Nonoz DAC page

[fedde]

I have made a page about my non-oversampling DAC. If you want to build a simple&cheap DAC that's quite OK have a look at:

http://home.student.utwente.nl/f.s.b...dio/nonoz.html

Fedde

[halcyon]

Have you considered alias-intermodulation distortion?

http://www.musaeus.co.uk/aespaper.htm

cheers,
Halcyon

[ultranalog]

No he hasn't, I pointed it out but Fedde seems sure he can judge IMD by ear...

[fedde]

Yes I have considered that problem. When used with a high bandwidth amplifier I believe that the problem is less worse than the changes the digital filter make to the sound.
There are better DACs than this one, but for the price it is very hard to beat.

Fedde

[ultranalog]

When used with a high bandwidth amplifier I believe that the problem is less worse than the changes the digital filter make to the sound.

The higher the bandwidth, the more chances for IMD.

Still waiting for that explanation of the 'changes' from the digital filter though.

How's about backing believe with measurements?

Remco

[jean-paul]

The higher the bandwidth, the more chances for IMD.

True, a non feedback tube amp seems to be the best to connect to such a DAC.

How about feeding your ears with good sound from a non os DAC and then talk further ...

If I connect it to my oscilloscope I see things that make me feel miserable.
When I connect it to my amp things are different. After building a non os DAC I experienced the feeling again why I once started with this hobby. Very natural sound and very direct. The highs are not the strongest point however. Overall the result is very good compared to other products. I did compare mine with a standard VRDS 10 and guess which of the two sounds better...

Build one, not necessarely Fedde's, it can be Richard Murdey's or Scott Nixon's and see that backing up belief that filterless non-os dac's should be bad will not be easy.

link:

http://www.geocities.com/rjm003.geo/...o/diy_dac.html

http://home.triad.rr.com/scottnixon/dac.htm

[alvaius]

Not only the amp, but also speakers have IM distortion.

Food for thought.

[fedde]

With the risk of being rude (!?), I would not advise building Richards DAC. It surprises me a little that you advise it, because it uses 9V for the TDA1543's. I thought you were against using the TDA's with such a high voltage !?!?
Furthermore there are some errors on the page and the I/V resistors are too high. (distortion !). Also the loop filter can be improved.

Scott Nixon's DAC seems very good though...

As I mentioned on my page, I didn't compare it to many DACs and CD-players. When comparing it to other equipment I felt that the DAC had something special. It was nice sounding, groovy and natural (musical) sound. Not better in all terms (detail, highs), but overall more pleasant. But this is just my humble opinion of course...

I do not care that the DAC misses a little treble, that is maybe even more comfortable to listen to.

I still have the opinion that digital filters alter the time resolution, they mix up data in the time domain. I have to do more research on the topic but I think that most of the filters used in audio systems have more taps than the 8 you mentioned. In the case of around 30 taps, impulses in the music (snares, piano attacks) get a lot of ripple. Also, musical instruments get mixed up. I think that any change in the signal, both digital and analog, changes the audible information. Of coarse, the whole thing is a trade-off. Good sounding systems can be made with both oversampling and non-oversampling. They will sound different in some ways however.

But anyway, the most important point is that you can build this DAC for very little money and for the money it is quite good...

Fedde

[jean-paul]

With the risk of being rude (!?), I would not advise building Richards DAC. It surprises me a little that you advise it, because it uses 9V for the TDA1543's. I thought you were against using the TDA's with such a high voltage !?!? Furthermore there are some errors on the page and the I/V resistors are too high. (distortion !). Also the loop filter can be improved.

You're right but I merely advised his page for the basic schematic and the hints for PCB design. Richard Murdey and Scott Nixon are the men that began with this schematic and Richard is the only one that shows different PCB designs. Everybody else followed this design and tweaked it. I did, you did certainly too.
Don not act like you invented this stuff.

