Go Back   Home > Forums > Blogs > abraxalito

Rate this Entry

Highly evolved converters - the next generation

Posted 8th September 2011 at 03:47 AM by abraxalito

Its been over a year since I blogged about DACs, so, long overdue, here's an update. I've abandoned my considerable development on the AD1955 because I found something that's more interesting - multibit. In the first instance - NOS.

A while back I bought a TDA1543 NOS DAC to play with, just to see what all the fuss was about with NOS. Plenty of people have waxed lyrical about the sound. At first listen, I wasn't impressed although it had a certain tonal richness in portraying orchestral instruments that was alluring. Bottom line - its soundstage was compressed front to back. This made it sound a little 'forward' - foreground instruments and voices were more highlighted compared to acoustic spaces ('background').

A second aspect which plagues pretty much all NOS DACs is their frequency response can hardly be termed 'accurate' - owing to the sinc function of first-order hold, they exhibit a significant HF roll-off which begins around 5kHz and reaches 3.2dB at 20kHz. So HF energy is missing.

Given that I'm an acoustic space addict and I at least aim for flat frequency response I put the NOS DAC back on the shelf, to use only as reference for deciding if a particular dac was 'sibilant' or not. Given that it has only passive I/V, its not subject to the electronic (opamp) glare of other designs with active I/V and active filters. So it's been just my benchmark against which I measured a DAC for sibilance but not used for serious listening.

Earlier this year I was browsing DDS solutions on Taobao and found a DDS vendor also selling these. AD603 was a part I'd read about in Analog Dialogue when I was working in vibration monitoring and had been fascinated by the inventiveness of the gain controlling element. But I'd forgotten the part number long ago so had to look this one up. On downloading the datasheet I figured this part might make a good back-end as volume control for a DAC. Then a friend suggested 'why not use it in I/V?'. Up until that point I'd not been a big fan of passive I/V, but here was a part which provided sufficient gain AND a volume control element. Two birds killed with one stone seemed so attractive I had to try it.

One of the reasons for wanting to try this combined amp and volume control is a drawback of passive I/V I haven't so far mentioned - the one that kept me devoted to active I/V for so long. That is the spectre of DAC output compliance. In short, current-out DACs are designed for very low impedance loads as any voltage variation on their outputs gives current variations. For this reason, low value I/V resistors are preferred, to keep the output voltage in the 10's of mV range. That's not enough to drive an amp though, so designers of stand-alone DACs like the TDA1543 one I tried use bigger resistors to get a bigger voltage. This leads to greater distortion. Incorporating the AD603's not insignificant gain (34dB minimum) allows the use of sub-100R resistors for I/V duty.

The pic shows my prototype lash-up - I went for shunt regulation, two TL431s per rail. The 2-pin connectors to the DAC I/V resistors are on the far side of the chips. To be continued...
Attached Thumbnails
Click image for larger version

Name:	P1010016.JPG
Views:	807
Size:	254.6 KB
ID:	483   Click image for larger version

Name:	AD603_FBS01.gif
Views:	802
Size:	8.1 KB
ID:	484  
Views 2154 Comments 4
Total Comments 4


  1. Old Comment
    Pano's Avatar
    Interesting chip! How is this going?
    At first I thought it had only 7 steps, pretty crude, but now see that it's continuous.
    The specs look great, but it's all Mhz stuff. How's it working for you at LF? Are you using the 100R input as the I/V resistor?
    Posted 29th October 2011 at 02:41 PM by Pano Pano is offline
  2. Old Comment
    abraxalito's Avatar
    Hi Pano - yeah its interesting. I've moved on from this chip now, looking at AD605 mainly because its differential input., using more than one DAC. LF noise I did wonder might be a problem but haven't noticed anything yet.

    The I/V resistor I'm using is generally lower than 100R - at the moment I have a 16DAC prototype and I'm using 7.5R. Yes the idea behind using a MHz spec'd part is to reduce RF intermodulation. It does seem to be giving a clean sound.
    Posted 31st October 2011 at 01:52 AM by abraxalito abraxalito is offline
  3. Old Comment
    Seriously look at the distortion figures for the Ad605, I can't imagine putting this in any audio equipment besides a phone line, what am I missing?
    Posted 13th January 2012 at 08:11 PM by regal regal is offline
  4. Old Comment
    abraxalito's Avatar
    Which distortion figures were you looking at? At full scale I measured something around 0.05% (predominantly 2nd and 3rd harmonics) which reduced considerably at lower level. That was before I went differential.
    Posted 14th January 2012 at 02:27 PM by abraxalito abraxalito is offline

New To Site? Need Help?
Copyright 1999-2017 diyAudio