Is the SAA7220P/B really that bad ?

Yup. OS is better, AEBE (all else being equal).
You're probably right. Unfortunately, my NOS setup was clocked better, but I hope to rectify that.
At the moment, I think I enjoy NOS more. Currently, my DAC doing 4x OS sounds okay, definitely different, with more treble detail, but it does remind me of a fairly standard CD player. My NOS setup definitely didn't sound like that.
However, that is more than likely due to the standard CD player-esque clock distribution I now have.

Cheers,
Phil
 
Nice you have solved the problem Phil.

For clock i have good results with a Tent XO, the PS is a "flea" on perfboard (with NE5534!) The gyrator feeding the PS is very nice. First here kwak won, (with XO on it's recommended PS) when xo sits on a flea, they come very close. But have also made a diy buffered cristal which looks nice on scope, waiting to be installed in SACD/DVD.

Also consider dem reclock with divider to 1541 @ 176kHz (Ecdesigns) I did all the clock bells-and-whisles on the 304.

If you have a good working bessel design i am interested. Wanted that for a while back but never came to it. (have young childen too, an 18 year old car with repair job's, TT, DVD, amp mods, pc jobs, all besides regular job :D )
 
hi Tubee,

My clock module is based on the Kwak clock. I've built a few in the past and I'm quite happy with it. Having said that, I have tried developing my own - purely to make it as cheap as possible, but also to see if there's anything clever really going on, or if its just the important things like well regulated power that makes the difference.

I did implement DEM reclocking, and it worked quite well, but i wasn't happy with my implementation, so I'm not using it at the moment. Did you hear a marked improvement? With OS as well (lets try and keep this on thread :) )?

I'm not even sure if its Bessel. Its simply a 3rd order filter, low group delay (probably Bessel then) 39kHz-ish group delay.
All credit goes to Ray (65hc) who designed it.

Its somewhere in the huge CD63 thread.
I'll check with Ray if he minds me posting it, and I'll post my take on it here.

Kids eh? At least my car is only 10 years old!

Cheers,
Phil
 
Sorry, for those unaware, ray is 65hc.

I haven't heard from him yet, but I've found his original post...

http://www.diyaudio.com/forums/showthread.php?postid=859397#post859397

If you ignore the tubestage, I implemented the passive filter on the left (there are two in the diagram as it is obviously a differential amplifier).

I don't think he'd mind. Its just a link to his original post. You'd find it if you searched anyway.

Cheers,
Phil
 
I have straightened things out a bit.
Its still a complete bird's nest of a mess, but I now have the clock module out of the transport and in the DAC. The clock module is feeding the decoder chip in the transport, the SAA7220, and the I2S reclocker I knocked up yesterday (which feeds the TDA1541A, not the 7220) - all clocked in parallel.
So now I feel I have a more equal contender to compare against my NOS setup.

Initially, my first impressions were lots more treble present, and a drastically reduced bass (however, my BBC monitors have a deliberately toned down treble to offer the impression of a better bass response, so maybe I'm just used to that kind of sound).
I admit I wasn't sure at first, but I probably aclimatised to the lack of bass, and the amp had warmed up, and with well recorded stuff, I admit I was impressed.

I think the imaging/soundstage may be better with NOS, however, the detail and 'instrument seperation' (I like it when you can pick out any instrument you like and follow it, it doesn't get obscured by louder instruments) seemed to be really good.

I was starting to like it!

I do worry that I need to be more careful with clock distribution, because I understand higher frequency clocks can be more sensitive to jitter, but I am impressed with the 7220 (it is much better today now the clocks have been, to some extent, sorted out).

It is a grower, but I when I set up my DAC properly with oversampling in, I will design in a way to select 4xOS and NOS to compare properly.

I haven't heard many other oversampling DACs (although I have heard a lot of delta-sigma and Bitstream DACs -which I don't like that much), but in answer to the thread question; I don't think the SAA7220 is really that bad.
 
Just looking at the datasheet, you can see its not the most sensible design, with only one supply and one ground pin (and so many unused pins!). Mind you, having worked for a major semiconductor company, I could easily imagine an accountant making the most major and fundamental design decisions. :bawling:

So, what would you recommend as perhaps the 'next step' beyond SAA7220 - for future reference (my to do list is a bit long)?
As a short term and easy improvement, I'm on the look out for a SAA7220P/B, but I'm interested in further suggestions.

Cheers,
Phil
 
philpoole said:

So, what would you recommend as perhaps the 'next step' beyond SAA7220 - for future reference (my to do list is a bit long)?
As a short term and easy improvement, I'm on the look out for a SAA7220P/B, but I'm interested in further suggestions.

The SM5842/3/7 and PMD100 are four to consider. The SM5842 is ridiculously expensive, the SM5847 doesn't come in DIL and the PMD100 is very hard to find. You might also want to try something other than the TDA1541.
 
TI/BB's DF1704 and PCM1704

rfbrw said:
The SM5842/3/7 and PMD100 are four to consider. The SM5842 is ridiculously expensive, the SM5847 doesn't come in DIL and the PMD100 is very hard to find. You might also want to try something other than the TDA1541.

TI/BB's DF1704 and PCM1704 are still avail. (e.g. via DigiKey). They are (alas) $$, DF1704 is avail. in SSOP only, and you need two PCM1704s for stereo.
 
Do you need 8x (or more for that matter) OS?
Surely, 4x makes the analogue filtering easy enough?

I understand it effectively adds, is it 1.5 bits per doubling of sampling rate, or something, but it must add all kinds of problems to the circuit construction. Greater susceptance to jitter, interference, etc
All work roundable, but is it worth the hassle?
 
I thought I might post an update to add to my conclusions.

Its been a few weeks since I finally got my DAC working with the SAA7220P/A and I finally procured an SAA7220P/B a few days ago.
Having listened to them for a while, I can't put my finger on it (and I refuse to go all WhatHifi on everyone), but I do think the B sounds better. Quite a bit better, but maybe not loads better (very scientific ;) ). I think the soundstage improves, and more complicated tracks are easier to follow (the Polyphonic Spree, love them or loathe them, have a lot going on and the sound often seems confused, this seems less so with the /B, for instance). I keep hearing subtle details that I hadn't before, which is nice.

Anyway, whichever version, 4x oversampling sounds a lot better than NOS, in my opinion. However, I think a lot of attention needs to be given to using a sensible clock, reclocking the output and keeping the SAA7220's power very seperate from everything else - or it will sound very average.
This is pretty much what has been concluded in this thread, but I have heard it myself, as I originally had this working using the 'standard' way, with a crystal attached to the chip and feeding the rest of the system, and no I2S reclocking (it didn't sound better than NOS - that's for sure), and sort those problems out made a big difference.

If I can get hold of other digital filters to try out (I'd love to try a PMD100, but can't justify the cost), I will, but for now, I am quite happy with this setup (I have to progress on to other parts of my stereo that need attention).