DIYHiFi.org

When will it be possible to buy the chip without jumping through hoops ?

You can get these from Shaw Electronics: shawsales@sbcglobal.net

Thanks
Gordon

_________________
Gordon
Chief Scientist
Wavelength Audio

Dustin,

Can we talk a little about the output conversion your thoughts and maybe a little about how bent some of us fools are about the use of or not use of opamps in the conversion process.

First for those who have not looked at this part. It is basically an 8 dac part. The part has seperate data inputs for each stereo pair. But singlular WCLK (word clock) and BCLK (bit clock). You can run the Crystal in a MCLK rates or others as Dustin has informed us.

So in stereo mode you basically tie all four data inputs in parallel so the quad dacs for each channel are running the same output. Each output is differential and so you can tie all the POSITIVE's and NEGATIVE's together.

But here's the interesting part... you can run these as they are in VOLTAGE mode or you can run them as standard CURRENT outputs. The voltage output has max SN of 120dB thd at -108dB, current output SN 132d thd -118dB. Though on the eval board and using my Prism dScope I got SN of better than 134dB and thd lower than -120dB. This is with all four left and right dacs in parallel and one of the reason's for better preformance as compared to a single channel.

~~~~~~

Ok Dustin... so each dac puts out 4.224ma PP or in the case of parallel outputs 16.896ma PP is that rights? I know on the eval board are AD797's but I think for that much current something a little heftier might be required. Have you evaluated other opamps for the i/v section or other forms?

I remember you telling me on the phone that the output impedance was like 753 (4 in parallel) or something like that. Can you expand on this a little again?

If we are looking at a discrete i/v is there anything we should look out for. We typically have been going into ground base amplifiers on the emitter. My basic idea was to use a NPN for the positive side and a matching PNP for the negative side. We usually use a discrete current source on both the emitter side and the collector side (better thd). In my case since I need single ended out that these will combine in the next stage before the output section.

In the past many have considered using say even a transformer to combine the + and - sides in voltage mode. Some even use moving coil stepup transforemrs as I/V (Why he got a patent on this I will never know) but as you may remember most moving coil cartridges are also current devices and therefore some designers use these moving coil step up transformers as I/V converters.

I was also considering maybe doing an all tube I/V stage using 6H30's with a push pull output transformer from plate to plate. Grounded grid using a fixed solid stage current source to assure matching of the current through the two tubes (thereby bettering the output transformer) and applying the positive and negative current outputs into the emitters for conversion.

I know... we are crazy...

Any comments?

Thanks
Gordon

_________________
Gordon
Chief Scientist
Wavelength Audio

You mean, it needs 2 samples (N, N-1) or 3 samples (N, N-1, N-2) ?

Actually, since the ASRC is after the oversampling filter, it operates on a highly-oversampled but very bandwidth-limited signal (ie. the intersample difference is likely to be quite smaller than the intersample difference at 44kHz) so the ASRC can be much simpler than ASRC operating on 44.1k signals. Which I think is the reason it apparently works well. It could even be linear interpolation... That's a good design decision...

Hi Peufeu,

Your are correct it needs 3 (N,N-1, N-2). By that way, I was reading around the site the other night I came across your "ninja" technique using the frying pan as a "reflow" I had a good laugh, thanks for that one.

Gordon, the output current is actually 0.924*4.224mA PP, One of the early datasheets had this error And I bet you have that sheet. WEll I guess late is better than never, hopefully I can be forgiven. About the I/V integrated/discrete battle, I havne't walked into this one and I think im just not experienced enough on this. Personally I think its a personal taste that no matter what someone else tells you, you will still like what you like, for whatever reason it may may, perhaps even a misguided reason. I used the AD797 for 1 reason in the beginning, I was told to design a DAC that outspecs the best AD DAC in THD, IDM, and DNR. So to compare apples to apples, I used the same opamps in the I/V stage when prototyping. THis design jsut got carried over and into the refference design. I did strip out the uneeded parts, and taylored it to the Sabre, but the AD797 just stuck around since I didn't need to change it to get the numbers I was shooting for. In my opinion its a VERY VERY good opamp. Im not really "tuning" the refference design anymore since I think it serves its purpose and that is to show the DAC performance itself, and let the customer (or DIY'ers) tweak it to their designs. I would like to see this DAC get put into some Hi-End gear since im proud of the part, and I believe the part measures well against all others, and I haven't heard 1 person come back and say it doesn't hold up to the hype. I waiting for that person and hope I can learn as much as I can from this part jsut out of personal interest. When I was back in grade 10 I spent physic's class deriving power formula's for my dual voice coil 15 inch subwoofer and how much power my amp could deliver when wiring in different configurations. I was trading off damping factor for power and being 15 at the time, guess which one won? It was around then I wanted to do audio design and even have my own company one day that designed car audio AMP's (seriously what high-school male doesn't want to design car AMP's for a living). I even emailed Rockford and MTX way back when and got no responce, but that never slowed me down. So I never did an AMP, but DAC's are cool too. maybe oneday.. Well not sure where im going with this rant about my past so Ill just end it there, hope that answered the questions I intended to when I first started typing this responce.

Dustin

Dustin,

So is it 4*0.924*4.224ma PP since we have the 4 dacs in parallel or am I missing something?

I have been really busy but have a guitar amp to build at 60W and a 140W bass amp to build real soon. The guitar amp will drive dual sealed Celestion UK Green Backs and the 140W bass amp will drive a single ported 15" TBD. They are both pretty cool designs using a autoformer as the differential mechanism and KT77's as output tubes (2 in the 60 and 4 in the 140).

Amps are fun to make but you designing for macro results. Digital is fun in that you have to deal with 2 realms for which most people are good at one and not at the other. no comment as too which I am better at ))

I have some really trick 10k:10k transformers I may layout this and just check it out and report back. I have to finish my regulator development stuff before I do this. Which should be sometime next week.

Thanks
Gordon

_________________
Gordon
Chief Scientist
Wavelength Audio