I simply forgot the 9 V and resistor values. They have been discussed in various fora.( but a newbie doesn't know that, could have given the links to them as well ). Like I advised you some time ago about the wrong resistor values in your DAC that result in too high distortion.That's why I gave both links to Richard's page and Scott Nixon's.
Scott is the person who has done a lot ( the most probably ) of research on this DAC. I am sure these two designs are the most built at this moment. All others are copies / combinations with other resistorvalues or other small differences.
The Monkeysect filter is an improvement that follows in the learning curve of building this DAC. I think you didn't implement it as well when you first build it. It's part of the tweaking.

[ultranalog]

How about feeding your ears with good sound from a non os DAC and then talk further ...

Out of arguments error.

Of course I tried non os (5 years ago, with 3 different dacs), but good sound from a non os is a contradictio in terminis.

Don't assume.

I did compare mine with a standard VRDS 10 and guess which of the two sounds better...

Who said you have to compare it to opamp gear?

Remco

[ultranalog]

I still have the opinion that digital filters alter the time resolution, they mix up data in the time domain.

Digital filters do not mix up data. Facts go before opinions.

I have to do more research on the topic but I think that most of the filters used in audio systems have more taps than the 8 you mentioned. In the case of around 30 taps, impulses in the music (snares, piano attacks) get a lot of ripple.

Regardless of the the # of taps, because of the oversampling nature of the filter, the ripple is always less than it would be in non-os. That's the wole point of filtering! I'm not wasting anymore time on it before I see some theoretical backing. You're an EE student, shouldn't be hard to prove if you're right. The rest of us keep with Nyquist and the cosine rules for determining IMD.

Remco

[jean-paul]

Out of arguments error.

Sorry but I am not out of arguments. Theoretical the non os dac's should be crap, but in real life they are not. I am well aware of the things you are saying and I agree technically.

But why do they sound so good then ???

Beter ask Zanden, Audio Note, Marantz and others the same question. Nobody can explain what the reason is they sound so good. Please don't come up that it is commercially good for them, other products create revenue as well.

Why would I need arguments anyhow if sound is OK ? To proove your theoretical and / or educational background ?

good sound from a non os is a contradictio in terminis.

OK, you are right Ultranalog, satisfied ? Talking about a "out of arguments" error...

[ultranalog]

Sorry but I am not out of arguments.

Well, let's have them.

Nobody can explain what the reason is they sound so good
But why do they sound so good then ???

If you're talking about the problems with usual gear, these are pretty easy to explain. But they do not occur in the oversampling filter. You are dismissing a system which you judged listening to the entire system, and blaming the wrong component.

The usual problem with mulitibit
converters is IMD in the I/V conversion.
Jocko made a fine circuit that greatly eliminates it be removing the nullor configuration that causes it.

Non-os is not a solution, it's lowering the clock to lower jitter problems. That's battling symptoms, not eliminating the problem.

Remove the intermodulation mechanism and listen again. It'll sound better. And how inconvenient, it will measure better as well.

Why would I need arguments anyhow if sound is OK ?

To 'make things better'? Isn't that what we all want to do?

Remco

[jean-paul]

Wasn't one of the pretty things of the TDA's and PCM63's that they didn,t suffer from these problems as bad as others do ?

I am very well able to judge my system as well as components thank you.

It'll sound better. And how inconvenient, it will measure better as well.

Please don't think I am one of the "don't measure but listen" zealots. Maybe you got me mixed up with somebody else ?

I think this discussion looses its character and no valuable info will come out anymore. Besides that the discussion should be continued in a new topic since it really doesn't have anything to do anymore with the Nonoz of Fedde's.

[ultranalog]

Wasn't one of the pretty things of the TDA's and PCM63's that they didn,t suffer from these problems as bad as others do ?

Exactly the opposite. The current-outputting multibit dacs need to see 0V at the output or their multiple current sources will mistime, causing spikes that are probably supersonic, but detrimental because they eat up energy and intermodulate easily in an environment filled with audio signals and numerous images.

Please don't think I am one of the "don't measure but listen" zealots.

Yeah, where would I get that idea...

I think this discussion looses its character and no valuable info will come out anymore.

It's not going the way you planned it?

Remco