1212m, 192khz, 22khz FR on Analog?

[MHz]

I'm no sound expert, though I'm looking at this card because I'd like to start exploring. I was looking over the specs and can't understand something. The card is a 192/24 sampler, but the analog input frequency response is 20-22000hz. Is there any point to be sampling a 22,000hz frequency at 192khz? Wouldn't 44.1khz sampling rate do just as well? Thanks for reading!

[Ethan Winer]

Originally Posted by MHz

Is there any point to be sampling a 22,000hz frequency at 192khz? Wouldn't 44.1khz sampling rate do just as well?

No, and Yes, in that order.

Welcome to the world of BS audiophile marketing.

--Ethan

[MHz]

Thank you for the response Ethan. I'm glad I waited to buy that interface. I've been looking all over for audio interfaces that are not overly expensive and all of them that I could find had at most a 22khz top end analog frequency response. Any recommendations on something between $100-$150 that is aimed at analog sampling rather than digital?

[ekb]

Originally Posted by Ethan Winer

No, and Yes, in that order.

Welcome to the world of BS audiophile marketing.

--Ethan

Isn't this really a textbook type of answer were everything is assumed to be ideal? To sample at 44.1 kHz, all content above 22.05 kHz needs to be filtered out in the analog domain, before sampling. That's a real steep analog filter which will have bad effects on the signal well below 22 kHz. On the other hand, if you sample at 192 kHz, you can use a much shallower analog filter starting around 22 Khz which only needs to be way down at 96 kHz. That will have much less effect in the audible range.

Ed

[MHz]

OK, let's take a possible real world example. Say I have a DVD-A outputting analog at 192/24. The audio interface specs say its analog inputs' frequency response is 20-22khz. If I go and sample this 192khz output from the DVD-A at 192khz and then run a FFT, will I see audio data above 22khz or will it be clipped?

[ekb]

Originally Posted by MHz

OK, let's take a possible real world example. Say I have a DVD-A outputting analog at 192/24. The audio interface specs say its analog inputs' frequency response is 20-22khz. If I go and sample this 192khz output from the DVD-A at 192khz and then run a FFT, will I see audio data above 22khz or will it be clipped?

I think that the specs on the card are too vague. If they really mean 20-22kHz then you will not see content above 22 kHz. But maybe they are just saying that it'll be flat out to 22 kHz with decreasing content above that. It depends on the details of how the filtering is done before the sampling.

Ed

[soho54]

The audio interface specs say its analog inputs' frequency response is 20-22khz.

This is the +/- ??db range, not the total operational range.

If I go and sample this 192khz output from the DVD-A at 192khz and then run a FFT, will I see audio data above 22khz or will it be clipped?

If there was +22khz data on the DVD-A it will be present on the FFT. Minus the cards own recording roll-off on top of the disks roll-off, of course.

[yuriv]

If the input signal has any energy from 22 kHz to 96 kHz, you’ll see it in your recording if you set your sample rate to 192 kHz. The specs in the 1212m manual say "±0.05dB 20 Hz – 20 KHz" for the analog line inputs. The -3dB point is a lot higher than 20 kHz. The 1212m is a fine card for $100. It sounds pretty good too. You’d be hard pressed to find anything that performs as well anywhere near that price. The 1212m has been out for a while now and the drivers are mature. The mixing app, called Patchmix, has flexible routing but may be hard for some people to learn.

[MHz]

OK guys, this makes a lot more sense now - I couldn't figure out why they'd build cards with high sampling capability but such low analog FR. There is also an external USB version called the 0202 which is essentially the same guts as the PCI card and I've read in numerous places that external is the way to go due to PC noise...so I think I'll go ahead and pick one up.

[yuriv]

I don’t think the 0202USB performs as well as the 1212m. The converters on the 1212m are better. The 0404USB is closer—I just picked one up and found its performance to be similar to my 1820m. It sounds very good, even as a standalone ADC/DAC. To me it sounds better than an MBox 2. Also I returned an M-Audio FastTrack Pro, which sounded dirty in comparison to the E-mu. The 0404USB drivers are a bit raw right now, and some people still have trouble getting the units to work on their computers. I suspect something similar for the 0202USB. I just got their beta drivers to work on my MacBook Pro.

On my PC, the 0404USB has over 3x the processor usage as the 1820m when it’s playing back or recording. I hope E-mu can improve the performance with future driver or firmware releases. I’m pretty sure the 0202USB has the same problem right now. Another drawback with the E-mu USB boxes is that they don’t come with Patchmix, but some people might think that’s a plus.

[MHz]

I've heard a lot of horror stories of the E-MU USB devices with Macs...thankfully that doesn't affect me since I use a Win32/AMDX2 platform instead. That's interesting about the USB device converters, though I'm tempted towards it so I can share it with my laptop too...decisions decisions...Does anybody have a 0202USB that can give some hands-on advice?

[Ethan Winer]

Originally Posted by MHz

I'm glad I waited to buy that interface. I've been looking all over for audio interfaces that are not overly expensive and all of them that I could find had at most a 22khz top end analog frequency response. Any recommendations on something between $100-$150 that is aimed at analog sampling rather than digital?

I said that a 192 KHz sample rate is silly, but I didn't say the sound card was not good. There's a big difference!

For PCI sound cards I'm a fan of the M-Audio line. I have a Delta 66, which has four ins and outs plus SPDIF. For stereo their Audiophile 2496 is an excellent value costing about half as much as the card you mentioned.

I also have a Presonus Firebox which is an external Firewire card. But it's got six outputs and costs more than a stereo card. Really, these days most sound cards are perfectly fine.

--Ethan

[Ethan Winer]

Ed,

> Isn't this really a textbook type of answer were everything is assumed to be ideal? <

Okay.

> To sample at 44.1 kHz, all content above 22.05 kHz needs to be filtered out in the analog domain, before sampling. <

This is not a problem these days where oversampling is used to simplify the filter requirements.

--Ethan

[hdgeek]

Originally Posted by MHz

OK guys, this makes a lot more sense now - I couldn't figure out why they'd build cards with high sampling capability but such low analog FR. There is also an external USB version called the 0202 which is essentially the same guts as the PCI card and I've read in numerous places that external is the way to go due to PC noise...so I think I'll go ahead and pick one up.

Noise isn't always a problem for the PCI version. I have the 0404 (more forward than the 1212 but that's what I like) and with my headphone amp and AT 900s the background is dead quiet even at max volume. The patch mix is a pain but once you have the settings you need saved for each situation, 96, 44, wave it's simple to change. If you don't use Foobar as a player I'd also recommend that using ASIO output.

[yuriv]

Originally Posted by Ethan Winer

I said that a 192 KHz sample rate is silly, but I didn't say the sound card was not good. There's a big difference!

For PCI sound cards I'm a fan of the M-Audio line. I have a Delta 66, which has four ins and outs plus SPDIF. For stereo their Audiophile 2496 is an excellent value costing about half as much as the card you mentioned.

If you’re recording music, 192k is silly. I rather like my results that come from 24-44 recordings. But if the application is to use the interface as a poor man’s data acquisitions card for a computer-based spectrum analyzer—which is what (I think) is wanted here—then the 192 k sample rate can sometimes come in handy. Didn’t we want to measure the ultrasonic content of an SACD? These new cheapo cards can turn your computer into a reasonably accurate and precise test instrument.

BTW, I also have the Audiophile 2496, which I bought about six years ago. It’s a good card and it still sounds cleaner than some of their newer stuff like the FastTrack Pro USB. But the new cards like the better E-mus and their own 24-192 have noticeably better performance. For example, the noise floor is a good 15-20 dB lower on the 1212m. You can really hear the difference when you crank up the gain on quiet recordings. Not bad for $100 shipped. I still like my old AP2496, but I really have to be careful with the levels (which I do anyway regardless of the card).

[MHz]

Yeah I'm not recording music at 192 (agreed, waste of hdd space), but instead am more interesting in sampling things and analyzing the spectrum data...it would also be interesting to see how many of my SACDs and DVD-As are really sampled as high as they should be, to evaluate amplifiers and players, and whatever else I come up with. I can learn so much just by having a capable sampler to mess around with.

I've decided on the external unit...the ability to share between computers is just too great to pass up. A PCI, while it may be slightly better, ties me to one computer, and PCI is on its way out anyway in favor of PCI-E.

[ekb]

Originally Posted by Ethan Winer

> To sample at 44.1 kHz, all content above 22.05 kHz needs to be filtered out in the analog domain, before sampling. <

This is not a problem these days where oversampling is used to simplify the filter requirements.

OK - I always wondered about that. So if I understand you correctly, the analog is gradually analog filtered, sampled at high rate, filtered further in the digital domain, and then decimated to the 44.1 sampling rate?

Ed

[CharlesJ]

Originally Posted by Ethan Winer

Ed,
> To sample at 44.1 kHz, all content above 22.05 kHz needs to be filtered out in the analog domain, before sampling. <

This is not a problem these days where oversampling is used to simplify the filter requirements.

--Ethan

These days? I remember CD players back in the late 1980s, soon after they came out, being advertised with 2x, 4x and 8x oversampling. Early 1990s it went higher. I am sure they knew the value of oversampling way back then, and just stopped advertising this basic design.

[Ethan Winer]

Ed:

> the analog is gradually analog filtered, sampled at high rate, filtered further in the digital domain, and then decimated to the 44.1 sampling rate? <

I'm not an expert with this stuff, but I believe the sampling is done 4x or 8x or whatever times higher than needed, using filters appropriate for those high sample rates, then every 4th or 8th or whatever sample is kept and all the rest discarded. Or something close to that.

Charles:

> Early 1990s it went higher. I am sure they knew the value of oversampling way back then, and just stopped advertising this basic design. <

Exactly.

--Ethan

[MichaelJHuman]

Did the amount of oversampling depend on the type of DAC? Wasn't there some players using 1-bit (delta-sigma modulation) DACs and others using more conventional DACs?

[MHz]

Thanks for all the input guys! I just received my 0202 and am thrilled with it...Upon sampling a SACD and running an FFT I did see audio data all the way to about 90khz, not bad I think. I always wondered if my receiver passed the super high frequencies through, and now I know. The 0202 USB is a well made, easy to use instrument.

[yuriv]

Originally Posted by MHz

Thanks for all the input guys! I just received my 0202 and am thrilled with it...Upon sampling a SACD and running an FFT I did see audio data all the way to about 90khz, not bad I think. I always wondered if my receiver passed the super high frequencies through, and now I know. The 0202 USB is a well made, easy to use instrument.

Cool. If you’re getting something at 90 kHz, it’s probably just noise. SACD/DSD has rising noise levels outside the audio band due to “noise shaping” as they call it. If you see a sharp drop-off before 96 kHz, it could be close to the corner frequency of the decoder’s LPF. My copy of Pohlmann says that 50 kHz – 100 kHz is typical. So when comparing the performances of SACD players using some kind of test disc, you might get lower S/N figures than CDs if you stick to a high sample rate. DVD-A spectra might look different. At least now you know something about your receiver; it’s not surprising—many power amps can reach 100 kHz, if not higher.

BTW, what CPU usage are you getting in Task Manager when using the 0202? How does it compare to onboard audio?

[MHz]

Yes, the data above 90khz is noise...I did a high pass and then changed the playback rate so that I could actually hear the just data up there and it sounded like crickets ;-) I *did* discover musical data on the SACD, along with some noise, at around 60khz. Anything above that was noise. Thing is, most of my DVD-As are sampled at 24/96, so there wouldn't be much above 48khz...

My system is a Athlon64-X2 @ 4200 on an Asus mobo. Processor use is not even noticable when recording or playing @ 192khz and stays at 0% in task manager. Don't know about the onboard, I never used it...I always used my external firewire video converter to get audio in at 48khz.