I'm trying that now (listening through DAC outputs to O2 headphone amp and Sony MDR-7506 headphones). Switched from 32bit/384kHz, High Quality to 32bit/352.8kHz, High Quality. There's very little, if any difference. But...
I have a recording with a very loud cymbal crash at one point (Joe Henderson, "Miles Ahead" from "So Near, So Far'). I couldn't swear to it, but I think that cymbal crash sounds ever-so-slightly less hissy and more like a cymbal with the oversampling set at 352.8kHz rather than 384kHz. The difference is tiny, but I think I can hear it (more like feel it).
Then again, if you gave me a double-blind test and asked me to guess which was which, I bet I'd come out being right only half the time, at best.
--
Half an hour later, I think I can reliably notice a difference. I like 16/44.1k upsampled to 32/352.8k better than 32/384k. The difference is barely noticeable.
--
I have a recording with a very loud cymbal crash at one point (Joe Henderson, "Miles Ahead" from "So Near, So Far'). I couldn't swear to it, but I think that cymbal crash sounds ever-so-slightly less hissy and more like a cymbal with the oversampling set at 352.8kHz rather than 384kHz. The difference is tiny, but I think I can hear it (more like feel it).
Then again, if you gave me a double-blind test and asked me to guess which was which, I bet I'd come out being right only half the time, at best.
--
Half an hour later, I think I can reliably notice a difference. I like 16/44.1k upsampled to 32/352.8k better than 32/384k. The difference is barely noticeable.
--
Last edited:
You would likely find yourself using a USB soundcard with a digital (SPDIF) input. The cheap(est) is based around CM6206LX and costs about $10. You would become quite familiar with the Linux "amixer, alsamixer" commands and possibly ".asoundrc and asound.conf".
Alloc mini BOSS?
miniBoss i2s dac for PI Zero
Though ISTR that may have been available a while ago....
@olsond3,
With the original Boss DAC, I found it to sound better in slave-mode on a Kali AND an Isolator, just like the original Piano 2.1 in dual-mono mode.
With the Boss 1.2, my results were more mixed. The Boss 1.2 has a very organic and seductive sound on an Isolator in my setups with good power supplies. Adding a Kali did improve the sonics a bit in typical 'audiophile' measures such as slightly increased apparent resolution, but that organic and seductive sound signature was diminished. I preferred it without the Kali myself, but could see how others might prefer how it sounded with a Kali.
My highly modified HiFiBerry DAC+Pro was similar to the original Boss... master mode notwithstanding, it was better on a Kali in slave mode.
I am clear that master mode can be made to work well and sound very good, based in my experiences with the decidedly NOT cheap Katana. I'm not clear why I heard those results on the 2 versions of the Boss. Given how many options there are when running the PCM51x2 chip in master mode AND how inexpensive the Boss 1.2 is, I don't plan to spend much time trying to understand what I heard, just recommend it as a very good 'cheap' option.
I'm personally not convinced master mode is a panacea based on my experiences with these DACs. I suspect that when running in master mode, the I2S signals coming out of the RPi have more jitter than the bit-clock signal sent into the RPi, with that jitter added by the RPi's processing. I've suspected that some gains could be made with flip-flop based synchronous reclocking just before the DAC chip, thought it is hard to implement that to work at higher sample rates. BUT the DAC chip's architecture and the specific implementation may matter most on whether master mode works well.
Just some thoughts. My recommendations remain:
Original Boss & Piano 2.1 (in dual-mono mode) on a Kali and Isolator for best sonics.
Boss 1.2 on just an Isolator for very good sonics preserving that DAC's musical character. OR add a Kali for better, but to my ears less musical sonics.
Buyers choice.
And of course, this was to my ears in my setups. YMMV.
Greg in Mississippi
With the original Boss DAC, I found it to sound better in slave-mode on a Kali AND an Isolator, just like the original Piano 2.1 in dual-mono mode.
With the Boss 1.2, my results were more mixed. The Boss 1.2 has a very organic and seductive sound on an Isolator in my setups with good power supplies. Adding a Kali did improve the sonics a bit in typical 'audiophile' measures such as slightly increased apparent resolution, but that organic and seductive sound signature was diminished. I preferred it without the Kali myself, but could see how others might prefer how it sounded with a Kali.
My highly modified HiFiBerry DAC+Pro was similar to the original Boss... master mode notwithstanding, it was better on a Kali in slave mode.
I am clear that master mode can be made to work well and sound very good, based in my experiences with the decidedly NOT cheap Katana. I'm not clear why I heard those results on the 2 versions of the Boss. Given how many options there are when running the PCM51x2 chip in master mode AND how inexpensive the Boss 1.2 is, I don't plan to spend much time trying to understand what I heard, just recommend it as a very good 'cheap' option.
I'm personally not convinced master mode is a panacea based on my experiences with these DACs. I suspect that when running in master mode, the I2S signals coming out of the RPi have more jitter than the bit-clock signal sent into the RPi, with that jitter added by the RPi's processing. I've suspected that some gains could be made with flip-flop based synchronous reclocking just before the DAC chip, thought it is hard to implement that to work at higher sample rates. BUT the DAC chip's architecture and the specific implementation may matter most on whether master mode works well.
Just some thoughts. My recommendations remain:
Original Boss & Piano 2.1 (in dual-mono mode) on a Kali and Isolator for best sonics.
Boss 1.2 on just an Isolator for very good sonics preserving that DAC's musical character. OR add a Kali for better, but to my ears less musical sonics.
Buyers choice.
And of course, this was to my ears in my setups. YMMV.
Greg in Mississippi
Greg Stewart, From your comments you clearly have no idea how an I2S bus or likely any digital bus works. I suspect you may have received most of your education on the topic from Stereophile magazine and the like. I suggest rather than wax poetic about differences you believe you hear, maybe go look at a data sheet for an I2S device and check out the bus timing diagrams for a DAC running in I2S master mode. You may notice that the DAC, when I2S master, clocks in the data at the times determined by the DAC clock. Any clocks on the slave are irrelevant to the quality or timing of the data received. The slave is commanded to change the data value on one clock edge and the master reads the stable value on the next master clock edge when it is guaranteed to be valid. The slave has no mechanism to cause jitter on the data received by the DAC when it is run in master mode. I suggest you try some double blind testing to see if you can actually hear any difference at all. All the flowery writing in the world about invalid testing doesn't make it valid.
The idea behind master mode is that the MCLK, BCLK and LRCK are not coming from the RPi - only the DATA is.
Basically the dac chip uses MCLK to generate the BCLK & LRCK signals that are then fed to the RPi. The RPi uses them to generate the DATA signal.
The DATA signal may no longer suffer from jitter due to timing issues originating from the RPi's not-optimal clocking mechanism, but it will suffer from (the RPi's) power supply induced jitter (yes, that too is a thing) plus RPi noise.
So, there exists no perfect solution. Only compromises. Engineering 101.
Look guys.
Allo sells the Boss 1.2 with 434 "femto" seconds jitter.
That's far below any audible impact. According to Allo, they usually run the
tests with their $11 SMPS and without reclocker or isolator.
olsond3 IMO has a point.
Allo themselves agreed though - in another thread - that the extreme low jitter numbers are
most probably not the actual issue.
To achieve these jitter and related noise values, they have to look at and improve the entire board.
They agreed that exactly pinpointing what's after all causing the audible changes/improvements
is almost impossible.
Apparently "femto second jitter" looks nice in any marketing brochure. To me the important part
is not the actual jitter value though, it's the underlying quality of the board/setup to achieve such a low level.
The funny discussions start when you try to explain things without having the in-depth knowledge about the stuff you talk about.
Identifing single-digit ps of jitter variations by ear I'd consider quite a challenge.
Pretty similar to hearing -1dB at 20kHz.
Seriously.
I pretty much trust Gregs listening test results. By swapping and combining boards,
power supplies and parts you'll cause tons of side effects.
IMO no doubt that these actions will change the result.
Talking about listening experiences by comparing this against that IMO is
a very valid approach. No question.
That's basically pretty much all, most of us around here can do.
It obviously bears the risk that you fool yourself here and there.
What should be avoided though:
Trying to explain what's going on. That puts you on slippery soil.
You'd better be carefool what you say.
Enjoy.
Allo sells the Boss 1.2 with 434 "femto" seconds jitter.
That's far below any audible impact. According to Allo, they usually run the
tests with their $11 SMPS and without reclocker or isolator.
olsond3 IMO has a point.
Allo themselves agreed though - in another thread - that the extreme low jitter numbers are
most probably not the actual issue.
To achieve these jitter and related noise values, they have to look at and improve the entire board.
They agreed that exactly pinpointing what's after all causing the audible changes/improvements
is almost impossible.
Apparently "femto second jitter" looks nice in any marketing brochure. To me the important part
is not the actual jitter value though, it's the underlying quality of the board/setup to achieve such a low level.
The funny discussions start when you try to explain things without having the in-depth knowledge about the stuff you talk about.
Identifing single-digit ps of jitter variations by ear I'd consider quite a challenge.
Pretty similar to hearing -1dB at 20kHz.
Seriously.
I pretty much trust Gregs listening test results. By swapping and combining boards,
power supplies and parts you'll cause tons of side effects.
IMO no doubt that these actions will change the result.
Talking about listening experiences by comparing this against that IMO is
a very valid approach. No question.
That's basically pretty much all, most of us around here can do.
It obviously bears the risk that you fool yourself here and there.
