A/D/A converters - help me cut through the BS

[bluebird]

No Scott, there wasn't any MOTU's in the test. I get your point.  8) But I would like the opinions being expressed here to have a little more procedural and testable weight behind them. After all music is subjective so I am not claiming to be the absolute truth in this matter.

That said, I did do AB/X tests maybe 9 or 10 years ago with two mastering engineers in a properly treated mastering room between a Digi003 and an Apogee PSX-100. The test was and analog source (tape machine) passively split to the A/D section  of each converter, looped to the D/A section and monitored. The digi003 had to have pro tools running and a loop through setup. The apogee simply needed a cord from the A/D to the D/A section and could run stand alone. The DA out of each converter went into a relay based AB/X box to the monitoring section of the console.

Again no one could tell the difference between the two converters when guesses were tallied. I believe the MOTU box in question is at least of the same build quality as the digi003 so I don't have a problem recommending that box.

I just want to be clear on the test procedures so people can better make a decision about a purchase...I'm just trying to help. Not discredit anyone.

Maybe it was common yesterday to clip an AD but in practice it is still done today.

Definitely, and if your talking clipping, that is a whole other subject and I think the differences are a lot more apparent. That would not really be something for technically sound tests to determine. I have never done a converter clipping sound test and would suppose it would consist of lining up a song clipped by a bunch of different converters in your DAW and just using your ears to pick the one you like. No special AB/X testing required.

I would love to hear descriptions of clipped converters as that would be subjective, and poetry would be a good tool in communicating the results. ;D

 

[seva]

nice that your expression has changed. Honestly, I am a bit tired to justify myself here. Do the tests and let me know what you think. After that we can talk about a DIY 10mhz clock !? ;D

I receive regular mixes of people working with a Motu24i / o. I also once owned the 2408mk3. You should consider a comparison with other converters. I heard good things about the Rosetta, Metric Halo is very good on the mac. If money is not the problem, you can simply order the converter and try it. Traders offer money back guarantee. please do the test with high-quality speakers (it can also be done with a headphoneannd good amp, - you can order the same, but then you can not judge the bass and the stereo image correctly -Sennheiser HD800 is a good choice)

When I was a teenager I had no money to build my studio. I always pushed the most important buying decisions backwards. I did not understand why I should buy good speakers or converters - if I only need to buy a Neve or SSL then I could blah blah too - that was a great waste of time

For me it was very difficult to do comparisons in the beginning because a big setup is needed. I did not have that at the time and had to remember the sound, so to speak. From a psychoacoustic point of view, man is constantly fooled. It still happens to me today that I test a device and make settings that sound great at first time then realize that it was in the bypass all the time - lol - you have to be careful not to be blinded by unconscious prejudice. So, yes - proper ABX is needed for many things. Nevertheless, I had a first impression with every converter. This is similar to many people. Terms as it sounds cold, digital, bass under-emphasized etc are then gladly used.

If you are looking for a converter for analog summing it can be more than looking only for ADDA. The metric halo is very nice because of the internal mixer. Depending on your DAW, there is a significant difference in the sound quality of using this mixer for grouping tracks. The mixer sounds very high resolution and gives your mixes a nice seperated Instrument sound - I think 24channel are not enough for a hole project. I worked with 2x 2882 and a mixer in this way. When I clocked this setup by an Eclipse, my opinion about clocking was formed. It was a big jump in terms of a wider stereo image. In the end I had problems with the Firewire ports and the need to work on the PC. The 2882 sounds a bit cold and I chose the Aurora and 2xRME HDSP AES cards. But that was a few years ago

I would say: room comes first, then speaker, then DA, then AD.
The answer is really simple. I just need much less time to make decisions and I can hear the analog technology properly. I do not need to build a fairchild or pultec if the UAD plugin performs better at the end of the day because I could not adjust the hardware well.

My first advice to clients who deliver bad mixes is to buying a HD800 and building a diy headphone amp.

