The more I TRY to comprehend this thread, I more confused I seemed to become.Yet clearly a major part of your confusion is because you've not educated yourself with the information which is available here. Messy I would tend to agree - that's the nature of forums and also my nature (ask my wife).
Earlier, I inquired about the diff between Ling and Phi. None was provided. Note that the word "Ling" is in title of this thread. Yet most contemporary references are about something called "Phi"?
Speaking of the thread title, " lingDAC - cost effective RBCD multibit DAC design" .... the "cost effective" is questionable given that Vendors Bazzar contains prices up to $91USD. See: PhiDAC hex kits with pre-built filters
Is that "cost effective"?
On the reverse side the only chip next to a battery of 8xTDA1387 is marked VHC574. It is an octal D-type flip-flop chip.
Hi! can you post 7th order and 3rd order filter schematic and the complete BOM? Thanks.
3rd and 7th order schematics
Here are the schematics for the two filter types. I don't have completed BOMs for these because the capacitors (and the inductor(s) for the 3rd order) are all select on test - so each completed filter is slightly different, depending on the precise values of each component.
@sajunky - that '574 is operating as an async reclocker. Its clock input is fed from the 50MHz osc.
Here are the schematics for the two filter types. I don't have completed BOMs for these because the capacitors (and the inductor(s) for the 3rd order) are all select on test - so each completed filter is slightly different, depending on the precise values of each component.
@sajunky - that '574 is operating as an async reclocker. Its clock input is fed from the 50MHz osc.
Last edited:
'NF' is short for 'no fit' - it means that capacitor isn't fitted. It is there for flexibility, so that the PCB can be used for other filter shapes with different cap values. For example narrower or wider filters can be built with the same board, or filters with different amounts of passband ripple.
Thanks a lot.
If its not too much to ask can you share a write up about assembling and testing this dac.
Regards
With hand soldering I recommend building the PCB in two stages - the left one-third of the PCB has the power supply and DAC parts, the rest of the PCB has the CLC filter plus output stages. Build the left side first and test it before populating the right side.
For testing you need to have a current limited power supply set to 12V and a DMM. The PSU current limit should be set about 200mA so that if you get an error the AD815 acting as regulator won't overheat too quickly. The regulator runs with two LEDs as a reference voltage of 4.5V - this voltage (on R4) is the first thing to check on initial power up. If that's OK then check the output voltages at B4 (ferrite bead) and R57 are as shown on the schematic. Lastly check the supply to the 74HC86 is 2.5V from the 431 shunt.
Assuming all those measurements check out, then move on to assembling the right side of the board.
For testing you need to have a current limited power supply set to 12V and a DMM. The PSU current limit should be set about 200mA so that if you get an error the AD815 acting as regulator won't overheat too quickly. The regulator runs with two LEDs as a reference voltage of 4.5V - this voltage (on R4) is the first thing to check on initial power up. If that's OK then check the output voltages at B4 (ferrite bead) and R57 are as shown on the schematic. Lastly check the supply to the 74HC86 is 2.5V from the 431 shunt.
Assuming all those measurements check out, then move on to assembling the right side of the board.
Non FS multiple frequency asynch reclocker is a bad idea, IHMO.
Not only in audio (I believe), but in the entire digital world. With increased sampling rates (like 192KHz) it would go even worse.
I didn't expect such things in the audio DIY industry, my bad. Thank you for enlightning me. I would rather pay for a version sans the expensive clock chip.
From datasheet
According to TDA1387 datasheet, we can do 48KHz*8=384KHz right? With such high sampling rate, can we just go with simple RC filters to in order to get smooth waves?
What is this "double speed" operation anyway? Does it have any advantage on implementation?
And this. I see some designs following this advice but some not (including yours). I see a 1uF ceramic cap in the middle of the row. Is this cap for this role? Can you share your opinion on independently decoupling each cap?
Attached photo is from another design.
According to TDA1387 datasheet, we can do 48KHz*8=384KHz right? With such high sampling rate, can we just go with simple RC filters to in order to get smooth waves?
What is this "double speed" operation anyway? Does it have any advantage on implementation?
And this. I see some designs following this advice but some not (including yours). I see a 1uF ceramic cap in the middle of the row. Is this cap for this role? Can you share your opinion on independently decoupling each cap?
Attached photo is from another design.
Attachments
Last edited:
At 8*fs the images will begin at 364kHz. To get say -60dB attenuation in just under 20X ratio would call for a 2nd order filter. Passive RC can't achieve that but in conjunction with an opamp its easily achievable.
I can only guess it means with an 88k2/96k recorded stream.
You mean independently decoupling each chip with a cap? I haven't done that but I've nothing against it. Yes, I use two caps for decoupling six chips - the 1uF ceramic and a 2700uF 'lytic.
On the photo there are two individdual caps per chip. One 1uF on a REF pin (7) and one 100nF on the Vdd pin (5). The same values as recommended on the datasheet. My portable Nobsound 8x TDA1387 DAC/HPA use exactly the same scheme. These bypassing caps are critical for SQ, some others use two large electrolythic caps per chip, but it requires mounting caps on the opposite PCB side of DAC chips.
actual owner of PhiDAC
I have to say I'm more of an audio enthusiast and audiophile than DYIer, but the thing you link is what most people here started with, realized all the problems it has and turned to phiDAC. What PhiDAC does is amazing and I guess true, not for everyone. I have one with 7th order filter, paired with analog Fisher CA-7700 amplifier, mains filter and what it produces is giving something to think about to DACs like Chord Mojo and more I listen to it, even to RME Adi2. Soundstage and separation is out of this world.
I guess linear power source would do another small step up.
I have to say I'm more of an audio enthusiast and audiophile than DYIer, but the thing you link is what most people here started with, realized all the problems it has and turned to phiDAC. What PhiDAC does is amazing and I guess true, not for everyone. I have one with 7th order filter, paired with analog Fisher CA-7700 amplifier, mains filter and what it produces is giving something to think about to DACs like Chord Mojo and more I listen to it, even to RME Adi2. Soundstage and separation is out of this world.
I guess linear power source would do another small step up.
@KorbenCZ. In a quoted text I was invetstigating reclocking in the audiophonics design. This matter is sorted out. They do use reclocking, but a primitive one, sort of things I would avoid. It is a marketting trick, trying to catch people with TXCO oscilator that (IHMO) does more harm than good. It is noted however that there is a proper bypassing of each TDA1387 chip individually.
Each one of us has an individual upgrade path. Many years ago the first DAC I could accept was in a Rotel CD player RCD971 (Burr-Brown PCM63 R2R chip), subsequently tweaked very hard for perfection - and it was working. Today, many years passed, I also stand for R2R design, nothing has changed. My current research has stopped at the Audio GD R2R11. It is because of lack of founds. With more founds it could be R28, R27... As all DACs you mention are very capable, but not a R2R design, I think you are also R2R guy. This is a tip for a further development. Enjoy it.
Each one of us has an individual upgrade path. Many years ago the first DAC I could accept was in a Rotel CD player RCD971 (Burr-Brown PCM63 R2R chip), subsequently tweaked very hard for perfection - and it was working. Today, many years passed, I also stand for R2R design, nothing has changed. My current research has stopped at the Audio GD R2R11. It is because of lack of founds. With more founds it could be R28, R27...
As all DACs you mention are very capable, but not a R2R design, I think you are also R2R guy. This is a tip for a further development. Enjoy it.
Last edited: