passive filters for DACs

[plp]

I am experimenting with a 0x oversampling DAC. Does anyone know of the actual configuration and component values of the CLC passive filter AudioNote uses in the DAC1.1 (0x oversampling DAC kit) While my design is not a copy of the AudioNote design in that it uses BurrBrown DACs and se i/v converter, I have experimented with a number of filter configurations but have not been happy with the results. I don't know whether this is a result of the filters tried, or an inherent trait of 0x oversampling. Apparently with the AudioNote filter, the inductor is not in series with the signal path (shunts to ground)

I would also be interested to know what chip AudioNote uses for L/R demux. Reading a number of Ben Duncan articles suggests digital ICs also have a sonic signature.

[hifi]

well im just about to build a new dac aswell with the cs43122 im figuring about using an transformer to couple it.

any opinion?

/micke

[yves]

SOwter has a tranformer that is 1:20 if I recall correctly.
http://www.sowter.co.uk.

If you check the audio asylum, there were some discussions
on this a while back, and crops up from time to time.
Audio Note has a patent on using the transformer as a
coupling and filter device for the dac output.

Pls let us know how you fare in your design, perhaps
share your circuit with us too?

Yves

[plp]

The suggestion of transformer coupling sounds interesting, but I have not seen it done on a se D/A conv like the BurrBrown. There was a design published in Glass Audio some time ago, but it used a Crystal D/A conv with balanced output, intergrated I/V conversion and oversampling.

The problem presented with 0x oversampling is the need to remove the 44.khz sampling frequency and a transformer alone can't do this. Adding a notch filter is not the answer for obvious reasons.

The AudioNote CLC filter starts to roll off @10khz, is 2db down @20khz and I think 17-20db down @44.1khz. It seems the secret of the AudioNote filter is that the signal is shunted by the inductor, not passed through it. Alternatively, 0x filters published on the web use multiple inductors and numerous capacitors which defeats the purpose of 0x oversampling, that being simple circuits.

The problem of using a coupling transformer in any design is the need for intergrated or external I/V conversion usually of IC design which sort of defeats many of the benefits offered by a tranformer in the first place. Admitted the only transformer coupled DACs I have seen use the Crystal chips with balanced output and intergrated I/V conversion, so I have not though about whether the entire analog stage could be done passively from the I output of the D/A conv, which in theory would be the ideal.

The advantages of 0x oversampling are an increase in soundstage and depth, a more solid dimensional image and a level of overall excitement which is bettered only by DVD-A. How to exploit those traits with an entirely passive analog stage is the million dollar question to which even AudioNote has note yet found the answer.

[tvi]

The Audionote Dac Filter Patent is here:
<a href="http://164.195.100.11/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PALL&p=1&u=/netahtml/srchnum.htm&r=1&f=G&l=50&s1='5,420,585'.WKU.&OS=PN/5,420,585&RS=PN/5,420,585">Digital to analog conversion devices</a>

Once again you will need a tif viewer, like Quicktime or a plugin like <a href="http://www.mieweb.com/alternatiff/">Alternatiff</a>, its free for personal use.

Hope this helps

Regards
James

[Hoffmeyer]

TNT-Audio has a 0 x oversampling DIY project at http://www.tnt-audio.com called Convertus, maby there's some inspiration to get from there.

[rol1]

http://www.coilcraft.com/pdf/doc200.pdf

A discussion on the effects inductance and capactance on common mode filters. I don't know if this will apply in the freq ranges you're working in but something may.

Also a non-oversampling DAC linked thru a link in JonT's post about the mj-amp.


http://www.homestead.com/_ksi0701961.../44k1_dac.html

[Edited by rol1 on 07-15-2001 at 11:24 PM]

[Bernhard]

The link to the audionote patent did not work for me.

Link to a schematic how they used the transformer ?

[hermanv]

Several points/opinions.

1. With 0x oversampling a complex filter is probably an absolute requirement. Almost certainly a series LC shunt to ground high Q notch at 44.1KHz. At least it’s easy to simulate.
2. Dual DACs, there's enough problem getting rid of 44.1KHz separating left from right would just make it a much bigger problem. Who wants sample and holds with all their problems?
3. A plain resistor to ground is a current to voltage converter; you must stay within the voltage compliance range of the DAC, check the data sheet.
4. If true purity is your goal only the very best passive parts will do, Teflon/tin or Polystyrene caps, maybe copper foil caps. A cheap little audio transformer just won't get the job done, you can't use ferrite or any steel alloy core (hysteresis) so you probably need to wind your own, or pay outrageous prices, silver wire? Vishay resistors?
5. Parts to ground i.e. in shunt with the signal cause exactly the same amount of problems as parts in series, if current flows through them and they have any kind of misbehavior that misbehavior will generate an opposing EMF that sums to the signal. There is no free lunch.

Still, well executed should sound great, whether it is the path of least resistance to good sound I'll leave up to you.

ps. Probably a bad choice if you have and like passive pre-amplifiers.

[rfbrw]

Originally posted by hermanv

separating left from right would just make it a much bigger problem.

Very easily done. There are two examples in this forum not counting the CD6000 and the Dacmagic.