If the budget allows it, you should think about the acoustics and speakers, then it's about the converters. My clients appreciated this advice very much and I received confirmation when they deliver better mixes. But of course there are believers and unbelievers - test it, hear it, measure it, explane it, DIY it better, let me know 

 

[mobdirt]

Might've been PCM63.

Thanks Andy
I remember tracking the rs7000 through that system and haven't been able to get the same results since, with different convertors..

 

[Matt C]

One thing I wonder about - people regularly claim that nicer converters and especially nicer clocks result in a wider, clearer stereo image.  It seems like it would be reasonably straight forward to directly measure the stereo cross talk as a comparison.  Has anyone tried this?

 

[scott2000]

One thing I wonder about - people regularly claim that nicer converters and especially nicer clocks result in a wider, clearer stereo image.  It seems like it would be reasonably straight forward to directly measure the stereo cross talk as a comparison.  Has anyone tried this?

This guy tested an 002 to a Mytek 192.....

https://www.gearslutz.com/board/gear-shoot-outs-sound-file-comparisons-audio-tests/175573-mytek-converter-v-s-digi-002-converter-b-test-here.html

I didn't check the gain and he didn't mention it so who knows if it's even reliable....I'm guessing not......Has to be some kind of variable....

Anyhow, I hear a difference..... I think this is where a lot of the confusion comes from is posts like that and from everyone's subjective opinions......

I really think it comes down to A/Bing something to really make that call even with test numbers.....

I have an old Tango converter I still use for stuff..... I did swap some smd caps for electrolytics in some areas but, it has a great sound to me....

 

[living sounds]

One thing I wonder about - people regularly claim that nicer converters and especially nicer clocks result in a wider, clearer stereo image.  It seems like it would be reasonably straight forward to directly measure the stereo cross talk as a comparison.  Has anyone tried this?

I don't think anyone can actually hear a difference between crosstalk of -70dB  and of -120dB. Hearing limitations, masking, room acoustics, speaker differences etc. should make this irrelevant.

But clocking can add components to the signal that enhance transients, certain parts of the spectrum, distortion etc. that will make fine details and thus room information more audible. It does not mean the signal is actually truer to the source, on the contrary.  Let's call it "pseudo detail".

I've had this experience with clocking, as well as with swapping op amps. "Pseudo detail" is to be avoided at all cost IMO, it will mess up a signal permanently.


This nonewithstanding, a clock that is actually more precise will provide better definition and and a more realistic representation of the room.  But it's not a given that this is what people are actually listening to.

 

[abbey road d enfer]

I don't think anyone can actually hear a difference between crosstalk of -70dB  and of -120dB. Hearing limitations, masking, room acoustics, speaker differences etc. should make this irrelevant.

I think that what people hear, or think they hear, is hard to translate into common language. Better stereo imaging does not imply better crosstalk; it has been shown that varying x-talk between -40 and -70dB does not change the perception of stereophony.

But clocking can add components to the signal that enhance transients, certain parts of the spectrum, distortion etc. that will make fine details and thus room information more audible. It does not mean the signal is actually truer to the source, on the contrary.  Let's call it "pseudo detail".

I've had this experience with clocking, as well as with swapping op amps. "Pseudo detail" is to be avoided at all cost IMO, it will mess up a signal permanently.

I'm afraid you've just opened your own can of worms! Look how dithering (i.e. introducing a deliberate amount of noise in the converters) actually improves vastly the resolution, particularly below the noise threshold.

 

[seva]

I can only confirm this op amp thing. Recent tests with the SPL PQ have been amazing in terms of transient behavior and 3D sound -even when bypassed all bands ???

 

[living sounds]

I'm afraid you've just opened your own can of worms! Look how dithering (i.e. introducing a deliberate amount of noise in the converters) actually improves vastly the resolution, particularly below the noise threshold.

I'm aware of that, but what I was talking about is distortion, it affects the signal, not the noise.

