Professional DAC vs HiFi DAC query

Hello there - just wondering if someone could enlighten me as to the differences between pro and domestic DACs? I use a Digi 002 as my main recording interface and have considered getting a Mytek or Apogee stereo DAC as there are many who claim the internal DAC in the 002 is rubbish. Is there any reason why a good quality balanced domestic DAC such as this wouldn't do the job? Many seem to use the same chips but I realize there's a bit more to it than that - thoughts?

http://cgi.ebay.com/Matrix-mini-i-24-192-Balanced-DAC-Headphone-Amp_W0QQitemZ350236731381QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item518bbc67f5&_trksid=p3286.c0.m14

this has clock recovery (crap) once fed with spdif or AES, the only local clock it has is for that usb codec (very bad) . You don want to do that, cause its jitter hell and these DA chips goin to suffer . Apogee has real clock regeneration thats difficult to "diy" or "chinaman", besides its expensive. Benchmark DAC-1 also has local clock , throwing away jitter like it should. I would check for "upsampling" converters on ebay, but last time i saw there wasnt anything interesting, really.

Besides the topic name , prawfessional vs hifi hurts my eye

Spendor,

interesting question. The main differences to look for between something like an 002 a Protools I/O 96 (just to keep it within the same brand in this case) is down to a few different things

In no specific order

• Converter choice - low cost vs performance. There can be a large difference in cost between low performance converters, and flagship ones.
• Clocking regeneration circuit - Playboss touched on this. Poor clocking regeneration typically leads to clock jitter (which is uneven sample clocks). Uneven clocks move the sampling instant back and forth in tiny increments. Not the end of the world for low frequencies, but disastrous for higher frequencies. USB is notoriously weak in the clock re-generation game, as it's own clock isn't a nice multiple of an audio frequency. There are several ways around this - but each add cost - which moves the  product out of the consumer/prosumer market
• The rest of the signal chain - low cost interfaces are likely to use low cost consumer op-amps in their circuitry, rather than higher performance, lower noise opamps that suit the converter and the performance of the whole system. To put it in perspective, an RC4558 opamp is $0.09 each in 1K volumes. It's a functional opamp that been used in a lot of consumer/prosumer applications. However, an OPA2134, comes in at $1.25 a peice. That's a 10x difference for the same functionality, but greater performance. Also, bear in mind that audio interfaces typically use buckets of opamps. Those costs can add up pretty quickly.

It's 6am in the morning here... and that's the best I have right now

/R

just to add... that little matrix DAC, from the pictures at least, looks like its using pretty good parts.
I spotted an Analog Devices DAC (not the best in the world, but a serious contender), a cypress USB 2.0 device, high performance caps, real electrolytics instead of tantalums for decoupling, and beautifully SMD'd caps and resistors.

This isn't a cheap n nasty chinese copy. I've seen a LOT worse than this.

However, wether it sounds any good. (and more importantly, sound good to YOU) is all a matter of taste.


One more thing (sorry, I'm rambling)
Converters are important in a studio, but a recording made with a good microphone, a good mic pre, and mixed well with good monitor speakers, on an 002, will sound better than a sh*tty microphone, sh*tty mic pre and crappy speakers on any audio converter system. Get the rest of your signal chain sorted first... THEN upgrade the converters.

I agree with Rochey about the parameters he listed that make a difference. I feel that often people go on and on about the coverters themselves (chips) when often the difference with commercially made stuff is in other areas - most notably the signal path in the analogue section. For example, the op amps used in different commercial units vary dramatically from brand and model, and as we know the difference between these little dudes can be night and day. Take a run of the mill prosumer interface like an M Audio Delta 1010 which are extremely common in home studios all over the world. There are companies which will mod them. The actual chip often doesn't even get changed but the mod company will fix up the analogue sections of both the AD and DA by replacing op amps and caps amongst other things. Again the differences are often dramatic if not astounding for relatively cheap units.

Clocks is an area I am honest enough to say I am out of my depth with. The previous posts I am sure explain it well.

