I don't know how much of an improvement it is, but I've ordered some boards using Dave's cousins design(with Dave's approval of course).
I don't know simply because I haven't tried the IC yet, the IC being SM5814.
Anyway, it's meant to replace the SAA7220 between for example a wm8804 and TDA1541A.
As the board house has minimum order quantities, I might have an extra. One of the ones not going to me or Dave is probably taken, but there might be one more depending on how many I will need.
Naturally I'll need to test my layout to make sure it works first.
I don't know simply because I haven't tried the IC yet, the IC being SM5814.
Anyway, it's meant to replace the SAA7220 between for example a wm8804 and TDA1541A.
As the board house has minimum order quantities, I might have an extra. One of the ones not going to me or Dave is probably taken, but there might be one more depending on how many I will need.
Naturally I'll need to test my layout to make sure it works first.
Hejsan Mayday, är i min stuga i Värmland nu så inte så långt bort i värden (sorry folks a little 'hillbilly talk')- yes if your board is a drop in sense that it has a pin out that could just mount in a socket the saa7220 occupies now = yes I would be eager to try (and thus buy board form you if any available)..as I don't think i want to go the NOS route
tack (again 'hillbillysh')
/Patrick
DEM
Hi Patrick, DEM... The capacitor at pin 16 and 17 are for the internal DEM oscillator in the TDA1541(A). The frequency of the DEM clock is dependent on the capacitor value. I use 120 pF PPS or silver Mica so that gets me a lock with the bit clock around 200 Khz. ECDesigns at one time championed a method where the clock was locked to the bit clock. He has abandoned that these days. He is using a very large capacitor value. You can find all that with the search function on this forum. I still like the 120 pF high DEM oscillator frequency. Some think that locking the bit clock and DEM clock are a bad idea. There are some papers on that subject as well as a Swedish man's website. Easy to find with a Google search. My take is why throw samples away? SAA7220... It is bad however if Arcam has cleaned up the supply it might be passable. I have yet to see a good implementation of that chip. Arcam's are not common here or at least I have never seen one. My big issue is it really isn't 16 bit, 12 more or less. So I opt for the better math precision in the SM5814. 18 bits.... It isn't something you can pop into a CD player. A standalone dac, yes. Drop in replacements for SAA7220... No. It is I2S in and I2S out. No digital filters have that except the SAA7220. The SM5814 is rare in that it can output I2S. Needs Sony mode in. 2's complement, MSB first, right justified data. In the old days we delayed the signal like 17 clock cycles and inverted another signal. Trouble is it is easy to hear. Not in a good way. 😱
ARCAM
If it is a CD player it won't work. Standalone dac you can make a work around. See my other response. My name is Dave...😉
PCM1704+DDNF+OPA627
Hi folks... I have been testing the new buffer for 6 hours today. Configuration: WM8804, DF1704, PCM1704, Pedja's discrete diamond no feed back I/V stage with voltage buffer mod using OPA627 buffer followed by a 2nd order passive transformer reconstruction filter (Edcor 600/600 2.5 Watt version with 4n7 Polystyrene cap on the primary. I have George's first order filter on the DDNF, 2.5K resistor with 470 pF Polystyrene capacitor. So have the first order 3 db more or less at 130 Khz the 2nd order filter is down about 1 db at 20 Khz. Sound with OPA627... Better then BUF03. Although it is easy to hear the improvement in the treble range I also hear an improvement across the entire audio band. I have the gain of the OPA627 at 2X, and it is a DIP version of the chip. The SOIC is a slightly different die and hence will sound different. I expected a darker sound. Not getting that at all. Excellent clarity, dynamics and powerful drive. It sounds reference quality as it now betters my Mark II TDA1541 based dac. My impression is it is close to the "state of the art". BUF03... It gets you most of the way. Meaning you get most of the details with a powerful driving quality. I'd rate it 6 out of 10. If your system is lean sounding and analytical the BUF03 will bring some warmth with it's class A amplifier operation. Kind of a matter of system matching. The OPA627 is a better match for my system as described in another post. Tomorrow 's experiment will be to listen to the triple stack AD844, with pin 5 TZ output to the OPA627 output stage. Whichever one is better will go into the final version of this dac. I think it is time to box up this project.
Hi folks... I have been testing the new buffer for 6 hours today. Configuration: WM8804, DF1704, PCM1704, Pedja's discrete diamond no feed back I/V stage with voltage buffer mod using OPA627 buffer followed by a 2nd order passive transformer reconstruction filter (Edcor 600/600 2.5 Watt version with 4n7 Polystyrene cap on the primary. I have George's first order filter on the DDNF, 2.5K resistor with 470 pF Polystyrene capacitor. So have the first order 3 db more or less at 130 Khz the 2nd order filter is down about 1 db at 20 Khz. Sound with OPA627... Better then BUF03. Although it is easy to hear the improvement in the treble range I also hear an improvement across the entire audio band. I have the gain of the OPA627 at 2X, and it is a DIP version of the chip. The SOIC is a slightly different die and hence will sound different. I expected a darker sound. Not getting that at all. Excellent clarity, dynamics and powerful drive. It sounds reference quality as it now betters my Mark II TDA1541 based dac. My impression is it is close to the "state of the art". BUF03... It gets you most of the way. Meaning you get most of the details with a powerful driving quality. I'd rate it 6 out of 10. If your system is lean sounding and analytical the BUF03 will bring some warmth with it's class A amplifier operation. Kind of a matter of system matching. The OPA627 is a better match for my system as described in another post. Tomorrow 's experiment will be to listen to the triple stack AD844, with pin 5 TZ output to the OPA627 output stage. Whichever one is better will go into the final version of this dac. I think it is time to box up this project.
Hi Dave
thanks for clarifying. OK see, so its a BIG board/work if anything would have to be made wrt true SAA7220 drop in : first a I2S to Sony (2s compl.) conversion -> then into SAA7220 + possible extra support/conversion stuff (same or different logic 5v TTL?) and pack/map all that into a footprint w a SAA7220 pin-out to the outside world. Well is way beyond my skills and time. So I stick w the BAD and maybe do a little more clean up around it to 'beautify' it.
/Patrick
Hi there fellow norseman 🙂
Or whatever they called us back in the "day".
No, it's not compatible as a drop in replacement in the socket.
I don't know of any IC that would fit the bill for that, Dave might though.
In Värmland, watch out for the pimped-out epa-traktorer haha.
Btw, It's swedish, though where you are atm it may sound like hillbilly-talk, sorry about being an a$$ about värmland, just couldn't help myself 😉
The board is 5x5cm, I could have made it smaller but wanted space enough for quality PS caps.
Using SMD's in 0603 size etc might make the above possible, especially using a two level/board solution connecting the boards via pinheaders. Though it's nothing I'd put alot of effort into as I no longer use CD-player as digital source.
I've got a SB Classic (V3), the one with slimdevices logo, and a raspberry pi3 with a Digi+ board(I2S to Spdif via WM8804) running piCoreplayer.
I haven't gotten to really put the Pi through the paces as none of my red baron DAC's are done yet. Working on enclosure and the regulated supplies(I'm waiting for a few tracking pre regulator boards I had made).
Giving me decent supplies and 4 rails, 2 pos and 2 neg, at a rather small footprint, 95x80mm IIRC.
Hejsan Mayday
if I understand you correctly, the board you have now has it own interfacing..but 'piggy backing' that on top of a bedding board w the SAA7220 pin out might be possible? The 2x+ and 2x- supplies you talk about is another board right? . The CD player has enough free volume behind the drive to put a dedicated PSU in.
Your 5cmx5cm board should fit and even your 9.5cmx8cm board might fit (see photo) as I could lay surrounding lytics down (and if same was possible on your board) - it being elevated by socket (not done yet) and a bedding board. Quess I have about 3-4cm free height above socket (will measure when I get home next week).
So If I understood you correctly and you think it is a feasible task for PCB design beginner to draft a bedding board layout -> SAA7220 pinout = then I am still in wrt your board (assuming reasonable cost so it does not 'drown' all the other parts I have in now)
Also I stubled across this thread
http://www.diyaudio.com/forums/digital-line-level/187816-question-about-receiver-dac-xtals-3.html
where abraxalito talks about an open source dig filter for SAA7220 drop in replacement - BUT have not been able to find any further info if it really materialized or not?
/Patrick
PS I just now notice I have P/B version of SAA7220 which should be the better (sounding) one as designed w only 0.5% bias (P/A was w 5% bias) and this put less stress on the interfacing to TDA1451. SO maybe that's why it really sounds pretty good as is. But since I now have rolled the sleeves up wrt modding - might go all the way possible w/o splitting the current Arcam audio board to pieces.
Attachments
I'm sorry, I think I need to write after my coffee kicks in.
No, unless you can "hack" the signal from whatever IC(s) are between the transport and the SAA7220P, and make it compatible with 16-bit right justified (IIRC), you won't be able to use the sm5814.
Atleast not to my knowledge, limited as it is.
No, unless you can "hack" the signal from whatever IC(s) are between the transport and the SAA7220P, and make it compatible with 16-bit right justified (IIRC), you won't be able to use the sm5814.
Atleast not to my knowledge, limited as it is.
I/V Test results
Hi Folks, I have spent most of the day testing. Did an A B and A tests of sorts. Switching I/V's takes time. First test was a fresh listen to the PCM1704 dac with the DDNF + OPA627. I changed over to the AD844 triple stack (pin 5) + OPA627. Then returned to the DDNF + OPA627. This is just my opinion and based on my system components. AD844 stack + buffer... Sounds very close to the DDNF. Similar amount of detail. Had a pervasive smooth character that is hard to put into words. At first I thought it had a bit to much energy in the lower mid range. Not so sure. It followed every CD change. At times it appeared to ever so slightly obscure low level details. Really subtle. Bass is very dynamic with great punch. Treble is clean with air around instruments like percussion, bells flutes etc. On a scale of 1 to 10, I give it about a 9. Excellent over all. DDNF + OPA627.... Plays a touch lower in level then the triple stack AD844. Just noticeable. They actually have the same 2.5K resistors and first order filters. Details are very easy to hear. Comes out of a silent background. Dynamics seemed bigger and quicker then the triple stack. Does not have that smooth quality either. Seems well balanced from bass all the way up into the treble. Clean and focused are the words that come to mind. On a scale of 1 to 10, I have to give it a 10. Best I/V I have come across. ADEL2020... Has a similar sound to the DDNF + OPA627. Clean, deep soundstage and very wide. A bit less detailed, on a scale of 1 to 10. I would give it a 6. The final version of the breadboarded PCM1704 will have the Pedja Rogic DDNF (matched transistors) + the OPA627 (gain = 2). If I use the Edcor filter I can likely reduce the output back to normal by using the center tap on the output winding for a 2:1 step down. Again this is just my opinion.
Hi Folks, I have spent most of the day testing. Did an A B and A tests of sorts. Switching I/V's takes time. First test was a fresh listen to the PCM1704 dac with the DDNF + OPA627. I changed over to the AD844 triple stack (pin 5) + OPA627. Then returned to the DDNF + OPA627. This is just my opinion and based on my system components. AD844 stack + buffer... Sounds very close to the DDNF. Similar amount of detail. Had a pervasive smooth character that is hard to put into words. At first I thought it had a bit to much energy in the lower mid range. Not so sure. It followed every CD change. At times it appeared to ever so slightly obscure low level details. Really subtle. Bass is very dynamic with great punch. Treble is clean with air around instruments like percussion, bells flutes etc. On a scale of 1 to 10, I give it about a 9. Excellent over all. DDNF + OPA627.... Plays a touch lower in level then the triple stack AD844. Just noticeable. They actually have the same 2.5K resistors and first order filters. Details are very easy to hear. Comes out of a silent background. Dynamics seemed bigger and quicker then the triple stack. Does not have that smooth quality either. Seems well balanced from bass all the way up into the treble. Clean and focused are the words that come to mind. On a scale of 1 to 10, I have to give it a 10. Best I/V I have come across. ADEL2020... Has a similar sound to the DDNF + OPA627. Clean, deep soundstage and very wide. A bit less detailed, on a scale of 1 to 10. I would give it a 6. The final version of the breadboarded PCM1704 will have the Pedja Rogic DDNF (matched transistors) + the OPA627 (gain = 2). If I use the Edcor filter I can likely reduce the output back to normal by using the center tap on the output winding for a 2:1 step down. Again this is just my opinion.
