As the old original thread went deleted or hidden.
My question about the asynchronous FIFO buffer is the following.
- there is no indication whether buffer over / under flow happens (sooner if low buffer latency used)... but sure will happen while listening, as those input and output clocks are any ppm apart and value differ over time. The temperature changes on source or play will be any different and will change may 10..50ppm.
- Easy to detect while playing an 10kHz sine ... as an "zack" noise to hear as missing or doubled samples on streaming
So how are your experiences in that matter?
Buffer underruns or overruns are always a potential problem with FIFOs. Most of the music I play is not too long, so there are periods of silence between songs and periods of clock family changes where a FIFO buffer could be reset. Therefore I have not noticed particular problems.
Moreover, IME there are other bigger problems with dacs that I would probably prioritize first, before focusing on an otherwise pretty well functioning FIFO buffer.
That said, my personal feeling is that FIFO buffering is not necessarily needed for well-implemented asynchronous USB. OTOH FIFO can sound better for SPDIF as compared to typical ASRCs, at least in many cases, and providing FIFO latency is not an issue.
Moreover, IME there are other bigger problems with dacs that I would probably prioritize first, before focusing on an otherwise pretty well functioning FIFO buffer.
That said, my personal feeling is that FIFO buffering is not necessarily needed for well-implemented asynchronous USB. OTOH FIFO can sound better for SPDIF as compared to typical ASRCs, at least in many cases, and providing FIFO latency is not an issue.
Yes. However, the main reason the FPGA based PCM->DSD converter needs something more or less like Andrea's FIFO board is mostly because of the galvanic isolation and final reclocking. The FIFO delay is kind of incidental in this particular case, we could design something that would work without it for the asynchronous USB case.
Things are running a little behind here, trying to get another project finished up. DAC upgrade maybe delayed another week.
Other projects are finished up enough that the DSD dac upgrades can start. At this point checking out latest FIFO firmware for RTZ, etc. Next step after this will to start on the hardware upgrades. Hoping get to the point of being able to do some listening tests pretty soon and reporting back with a opinion on current version of the dac, including with the tantalum nitride resistor option.
Reason for posting about this is since there is a prototype here maybe I will be the first person to be able to express an opinion. Hopefully other people will also have an opportunity to listen before too long so there can be a variety of opinions to consider.
Mark
Reason for posting about this is since there is a prototype here maybe I will be the first person to be able to express an opinion. Hopefully other people will also have an opportunity to listen before too long so there can be a variety of opinions to consider.
Mark
Said I would report back at some point with some listening impressions, although the following is preliminary:
The DSD dac has been listened to in SE output mode using the old and the new resistors. The results here are different from the results found in listening sessions in Italy.
Here, we find the old resistors to be in our judgement superior. (Note: There is also a DAC_Lite with original resistors that has been set aside for now to focus on the DSD dac.)
There has been some discussion about why resistor sound impressions are not in agreement. Here where I am the speakers are large ESL panels. Where Andrea is the system uses high-sensitivity horns. Both locations are using solid-state amplifiers. Don't know anything more about the electronics being used in Italy.
Anyway will do another listening comparison here A/B comparing the DSD dac in SE output mode with both resistor options. If there is interest maybe we can share some specific pros and cons for each resistor type sound characteristics.
After that the dac will be reconfigured for balanced mode and tried with Andrea's latest discrete balanced to SE output stage. The output stage in use now is simply a transformer-coupled-input class-A line stage with volume control. The particular transformer is developmental with no more details available.
Regarding PCM versus DSD dacs, I was told there is much more interest in PCM than DSD, which seems somewhat puzzling. Would anyone care to give an opinion as to why DSD may be of less interest? Is SQ not that important? Does PCM seem like a historically more trusted sound, whereas DSD is unfamiliar? Maybe I am missing something here?
Also, I hope it isn't that nobody believes a word I have said about our findings here?
The DSD dac has been listened to in SE output mode using the old and the new resistors. The results here are different from the results found in listening sessions in Italy.
Here, we find the old resistors to be in our judgement superior. (Note: There is also a DAC_Lite with original resistors that has been set aside for now to focus on the DSD dac.)
There has been some discussion about why resistor sound impressions are not in agreement. Here where I am the speakers are large ESL panels. Where Andrea is the system uses high-sensitivity horns. Both locations are using solid-state amplifiers. Don't know anything more about the electronics being used in Italy.
Anyway will do another listening comparison here A/B comparing the DSD dac in SE output mode with both resistor options. If there is interest maybe we can share some specific pros and cons for each resistor type sound characteristics.