As also explained above, I would rather have a nice mic or two, an orgasmic pre, and maybe a nice warm compressor going into a home studio like converter than the other way round.

we have no adequate low jitter sources, nor adequate interfacing. This  24/ 192khz spdif thing is a joke because of that , and 16/48 usb is a joke anyway. Using an external DA,  decent clock regeneration or "upsampling" (= ASRC) is a must if you are looking for those 16+ bits.
An el-cheapo Emu 1820m might have the same converter IC-s as a DIGI192, but it wont have this:

http://akmedia.digidesign.com/support/docs/192ClockJitter_30957.pdf

hence those 16+ bits are lost , or better yet , contain nothing but noise, "dancing bits" as someone called : ))

Rochey said:

Uneven clocks move the sampling instant back and forth in tiny increments. Not the end of the world for low frequencies, but disastrous for higher frequencies.

Click to expand...

high frequencies are screwed once the incoming jitter is random. With the (sub)standard clock recovery, you get data correlated jitter, and that means AHARMONICS. Sounds (and looks) a little bit worse than "oh well, 10khz+ is overrated anyway"  ;D

playboss said:

high frequencies are screwed once the incoming jitter is random. With the (sub)standard clock recovery, you get data correlated jitter, and that means AHARMONICS. Sounds (and looks) a little bit worse than "oh well, 10khz+ is overrated anyway"  ;D

Click to expand...

Thank you PB, I confess that I didn't know that.
Do you have any links to somewhere I can read a little more about that?

thanks again

Dafydd

yes, 11/62 http://www.theaudiocritic.com/back_issues/The_Audio_Critic_21_r.pdf

Spendor said:

Hello there - just wondering if someone could enlighten me as to the differences between pro and domestic DACs? I use a Digi 002 as my main recording interface and have considered getting a Mytek or Apogee stereo DAC as there are many who claim the internal DAC in the 002 is rubbish.

Click to expand...

Hi Spendor,

The converters in the 002 aren't really all that bad and I'd be shocked if they are the weakest link in your listening chain.  Try not to get sucked into the Gearslutz BS (I've been there).  What monitors are you using and is your workspace acoustically treated and dialed in?  The average home mixing environment has huge peaks and dips in frequency response and RT at various frequencies.  I use a Benchmark DAC-1 feeding very big and very wonderful PMC MB2s for my mastering work.  The DAC quality is only relevant because my speakers are good and my room is reasonably well sorted at least at my listening position.

Cheers,
Ruairi

i want to add to that.... people get sucked into the audio interface issue based on cost -- it's cheaper and smaller to try and fix things with an audio interface, than to fix the acoustics of the room etc.

If you really do want to improve the sound of the system - get one of those reflection filters, a good mic, and a good pre (with an integrated ADC) -- I suggest the ISA828.

If your capturing crap on your input, then the best DAC in the world will just sound... crap.

Disclaimer: I have no more information on the Matrix DAC than what I can see from the pictures in the eBay listing. If anyone has more, please link here.

playboss said:

we have no adequate low jitter sources,

Click to expand...

How do you know? I can't make out all the chips in there, from the packages it's not impossible that one is an ASRC, possibly with an integrated AES receiver, possibly fed by that clock module. Admittedly if this wee beastie had an ASRC the folks at Matrix would have likely mentioned it in their marketing material.

playboss said:

[...]"upsampling" (= ASRC) [...]

Click to expand...

Upsampling and ASRC are most definitely not the same, and it's quite possible to have one without the other. I really hope that these are not being treated as equivalent in Marketese.

playboss said:

high frequencies are screwed once the incoming jitter is random. With the (sub)standard clock recovery, you get data correlated jitter, and that means AHARMONICS.

Click to expand...

No.

I am not aware of any research showing such effects occurring in common data interfaces and either analog or digital clock recovery circuits that would be applicable to audio interfacing. Sure, with specially crafted sequences and weakened (overwide) PLLs you can show some effect, but nothing that'll show up in actual equipment in actual usage scenarios. I am more than willing to be proven wrong about this by publicly published peer-reviewed research results (as opposed to sales white papers).

