TI / Burr-Brown released the 96k DF1704 in early 2001, and the 192k DF1706 a few years later, as a companion DF's for their "high-rated" PCM1704 DAC.
https://www.ti.com/lit/ds/symlink/df1704.pdf
https://media.digikey.com/pdf/Data Sheets/Texas Instruments PDFs/DF1706.pdf
What's a bit odd is that the datasheets for both the 1704 and 1706 state: " COMPANION DIGITAL FILTER FOR THE PCM1704 24-BIT AUDIO DAC"
Is there anything inherent to these DFs that makes them compatible only -- or simply ideal for use -- with PCM1704s?
Or are they generic enough to be used with other DACs?
I have a few of each in my parts bin. But I no longer have the PCM1704s. It would be interesting to use the DF1704 for another DAC in my bin, say a multi-bit dac that has no OS built in.
EDIT: Just checked my parts bin. I do have some TI and other manuf DACs that will allow their internal DFs to be bypassed. E.g., PCM1794. This might be an interesting experiment for the DF1704/6.
https://www.ti.com/lit/ds/symlink/df1704.pdf
https://media.digikey.com/pdf/Data Sheets/Texas Instruments PDFs/DF1706.pdf
What's a bit odd is that the datasheets for both the 1704 and 1706 state: " COMPANION DIGITAL FILTER FOR THE PCM1704 24-BIT AUDIO DAC"
Is there anything inherent to these DFs that makes them compatible only -- or simply ideal for use -- with PCM1704s?
Or are they generic enough to be used with other DACs?
I have a few of each in my parts bin. But I no longer have the PCM1704s. It would be interesting to use the DF1704 for another DAC in my bin, say a multi-bit dac that has no OS built in.
EDIT: Just checked my parts bin. I do have some TI and other manuf DACs that will allow their internal DFs to be bypassed. E.g., PCM1794. This might be an interesting experiment for the DF1704/6.
Last edited:
Can be used with anything beginning from PCM56, PCM61, PCM67, AD1850-60, AD1862, AD1865, PCM1702 and more.
How about TDA1541(a)? There seems to be some conflicting info on whether the 1541(a) can handle 8x OS.
See:
https://www.audialonline.com/blog/another-story-about-tda1541a-384-khz/
https://www.bramjacobse.nl/wordpress/?p=2834
But based on this old DIYA thread, it seems the DF1704/6 will work:
https://www.diyaudio.com/community/threads/tda1541a-running-in-96khz.43484/
Not sure about ultimate results. May work and measure okay ... but not sound so great (compared to SAA7220, or interpolator of similar vintage)
Are you asking these questions because you’re planning a DAC project, or merely out of intellectual curiosity?
Both/neither. "Curiosity"--intellectual or not-- can lead to plans.
Proof of concept ... dreams and magic ... UFOs ... again ...
All that said, I do have a few SSOP hobby boards that I can drop the 1704/6 on to.
It should be some sort of very simple (= low effort) contraption such as one that plugs into a 8-pin DAC socket of, say, a now-vintage CDP using a TDA1543.
The mclk should come from a unmodded CDP, like the one I have from 1989.
Some possible snags with using the DF1704 with older DACs ... looking at the DF1704 datasheet, I noted that while it will take an I2S input, the DF1704 will not output I2s:
Unclear what the DF1704 is spitting out is compatible with a DAC that only accepts I2S. E.g. TDA1387A: "I2S-bus input format (time multiplex, two’s complement, TTL)"
If not ...
The only viable DAC I have is the TDA1545A. I have several in my parts bin and may need two (see below)
https://pdf1.alldatasheet.com/datasheet-pdf/view/19218/PHILIPS/TDA1545.html
Which notes ...
The other potential snag is that the DF1704 outputs separate DOR and DOL data signals -- to feed two mono DAC chips. There seem to be no programmable option for a single Data (stereo) output. Correct?
I could run DOL to the one TDA1545A and DOR to a second TDA1545A. And then just use one output (l or R) from each TDA1454A to make stereo. Yes or no?
Or is there a clean way to combine the DOR+DOL using glue logic, and then send that to a single TDA1545A?
Unclear what the DF1704 is spitting out is compatible with a DAC that only accepts I2S. E.g. TDA1387A: "I2S-bus input format (time multiplex, two’s complement, TTL)"
If not ...
The only viable DAC I have is the TDA1545A. I have several in my parts bin and may need two (see below)
https://pdf1.alldatasheet.com/datasheet-pdf/view/19218/PHILIPS/TDA1545.html
Which notes ...
The other potential snag is that the DF1704 outputs separate DOR and DOL data signals -- to feed two mono DAC chips. There seem to be no programmable option for a single Data (stereo) output. Correct?
I could run DOL to the one TDA1545A and DOR to a second TDA1545A. And then just use one output (l or R) from each TDA1454A to make stereo. Yes or no?
Or is there a clean way to combine the DOR+DOL using glue logic, and then send that to a single TDA1545A?
Last edited:
DF1704/6 won't work with any of the Philips DACs commonly known (maybe SAA7350).
A more elegant route is to combine the DF with some known-good DAC projects that natively support the data format, just visit the thread from MIRO1360 and the work he has done with vintage AD and BB DACs.
At least the DF natively support I2S input.
A more elegant route is to combine the DF with some known-good DAC projects that natively support the data format, just visit the thread from MIRO1360 and the work he has done with vintage AD and BB DACs.
At least the DF natively support I2S input.
As noted in TDA1543/1545 thread, I did get the 1545A + DF1704 to mostly work, with very good resulting fidelity. No glue logic used.
Still have some (as of yet, unresolved) issues with a faint 2k whine and LSB distortion at low recorded levels (fade ups/downs), The faint whine can be heard with CDP in pause or stop mode as well.
May have to go into Software Mode to tweak the DF1704 data timing. In Hardware Mode, I did play with increasing WL (18-24 bit) but that added considerable distortion to MSBs.
All that said, using std. 16-bit WL, the sonics of 1545A + DF1704 are comparable to some of the best DAC project devices I have built.
The LSB distortion may be compared to playing a worn but well-cleaned LP on a properly aligned TT/cart. with something like a Nagaoka MP-110 (it can penetrate beyond the wear to deeper layers of groove). The point being that although some LP wear distortion can be heard at low levels, it is tolerable, and is almost transparent at normal music levels.
The LSB issue is a work in progress.
Still have some (as of yet, unresolved) issues with a faint 2k whine and LSB distortion at low recorded levels (fade ups/downs), The faint whine can be heard with CDP in pause or stop mode as well.
May have to go into Software Mode to tweak the DF1704 data timing. In Hardware Mode, I did play with increasing WL (18-24 bit) but that added considerable distortion to MSBs.
All that said, using std. 16-bit WL, the sonics of 1545A + DF1704 are comparable to some of the best DAC project devices I have built.
The LSB distortion may be compared to playing a worn but well-cleaned LP on a properly aligned TT/cart. with something like a Nagaoka MP-110 (it can penetrate beyond the wear to deeper layers of groove). The point being that although some LP wear distortion can be heard at low levels, it is tolerable, and is almost transparent at normal music levels.
The LSB issue is a work in progress.
But why would you do so? 1545A is for sure no match to a AD1862/AD1865 - while they would fit a DF1706 perfectly. For I2S, the Sony CXD1244S is probably the best 4x DF available that is directly compatible, why not go this route if it has to be a Philips DAC?
I'm not "sure" about anything, anywhen, anywhere, anymore, ... "thus quoth The Raven, nevermore".
The 1545A is surprisingly good in the weird pair arrangement I've stumbled into with the DF1704.
I don't have a Sony CXD1244S (or, unfortunately, any Sony DACs or DFs in my parts bin). A good, reasonable-$ source of these would be handy. As would be a source for various Yamaha "YAC" or "YM" devices.
I do have two PMD100s, which I experimented with years ago with a TDA1541A (. mixed results -- can't find the PMD100s, been misplaced for years). Also have a SM5843AP1. Both those no native I2S out.
I do have AD1862 (pair) -- and used to own PCM63 (pair) and PCM1704 (pair). Not interested in those at the moment. The call from Nature must be ... ahem .. natural.
Maybe next: PCM1794 (with DF1704 external filter) , AD1853 or AD1955 are next. As is a very high possibility of SAA7220A + TDA1387T. Or: SAA7220A + TDA1543.
=========
BTW: check the eBay prices and/or Ali availability of SAA7220 or PMD100 sometimes. Maybe the news is spreading.
Last edited:
I was able to reduce the 2k whine considerably by switching to the SLOW filter (pin 27 on DF1704). The new 3-pos DIP switch allow one to toggle the filter, and also output bit rate. Minor LSB distortion remains when signal level is very low. Maybe, segregating and significantly cleaning up pwr to 1704 and 1545's will help. Maybe.
The 1545A is extremely good bang for the buck given its very diminutive price. It gets better with more current (try it on 2mA) and better still paralleled which improves the SNR even further.
Update ...
Playing around with cleaning up power ... using dedicated 5v regulators for DF and DACs , which ran into some issues that I have postpone that part of the project. I came back to the exact same arrangement as in photos ... and, like magic, no LSB distortion, and no whine at any filter setting!! The minor issues I noted were likely due to rushed veroboard soldering or related protoytype hacking issues.
IOW ... all normal now.
It seem that the DF1704's "MSB-first" design works for tda1545A. And using only one channel for two 1545's is just fine.
Playing around with cleaning up power ... using dedicated 5v regulators for DF and DACs , which ran into some issues that I have postpone that part of the project. I came back to the exact same arrangement as in photos ... and, like magic, no LSB distortion, and no whine at any filter setting!! The minor issues I noted were likely due to rushed veroboard soldering or related protoytype hacking issues.
IOW ... all normal now.
It seem that the DF1704's "MSB-first" design works for tda1545A. And using only one channel for two 1545's is just fine.
I purch'd about 6 well over 14 years ago. Orig for the diyparadise projects -- which are just okay. Also tried to parallel simply by stacking, ala dddac 1543 experiment. Did not like that at all.
How does one give it 2mA -- I assume the CD492's defalt 5v regulation , meant for 1543, supplies plenty of juice.
The DS talks about how to give the chip 2mA output current. Use 2 * 11k resistors from VDD to pin7, rather than the default of 22k + 33k.
Not around here. Especially if you plan on stacking them. And to waste a channel to boot.
Hmm ... I was thinking about how to best "terminate" the unused channel. Just leave it open, or some cap to ground?
Out of curiosity: Experimented with plugging in a 2 x 1387's into same rig, but with the Vref pin to gnd via 1uF; not the 1545 config. And noise, as expected perhaps -- audio was barely discernible over much distortion.