Using a lot of 'glue' logic to correct for the 11.3uS inter-channel offset of a 44.1kHz sample rate is controversial. IIRC the distance offset works out to something around 0.15 in./3.8mm. So, if your head isn't locked equidistant from your speakers to within 0.15in., the correction delay is a waste. Not only would the speakers have to be that equidistant from your ears, your head would have to remain absolutely stationary as well. Of course, that won't happen. Which should give you some idea of how insignificant the correction is for listening. Although, I haven't evaluated the significance for headphone listening. Also, additional logic chips for the correction generates additional board ground noise, which may increase jitter.Miro, did you look at what Doede Duma did on DDDAC? There is a similar schematic, the same shift registers, I2S input. The DAC IC is PCM1794 also in NOS mode. I'm not expert enough to talk about it, because I'm a mechanical engineer.
You assume correctly. The delay becomes progressively shorter as the sample rate increases. Formula is: 1/Fs (Fs being the sample rate), divided by 2.
192kHz > 2.6uS (millionths of a second) = less than 1mm of offset
384kHz > 1.3uS (millionths of a second) = less than 0.5mm of offset
FYI - I’ve listened to an inter-channel delay uncorrected AD1865 based DAC for many years without noticing anything lacking in the soundstage or imaging.
Last edited:
SM5813 output voltage is VDD/2 = 5/2 = 2.5V minimum (can be more and probably is) ... PCM63 minimum for Vih is 2.4V, but Vih minimum for MC74HCU04N is 3.6V ... it looks like they removed a potential issue and added another trouble to this circuit... On the one hand, they eliminate errors resulting from voltage drops on quite long I2S signal paths, about 30 cm, but on the other hand, they cause some delays due to the imperfection of the MC74HCU04N circuits.If we don't have a filter chip, then there must be shifter registers. And a short wired connection.
I tried 1mS delay in audacity for one channel - it is almost inaudible. But when I know that the DAC has an 11uS offset (100x less), my brain is amplifying this difference unbearable
... analogy to the most expensive delta-sigma DAC which must sound better
Miro, your DAC runs five laps around them. Plus it's rock solid - I haven't turned it off for at least 100 hours and no glitches.analogy to the most expensive delta-sigma DAC which must sound better
You can't hear scrolling with the mouse, the sound of the fans and HDD from the PC - a mobile phone nearby like with ordinary Chinese DAC garbage....This is perfection....
I've tried DAC on everything......MacOS, Linux, Windows...On everything it plays exactly the same. On delta sigma there are bigger differences, depending on the system settings - here is not... Is it a real time kernel or not - whether it is ASIO or not - Album Player - whatever the driver or playback method is - from RAM - directly from the disk - everything is esoteric - even online radio stations.
Last edited:
Thank you for this detailed review and testing 🤩
This type of DAC is the way how to do DACs for listening. I can't understand why people let these chips/technology die and they decided for delta-sigma *****
Delta-sigma has its place for example in measuring instruments, mobile phones due to miniaturization, and so on.
Problem is, that many people have no choice when shopping, because everything is now delta-sigma. This "older" better sounding alternative was pushed off and dealers started selling devices according to parameters ... 32-Bit/768kHz with the most advanced digital filters must sound better than a shy 20-bit chip, not?
This type of DAC is the way how to do DACs for listening. I can't understand why people let these chips/technology die and they decided for delta-sigma *****
Delta-sigma has its place for example in measuring instruments, mobile phones due to miniaturization, and so on.
Problem is, that many people have no choice when shopping, because everything is now delta-sigma. This "older" better sounding alternative was pushed off and dealers started selling devices according to parameters ... 32-Bit/768kHz with the most advanced digital filters must sound better than a shy 20-bit chip, not?
miro,…I can't understand why people let these chips/technology die and they decided for delta-sigma *****
Sigma-Delta took over the DAC chip market because it offered a much cheaper means to achieve a given level of measured converter linearity. As you probably know, full multi-bit-resistor quantizers require very precise tolerance resistors to achieve low distortion. The greater the quantizer bit-depth, the greater the resistor precision, and value stability over temperature, that’s required. Ultra high-precision stable resistors are very costly to realize, or are simply impractical beyond a certain point. That was especially true back when Sigma-Delta was introduced, which coincided with a demand for converters of greater than 16-bit resolution for various reasons. Some reasons technical, some reasons market driven.
Sigma-Delta replaced the imprecision of reasonable cost resistors, with a fully digital process that essentially converts linearity errors into ultrasonic noise. Subjective sound quality played no role, only objective measurements. What device manufactures wanted was the least costly solution that objectively met their performance objectives. You still see the attitude today that SDM must sound superior simply because it’s basic specifications are superior.
Last edited:
finger tap my leg, foot tap the floor, shaking head, idiot smile...... i dont know why.I can't stop shaking my head when I listen to this DAC.
btw, technically tam lin is right, so i wired diyinhk ad1862 again, and compare with shift registers 1862, very little hair different.
yes yes yes, even play low resolution 56kbps or 64kbps m3u8 or mp3 online radio compare to delta sigma, sound still good...... i dont know why.Miro, your DAC runs five laps around them. Plus it's rock solid - I haven't turned it off for at least 100 hours and no glitches.
You can't hear scrolling with the mouse, the sound of the fans and HDD from the PC - a mobile phone nearby like with ordinary Chinese DAC garbage....This is perfection....
I've tried DAC on everything......MacOS, Linux, Windows...On everything it plays exactly the same. On delta sigma there are bigger differences, depending on the system settings - here is not... Is it a real time kernel or not - whether it is ASIO or not - Album Player - whatever the driver or playback method is - from RAM - directly from the disk - everything is esoteric - even online radio stations.
You need to have superhuman hearing to hear the delay anomalies at an 11uS offset.
Or that you evolved into a bat.
Otherwise, everyone lived for years with radio tuners that had a larger offset. Latency in live performance occurs naturally from the speed of sound about 3 milliseconds to travel 1 meter.
you can try bypass shift registers and change the jlsound i2susb jumper, but if me, i dont want to move anything, happy listen, keep shaking head.
JLSound has a Xilinx XC2C32A-6CPG56C CoolRunner-II CPLD chip on it
Why would it be better?
https://www.mouser.in/datasheet/2/903/ds310-2907205.pdf
Look the table inside pdf and speed
and 3.4 ns shifter
https://eu.mouser.com/datasheet/2/916/74AHC_AHCT164-1541876.pdf
Last edited:
Hi Ken
I think that in this format case when LE triggering both chanels data in the same time - there is no that kind of phase shift.
(Other thing is that in some designs LE for other channel is inverted and this inverter delay at the specific F making the shift equivalent in distance from the speaker...)
Data are already loaded in the the internal serial-to-parallel register... So if it is the case of delay in the BCK to DATA line it is not happening, because all the bits from serial to parallel has different delay
first smallest last the most.But it is not from the merit for the analog delay issues.
.
Next these parallel datas going to thermomether encoding for first 3 bits (AD1862) and 4 MSB in the AD1865. The thrmomether encoder addinig also a significant "delay" to other bits that are not managed. Again no analog delay. And it is done all with classic logic ICs inside the DACs.
.
realtion between BCK and data is locked and it is not so critical as long as in the margins of specs.
This is actually a hacking the old chip.
We do this at our own risk - because it is not supported according to the datasheet.
Even the direct JLSound method is hacking - emulation.
There is no better or worse method when hacking this chip because both are against the rules and logic.
Last edited:
BTW
The true R2R Ladder resistor DAC is PCM54 and PCM55 (up to +-12V PS)
http://www.suzushoweb.com/pdf_file/454f054a0a228.pdf
No segmented concept, not even derial to parallel register inside, nothing additionaly only core
Probably AD had some equivalents?
The true R2R Ladder resistor DAC is PCM54 and PCM55 (up to +-12V PS)
http://www.suzushoweb.com/pdf_file/454f054a0a228.pdf
No segmented concept, not even derial to parallel register inside, nothing additionaly only core
Probably AD had some equivalents?