Very good work Torchwood, I can't remember where I got it, (maybe Borboletti??) but here is a discrete version of the 844 if someone has the time to play around with it.
Cheers George
Cheers George
Attachments
Last edited:
Very interestesting, I'll have a look at that schematic.
Any idea which BJT's are used?
Hi,
that's a rather basic schematic, good for circuit analyzis.
There are several points this schematic may be more elaborated, starting from omittance of -for many applications- unnessecessary filter parts, decreasing the stage's input impedance with a CFP, or adding dc-trimming or dc-servoes, or changing to cross couling the output buffer, etc etc.
One could use the dual matched BC847(suffix BS or DS) from Diodes, NXP etc. throughout or dual matched versions of the BC807/817 for the current mirrors.
If one needs more power for Q51/52 the Zetex/Diodes FZT651/751 are fine.
jauu
Calvin
that's a rather basic schematic, good for circuit analyzis.
There are several points this schematic may be more elaborated, starting from omittance of -for many applications- unnessecessary filter parts, decreasing the stage's input impedance with a CFP, or adding dc-trimming or dc-servoes, or changing to cross couling the output buffer, etc etc.
One could use the dual matched BC847(suffix BS or DS) from Diodes, NXP etc. throughout or dual matched versions of the BC807/817 for the current mirrors.
If one needs more power for Q51/52 the Zetex/Diodes FZT651/751 are fine.
jauu
Calvin
I simulated the circuit more or less as is. I did swap 670R on the input to 680R and added 2x 680uF per rail.
I used BC547B/557B for BJT's.
Rails: +/-15Vdc
I did transient and FFT for out1 and for out2, assuming out1 would be TZ pin of the IC?
I used BC547B/557B for BJT's.
Rails: +/-15Vdc
I did transient and FFT for out1 and for out2, assuming out1 would be TZ pin of the IC?
Attachments
Last edited:
PCM1704+DDNF+OPA627
Took the dac over to my cousin's on Monday. We listened to it for several hours. He has the Acarian Alon 4 loud speakers, Bride of Zen balanced pre, and a B&O Cold amp with input board revisions. The sound was very much what I heard in my system. Always nice to get a second data point. Ordered an R core power transformer and looking at enclosures for the new dac. I changed the output tap on the transformer based filter so the Edcor is doing a 2:1 step down. Sound quality remained and level dropped to what I need in my system. Considering trying Lundahl LL1690's this week from my defunct CS4398 dac. Actually I can't think of anything that would improve it in all honesty. 😉
Took the dac over to my cousin's on Monday. We listened to it for several hours. He has the Acarian Alon 4 loud speakers, Bride of Zen balanced pre, and a B&O Cold amp with input board revisions. The sound was very much what I heard in my system. Always nice to get a second data point. Ordered an R core power transformer and looking at enclosures for the new dac. I changed the output tap on the transformer based filter so the Edcor is doing a 2:1 step down. Sound quality remained and level dropped to what I need in my system. Considering trying Lundahl LL1690's this week from my defunct CS4398 dac. Actually I can't think of anything that would improve it in all honesty. 😉
Lundahl based filter
I was wrong about the Lundahl LL1690's. Had to adjust to 3n3 Polystyrene capacitor. It is better in every way, clean, loud, dynamic and more resolution. I had to figure out how to get a 2:1 step down. Funny thing is it seems just as loud as the Edcor's at 1:1. 😱
Attachments
I told you I thought they might sound better.
I guess there's a reason people like them, I'll have to treat myself to a pair of Lundahl's one of these days.
Congrats on the success 🙂
Rowland Aeris Dac uses them somewhere in the analog stage.
http://www.my-hiend.com/leoyeh/2011b/IMG_9653cc.jpg
They also use the hybrid AD1853 dac, this is supposed to be so many bits multibit and delta Sigma, does anyone know?
Cheers George
http://www.my-hiend.com/leoyeh/2011b/IMG_9653cc.jpg
They also use the hybrid AD1853 dac, this is supposed to be so many bits multibit and delta Sigma, does anyone know?
Cheers George