What should be avoided though:
Trying to explain what's going on. That puts you on slippery soil.
You'd better be carefool what you say.
Enjoy.
Last edited:
Yatsushiro: You sir are correct. I had a good start, "People's favorite word is their name" but went off track a bit after that. Ha.
As an engineer I come here to read what is true and supported by evidence collected using a scientifically valid method so I can build ever better equipment. Engineering. When I want to read fantastic claims in pages of poetry about mythical differences without any supporting evidence, likely mechanism or root cause, I go read Stereophile. When someone comes on here and starts making confident claims based on confirmation bias and has a hypothesis completely divorced from the design at hand I get a bit disappointed. If you want to know the truth, you need to use a method that has been shown to reveal it reliably.
As an engineer I come here to read what is true and supported by evidence collected using a scientifically valid method so I can build ever better equipment. Engineering. When I want to read fantastic claims in pages of poetry about mythical differences without any supporting evidence, likely mechanism or root cause, I go read Stereophile. When someone comes on here and starts making confident claims based on confirmation bias and has a hypothesis completely divorced from the design at hand I get a bit disappointed. If you want to know the truth, you need to use a method that has been shown to reveal it reliably.
Dimdim: "The DATA signal may no longer suffer from jitter due to timing issues originating from the RPi's not-optimal clocking mechanism, but it will suffer from (the RPi's) power supply induced jitter (yes, that too is a thing) plus RPi noise."
I don't follow. The Boss 2.1 has an excellent power supply, and if you don't trust that it has the option to power it separately from the RPi power when a jumper is removed. What is the "plus RPi noise" and mechanism for transferring noise you are speaking of?
What do you think is the audible limit for jitter anyway?
I don't follow. The Boss 2.1 has an excellent power supply, and if you don't trust that it has the option to power it separately from the RPi power when a jumper is removed. What is the "plus RPi noise" and mechanism for transferring noise you are speaking of?
What do you think is the audible limit for jitter anyway?
I quite understand. An Aeronautical Engineer by education and career, I'm here mostly to try and get a better understanding of electronics (always my weak point) on the back of building/assembling my own equipment. I like facts, figures and science, but a bit of poetry and artistic licence every now and then is quite enjoyable, and arguably adds a bit of colour to an otherwise black and white environment. It wasn't so much what you said, I think more to the way I read it
It's not the Boss' power supply that I'm worried about, it's the RPi's.
The RPi even when powered by a very good, low noise 5V linear power supply still has a number of on-board switch mode power supplies that inevitably pollute pretty much everything. This "pollution" is not severe enough to cause actual problems in the operation of the RPi but when it is coupled to our DACs through the I2S lines it can (and most of the time, does) cause degradation in the SQ.
This is the main reason that the RPi sounds better with a linear power supply vs. a run-of-the-mill SMPS. Noise signature. It matters even in the most carefully implemented systems. It's also the reason that different linux distributions have slightly different SQ.
We're not talking about night-and-day differences here, and you will need a good (read: resolving) sound system to even notice them, but once you do, there's no going back. You have to chase the rabbit down the rabbit hole.
These days I'm doing a p.o.c. board for a CM3-based "audio grade" RPi. It is looking promising, with noise on the I2S lines looking significantly less than that on a "standard" RPi's lines, both of them using the same good quality power supply.
Still, it's too early to say whether these electrical differences will translate to a significant SQ improvement. Time (and testing) will tell. There are still too many variables, plus my p.o.c. board is extremely far from being optimized for audio SQ.
Attachments
Here we are again.
Comparing a cheap "noisy" SPMS with a quiet high quality linear PS !?!?
What does this tell you? NOTHING! Ah No. It could mean that DimDim got stuck in one of his rabbit holes - to stay within his own chosen terminology.
Take a quiet SMPS and a quiet (same noise level) linear PS.
And then ask yourself - is there a difference - if so - what makes that difference!
It's not about SMPS vs. linear PS. It's about the resulting performance.
I could dig out several statements where "linear betters SMPS" is not being agreed upon.
And that includes my experience - especially for a PS dedicated to the RPI only.
Bottom line: It's just DimDims home setup he's talking about! And it's just his RPI that is sounding better in his setup.
Polluting everything!?!? Interesting.
I'm sure you measured this "onboard switch-mode power supplies" originated pollution and it's impact on "everything".
There are no more power supplies onboard btw! There are DC/DC converters or regulators.
Good to have you around. Finally you found the answer why operating systems have an impact on sound experience.
The whole audio world is looking for that answer for more than a decade.
You have to chase the rabbit down the rabbit hole.
Getting stuck with your head in rabbit holes while chasing rabbits I consider counter productive though.
There's a huge risk you'll loose the big picture. The rabbit might leave at a different exit.
Just look at the Allo Datasheets. It's pretty well known what's left of
that overwhelming noise and jitter. Close to nothing.
And here you can read a nice story about the iFi iPower (SMPS) noise and related measurements.
Enjoy.
Here we are again.
Comparing a cheap "noisy" SPMS with a quiet high quality linear PS !?!?
What does this tell you? NOTHING! Ah No. It could mean that DimDim got stuck in one of his rabbit holes - to stay within his own chosen terminology.
Take a quiet SMPS and a quiet (same noise level) linear PS.
And then ask yourself - is there a difference - if so - what makes that difference!
It's not about SMPS vs. linear PS. It's about the resulting performance.
I could dig out several statements where "linear betters SMPS" is not being agreed upon.
And that includes my experience - especially for a PS dedicated to the RPI only.
Bottom line: It's just DimDims home setup he's talking about! And it's just his RPI that is sounding better in his setup.
Deja Vu happening here..
Like I said, I am indeed comparing a linear power supply to a run-of-the-mill SMPS. I am very specific about what I am comparing to what.
Why am I doing that? Quite simply because that is what I have on hand and/or what I can design and build.
I do not plan to become an expert in SMPS design. I see no reason to do so, since linear power supplies are so much easier to "get right". See Allo's attempts at a "proper" SMPS and how many man hours they have sunk to it (with no finished product to show yet).
As a matter of fact, I am very close to doing just that. Seeing on my oscilloscope "fuzzy" (RPi 3) vs. "less fuzzy" (my CM3 design) I2S lines does point to exactly that.
RPi 3:
My RPi:
Mind you, they were both powered by the same linear power supply. The main difference is that on my Pi the 1.8 & 3.3V rails get generated by linear power supplies.
Semantics. A DC/DC converter that uses PWM (like the one that is used on the RPi) is very different to an LDO regulator DC/DC converter.
This answer has been around for a few years now, but many people just don't "believe in it".
Getting stuck with your head in rabbit holes while chasing rabbits I consider counter productive though.
There's a huge risk you'll loose the big picture. The rabbit might leave at a different exit.
There are many holes with many exits. Unfortunately, one man can only check one hole at a time.
You know very well that we can't really measure how things sound, and we do not have all of the explanations regarding why they sound the way they do. In theory, once you've got everything right, you will truly hear the actual sound of a specific dac chip. But you know that that will never really happen. Because engineering.
That was an interesting read. Measuring stuff with any degree of certainty is definitely tough. One does have to beware of a million different things. And it's the ones that you are not aware of that get you.
Allo Boss HW Rev 1.1
Interesting thread for a newbie like me. Setting up a RPi streamer is on my bucket list for 2019, but this is totally new ground for me. I am seriously considering buying a Allo Boss Rev 1.1 (old stock), currently available at a very good price locally, just to get things going. From what I could find the current HW version 1.4 has the following improvements over the original 1.1 version:
1) improved clocking (making use of the new “audio grade” “SDA” oscillators from NDK),
2) improved powering scheme (including the addition of a USB type C port for dedicated external power as well as a new layout),
3) and improved filtering capacitors in the output stage.
What would you recommend in terms of power supplies (or other mods) to get the most out of this older 1.1 version of the Boss? Should I just get a (good?) single supply, or should I consider two supplies to power the RPi and Boss separately? And how many amps do these supplies need to be? I read somewhere that 3A is recommended for the RPi. And are there diy options for these supplies?
Interesting thread for a newbie like me. Setting up a RPi streamer is on my bucket list for 2019, but this is totally new ground for me. I am seriously considering buying a Allo Boss Rev 1.1 (old stock), currently available at a very good price locally, just to get things going. From what I could find the current HW version 1.4 has the following improvements over the original 1.1 version:
1) improved clocking (making use of the new “audio grade” “SDA” oscillators from NDK),
2) improved powering scheme (including the addition of a USB type C port for dedicated external power as well as a new layout),
3) and improved filtering capacitors in the output stage.
What would you recommend in terms of power supplies (or other mods) to get the most out of this older 1.1 version of the Boss? Should I just get a (good?) single supply, or should I consider two supplies to power the RPi and Boss separately? And how many amps do these supplies need to be? I read somewhere that 3A is recommended for the RPi. And are there diy options for these supplies?
I'm curious as to all these experiences with the various setups and what kind of performances you others get when it comes to track changes and other actions and if any of you experience clicks or pops in the slightest. In my setup, everything is under one hood and there's no volume aside of the hardware built into the DAC, so my amplifier is running wide open and allows me to experience even the slightest anomaly. For me, I've had to go the extra step and use balanced out DAC to get rid of the artifacts that are produced when the sound system goes through it's on and off type of operations and other.. knowing and experiencing what I have so far, I couldn't do a single ended design anymore using the pi and it's current software etc...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.