I am bent on getting a board or a few if needed to build myself a small tinkering platform for DAC experiments. Based on what I understand for now, it looks like the power supply or supplies will be very important, a USB-2-I2S interface as well.
I may experiment with or without DAC chips: for instance, I would like to play with the DSD bitstream and filter it passively for a start.
The thing is: I'm not sure if for example just getting an Amanero or similar interface suffices for me to tinker with or if I need to add additional boards.
In the future I would certainly like to tinker with multi-channel DSD as well and perhaps even do some digital operations on the bitstreams, so for this, I am planning to look into affordable FPGA boards.
I already have high-res and DSD material, including the software to do bit-perfect output through asynchronous USB (I'm using Audirvana+ on a Mac, which I love to bits).
Let me know your thoughts, gentlemen. Thanks.
I may experiment with or without DAC chips: for instance, I would like to play with the DSD bitstream and filter it passively for a start.
The thing is: I'm not sure if for example just getting an Amanero or similar interface suffices for me to tinker with or if I need to add additional boards.
In the future I would certainly like to tinker with multi-channel DSD as well and perhaps even do some digital operations on the bitstreams, so for this, I am planning to look into affordable FPGA boards.
I already have high-res and DSD material, including the software to do bit-perfect output through asynchronous USB (I'm using Audirvana+ on a Mac, which I love to bits).
Let me know your thoughts, gentlemen. Thanks.
I haven't played with DSD myself although I did attempt a 'discrete' version of Philips' TDA1547 many, many years ago.
My thoughts would be 1) jitter is going to be a major problem and 2) use an LC filter, at least 4th order.
To address 1) I'd suggest going for an analog FIR filter by using a shift register as delay line and summing the outputs. I believe Bruno Putzeys is using something similar on his PWM-based DAC.
My thoughts would be 1) jitter is going to be a major problem and 2) use an LC filter, at least 4th order.
To address 1) I'd suggest going for an analog FIR filter by using a shift register as delay line and summing the outputs. I believe Bruno Putzeys is using something similar on his PWM-based DAC.
Thanks for the pointers, Abraxalito. I'll certainly also use an external clock and experiment with jitter-reduction with or without an FPGA board. That analog FIR Filter is intriguing.
I've been doing some head-scratching research on a board by ST Micro, and right after posting, it seems I could even get it to some USB-2-I2S, not that I'm anywhere near being able to code that tonight, but heck, I want to try
Actually, having seen the PS Audio technical presentation by Smith, I think dabbling in proper PCM decoding could be a worthwhile pursuit, even with RedBook.
My main interests though remain DSD, multi-channel as well as integration with tubes.
I've been doing some head-scratching research on a board by ST Micro, and right after posting, it seems I could even get it to some USB-2-I2S, not that I'm anywhere near being able to code that tonight, but heck, I want to try
Actually, having seen the PS Audio technical presentation by Smith, I think dabbling in proper PCM decoding could be a worthwhile pursuit, even with RedBook.
My main interests though remain DSD, multi-channel as well as integration with tubes.
I asked him over on CA how he dealt with noise modulation which seems to be inherent in DSD (based on the Lipshitz/Vanderkooy paper from way back) and he got rather political saying he wasn't going to answer any more of my questions. So obviously there's something to hide there
Ah, that was you! I just read that thread a few days ago. I'm not sure why he decided not to reply to you, but I remember that in the thread.
So counting PS Audio's DirectStream DAC, Putzey's Mola-Mola, the Lampizator, the Phasure and the Trinity, that's a good set of innovative and interesting DAC approaches.
So counting PS Audio's DirectStream DAC, Putzey's Mola-Mola, the Lampizator, the Phasure and the Trinity, that's a good set of innovative and interesting DAC approaches.
For myself I don't see anything particularly innovative about the PS audio latest. Reading their white paper sent the needle on my hype-o-meter wrapping itself around the end stop. I am though curious about what Berkeley Audio will come up with for their latest offering. That strikes me as one interesting device to watch...
If I remember correctly, B. Audio is doing the opposite, i.e. converting to PCM?
For me, on the contrary, there is a lot about Ted Smith's approach which are rather innovative (not saying somebody else didn't try similar things before), but I like the FPGA approach (here again, you can say that Chord was doing that too for instance), but he seems to be meticulous in choosing and implementing his algorithms, like validating things with numerical analysis first. This whole meticulousness while hacking/optimizing appears to be throughout his implementation. There's a lot to grasp from what he did, that technical presentation video is rife with little details.
This attention to detail is very similar to Peter's Phasure and Dietmar's Trinity DAC mindset.
To come back to my initial question: let's say I get an Amanero interface. Can I tap directly into the I2S side to get the 1-bit DSD bitstream?
For me, on the contrary, there is a lot about Ted Smith's approach which are rather innovative (not saying somebody else didn't try similar things before), but I like the FPGA approach (here again, you can say that Chord was doing that too for instance), but he seems to be meticulous in choosing and implementing his algorithms, like validating things with numerical analysis first. This whole meticulousness while hacking/optimizing appears to be throughout his implementation. There's a lot to grasp from what he did, that technical presentation video is rife with little details.
This attention to detail is very similar to Peter's Phasure and Dietmar's Trinity DAC mindset.
To come back to my initial question: let's say I get an Amanero interface. Can I tap directly into the I2S side to get the 1-bit DSD bitstream?
Yes, in my understanding the BA is converting all to PCM.
There's no doubt the guys you mention are paying attention to details. The doubts in my mind are concerned with whether they're the details that matter. What I hear in listening to DSD-recorded CDs is a kind of artificial spaciousness (auditory smoke, using an analogy with film) amd HF cloudiness. As for the Phasure and the Trinity they both use the PCM1704 which brings the inherent weakness of the R2R architecture (glitchiness) to the table which neither designer seems aware of.
There's no doubt the guys you mention are paying attention to details. The doubts in my mind are concerned with whether they're the details that matter. What I hear in listening to DSD-recorded CDs is a kind of artificial spaciousness (auditory smoke, using an analogy with film) amd HF cloudiness. As for the Phasure and the Trinity they both use the PCM1704 which brings the inherent weakness of the R2R architecture (glitchiness) to the table which neither designer seems aware of.
It all depends on what you're looking for, isn't it? If someone has released a product with 10 innovations but you're focused on a single thing, it won't matter to you, will it?
I have more of a holistic view in the sense that I perceive the listening equipment and surrounding as a chain to be optimized. Therefore, anything helping anywhere in the chain and which adds up to a perceived effect is valuable.
I have more of a holistic view in the sense that I perceive the listening equipment and surrounding as a chain to be optimized. Therefore, anything helping anywhere in the chain and which adds up to a perceived effect is valuable.
Sure it depends on what each person's looking for. For me, its SQ first and foremost determined not by measurements but from listening satisfaction. When people focus on the wrong things its a problem for SQ because it means they've not put enough effort into the things that really matter. Since engineering is always about optimizing the available resources, focussing on irrelevancies means money's wasted.
Most of what I see in 'high end' digital audio these days looks to me to be straining at gnats but swallowing camels. Take for example the Phasure, Peter selects for THD with his DACs but doesn't have a story (that I'm aware of) for why lowest THD (and we're talking down at the 0.001% level) matters so much. So this focus on THD means other areas get neglected as all designers only have limited attention. Ditto Bruno's work getting his DAC's THD down to immeasurable levels. He admits its a matter of pride (therefore not SQ).
Most of what I see in 'high end' digital audio these days looks to me to be straining at gnats but swallowing camels. Take for example the Phasure, Peter selects for THD with his DACs but doesn't have a story (that I'm aware of) for why lowest THD (and we're talking down at the 0.001% level) matters so much. So this focus on THD means other areas get neglected as all designers only have limited attention. Ditto Bruno's work getting his DAC's THD down to immeasurable levels. He admits its a matter of pride (therefore not SQ).
Last edited:
True, sometimes manufacturers play the numbers game - how long till we have 24 blades on a razor for a quick two-swipe full shave, eh?
On the other hand, since listening is subjective, you probably will recognize more or less two factions: those focusing on frequency response and those focusing on other things like soundstage, realism or what not.
It's not that one should go against the other conceptually, but in reality it does: you have for instance, people (more consumers than manufacturers here) focusing too much on frequency response, and hence missing out completely on what equipment provides for better dynamics, soundstage and so on.
Coming back to my pursuits: testing the STM32 Nucleo board is trying: ST's website is a mess and the toolchain for Mac OS X is lacking. I can't even update the firmware. Managed to get a little open source toolchain running, but then it seems it doesn't flash reliably. Now, that's a problem.
Furthermore, I've seen some example builds for USB Audio which could have been helpful as they are for the same family of uC, the STM32F4 / Cortex-M4, but the example mentioned having to set a clock at 180 MHz or so to get a proper sample rate for audio. The snag is I don't think my board goes that high.
I guess I could still dabble with it and some sub-standard PDM decoding...
I saw another board on eBay for around $180 using XMOS and doing DSD 2x, but I'm sure that if I do this, I'll soon be disappointed not being able to experiment with multi-channel. There's the Buffalo III but it's out of my budget and out of stock too.
On the other hand, since listening is subjective, you probably will recognize more or less two factions: those focusing on frequency response and those focusing on other things like soundstage, realism or what not.
It's not that one should go against the other conceptually, but in reality it does: you have for instance, people (more consumers than manufacturers here) focusing too much on frequency response, and hence missing out completely on what equipment provides for better dynamics, soundstage and so on.
Coming back to my pursuits: testing the STM32 Nucleo board is trying: ST's website is a mess and the toolchain for Mac OS X is lacking. I can't even update the firmware. Managed to get a little open source toolchain running, but then it seems it doesn't flash reliably. Now, that's a problem.
Furthermore, I've seen some example builds for USB Audio which could have been helpful as they are for the same family of uC, the STM32F4 / Cortex-M4, but the example mentioned having to set a clock at 180 MHz or so to get a proper sample rate for audio. The snag is I don't think my board goes that high.
I guess I could still dabble with it and some sub-standard PDM decoding...
I saw another board on eBay for around $180 using XMOS and doing DSD 2x, but I'm sure that if I do this, I'll soon be disappointed not being able to experiment with multi-channel. There's the Buffalo III but it's out of my budget and out of stock too.
Hello,
if you want you can consider this: as a possible starting point. Sorry, not in english.
if you want you can consider this: as a possible starting point. Sorry, not in english.
There was a "analog FIR" monolithic DSD DAC--DSD1700 of Burr Brown.
You may be able to estimate the filter coefficients from the FR graph in the datasheet. (or with PCM179x , WM8741)
You must change DSD signal "return to zero" at at analog conversion,
because of inequality of rise time and fall time of pulses.
DSD1700 is 75% of duty at each H output according to the datasheet.
Much older PDM DACs did similarly --CS4303 etc..
PSU is extremely important. its noise directly affect SNR.
You may be able to estimate the filter coefficients from the FR graph in the datasheet. (or with PCM179x , WM8741)
You must change DSD signal "return to zero" at at analog conversion,
because of inequality of rise time and fall time of pulses.
DSD1700 is 75% of duty at each H output according to the datasheet.
Much older PDM DACs did similarly --CS4303 etc..
PSU is extremely important. its noise directly affect SNR.
giefe,
Thanks, that's an awesome project and similar to what I would like to do. Not sure the STM32 Nucleo board can be good enough though. I may have a little board that could do the trick, we'll see.
Actually, saw that article from another forum, then went to search for the Nucleo and DSD output, which brought me back here, LOL.
Thanks, that's an awesome project and similar to what I would like to do. Not sure the STM32 Nucleo board can be good enough though. I may have a little board that could do the trick, we'll see.
Actually, saw that article from another forum, then went to search for the Nucleo and DSD output, which brought me back here, LOL.
http://www.webalice.it/meneghettig/Un DSD player autocostruito.pdf
Regarding a suitable usb to i2s&dsd solution for multichannel. The known exadevices is of course great, but you should also consider the UsbPal from Rigi Systems.
What you might also have read is the open hardware design from Miska/Signalyst. There's a lot to think through and learn from those designs.
I am busy with some research as well, as i feel and hear it's the way to go.
Cheers!
What you might also have read is the open hardware design from Miska/Signalyst. There's a lot to think through and learn from those designs.
I am busy with some research as well, as i feel and hear it's the way to go.
Cheers!
What's "DIY"?
Since Sony's RE-support of DSD (via their latest Walkman and headphone amp) , as well as other commercial (read: NON-DIY!!!!!!! ... the BEST AUDIO PATHOLOGY!!!!!) stuff from Audio-Technica, iDSD, Head-Direct (HiFiMan), AK, LightHarmonic, et. al. ... as well as the DSD-dedicated seminars (RMAF)/CanJam and lotsa new stuff at CES, KickStarter/indiegogo ... the best "experimental platform" may be to simply buy one of the cheap COMMERCIAL DSD offering (probably lots more coming soon) ... rip the puppies apart ... and DIY to your heart's content.
Q: What's "DIY"?
A: BTDT (been there, done that)
Since Sony's RE-support of DSD (via their latest Walkman and headphone amp) , as well as other commercial (read: NON-DIY!!!!!!! ... the BEST AUDIO PATHOLOGY!!!!!) stuff from Audio-Technica, iDSD, Head-Direct (HiFiMan), AK, LightHarmonic, et. al. ... as well as the DSD-dedicated seminars (RMAF)/CanJam and lotsa new stuff at CES, KickStarter/indiegogo ... the best "experimental platform" may be to simply buy one of the cheap COMMERCIAL DSD offering (probably lots more coming soon) ... rip the puppies apart ... and DIY to your heart's content.
An externally hosted image should be here but it was not working when we last tested it.
Q: What's "DIY"?
A: BTDT (been there, done that)
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.