are you serious? simply because audio as entertainment is all about perception. for instance human beings do not like odd harmonics, and if a given component in a given design will reduce them even on expense of rising even harmonics (to the certain limit of course) the whole thing will "sound better".
since the original topic is AK4499 design, perhaps it is worth to listen to someone who has had an opportunity to find out how actually this beast sounds?
Datasheet measurements are steady state. They are not the whole picture for audio, all audio circuits are operated under dynamic conditions.
I'm not standing up for the 7805, a RF engineer once showed me how a 1 volt 1uS square pulse went absolutely straight through a 7805 completely unattenuated. 1Mhz is LF in DACs
The rabbit hole goes deep.... a RF engineer once showed me how a 1 volt 1uS square pulse went absolutely straight through a 7805 completely unattenuated. 1Mhz is LF in DACs
exactly! do not underestimate the power of passive filtering !
Guys,
Please let's not get too excited about the meaning of one word. I checked Webster's and I think my use of the word test was not out of bounds for English.
However, I agree we are not trying to see if we can meet any specs at this point, we are just trying to decide whether we find one voltage regulator to be audibly preferable to another.
Simon: I will get to RF problems sometime later in another context. Not an audible problem with the eval board implementation in the area around the 7805's that I can hear.
One reason I strongly advocate for starting with an eval board and trying to get the best sound out of it that one is able to do is that when a prospective new board design is created, how will anyone know if it sounds as good as it should, as good as it might, or better than ever expected? (assuming use of skilled listening)
Please let's not get too excited about the meaning of one word. I checked Webster's and I think my use of the word test was not out of bounds for English.
However, I agree we are not trying to see if we can meet any specs at this point, we are just trying to decide whether we find one voltage regulator to be audibly preferable to another.
Simon: I will get to RF problems sometime later in another context. Not an audible problem with the eval board implementation in the area around the 7805's that I can hear.
One reason I strongly advocate for starting with an eval board and trying to get the best sound out of it that one is able to do is that when a prospective new board design is created, how will anyone know if it sounds as good as it should, as good as it might, or better than ever expected? (assuming use of skilled listening)
Last edited:
Since I have said as much as I have about the regulators I am currently using for some things (LT1083), I should probably also say I an using R-core power transformers with individual windings (no center-taps).
In addition, incoming AC to the R-cores is currently sequenced manually with some toggle switches to meet AK4499 data sheet requirements. Some DC power is also switched. (A project is underway to automate power sequencing and also to automate dac chip register setting with the Arduino.)
Moreover, AC power is cleaned up though use of one of Richard Marsh's Monster HTPS 7000 MKii power conditioners. IME they do audibly work as claimed, and are especially good for bench use where multiple bench power supplies, a laptop, and a headphone amp or power amp may all be in use at the same time. (Naturally for me, I was skeptical until I tried it experimentally.)
In addition, incoming AC to the R-cores is currently sequenced manually with some toggle switches to meet AK4499 data sheet requirements. Some DC power is also switched. (A project is underway to automate power sequencing and also to automate dac chip register setting with the Arduino.)
Moreover, AC power is cleaned up though use of one of Richard Marsh's Monster HTPS 7000 MKii power conditioners. IME they do audibly work as claimed, and are especially good for bench use where multiple bench power supplies, a laptop, and a headphone amp or power amp may all be in use at the same time. (Naturally for me, I was skeptical until I tried it experimentally.)
Last edited:
Datasheet measurements are steady state. They are not the whole picture for audio, all audio circuits are operated under dynamic conditions.
I'm not standing up for the 7805, a RF engineer once showed me how a 1 volt 1uS square pulse went absolutely straight through a 7805 completely unattenuated. 1Mhz is LF in DACs
Apparently you haven’t read the datasheet either, or are unable to.
You left out confirming the observation.
Somehow you are sure about the subject that he won't disclose.
Oh, I see, nothing about Markw4 repeating same old claims but you want to be critical of those who react to Markw4.
The devices in debate is for audio reproduction. Not the same as musical performance in front of listeners or recording device.
Do you know what that opportunity was? Since he won't answer when I asked, maybe you can ask him on the details of that.
I want to keep this project as open and modular as possible. Definitely want a USB input for my own use, but I want a I2S input, likely hidden from back panel connectors for people who want to integrate their own source, and then on the back panel, AES/EBU, Optical, BNC Coax, and RCA coax, to offer as much flexibility as possible. I'd be lying if I said this project wasn't slightly inspired by the modularity and i/o options of the Schiit Yggdasril, but the internals are going to be radically different.
Do you want to use the clocks in something like a I2SoverUSB board, or separate clocks closer to the dac chip?
In the first case above the USB board will switch the clocks. If you want to design in your own clocks, we can talk about ways to do that. I can post of pics of implemenations that are reasonably good, and discuss how they work. ...Unless of course one of the critics wants to expound on the subject first
In the first case above the USB board will switch the clocks. If you want to design in your own clocks, we can talk about ways to do that. I can post of pics of implemenations that are reasonably good, and discuss how they work. ...Unless of course one of the critics wants to expound on the subject first
Do you want to use the clocks in something like a I2SoverUSB board, or separate clocks closer to the dac chip?
In the first case above the USB board will switch the clocks. If you want to design in your own clocks, we can talk about ways to do that. I can post of pics of implemenations that are reasonably good, and discuss how they work. ...Unless of course one of the critics wants to expound on the subject first
I can show a fo=25.576MHz clock with under -170dBc/Hz @100kHz offset, -150dBc/Hz @1KHz offset and -140dBc/Hz @10Hz offset, which is about 0.2pS RMS jitter @fo, with noise corner frequency of about 20KHz. But you have to show me first why something like this is required for any DAC to sound good.
Or you can buy an even better one from Wenzel HF ULN | for big bucks.
If the sound quality isn't the issue in your DAC pursuit, doesn't it make sense for you to advise Markw4 to keep the sound quality claims out of discussion? If the sound quality is the issue and when he keeps bringing up unsubstantiated claims, doesn't it make sense for you to inquire about that claim? Just thinking out loud.
Getting back to reality...
I2SoverUSB can accept external clocks at 45/49MHz. There is no input on it for user selection of clock frequencies, they are selected automatically based on incoming USB audio sample rate. That means its clocks might not be suitable for use with AK4137 since we probably want to be able to select the reference clock frequency for it.
If the master clocks are to be included on the dac board, the will need to 45/49MHz clocks to work with I2SoverUSB. Therefore, a clock divider would be needed to derive 22/24MHz for AK4134 and AK4499 (except at the highest AK4499 PCM sample rates).
If using an Amanero USB board with local master clocks on the dac board, Amanero requires 22/24MHz clocks so no divider would be needed. However, Amanero boards do not include galvanic isolation, which means that if desired it would have to be implemented into the dac board design.
Another option might be to use I2SoverUSB to provide master clocks when playing from USB, and then switch to another master clock for use with AK4137 (only one clock is really needed for it). Switching could possibly be provided by a fast, small SMD LVCMOS multiplexer chip. In any case, I2S signal lines would have to be switched between the USB board and the AK4137 source devices.
In any case, appropriate PCB design with multiple layers and including at least one ground plane is a practical necessity, IMHO. Again, the book, Electromagnetic Compatibility Engineering, should be studied and understood. Perhaps worth mentioning, the AK4499 eval board uses 4-layers. The dac chip daughter board adds another 4-layers in that area. Some components are mounted on the bottom of each board.
I2SoverUSB can accept external clocks at 45/49MHz. There is no input on it for user selection of clock frequencies, they are selected automatically based on incoming USB audio sample rate. That means its clocks might not be suitable for use with AK4137 since we probably want to be able to select the reference clock frequency for it.
If the master clocks are to be included on the dac board, the will need to 45/49MHz clocks to work with I2SoverUSB. Therefore, a clock divider would be needed to derive 22/24MHz for AK4134 and AK4499 (except at the highest AK4499 PCM sample rates).
If using an Amanero USB board with local master clocks on the dac board, Amanero requires 22/24MHz clocks so no divider would be needed. However, Amanero boards do not include galvanic isolation, which means that if desired it would have to be implemented into the dac board design.
Another option might be to use I2SoverUSB to provide master clocks when playing from USB, and then switch to another master clock for use with AK4137 (only one clock is really needed for it). Switching could possibly be provided by a fast, small SMD LVCMOS multiplexer chip. In any case, I2S signal lines would have to be switched between the USB board and the AK4137 source devices.
In any case, appropriate PCB design with multiple layers and including at least one ground plane is a practical necessity, IMHO. Again, the book, Electromagnetic Compatibility Engineering, should be studied and understood. Perhaps worth mentioning, the AK4499 eval board uses 4-layers. The dac chip daughter board adds another 4-layers in that area. Some components are mounted on the bottom of each board.
Last edited:
But there is some catch, right? Otherwise why haven't we heard of this clock already?
No catch. I just gave a link to even better (-180dBc/Hz) phase noise oscillators from Wenzel that could be used as a PLL reference.
It is an ASRC. It bridges dissimilar clock domains. You only need the one clock.