, by SirNickity

User metadata
Rank Oldbie
Rank
Oldbie

I would consider that to be nearly inaudible bandwidth limiting. It could be better, but -1dB 30-15kHz isn't bad at all. It's certainly flat enough across the audible spectrum for the needs of your average consumer, and around -80dB noise floor is totally fine.

This makes me wonder what all the severe complaints are about. (I would suspect bad mixer settings account for most of it.) I'm also curious how the SB Pro compares -- at least within its limited specifications.

Reply 86 of 129, by Am386DX-40

User metadata
Rank Member
Rank
Member
Cloudschatze wrote:

The "A" mixer chip versions, as found on the tested CT1770, CT2740_1, and CT2290 cards, appear to result in the better THD and IMD + Noise metrics than the non-"A" mixer chips.

You sure it's the mixer chip revision? Take a look at this: My Sound Blaster 16 CT2230 Review

Reply 87 of 129, by Scali

User metadata
Rank l33t
Rank
l33t
Cloudschatze wrote:

Couple of observerations/points, based on just this sample set:

My main observation here is that the CT1770 and CT2290 seem to differ considerably from the others in their frequency response.
Also, the CT1770 seems to have to pay for that by being the only card with considerably worse SNR than any other card.

Another observation is that the CT1740 seems to have completely horrible stereo crosstalk. This is something I know from my SB Pro. If you hard-pan audio to the left, you can actually hear a distorted version of it quite well in the right speaker, and vice versa. Especially annoying with 4ch MOD playing.

Getting back to my earlier remark that 'a really bad SB16 would be around 70 db or worse'... Well, the CT1770 is worst, and at -78.8db it's still better than 70 db, so still able to reproduce more than 12-bit digital audio accurately, in theory that is.

http://scalibq.wordpress.com/just-keeping-it- … ro-programming/

Reply 88 of 129, by SirNickity

User metadata
Rank Oldbie
Rank
Oldbie

-60dB crosstalk is still better than vinyl and tape, IIRC. By that time we had transitioned to CD pretty thoroughly, but still. If you are hearing a single channel in a MOD playing on one channel, a ghost of that sound playing in the right channel at -60 would be inconsequential. Maybe the Pro was worse?

Reply 89 of 129, by Scali

User metadata
Rank l33t
Rank
l33t
SirNickity wrote:

-60dB crosstalk is still better than vinyl and tape, IIRC.

Depends on how it manifests itself I suppose. Vinyl basically uses a 'hack' to create stereo, so the separation is never going to be great, but at the same time it is a very 'natural' sounding signal.
In the case of the SB Pro it actually distorts too, so it's annoying.

SirNickity wrote:

If you are hearing a single channel in a MOD playing on one channel, a ghost of that sound playing in the right channel at -60 would be inconsequential.

A distorted ghost is worse than a clear ghost.
Aside from that, it's pretty crap because playing MODs on a real Amiga with 1985 technology doesn't have these limitations.

SirNickity wrote:

Maybe the Pro was worse?

Probably. But as I said before, the best way to judge these cards is by playing reference material on them.

http://scalibq.wordpress.com/just-keeping-it- … ro-programming/

Reply 90 of 129, by rasz_pl

User metadata
Rank l33t
Rank
l33t
Cloudschatze wrote:

So, here's a small preview of things to come.

are those stock setting, or with massaged mixer?

Open Source AT&T Globalyst/NCR/FIC 486-GAC-2 proprietary Cache Module reproduction

Reply 91 of 129, by Cloudschatze

User metadata
Rank Oldbie
Rank
Oldbie
SirNickity wrote:

I'm also curious how the SB Pro compares -- at least within its limited specifications.

Despite using an appropriate 8-bit, 44kHz (mono) test WAV, the figures that RMAA is reporting for the SBPro2's noise level and dynamic range are nonsensical. I'll have to revisit this, as I'm either doing something wrong that I have yet to account for, or the software is simply unable to interpret the data correctly.

On an unrelated note, I figured I'd check to see how a treble value of 0 (-14dB) on the SB16 compares to the default 3.2kHz cutoff filter of the SBPro/Pro2.

Sound Blaster Pro 2, 3.2kHz cutoff filter - 8-bits per sample MONO 44.100 kHz PCM
SBP_32f.png

Sound Blaster 16, -14dB Treble - 8-bits per sample MONO 44.100 kHz PCM
SB16_TRf.png

Comparison recordings of the above:

Gotta love the "approximate" 44.1kHz playback rate of the Sound Blaster Pro 2 example...

Unsurprisingly, the filtering effect of the SB16's treble reduction isn't nearly as pronounced as the Pro's low-pass filter. At the same time, Creative's suggestion of reducing the treble to mask some of the 8-bit playback issues seems slightly less lame. That is, it's still lame, but more effective than doing nothing at all.

Am386DX-40 wrote:

You sure it's the mixer chip revision? Take a look at this: My Sound Blaster 16 CT2230 Review

Of the cards represented, only the CT2290 bears a CT1703 CODEC. The commonality in this sample set, as relates to the THD/IMD metrics, is an "A" mixer chip (CT1745A, and CT1745A-S). Preliminarily speaking, dash-following suffixes don't appear to be particularly meaningful (e.g. an "-S"), but a larger set of metrics and comparisions should help clarify that.

James' observation of mixer channel settings of -2dB producing the optimal SNR is pretty spot-on, and is reflected in what I'd arrived at for the tested mixer configuration:

/MA:240;240;50
/VO:240;240;50
/MI:0;0;50
/CD:0;0;50
/LI:0;0;50
/MIC:0
/SP:0
/TR:128;128;50
/BA:128;128;50
/IPL:MIC- CDR- LIR- LIL- MIR- MIL-
/IPR:MIC- CDR- LIR- LIL- MIR- MIL-
/OPS:MIC- CDR- CDL- LIR-
/AGC:-
/IPG:1,1
/OPG:1,1
/SE:- (where applicable)

Scali wrote:

Another observation is that the CT1740 seems to have completely horrible stereo crosstalk.

Where the tested CT1740 is a pretty early version (rev. 3 PCB, CT1745 mixer), I suspect that same-model cards with later PCB designs and/or a CT1745A may fare better in this regard. I have a couple of additional CT1730/1740 cards to test yet.

rasz_pl wrote:

are those stock setting, or with massaged mixer?

Answered above. Recall that I'm trying to establish an "optimal" baseline.

Reply 92 of 129, by Scali

User metadata
Rank l33t
Rank
l33t
Cloudschatze wrote:
Comparison recordings of the above: […]
Show full quote

Comparison recordings of the above:

I think especially that SB Pro 2 recording sums up just how much Creative ripped off their customers.
I mean, even back in 1991, this was a completely horrible reproduction of a 44.1 kHz 8-bit mono PCM file.
And considering that the SB Pro 2 was the flagship card at the time (with a pricetag to boot), there just are no words to describe this.

Heck, I've heard more hi-fi quality from a C64, which isn't even supposed to be able to play PCM data at all... But just listen to the clarity here:
https://youtu.be/bIM4p0uL6Pw?t=452

http://scalibq.wordpress.com/just-keeping-it- … ro-programming/

Reply 93 of 129, by SirNickity

User metadata
Rank Oldbie
Rank
Oldbie
Scali wrote:

A distorted ghost is worse than a clear ghost.

Figured I would give this a shot, so I opened up my DAW, used an instance of Massive to create a sine wave, then split that into two channels. One panned hard left at +0dB, the other going through a guitar amp and cab simulator with the Drive knob turned up to 11, level-matched and then panned hard right at -60dB. Now, I'm in an office with a few routers and switches powered on in a rack, and using iPhone earbuds, so not exactly a sound booth with studio monitors. But if I mute or unmute the -60 track, I can't tell the difference. If I mute the 0dB track and solo the -60 track, it's a hearing test, but it's there.

Thinking back to when I had an SB Pro as my first sound card, I was using a set of Koss speakers powered from 4x C-cell batteries or a Radio Shack wall wart, we were still watching a 13" Sony TV from the mid 80s with a mono VCR, and I was the only kid I knew with a portable CD player. So, if the Pro could perform even close to the noise and crosstalk specs of the SB 16, it would have been pretty decent for the time. I doubt it does, though. I'm sure it's worse. 😉

Scali wrote:

Aside from that, it's pretty crap because playing MODs on a real Amiga with 1985 technology doesn't have these limitations.

No, it has its own set of limitations. 🤣 Seriously though, wasn't it designed by Ensoniq? It was basically a cost-reduced sampler with 4-channel independent variable sample rates. It could do the "MOD" thing very well, but it's a little like comparing the SNES SPC to the Genesis / MD 2612+76489. They were both better at some things, worse at others. I know a lot of folks put the Amiga up on a pedestal, but as someone without an emotional attachment to that platform, I just don't get it.

Reply 94 of 129, by Cloudschatze

User metadata
Rank Oldbie
Rank
Oldbie
Scali wrote:

I think especially that SB Pro 2 recording sums up just how much Creative ripped off their customers.
I mean, even back in 1991, this was a completely horrible reproduction of a 44.1 kHz 8-bit mono PCM file.

That's with the filter on, mind you, which is the default setting. For higher-rate playback, it's recommended that the filter be completely disabled.

Reply 95 of 129, by SirNickity

User metadata
Rank Oldbie
Rank
Oldbie
Scali wrote:
Cloudschatze wrote:
Comparison recordings of the above: […]
Show full quote

Comparison recordings of the above:

I think especially that SB Pro 2 recording sums up just how much Creative ripped off their customers.
I mean, even back in 1991, this was a completely horrible reproduction of a 44.1 kHz 8-bit mono PCM file.

Obviously, as mentioned, the sample rate isn't terribly accurate. The whole "time constant" scheme is sloppy, but again, given the era and the intended purpose....

Other than that, I hear a bit of a boost in the mids that makes it more brash than the SB16 version. I wonder if that has something to do with the 3.2kHz filter?
The graph seems to show a bit of a hump around 1-3kHz. Too wide-band to be filter resonance. Maybe it combines a high shelf with a lowpass filter to make up some of the "highs" that will still pass through, so it doesn't sound quite as dull.

The hardware manual says that both 3.2kHz and 8.8kHz filters are available during recording, but only the 3.2kHz filter is available for playback. Either can be disabled. 3.2kHz is recommended for 18kHz and below, 8.8kHz for higher. In this case, the filter shouldn't be on -- except of course if you were testing its effect on bandwidth for the purpose of comparing to the Creative Labs suggestion of using the treble cut. 😉

Reply 96 of 129, by Scali

User metadata
Rank l33t
Rank
l33t
SirNickity wrote:

If I mute the 0dB track and solo the -60 track, it's a hearing test, but it's there.

My point was rather that when it's not distorted, and you hear some of the left channel bleed in the right channel, it effectively just makes the stereo image a bit smaller, but is otherwise not an issue (like how vinyl works).
But if you can hear the other channel (which depends on what the stereo separation is, perhaps -60 is still 'too good', SB Pro was likely better, or perhaps my hearing is/was more sensitive), and it is a distorted version, it doesn't just make the stereo image smaller, it also adds well, distortion. This is perceived as annoying/bad quality.

SirNickity wrote:

So, if the Pro could perform even close to the noise and crosstalk specs of the SB 16, it would have been pretty decent for the time.

Why? Because you used crap amp/speakers?
That's not what the SB Pro was aimed it. It was aimed at semi-professional musicians (mine even included the MIDI kit and Voyetra sequencer software).
They would have better amps and speakers, and would expect considerably better, even in 1991, especially at that price.

SirNickity wrote:

Seriously though, wasn't it designed by Ensoniq?

No.
I think you're confused with the C64 SID chip? It was designed by Bob Yannes, who later left Commodore, and was one of the founders of Ensoniq. The SID was indeed a scaled-down version of his original design for a synthesizer.
The Amiga and its Paula chip have no link to Bob Yannes, Ensoniq or even Commodore itself. As you may know, the Amiga was designed by former Atari employees, and Commodore bought it when the system was already in its final stages.

http://scalibq.wordpress.com/just-keeping-it- … ro-programming/

Reply 97 of 129, by Scali

User metadata
Rank l33t
Rank
l33t
Cloudschatze wrote:

That's with the filter on, mind you, which is the default setting. For higher-rate playback, it's recommended that the filter be completely disabled.

I hope it can redeem itself without the filter... but I'm afraid this was not all just filter that we heard.

http://scalibq.wordpress.com/just-keeping-it- … ro-programming/

Reply 98 of 129, by SirNickity

User metadata
Rank Oldbie
Rank
Oldbie
Scali wrote:

But if you can hear the other channel (which depends on what the stereo separation is, perhaps -60 is still 'too good', SB Pro was likely better, or perhaps my hearing is/was more sensitive), and it is a distorted version, it doesn't just make the stereo image smaller, it also adds well, distortion. This is perceived as annoying/bad quality.

Given the sample case here -- either a 4-channel MOD with hard-panned samples, or my sine wave test with channels panned L/R -- stereo separation and equivalent levels of signal and noise are effectively the same. Either way, if the noise -- distorted or perfect -- is -60 down, it's going to be difficult to perceive. Especially in anything less than critical listening conditions. However... again... I doubt the SB Pro is on-par with the -60dB crosstalk figure of the SB 16 above. So it's likely to be a bit worse than this. I'm honestly curious, so I'll just have to try it and see.

Scali wrote:
Why? Because you used crap amp/speakers? That's not what the SB Pro was aimed it. It was aimed at semi-professional musicians (m […]
Show full quote
SirNickity wrote:

So, if the Pro could perform even close to the noise and crosstalk specs of the SB 16, it would have been pretty decent for the time.

Why? Because you used crap amp/speakers?
That's not what the SB Pro was aimed it. It was aimed at semi-professional musicians (mine even included the MIDI kit and Voyetra sequencer software).
They would have better amps and speakers, and would expect considerably better, even in 1991, especially at that price.

Easy, tiger. 😀 You seem to be passionately against Creative, or at least the SB Pro. Fair enough. I'm just trying to find out how bad the situation ACTUALLY is, not the rhetoric that gets thrown around by people who just don't like Creative or had a bad experience with a Sound Blaster at some point. I'm also willing to throw them a bone. Yes there's better stuff out there, even for the time. But I think reading too much into the "Pro" part of the name is probably not wise.

I also got the SB Pro 2.0 kit with Voyetra Sequencer Plus "Pro" (ahem), and the MIDI cables, etc. etc. I think I paid (ok, I was a kid, so read: my dad paid) about $150 for it? Not spare change, but given the crappy old Ad Lib was running around $60-80 at the time.... ehh.

I really think you have to be careful with how much you associate this hardware with professional music equipment. OTOH, even then musicians were still using grainy Akai samplers. Better stuff existed. Worse stuff existed. It was a mixed bag. I still don't see anyone taking the SB Pro seriously. IMO, it was a gaming card, despite all the marketing claims on the box telling you that you could be the next Herbie Hancock with the included software bundle. It's not a reference DAC, but it could play those 8kHz sound clips from games on 3.5" floppies just fine. It also does OPL2 in all its crappy honky-blurby glory. It wasn't until the EMU8K that you could even kind of look at a Creative Labs card as a professional instrument. Even that's a stretch though. But it was cheaper than an Emulator or a Fairlight.

Scali wrote:

I think you're confused with the C64 SID chip? It was designed by Bob Yannes, who later left Commodore, and was one of the founders of Ensoniq. The SID was indeed a scaled-down version of his original design for a synthesizer.
The Amiga and its Paula chip have no link to Bob Yannes, Ensoniq or even Commodore itself. As you may know, the Amiga was designed by former Atari employees, and Commodore bought it when the system was already in its final stages.

Not confused. I didn't even remember that MOS had anything to do with Ensoniq. I just thought there was some lineage there on the Paula. I dunno where that came from. I don't know a ton about it because I've never really found the Amiga to be that interesting.

Reply 99 of 129, by Scali

User metadata
Rank l33t
Rank
l33t
SirNickity wrote:

But I think reading too much into the "Pro" part of the name is probably not wise.

I wouldn't make too many assumptions about me if I were you.

SirNickity wrote:

I also got the SB Pro 2.0 kit with Voyetra Sequencer Plus "Pro" (ahem), and the MIDI cables, etc. etc. I think I paid (ok, I was a kid, so read: my dad paid) about $150 for it? Not spare change, but given the crappy old Ad Lib was running around $60-80 at the time.... ehh.

As I said before, I paid 600 guilders for the SB Pro 2, and not much later, I bought an Amiga 600, also for 600 guilders.
600 guilders for just a sound card was a pretty insane price. I think the SB 2.0 cost about half that, which was still expensive as hell for what it was.

I was/am a musician, and I bought both for recording/editing music. I wanted the SB Pro for the MIDI and the fact that it could do stereo.
I wanted the Amiga because I wanted to use the real ProTracker on a real Amiga, since PC trackers at the time were horrible (this was before Scream Tracker 3 and FastTracker II).

I can say I was horribly disappointed by the sound quality of the SB Pro, and even more so once I got my Amiga 600.
I got a GUS later, which made up for it. Friends bought SB16s and AWE32s... which, even though they weren't quite as bad as the SB Pro, still couldn't hold a candle to my GUS or the PAS16 that a friend had.
And both the GUS and the PAS16 were cheaper than the SB16, so what excuse does Creative have really? You're completely missing the point, if you're just brushing it all off as "yea, but it was made for games".
Creative were basically criminals. They overcharged their customers, and used dirty tricks to screw the competition over (such as the AdLib Gold story).

http://scalibq.wordpress.com/just-keeping-it- … ro-programming/