(Playboss, I'm not trying to pick on you, but you've made a few very broad, very strong and very unsubstantiated claims. Please back them up with hard data or show some nuance).

Having said that, I have some concerns about the Matrix DAC as shown. I'm not too happy that at least one of the converter chips looks to be no more than a few cm from the power transformer and rectifier, plus both converters are rather close to the VFD given that VFDs have a habit of spewing wideband EMI. Also, unless they've played very complex (and expensive) shielding games, the power supply return current for the digital subsystem would appear to flow through the analog subsystem, something you'd normally want to avoid.

To answer the original question: since both professional and HiFi are ill-defined and unprotected labels, there is no way of telling whether a randomly selected Pro converter is always better, equal or worse than a randomly selected HiFi converter. There are jewels and stinkers in either category.

JDB.
[plus the sad truth is that many of your customers'll be listening to your product on their iPod or their computer speakers,  so in the end it'll have to sound good on those. And I'm sure we've all had mixes sounding great on the studio monitors but completely uninspiring on cheap consumer equipment]

playboss said:

yes, 11/62 http://www.theaudiocritic.com/back_issues/The_Audio_Critic_21_r.pdf

Click to expand...

The article by Bob Adams, you mean? While it's a very interesting overview of the state-of-the art in jitter and sample rate conversion fifteen years ago, I fail to see any strong mention of the link between imperfect clock recovery in modern receiver chips and deterministic clock jitter, never mind aharmonic distortion (unless you're referring to his offhand remark on page 21 col 2).

For those who want to learn more about the basics behind asynchronous sample rate conversion, read this paper (start at Section 5, p.19 if you're short on time), and then the AD1890 data sheet.

Rochey said:

a good pre (with an integrated ADC) -- I suggest the ISA828

Click to expand...

For a second there I thought this was a subtle plug for a new all-in-one TI chip, integrating a PGA2500 and a PCM4222. But nooooo....

JDB.

Sometimes there's posts or discussions here that make me twice as good/knowledgeable at my job/passion.  This last post is reminiscent of this, almost like when John Hardy showed up to talk about potting DOAs.  At this rate I'll never stop checking this board 25+ plus times a day... 

Thank you all for your comments. First of all, I have addressed many of the equipment issues raised. I have a good signal chain up to the Digi 002 including 2 Neumann u67's, AKG 414 etc. i also have built 2 Hamptone mic pres, 4 Baby Animals and am about to finish a Great River MP1 - these can all be fed into my apogee AD or line level when recording multiple takes. I have also acoustically treated my room and have Quested monitors. What prompted this was viewing a Youtube video from Black Lion audio which detailed the differences after their mods - I was thinking that if I used the SPDIF output of the Digi002 through a DAC I may have improved monitoring and this seemed like a possible answer. A Mytek DAC would of course eat this for breakfast but its a money thing and times are tough....... I guess the proof is in the pudding but it sounds like a bit of a gamble.
Cheers!

"aes receiver with integrated asrc"  
...Treachery! Thats what im working on  ;D
jdbakker you wrote me a satire , while I probably saw more jitter plots than whitepapers, really. They actually DO call asrc "upsampling" nowdays, and I wouldnt hurry to call the Adams paper 15yr old state of art .

Anyway heres another one,but wait ,  I dont know how old it is and I dont have time for such arguments really:
http://www.tcelectronic.com/Media/frandsen_travis_2006_clean_clocks_tc(1).pdf

http://www.scalatech.co.uk/papers/aes93.pdf
The rest you should buy at AES.

oh, and any self respecting ASRC has upsampling before the actual interpolation, so the name is justified, unlike the 44khz "NOS" thing with those terrible, junk Philips TDA chips, that cannot be justified anyway.  

playboss said:

"aes receiver with integrated asrc" 
...Treachery! Thats what im working on  ;D

Click to expand...