After that the dac will be reconfigured for balanced mode and tried with Andrea's latest discrete balanced to SE output stage. The output stage in use now is simply a transformer-coupled-input class-A line stage with volume control. The particular transformer is developmental with no more details available.
Regarding PCM versus DSD dacs, I was told there is much more interest in PCM than DSD, which seems somewhat puzzling. Would anyone care to give an opinion as to why DSD may be of less interest? Is SQ not that important? Does PCM seem like a historically more trusted sound, whereas DSD is unfamiliar? Maybe I am missing something here?
Also, I hope it isn't that nobody believes a word I have said about our findings here?
Last edited:
Thank you for the reply. But PCM can be converted to DSD. And DSD sounds better in almost all respects. So why buy a fancy dac if convenience is more important than SQ? I don't get it.
EDIT: Also, there is a hardware 'simple DSD converter' project in the forum that Andrea and I both use if we want to play PCM through the DSD dac directly.
Moreover, there is software that supports real time PCM->DSD256 conversion. Such software functionality is becoming a more common feature of player apps, since there is enough consumer demand for it from people who prefer the sound.
EDIT: Also, there is a hardware 'simple DSD converter' project in the forum that Andrea and I both use if we want to play PCM through the DSD dac directly.
Moreover, there is software that supports real time PCM->DSD256 conversion. Such software functionality is becoming a more common feature of player apps, since there is enough consumer demand for it from people who prefer the sound.
Last edited:
Those of us who have listened to both of Andrea's dacs have found the DSD dac to sound better in almost all respects. The PCM dac is preferred by some for making the midrange sound more prominent. The DSD dac is better at precise/accurate reproduction (in a good way, not-sterile), wider deeper soundstage, more precise imaging, and maybe some other things. To me the DSD dac also sounds lower distortion (including possibly of the non-PSS type).
Therefore it would seem to make sense that for some people, if a more prominent midrange were to be of overriding importance, then the PCM dac would be the better choice.
Regarding convenience, there is a hardware PCM->DSD256 project in the forum with gerbers, BOM, and firmware. Don't see how that would be an inconvenience other than the initial assembly needed?
Therefore it would seem to make sense that for some people, if a more prominent midrange were to be of overriding importance, then the PCM dac would be the better choice.
Regarding convenience, there is a hardware PCM->DSD256 project in the forum with gerbers, BOM, and firmware. Don't see how that would be an inconvenience other than the initial assembly needed?
Last edited:
Since there is the hardware 'Simple DSD Converter' project, then doesn't the 'convenience' argument for PCM tend to evaporate?
If so; and if Andrea's DSD dac does sound better in almost all ways; then it starts to look more likely there is some deeper or more fundamental reason people still mostly choose the PCM version over DSD?
Maybe its that 'PCM' is a more familiar and comfortable term, so it feels like a safer bet?
If so; and if Andrea's DSD dac does sound better in almost all ways; then it starts to look more likely there is some deeper or more fundamental reason people still mostly choose the PCM version over DSD?
Maybe its that 'PCM' is a more familiar and comfortable term, so it feels like a safer bet?
Probably some truth there. I for one am still very interested in hearing reports about best sound from DSD even though I'm staying with PCM. I was most interested in direct DSD player using tubes by Lucas Fikus and then the valve DAC from Linear Audio by MarcelvdG. Both claimed sound qualities that left PCM in the dust. I've decided to stick with old and outdated technology of the TDA1541a playing 44.1 PCM. I know it will never play as well, but I don't want to mess with converting the entire library or real time conversion. Still keen to hear about the state of the art, so keep up the good work.
Well, if and when a Sonic Empire dac comes out, it may be that it will come with both PCM and DSD. You going to not use the DSD part?
EDIT: Just as a thought experiment, please let me pose a hypothetical: What if Andrea offered to send around DSD dac board (and maybe a set of faster clocks) that can be driven from the existing FIFO buffer. People can try it for a couple of weeks, then pay for shipping to the next person who wants to try it. Would that still be too much of a risk and or too much inconvenience? IOW, is it more that people are wedded to PCM, or does PCM just feel like a safer bet given the DSD dac is more of an unknown?
EDIT: Just as a thought experiment, please let me pose a hypothetical: What if Andrea offered to send around DSD dac board (and maybe a set of faster clocks) that can be driven from the existing FIFO buffer. People can try it for a couple of weeks, then pay for shipping to the next person who wants to try it. Would that still be too much of a risk and or too much inconvenience? IOW, is it more that people are wedded to PCM, or does PCM just feel like a safer bet given the DSD dac is more of an unknown?
Last edited: