IMHO, I don't see it as polarized as you suggest. I have a few DSD tracks, a few Hires PCM and thousands of 44.1 redbook tracks. Until I made the choice to use just one 5mHz clock I'd play them all. I have mostly ripped into WAV but also have a few FLAC. Really don't have a strong opinion on file format. I just found that optimizing for 44.1 .WAV playback can be very good, and there are a few advantages to minimal processing on a Rpi during playback. And CD's are cheap/free.
I'd be delighted to use Andea's DSD board for a few days. It would come down to how much money and hassle not religious dogma.
I'd be delighted to use Andea's DSD board for a few days. It would come down to how much money and hassle not religious dogma.
Okay, thank you for the reply.
Most of my collection is redbook too. It just happens to get played back on a DSD256 RTZ FIRDAC, as opposed to played on a Multibit R2R dac. The fact that the music happens to be stored on PMC coded media is incidental.
EDIT: Don't want to give the impression I use the DSD dac for everything just because that's the only dac here. There is a DAC_Lite board too.
Most of my collection is redbook too. It just happens to get played back on a DSD256 RTZ FIRDAC, as opposed to played on a Multibit R2R dac. The fact that the music happens to be stored on PMC coded media is incidental.
EDIT: Don't want to give the impression I use the DSD dac for everything just because that's the only dac here. There is a DAC_Lite board too.
Last edited:
Nobody expressed any interest in resistor types, so won't go into that. If anyone changes their mind later, PM would be okay.
Leaving aside the above questions for now, I probably should say about about the reproduction system here. Speakers are large panel electrostatics made by Sound Lab. They are extremely low distortion as speakers go, which means they give a pretty clear picture of the effects of upstream electronics. If the electronics have problems, and or if the electronics are about as problem free as we know how to do, either way the speakers tend to show it plainly. IOW, it might be said that the speakers are "unforgiving" of any upstream problems, where the term forgiving is sometimes used to refer to an audio device obscures problems, hopefully so that the problems are less audible and or less objectionable. However, unforgiving usually implies some obscuration/coloration of sound in some way that makes upstream devices and thus reproduced music sound more or less the same in some way. Another term for obscuration might be "masking," where the masking is not limited to frequency masking.
To cut to the chase, here is what my friend said about the tantalum nitride resistors as they sound on the system here: "At first it sounds great. Then you realize something is wrong. Its like there is a thin veil over the sound, and the top end is rolled off." Later after we turned the volume down quite a bit so we could converse comfortably, at some point he said, "You notice it a lot more a low volume."
For myself, I understood what he meant by all of his comments.
We removed all the de-glitching and or LP filter caps after the dac analog resistor array output. That mostly fixed the high end roll off sound.
When I listened to the tantalum resistors alone, I turned the volume down pretty low at put my ear up close to the speaker where there is little room sound and little air attenuation. It sounded to me like the "veiled" effect was related to some kind of (signal correlated) current noise and or excess noise that was obscuring some lower level details. At the same time it sounded like the midrange was more full and prominent, while also sounding more distorted in a way. There was digital graininess in the midrange that reminded me of the Telarc vinyl records that were originally recorded to digital.
Later when my friend and I were talking about the sound, we agreed that the quality of reproduction was diminished by the tantalum resistors. I suggested at the normal listening position it sounded kind of like mp3, but my friend responded with, "Its not that bad."
Moving along then, it also turns out there is a history of people using things like tantalum resistors and or Mundorf caps to shape the sound of a system more to their liking, even if it amounts to adding some kind of distortion (not necessarily nonlinear distortion that shows up well on an FFT), and or masking. In the case of Mundorf caps I kind of suspect they might sometimes be used as something like a "harmonic exciter" type of effect, to synthesize some high end when it sounds "too rolled off" or otherwise a bit muddy or veiled (although I haven't looked into that possibility in any detail).
Regarding the history of tantalum resistors for audio, there is a thread here in the forum no that subject: https://www.diyaudio.com/community/threads/tantalum-resistors-whats-so-different.182611/post-3741219
Also, I might mention that it is known that resistors can exhibit something called "excess noise" which is noise beyond the thermal noise associated with electrical resistance from any source. The latter noise is a function of temperature, resistance, and the bandwidth of interest. Some info on noise in thin film and metal film resistors: https://arxiv.org/pdf/2109.02448.pdf#:~:text=Excess noise arises from low,density, where α ≈1.
Now, having said all the above, IIUC the system used for listening sessions around where Andrea is located uses high sensitivity horn speakers. They are very different sounding as versus electrostatic panels, although some horns can be perceived as sounding quite "good." OTOH the panels are much less distorted, more unforgiving, and more transparent of upstream electronics.
In addition, it may be that the type of sound preferred by the listening session attendees in Andrea's location may be more of the midrange-prominent type. To some people its more important than other things such as soundstage, precision, more realistic HF such as for cymbals, tight well-defined bass, etc. OTOH, here in Auburn we value all those things I just mentioned. We would like a touch more of full, warm midrage too, but we don't think resistor shaping of the sound is the right way to get there.
Changing the subject some, the next thing we are planning to do here is restore the DSD dac to using all the the same type of SMD metal thin film resistors. However, this time they will be .05% tolerance Susumu RG. Then we will add Andrea's discrete balanced to SE converter/ output stage to the dac and evaluate the sound in that configuration.
To cut to the chase, here is what my friend said about the tantalum nitride resistors as they sound on the system here: "At first it sounds great. Then you realize something is wrong. Its like there is a thin veil over the sound, and the top end is rolled off." Later after we turned the volume down quite a bit so we could converse comfortably, at some point he said, "You notice it a lot more a low volume."
For myself, I understood what he meant by all of his comments.
We removed all the de-glitching and or LP filter caps after the dac analog resistor array output. That mostly fixed the high end roll off sound.
When I listened to the tantalum resistors alone, I turned the volume down pretty low at put my ear up close to the speaker where there is little room sound and little air attenuation. It sounded to me like the "veiled" effect was related to some kind of (signal correlated) current noise and or excess noise that was obscuring some lower level details. At the same time it sounded like the midrange was more full and prominent, while also sounding more distorted in a way. There was digital graininess in the midrange that reminded me of the Telarc vinyl records that were originally recorded to digital.
Later when my friend and I were talking about the sound, we agreed that the quality of reproduction was diminished by the tantalum resistors. I suggested at the normal listening position it sounded kind of like mp3, but my friend responded with, "Its not that bad."
Moving along then, it also turns out there is a history of people using things like tantalum resistors and or Mundorf caps to shape the sound of a system more to their liking, even if it amounts to adding some kind of distortion (not necessarily nonlinear distortion that shows up well on an FFT), and or masking. In the case of Mundorf caps I kind of suspect they might sometimes be used as something like a "harmonic exciter" type of effect, to synthesize some high end when it sounds "too rolled off" or otherwise a bit muddy or veiled (although I haven't looked into that possibility in any detail).
Regarding the history of tantalum resistors for audio, there is a thread here in the forum no that subject: https://www.diyaudio.com/community/threads/tantalum-resistors-whats-so-different.182611/post-3741219
Also, I might mention that it is known that resistors can exhibit something called "excess noise" which is noise beyond the thermal noise associated with electrical resistance from any source. The latter noise is a function of temperature, resistance, and the bandwidth of interest. Some info on noise in thin film and metal film resistors: https://arxiv.org/pdf/2109.02448.pdf#:~:text=Excess noise arises from low,density, where α ≈1.
Now, having said all the above, IIUC the system used for listening sessions around where Andrea is located uses high sensitivity horn speakers. They are very different sounding as versus electrostatic panels, although some horns can be perceived as sounding quite "good." OTOH the panels are much less distorted, more unforgiving, and more transparent of upstream electronics.
In addition, it may be that the type of sound preferred by the listening session attendees in Andrea's location may be more of the midrange-prominent type. To some people its more important than other things such as soundstage, precision, more realistic HF such as for cymbals, tight well-defined bass, etc. OTOH, here in Auburn we value all those things I just mentioned. We would like a touch more of full, warm midrage too, but we don't think resistor shaping of the sound is the right way to get there.
Changing the subject some, the next thing we are planning to do here is restore the DSD dac to using all the the same type of SMD metal thin film resistors. However, this time they will be .05% tolerance Susumu RG. Then we will add Andrea's discrete balanced to SE converter/ output stage to the dac and evaluate the sound in that configuration.
Last edited:
Yes and Yes. Resistors have played a key role in shaping the sound of my system. More so than capacitors. Find your comments about Tantalum interesting. I found very similar results using them in a simple I/V resistor. Metal film were ok. Tantalum not good. Tried the vintage Audionote non magnetic tantalum and they were terrible. Veiled and distorted. Rolled my own Manganin which are wonderful.
Last edited:
For I/V, manganin. Bulk foil worked well in tube amps. Feedback resistor is a sensitive spot. Found a Caddock TF020 Ultra Precision Film Resistor to be very good. (seems no longer available). Copper vs silver easy to hear. I like solid silver for signal. And I disliked almost all resistors in attenuators. Optocouplers ok, autoformers far better.
I think most of us here have their experience about resistors sound.
I agree about tantalum resistors, i find there's something wrong in their sound, something "false".
I think z-foil and some types of wirewound are the best sounding, manganin are very good too. Susumu RG good enough
I agree about tantalum resistors, i find there's something wrong in their sound, something "false".
I think z-foil and some types of wirewound are the best sounding, manganin are very good too. Susumu RG good enough
Hi Mark et al,
With reference to your earlier post(s) about why people may not playback using DSD I would like to offer my perspective on this. However, before doing so I would like to say that in my recent trials with simple DSD conversions I have experienced DSD playback to - possibly fundamentally - be different than what I have experienced with PCM playback. And to my ears in some respects very much more musical and natural. There appears to be a seamlessness and coherence in DSD playback that I yet have to hear in PCM playback. Without knowing I guess that this may be due to the basic simplicity of the DSD signal structure which - in its more simpler forms - does not require complex circuitry or the split of the signal between more processing devices. Indeed, I have never heard anything close to what DSD can do when it comes to nuance - be it tonal nuance, spatial nuance, speed nuances, inter-instrument nuances, well .. etc. .. I have also found that in some contexts DSD can sound almost beautiful - which is highly desirable ...
So I personally will progress with DSD when it is feasible for me to do so.
That said some of the reasons why I have not previously been playing back via DSD are the following:
- I have compared JRiver, Foobar, and HQPlayer DSD playback SQ. And while JRiver is capable of playing back DSD256 on a ~ 15 years old Lenovo laptop with 4 GB RAM w/o dropouts JRiver's processing to my ears is less detailed or resolved than HQPlayers. I also tried to measure the distortion of a sine signal played back via Jriver DSD and HQPlayer DSD (5adsm I think they are called? filters) and JRiver's distortion level was some 6-10 dBs higher. Regarding Foobar I never could make it work - sounds distortion and with noises surrounding the actual music.
Also, although it could be tempting to use a hardware-based PCM-to-DSD converter, my feel just is that they likely will not be able to achieve the SQ level of HQPlayer which may have available the processing power of a potentially powerful computer.
So this leaves me with HQPlayer as the main processing "unit" (should others have found a SQ wise equal alternative to HQPlayer I'd appreciate hearing about it, though).
- Being able to play back HQPlayer at DSD256 - or preferably higher sample rates, just for future proofing - with what appears to be the preferred "filters" as I understand it requires quite a substantial computer. A computer which thus may/might be noisy, may consume some energy, or may take up (visual) space. Thus it may be feasible to place it in another room - which may then introduce more "system complexity" when routing the processed DSD signal to the DSD DAC. PCM playback - also via HQPlayer - on the other hand appears to be much less taxing on the computer.
- Additionally, when considering how to get a signal from e.g. the internet to the DSD DAC it appears to me that either something like Roon or JRiver (or?) is required to "communicate" with the internet - when streaming music files. This incurs a higher cost and more software setup complexity. To me less desirable. When I want to play back via PCM I just connect the LAN cable for the internet and the USB cable for the PCB DAC to my normal laptop and then I can immediately play back essentially anything. Very straightforward and in practice it has proven to also be a very reliable solution, i.e. very few drop-outs etc.
- And with the above in mind: When listening to music I prefer it to be a very straightforward and relaxing process. I actually want/desire my audio setup to be completely reliable and "just work" so that I can enjoy the music and completely forget about the technical issues. And this probably is a main consideration in terms of not yet having implemented DSD. I may be mistaken here but my feel is that I can't just "flip a switch" and start listening when playing DSD ...
So, these are my reasons for not yet having implemented DSD for home listening. Yet, as I wrote above, given what I have recently heard from DSD playback I will progress with DSD in due time.
Cheers, Jesper
A P.S.: I concur with the tantalum nitride resistors - with the caveat that I have only heard the TN resistor type that PPY used in his DSC 2.x a couple of years ago. To my memory it was an 8*resistor in an SOIC 16 package - or something like this. To my ears not as open as e.g. the Z-foils, and somewhat veiled and congested/muddled.
To Markw4: Have you considered the Susumu RS as an alternative to their RG series? According to Susumu the RS series should be particularly designed for audio and have lower noise. Haven't listened to them yet, though. To my ears the Susumu RG may sound a bit "silvery" - but I have yet to find an SMD resistor that to my ears is as good as the Z-foils.
With reference to your earlier post(s) about why people may not playback using DSD I would like to offer my perspective on this. However, before doing so I would like to say that in my recent trials with simple DSD conversions I have experienced DSD playback to - possibly fundamentally - be different than what I have experienced with PCM playback. And to my ears in some respects very much more musical and natural. There appears to be a seamlessness and coherence in DSD playback that I yet have to hear in PCM playback. Without knowing I guess that this may be due to the basic simplicity of the DSD signal structure which - in its more simpler forms - does not require complex circuitry or the split of the signal between more processing devices. Indeed, I have never heard anything close to what DSD can do when it comes to nuance - be it tonal nuance, spatial nuance, speed nuances, inter-instrument nuances, well .. etc. .. I have also found that in some contexts DSD can sound almost beautiful - which is highly desirable ...
So I personally will progress with DSD when it is feasible for me to do so.
That said some of the reasons why I have not previously been playing back via DSD are the following:
- I have compared JRiver, Foobar, and HQPlayer DSD playback SQ. And while JRiver is capable of playing back DSD256 on a ~ 15 years old Lenovo laptop with 4 GB RAM w/o dropouts JRiver's processing to my ears is less detailed or resolved than HQPlayers. I also tried to measure the distortion of a sine signal played back via Jriver DSD and HQPlayer DSD (5adsm I think they are called? filters) and JRiver's distortion level was some 6-10 dBs higher. Regarding Foobar I never could make it work - sounds distortion and with noises surrounding the actual music.
Also, although it could be tempting to use a hardware-based PCM-to-DSD converter, my feel just is that they likely will not be able to achieve the SQ level of HQPlayer which may have available the processing power of a potentially powerful computer.
So this leaves me with HQPlayer as the main processing "unit" (should others have found a SQ wise equal alternative to HQPlayer I'd appreciate hearing about it, though).
- Being able to play back HQPlayer at DSD256 - or preferably higher sample rates, just for future proofing - with what appears to be the preferred "filters" as I understand it requires quite a substantial computer. A computer which thus may/might be noisy, may consume some energy, or may take up (visual) space. Thus it may be feasible to place it in another room - which may then introduce more "system complexity" when routing the processed DSD signal to the DSD DAC. PCM playback - also via HQPlayer - on the other hand appears to be much less taxing on the computer.
- Additionally, when considering how to get a signal from e.g. the internet to the DSD DAC it appears to me that either something like Roon or JRiver (or?) is required to "communicate" with the internet - when streaming music files. This incurs a higher cost and more software setup complexity. To me less desirable. When I want to play back via PCM I just connect the LAN cable for the internet and the USB cable for the PCB DAC to my normal laptop and then I can immediately play back essentially anything. Very straightforward and in practice it has proven to also be a very reliable solution, i.e. very few drop-outs etc.
- And with the above in mind: When listening to music I prefer it to be a very straightforward and relaxing process. I actually want/desire my audio setup to be completely reliable and "just work" so that I can enjoy the music and completely forget about the technical issues. And this probably is a main consideration in terms of not yet having implemented DSD. I may be mistaken here but my feel is that I can't just "flip a switch" and start listening when playing DSD ...
So, these are my reasons for not yet having implemented DSD for home listening. Yet, as I wrote above, given what I have recently heard from DSD playback I will progress with DSD in due time.
Cheers, Jesper
A P.S.: I concur with the tantalum nitride resistors - with the caveat that I have only heard the TN resistor type that PPY used in his DSC 2.x a couple of years ago. To my memory it was an 8*resistor in an SOIC 16 package - or something like this. To my ears not as open as e.g. the Z-foils, and somewhat veiled and congested/muddled.
To Markw4: Have you considered the Susumu RS as an alternative to their RG series? According to Susumu the RS series should be particularly designed for audio and have lower noise. Haven't listened to them yet, though. To my ears the Susumu RG may sound a bit "silvery" - but I have yet to find an SMD resistor that to my ears is as good as the Z-foils.
Last edited:
Hi Jesper,
You raise a number of issues that I could respond to one by one in some detail. But it would take a lot of writing to explain why a number of your intuitions turn out not to be the so much the case in practice. Some short answers below. If you need more detail on anything or if I overlooked responding to some particular concern, please let me know.
Short answer about HQ Player versus Simple DSD Converter is that the former made rather inefficient use of a powerful PC, whereas the latter is about as optimized as it can get for efficient high quality conversion. Thus Simple DSD Converter can sound very good indeed. Now, if we are talking about the very best HQ Player can do on an SOA PC that is water cooled (both CPU and GPU) for quiet operation in a listening room, then HQ Player can probably do a little better than Simple DSD Converter. If cost of a PC to get that bit of extra SQ is an issue, the very low cost solution is to buy a big disk drive and use offline conversion over time to convert your PCM media to DSD256. So far as I can tell the freeware PCM-DSD converter Windows App competes on SQ with the best I have heard from HQ Player so far.
Regarding Simple DSD Converter, it does benefit from cleaning up and reclocking the FPGA outputs. Andrea's FIFO board does that if using his DSD dac.
Solution to the issue about foobar and DSD for me is not to use foobar. I use a freeware app instead. Its called PlayPcmWin. Downside is that it requires a fair amount of memory and is a little slow to start playing (as is sometimes the case for HQ Player, depending).
Regarding cost, Andrea's dac is not low cost. By the time someone has everything needed the cost can add up to something substantial. That said, the purpose of Andrea's stuff is for it to be the best it can be (which in some cases may amount to the best he knows how to do at the time). Cost is secondary to SQ. People who buy, or who are considering Andrea's products should be aware of that.
Regarding Susumu RG versus RS, looks to me like RG is the top of the line. RS appears to be a lower grade of RG that they figure is good enough for audio. Selection guide at: https://www.susumu.co.jp/usa/product/selection-guide.php
Regarding RG metal thin film versus something like Z-foil, for a dac its very important the resistors are extremely well matched and that they track thermally. Also the resistors need to be suitable for use in an RF environment at a few tens of MHz and with fast clock edges. In that case RG may be close to as good as we can do. Did see Vishay has some .01% metal thin film parts in stock at Mouser, but the cost is high enough in the quantities I would need (128 resistors for one dac @ maybe around $2-$3/ea) so that I opted for RG for this experiment.
You raise a number of issues that I could respond to one by one in some detail. But it would take a lot of writing to explain why a number of your intuitions turn out not to be the so much the case in practice. Some short answers below. If you need more detail on anything or if I overlooked responding to some particular concern, please let me know.
Short answer about HQ Player versus Simple DSD Converter is that the former made rather inefficient use of a powerful PC, whereas the latter is about as optimized as it can get for efficient high quality conversion. Thus Simple DSD Converter can sound very good indeed. Now, if we are talking about the very best HQ Player can do on an SOA PC that is water cooled (both CPU and GPU) for quiet operation in a listening room, then HQ Player can probably do a little better than Simple DSD Converter. If cost of a PC to get that bit of extra SQ is an issue, the very low cost solution is to buy a big disk drive and use offline conversion over time to convert your PCM media to DSD256. So far as I can tell the freeware PCM-DSD converter Windows App competes on SQ with the best I have heard from HQ Player so far.
Regarding Simple DSD Converter, it does benefit from cleaning up and reclocking the FPGA outputs. Andrea's FIFO board does that if using his DSD dac.
Solution to the issue about foobar and DSD for me is not to use foobar. I use a freeware app instead. Its called PlayPcmWin. Downside is that it requires a fair amount of memory and is a little slow to start playing (as is sometimes the case for HQ Player, depending).
Regarding cost, Andrea's dac is not low cost. By the time someone has everything needed the cost can add up to something substantial. That said, the purpose of Andrea's stuff is for it to be the best it can be (which in some cases may amount to the best he knows how to do at the time). Cost is secondary to SQ. People who buy, or who are considering Andrea's products should be aware of that.
Regarding Susumu RG versus RS, looks to me like RG is the top of the line. RS appears to be a lower grade of RG that they figure is good enough for audio. Selection guide at: https://www.susumu.co.jp/usa/product/selection-guide.php
Regarding RG metal thin film versus something like Z-foil, for a dac its very important the resistors are extremely well matched and that they track thermally. Also the resistors need to be suitable for use in an RF environment at a few tens of MHz and with fast clock edges. In that case RG may be close to as good as we can do. Did see Vishay has some .01% metal thin film parts in stock at Mouser, but the cost is high enough in the quantities I would need (128 resistors for one dac @ maybe around $2-$3/ea) so that I opted for RG for this experiment.
Last edited:
The above comment apparently came to the attention of Andrea. He said he would not want to do exactly as I described because not everyone has all the other stuff needed to use the dac board, and because the board might not stand up the well to wear and tear of that type of usage.
However he did suggest an alternate idea, which would be if he supplied enough of his products to assemble one finished DSD dac-in-a-box that could be shipped from person to person to try. But there is a catch: someone would have to assemble the boards into a case to make a finished dac box, where the 'someone' would be acting in a volunteer capacity and likely look a lot like me. Have to think about that one. Or at least I would have to think about it if there were enough interest from people in trying it, given they would have to pay for shipping to the next person (presumably within one continent, not for overseas shipping). If insufficient interest, then nothing more to think about on that subject
EDIT: Probably such a dac box would not be exactly small and lightweight to ship. Would need a steel case for proper shielding, there would be multiple transformers, and the box would have to be big enough to fit all the boards without stray coupling problems between them.
Last edited:
Hi Mark ... & thanks again for your feedback
As in my post above I was mainly intending to share my perspective on DSD playback (and I reckon other people will have their unique perspectives as well) I will comment only briefly here to your posts above:
* I suppose the Simple DSD converter you are referring to is this one? https://www.diyaudio.com/community/threads/simple-dsd-modulator-for-dsc2.370177/
If this is the case ... Hmmm ... it may/might be interesting, however, wouldn't it require loading some software to it every time it is powered on? As far as I can see it is FPGA-based and I reckon the loaded program is erased if the power disappears ... right?
* Regarding the Susumu resistors I tend to interpret the figure & information you link to somewhat differently. As I understand it, the RG is their very good base resistor type - with a given set of specifications - and then the branched out resistor types improve on the RG's basic specs in the ways indicated in the arrow associated with each resistor type ... Thus, and including the information on the Susumu RS resistor page, I would still consider the RS to have lower noise than the RG. That said, however, I have tried to obtain more information about this from Susumu - yet without a useful reply.
BTW - the Vishay 0.01% resistors you found at a reasonable price - do you have a link to them? Couldn't find anything searching Mouser yesterday ...
* Regarding sending around one of Andrea's DSD converters I find this to be a very fine idea but since I prefer to implement things "my own way", so to say, my feedback would be that I am not interested in the solution you suggest (although also a fine, yet as I read your comments, slightly self-sacrificing offer ).
* Regarding playback softwares for DSD it is imperative to me that I can use the software to stream music - or videos - from the internet without an audible delay between sound and image. And it needs be a simple, straightforward and reliable - no hassle - setup to use. JRiver can do this & to my knowledge also Roon. In case I should consider PlayPcmWin it should also be able to do this ...
Well, my brief comments on this today. Thanks again for your feedback, Mark - & not least have a good day ...
Cheers, Jesper
As in my post above I was mainly intending to share my perspective on DSD playback (and I reckon other people will have their unique perspectives as well) I will comment only briefly here to your posts above:
* I suppose the Simple DSD converter you are referring to is this one? https://www.diyaudio.com/community/threads/simple-dsd-modulator-for-dsc2.370177/
If this is the case ... Hmmm ... it may/might be interesting, however, wouldn't it require loading some software to it every time it is powered on? As far as I can see it is FPGA-based and I reckon the loaded program is erased if the power disappears ... right?