 

[seva]

A very interesting thing comes out in this discussion. I know about consulting activities for some software companies and people who have studied electronics. It is the teachability of people. If you think you know everything, the teachability is 0%. If you doubt your own knowledge, it rises to 100%.

It is the same as with eg Vovox cables. We use two pieces of it in the signal chain if a recording can benefit from it. It's really exhausting to talk about it with a student. Interestingly, my old physics professor (a mix of Jesus and a hippy) told me in the early years something about things that are ridiculed in many discussions. I test the things if they sound good, that's right!

The reason for this talk is that it annoys me very much when I buy a product that is not really proven in the end practice. What a s...t to buy a socket in the hardware store and then breaks off something or you can not get the cables right in because a little space is missing. I would like to talk to the product developers and see how they can use their own products. The best products are invented by practice-proven engineers!

 

[Michael Tibes]

My ultimate goal is to build a 32-channel Soekris DAC.

Have you heard his converters? Do they sound better to you than the usual delta-sigma?

Michael

 

[JohnRoberts]

Arguing with people about what they hear on the WWW is a black hole for time.

@ Seva, doubting your own knowledge can be useful to make you more receptive to acquiring new information, but doubting established science can make one receptive to voodoo and hocus pocus.  Some tiny fraction of established science could be improved upon but is more correct than not.

JR

 

[seva]

Hi John, right! There is a lot of marketing behind recommendations and tests. Every marketing needs a story. I really dont wanna be the alchemist here . I visit this forum to further educate myself in the electronics. I appreciate your opinion very much because it is all of you who help me with my attempts to go and annoying beginner questions. My big weakness is the math that is unfortunately necessary for scientific arguments - I'm working on it. promised! But I am nearly free of objective opinions - observations are needed for many things. As far as I know, the scientific field is not really well researched about hearing. You know that the body consists of water and that the skin may be sensitive to the sonic. The brain is very sensitive to timing. The ear is coupled to the brain of which we do not use much. I think there is still a lot to explain for me. For me this topic is more like the ocean where MP3 is the nanoplastic. It can be very interesting to come closer to certain areas. Acoustics is a big topic for me. There are really interesting experiments in which it has been shown that, for example, an owl can not fly in a fully acoustical dead room - missing room information. A similar phenomenon occurs to people in acoustically poorly optimized rooms. The brain has an extremely good memory of sound events and can also orient itself without light by ears. The brain tries to orient itself in rooms where acoustical informations are missing and this loop repeats constantly - thereby the brain tires. The result is a tired listener. The reproducing of information from MP3 results in similar observations.

 

[JohnRoberts]

Hi John, right! There is a lot of marketing behind recommendations and tests. Every marketing needs a story. I really dont wanna be the alchemist here . I visit this forum to further educate myself in the electronics. I appreciate your opinion very much because it is all of you who help me with my attempts to go and annoying beginner questions. My big weakness is the math that is unfortunately necessary for scientific arguments - I'm working on it. promised! But I am nearly free of objective opinions - observations are needed for many things. As far as I know, the scientific field is not really well researched about hearing. You know that the body consists of water and that the skin may be sensitive to the sonic. The brain is very sensitive to timing.

Brain evolved sensitivity to arrival time differences between our two ears to judge rough direction a sound was coming from to aid our survival in caveman days.

The ear is coupled to the brain of which we do not use much.

some more than others, but the brain is constantly working to make sense of the flood of aural and visual data bombarding us. We discard a major fraction of this input data.

I think there is still a lot to explain for me. For me this topic is more like the ocean where MP3 is the nanoplastic. It can be very interesting to come closer to certain areas. Acoustics is a big topic for me.

another topic is psycho-acoustics or the study of how we interpret sound.

There are really interesting experiments in which it has been shown that, for example, an owl can not fly in a fully acoustical dead room - missing room information. A similar phenomenon occurs to people in acoustically poorly optimized rooms. The brain has an extremely good memory of sound events and can also orient itself without light by ears.

the brain has relatively good short term memory of things like sound levels, poor long term memory.

The brain tries to orient itself in rooms where acoustical informations are missing and this loop repeats constantly - thereby the brain tires. The result is a tired listener. The reproducing of information from MP3 results in similar observations.

??  The whole concept of stereo playback is replacing temporal cues with relative level differences. (pan pots vary level not arrival time). 

Stereo has been well regarded for a long time and still in use.

JR

 

[seva]

The brain has an extremely good memory of sound events room informations and can also orient itself without light by ears.

Could it be that the brain perceives the precise timing as less exhausting and suppresses measurable unwanted information? Where would be the technical limit currently. I have seen some 10mhz OCX at mouser. Too expensive than I could test it. I am afraid to order something without knowing testimonials. I have also seen on ebay some used clocks from gps. some diy clocks are around but looks not so reliable for any reasons.

 

[Andy Peters]

I have seen some 10mhz OCX at mouser.

Pedantic note: you surely mean 10 MHz.

Too expensive than I could test it.

Come on, only $80! (not including everything else needed to make it work.)

I am afraid to order something without knowing testimonials. I have also seen on ebay some used clocks from gps. some diy clocks are around but looks not so reliable for any reasons.

I have read discussions about converters which sync to rubidium 10 MHz reference clocks and apparently the results are ... mixed. As in all of these things, there is circuitry including a PLL to synthesize the necessary audio-rate modulator clocks (24.576 MHz for the 48 kHz family, 22.5792 MHz for the 44.1 kHz family), and it is the quality of the synthesizers that ultimately determines performance.

Or, put another way, why would you think that using a synthesizer to create a clock would be better than just using a crystal oscillator at the desired frequency?

And, just to throw a spanner in the works, there are clock-synthesizer chips that guarantee jitter at levels below what we can reasonably measure (femtoseconds).

 

[ruairioflaherty]

Seva,

To tag on to Andy's post, you are mixing up a whole bunch of unrelated or barely related concepts and phenomenon.

1) Higher clock rates/sample rates do not give higher temporal accuracy.  Remember that the signal that enters an A/D and leaves a D/A is completely analog and continuous - no stair steps, no discrete pieces of audio information.

Watch these videos to help gain an understanding - https://www.xiph.org/video/

2) Despite what various "Pro" Audio companies will have you believe a 10MHz clock is a dreadful place to start when you need to derive a typical audio sample rate clock. 

Read here for an excellent paper that expands on Andy's points - https://www.grimmaudio.com/site/assets/files/1088/picoseconds_or_ppm.pdf

3) Audio in a typical AD is sampled at much higher rates than 44.1 kHz or 48 kHz and then decimated down to the desired sample rate.

4) What does matter a lot for audio quality is short term stability of the clock, we want the lowest possible short term jitter for accurate sampling and reconstruction.  Interestingly that is something that Atomic Rubidium clocks are rather bad at (we are not interested in their long term accuracy).

The best clock is a selected crystal oscillator as close as possible to the converter chip on a well laid out PCB.

 

[Gold]

I have read discussions about converters which sync to rubidium 10 MHz reference clocks and apparently the results are ... mixed.

Rubidium clock addresses an non problem. Audio doesn't need an absolute time reference it needs the time slices to be evenly spaced.

I'm not sure about the technical issues but I bought a MOTU 828 because it was D.C. coupled. It was to make an automated locked groove contraption for vinyl. Put a MIDI note at the start and end of the loop. Route the notes to different outputs to control cutterhead drop and cutterhead lift. Never got around to it though.

 

[living sounds]

Have you heard his converters? Do they sound better to you than the usual delta-sigma?

No, when I get around to it I will built a stereo unit first and do a lot of testing. I do own various types of linear converter, and some of them sound very good indeed.

 

[living sounds]

The ear is coupled to the brain of which we do not use much.

I hope you were not implying we don't use much of our brains, because that is a myth.