There are a few chips on the market that combine those functions in one package; the TI SRC4382/4392 and Cirrus/Crystal CS8420/8422 come to mind, I'm sure there are more.

playboss said:

They actually DO call asrc "upsampling" nowdays

Click to expand...

Who are 'they' here? While you can do upsampling with an ASRC, you can do downsampling too; treating the two terms as equivalent is like calling a screwdriver a knife.

playboss said:

Anyway heres another one,but wait ,  I dont know how old it is and I dont have time for such arguments really:
http://www.tcelectronic.com/Media/frandsen_travis_2006_clean_clocks_tc(1).pdf

http://www.scalatech.co.uk/papers/aes93.pdf
The rest you should buy at AES.

Click to expand...

Yes, I've read those papers, but I fail to see their relevance. I agree that there's inter-symbol interference causing edge jitter on AES links due to the closing of the eye, and that paper selling JET gives a more modern yet rather limited overview of jitter (which is understandable, as they're trying to show how much their New! Improved! scheme helps).

What I was arguing with is this:

playboss said:

With the (sub)standard clock recovery, you get data correlated jitter, and that means AHARMONICS.

Click to expand...

In modern receiver chips, the actual content of the incoming data stream has very little influence on the jitter spectrum of the recovered clocks. True, the further the eye is closed (and in AES that tends to mean longer cables, or more reflections in optical S/PDIF) the more jitter a simple receiver will produce, but none of that is data correlated for a 'normal' data stream. How all this would lead to AHARMONICS rather escapes me, I'm afraid.

playboss said:

oh, and any self respecting ASRC has upsampling before the actual interpolation, so the name is justified,

Click to expand...

Again I fail to see your point. While most ASRC cores do indeed have an upsampler-interpolator-resampler-decimation filter-decimator signal chain, this is purely to keep the number of polyphase filters/coefficients to a manageable level to reduce chip area and not because of any properties of the underlying signal processing. So why call the whole chain by the name of the first element? The signal path of a GSSL starts with a line receiver, yet we call the whole a compressor.

playboss said:

unlike the 44khz "NOS" thing with those terrible, junk Philips TDA chips, that cannot be justified anyway. 

Click to expand...

...I agree, but let's not start that discussion here, thank you very much.

JDB.

if you forgive me im going to leave jitter out as well,
hehe, the cs8422 was new to me, and you know we tought about offering such src4392 module on ebay but we grew out the chip. I think if the people were better educated , such SRC receiver would sell. The texas chip looks better than the cirrus , mos definately. However, lets just argue about the
Matrix from china.
With clock recovery from a personal computer you goin to get jitter towards >1 nanosec. Is the ad1955 switched capacitor dac therefore less sensititve to jitter? No it isnt switched capacitor outpu stage but pure continous current, therefore you wont even hit 16bits.

This dac is a joke for that money, and leaves ad1955 features unused, like the builtin volume control (not the usual d-s vol. c. ) .

I run a Digi002 ProTools LE setup. I find the conveters to be grainy and just plain bad. I also found the imaging to be smeared with the internal clock. I originally bought an Apogee Rosetta. This dramatically improved the A/D-D/A but when clocking PT with it, the top end was really harsh and the midrange seemed scooped.

I recently decided to drop the cash on an Apogee Big Ben and and Lynx Aurora8, and I couldn't be happier. The Aurora8 (clocked internally) has a wonderful midrange and is very "true" to the source. When clocked with the Big Ben, the top end seems to get a little smoother; which can be a good or bad thing depending.

The most noticable improvement is clocking PT LE with the Big Ben. Bottom end tightens up and the low midrange becomes more defined. And the overall seperation of instruments and stereo image widens. It really was/is a purchase I am very happy with. And even clocking PT with the Aurora8 was an improvement, more so than the Rosetta. I just like the Big Ben best.

Cheers.

AFAIK...the Aurora is the best bang for the bucks...