* Regarding the Susumu resistors I tend to interpret the figure & information you link to somewhat differently. As I understand it, the RG is their very good base resistor type - with a given set of specifications - and then the branched out resistor types improve on the RG's basic specs in the ways indicated in the arrow associated with each resistor type ... Thus, and including the information on the Susumu RS resistor page, I would still consider the RS to have lower noise than the RG. That said, however, I have tried to obtain more information about this from Susumu - yet without a useful reply.
BTW - the Vishay 0.01% resistors you found at a reasonable price - do you have a link to them? Couldn't find anything searching Mouser yesterday ...
* Regarding sending around one of Andrea's DSD converters I find this to be a very fine idea but since I prefer to implement things "my own way", so to say, my feedback would be that I am not interested in the solution you suggest (although also a fine, yet as I read your comments, slightly self-sacrificing offer
).* Regarding playback softwares for DSD it is imperative to me that I can use the software to stream music - or videos - from the internet without an audible delay between sound and image. And it needs be a simple, straightforward and reliable - no hassle - setup to use. JRiver can do this & to my knowledge also Roon. In case I should consider PlayPcmWin it should also be able to do this ...
Well, my brief comments on this today. Thanks again for your feedback, Mark - & not least have a good day ...
Cheers, Jesper
Hi Jesper,
Simple DSD Converter is FPGA based. However, an FPGA is a reconfigurable logic device. It can be programed to have the logic of a CPU, or of a DSP, or lots of things. That configuration is nonvolatile until such time as the FPGA is reconfigured again.
Suppose, for example, an FPGA was configured as a CPU. Does a CPU store its bootloader program internally? Some do and some don't, right? You could configure a FPGA to work either way, at least in principle. If a CPU bootloader program is external then it may be stored in an EEPROM. Well, the FPGA used in the DSD converter has the ability to load data from an EEPROM when it is powered up, which is exactly what it does. Therefore it powers up ready to go.
Also, the latency of the FPGA is pretty low. Should be fast enough to use the dac to play audio in sync with video on a computer.
Similarly, for real time streaming just use Simple DSD Converter without a PC if you want. If using HQ player for streaming the HQ Player can run on a PC in your closet, do the DSD conversion and then send it off to a streamer that feeds the dac. People do that kind of stuff all the time.
Regarding Susumu resistors, they have specifications. According to the specifications RG are tighter tolerance and better temperature tracking than RS. As explained before a dac needs exactly those things. Other than the specs both resistors are described as having the same general characteristics, such as high linearity, low noise, high resistance to humidity, etc.
Regarding noise, there are two types of resistor noise, thermal and excess. Thermal noise is a function of the resistance value. For the dac the resistance is right around 5k. Excess noise is caused by current flowing in the resistor material and or things like end cap connection design. For RG and RS, they are both of the same basic Susumu family, which is called RG. Would be kind of surprising if they had different excess noise properties while still being part of the same family. Therefore my interpretation is that the main difference is that Susumu is trying to make it easy for people looking for audio resistors to figure out what specs makes sense for that application. I believe RS is way for Susumu to provide some clarity as to their recommendation for what they think should be good enough. In other words RS is more of a branding designation rather than a differentiator from standard RG family composition and construction.
Cheers,
Mark
Simple DSD Converter is FPGA based. However, an FPGA is a reconfigurable logic device. It can be programed to have the logic of a CPU, or of a DSP, or lots of things. That configuration is nonvolatile until such time as the FPGA is reconfigured again.
Suppose, for example, an FPGA was configured as a CPU. Does a CPU store its bootloader program internally? Some do and some don't, right? You could configure a FPGA to work either way, at least in principle. If a CPU bootloader program is external then it may be stored in an EEPROM. Well, the FPGA used in the DSD converter has the ability to load data from an EEPROM when it is powered up, which is exactly what it does. Therefore it powers up ready to go.
Also, the latency of the FPGA is pretty low. Should be fast enough to use the dac to play audio in sync with video on a computer.
Similarly, for real time streaming just use Simple DSD Converter without a PC if you want. If using HQ player for streaming the HQ Player can run on a PC in your closet, do the DSD conversion and then send it off to a streamer that feeds the dac. People do that kind of stuff all the time.
Regarding Susumu resistors, they have specifications. According to the specifications RG are tighter tolerance and better temperature tracking than RS. As explained before a dac needs exactly those things. Other than the specs both resistors are described as having the same general characteristics, such as high linearity, low noise, high resistance to humidity, etc.
Regarding noise, there are two types of resistor noise, thermal and excess. Thermal noise is a function of the resistance value. For the dac the resistance is right around 5k. Excess noise is caused by current flowing in the resistor material and or things like end cap connection design. For RG and RS, they are both of the same basic Susumu family, which is called RG. Would be kind of surprising if they had different excess noise properties while still being part of the same family. Therefore my interpretation is that the main difference is that Susumu is trying to make it easy for people looking for audio resistors to figure out what specs makes sense for that application. I believe RS is way for Susumu to provide some clarity as to their recommendation for what they think should be good enough. In other words RS is more of a branding designation rather than a differentiator from standard RG family composition and construction.
Cheers,
Mark
Hi Jesper,
From taking another look at RG and RS datasheets, it appears that for RS they do claim "improved low noise thin film character even further." That would have to refer to excess noise, of course, but its not clear exactly what they mean by 'character.' Can't tell if they are referring to the sound of the excess noise being less objectionable in some way. Unfortunately they don't go into any more detail than the above quote. Also unfortunately, the other attributes we want for a dac are not as good as for the RG type I selected.
Also it turns out that some more expensive high end discrete resistor dacs use .01% tolerance parts, which is 10x better than RS can do. According to some claims, the improvement is sound is only slight. That said, don't know what someone else's 'slight' sounds like on my system.
OTOH, some manufactures use MELF resistors for discrete resistor dacs. IIRC they are limited to .1% tolerance.
Bottom line for me is probably this is like some other things in high end audio: You have to do some work and maybe spend some money to find out what you think works best for a particular application. Published specs and claims from manufacturers don't always tell the whole story.
From taking another look at RG and RS datasheets, it appears that for RS they do claim "improved low noise thin film character even further." That would have to refer to excess noise, of course, but its not clear exactly what they mean by 'character.' Can't tell if they are referring to the sound of the excess noise being less objectionable in some way. Unfortunately they don't go into any more detail than the above quote. Also unfortunately, the other attributes we want for a dac are not as good as for the RG type I selected.
Also it turns out that some more expensive high end discrete resistor dacs use .01% tolerance parts, which is 10x better than RS can do. According to some claims, the improvement is sound is only slight. That said, don't know what someone else's 'slight' sounds like on my system.
OTOH, some manufactures use MELF resistors for discrete resistor dacs. IIRC they are limited to .1% tolerance.
Bottom line for me is probably this is like some other things in high end audio: You have to do some work and maybe spend some money to find out what you think works best for a particular application. Published specs and claims from manufacturers don't always tell the whole story.
Last edited:
... "Published specs and claims from manufacturers don't always tell the whole story." ... Fully agree, Mark.
One more comment on Susumu resistors: I measured some RR1220 types of the same nominal value and found that resistors from the same "strip" typically were exceptionally well matched (to my memory around or better than 0.01%) and only maybe 5-10% of them were outside of this very precise value. And these 5-10% were then closely grouped at one or a couple of different slightly offset values. So in general my guess would be that Susumu resistors can be very closely matched if one has a high resolution ohmmeter at hand.
Cheers - Jesper
One more comment on Susumu resistors: I measured some RR1220 types of the same nominal value and found that resistors from the same "strip" typically were exceptionally well matched (to my memory around or better than 0.01%) and only maybe 5-10% of them were outside of this very precise value. And these 5-10% were then closely grouped at one or a couple of different slightly offset values. So in general my guess would be that Susumu resistors can be very closely matched if one has a high resolution ohmmeter at hand.
Cheers - Jesper
Back again to talk about where things are at for dac resistors so far. Just tried Susumu RG with the dac connected in SE mode. Have decided these resistors are not good enough. The issue with resistor type seems mostly to have to do with Excess Noise characteristics. There are SMD metal foil resistors in .01% tolerance and with better noise performance that could be tried. However, it turns out that metal thin film resistors like those that come with the dac and that I have been trying cost maybe on the order of 50-cents each. OTOH, SMD metal foil resistors are more in the $8 - $35 each price class. Price depends on which model resistor and on quantity. Minimum number of resistors needed is 64, which is enough for SE operation. That's the brief summary.
Wrote down some brief and preliminary notes to self last night after listening to RG. I will share the notes here for anyone interested:
-----------------------------------------------------------------------------------------------------------
Most excess noise is believed to be due to current noise. Current noise tends to be 1/f noise.
End cap connection noise, for example, may or may not be of that same type.
Current noise is lower for metal foil than for metal film or metal thin film resistors.
Current noise is a signal correlated noise, and for this type of FIRDAC the noise is correlated with the audio such that the more positive-going the dac output is the more noise there is, and the more negative going, the less noise. That's because current only flows in the resistors when they pulling output voltage up.
The audible effect is that the sound may be blurred/masked, dark level between instruments and voices is only semi-gray, and the overall sound tends toward soft and muted, not as much of percussive, dynamic, and or detailed.
Current noise isn't normally specified for metal film and metal thin film resistors, presumably because the numbers don't look that great.
Metal foil resistors may have a current noise specification, if they are claimed to be good it might be around: <-40dB (by some standardized measurement procedure). Compare that to HD at -120dB. What a difference, eh?
Metal thin film resistors can be had for may around 50-cents each, or something more or less on that order given the quantities needed for this dac. OTOH 805 metal foil are at best closer to $8 - $10 each in similar quantities. Maybe those resistors can be used for a $100,000 DCS dac or something, but probabably not for a diy'er affordable dac board. Andrea's products are already kind of on the high end side for hobbyist board integration projects.
Regarding the Yageo resistors that come by default with the DSD dac, they don't sound so muddled in noise as some others but they don't sound exactly natural either. Have to listen and think about it some more but wondering if some of their excess noise is of a brighter, edgier type that helps to cut through what would otherwise perceived as a bit more of correlated 1/f noise muddy/blured/soft/gray-not-black-level type sound.
Will try Panasonic metal thin film next. Still thinking about what to do next if not satisfied with them.
-----------------------------------------------------------------------------------
Despite the possibly gloomy looking notes, there is no way I can imagine going back to an IC dac such as AK4499EX or TDA1541A. No matter the problems with discrete resistors dacs, if well designed they are still IMHO the better overall choice. There are reasons some of the high audio manufacturers have been abandoning IC dacs and going with discrete resistor dacs instead. Some of those dacs are doing conversion of PCM->DSD256 in FPGAs. So anyway, I'm going to keep investigating this stuff for a bit longer.
Wrote down some brief and preliminary notes to self last night after listening to RG. I will share the notes here for anyone interested:
-----------------------------------------------------------------------------------------------------------
Most excess noise is believed to be due to current noise. Current noise tends to be 1/f noise.
End cap connection noise, for example, may or may not be of that same type.
Current noise is lower for metal foil than for metal film or metal thin film resistors.
Current noise is a signal correlated noise, and for this type of FIRDAC the noise is correlated with the audio such that the more positive-going the dac output is the more noise there is, and the more negative going, the less noise. That's because current only flows in the resistors when they pulling output voltage up.
The audible effect is that the sound may be blurred/masked, dark level between instruments and voices is only semi-gray, and the overall sound tends toward soft and muted, not as much of percussive, dynamic, and or detailed.
Current noise isn't normally specified for metal film and metal thin film resistors, presumably because the numbers don't look that great.
Metal foil resistors may have a current noise specification, if they are claimed to be good it might be around: <-40dB (by some standardized measurement procedure). Compare that to HD at -120dB. What a difference, eh?
Metal thin film resistors can be had for may around 50-cents each, or something more or less on that order given the quantities needed for this dac. OTOH 805 metal foil are at best closer to $8 - $10 each in similar quantities. Maybe those resistors can be used for a $100,000 DCS dac or something, but probabably not for a diy'er affordable dac board. Andrea's products are already kind of on the high end side for hobbyist board integration projects.
Regarding the Yageo resistors that come by default with the DSD dac, they don't sound so muddled in noise as some others but they don't sound exactly natural either. Have to listen and think about it some more but wondering if some of their excess noise is of a brighter, edgier type that helps to cut through what would otherwise perceived as a bit more of correlated 1/f noise muddy/blured/soft/gray-not-black-level type sound.
Will try Panasonic metal thin film next. Still thinking about what to do next if not satisfied with them.
-----------------------------------------------------------------------------------
Despite the possibly gloomy looking notes, there is no way I can imagine going back to an IC dac such as AK4499EX or TDA1541A. No matter the problems with discrete resistors dacs, if well designed they are still IMHO the better overall choice. There are reasons some of the high audio manufacturers have been abandoning IC dacs and going with discrete resistor dacs instead. Some of those dacs are doing conversion of PCM->DSD256 in FPGAs. So anyway, I'm going to keep investigating this stuff for a bit longer.
Last edited:
I don't know if they can fit your requirements, but i would suggest having a look at Rhopoint (wirewound Econistors are very good, but they have other good SMD resistors too) and Isabellenhutte (mainly manganin) resistors, maybe they can cost less than metal foils when purchased in good quantities but still be good enough for your needs. Both are close in clarity/contrast to Z-foils in my tests, even if not at the same level... The problem is how much "close" is enough...
Hi all,
FYI a link to an extensive investigation on various resistors' current noise spectra:
https://dcc.ligo.org/public/0002/T0900200/001/current_noise.pdf
As far as I can see - and the color resolution of my monitor allows - the Vishay TNPW types should be lowest noise SMD resistors. I have tried those, however, "for some reason" their sound to my ears is sort of grainy and grayish.
@Luca72c: Would you happen to have a link to both the Rhopoint & Isabellehütte SMD resistors? I have previously searched for isabellehütte SMD resistors (with manganin) but didn't find any ...
Cheers, Jesper
FYI a link to an extensive investigation on various resistors' current noise spectra:
https://dcc.ligo.org/public/0002/T0900200/001/current_noise.pdf
As far as I can see - and the color resolution of my monitor allows - the Vishay TNPW types should be lowest noise SMD resistors. I have tried those, however, "for some reason" their sound to my ears is sort of grainy and grayish.
@Luca72c: Would you happen to have a link to both the Rhopoint & Isabellehütte SMD resistors? I have previously searched for isabellehütte SMD resistors (with manganin) but didn't find any ...
Cheers, Jesper
Last edited: