You know what? Someone said I was rude to criticize anything Ian makes in a thread originated by him. I should have told my critic, that's nothing, just look at what I have to put up with.
Also, I think its important to recognize that the guys building RPi hat dac stacks to get better sound are not as interested in looking at graphs and comparing numbers as much as they are interested in good sound. If a $500, or whatever, Topping dac satisfied them then they wouldn't still be looking for better SQ. Trying to convince them they can't hear any difference blindfolded is not going to work. Every time you criticize me for weighting SQ over measurement numbers, you are also insulting all those other people too, because they feel the same way I do: The numbers don't tell the whole story of how something sounds.
Also, I think its important to recognize that the guys building RPi hat dac stacks to get better sound are not as interested in looking at graphs and comparing numbers as much as they are interested in good sound. If a $500, or whatever, Topping dac satisfied them then they wouldn't still be looking for better SQ. Trying to convince them they can't hear any difference blindfolded is not going to work. Every time you criticize me for weighting SQ over measurement numbers, you are also insulting all those other people too, because they feel the same way I do: The numbers don't tell the whole story of how something sounds.
Last edited:
Isn't telling the truth about something a defense against accusations of libel? I'm not out to hurt your feelings, and I'm not out to make money off you or your followers. Just here to tell the truth as I see it.
Okay. As the copycat that you are, are you now going to take the rest of what you have to say to your own thread?
I will also just observe that what you are really doing is mocking me. And mocking is a well recognized type of derision. https://www.scirp.org/journal/paperinformation?paperid=116357 If you would try to follow the intent of the forum rules you would probably refrain from that.
I will also just observe that what you are really doing is mocking me. And mocking is a well recognized type of derision. https://www.scirp.org/journal/paperinformation?paperid=116357 If you would try to follow the intent of the forum rules you would probably refrain from that.
Last edited:
If we could get back on topic regarding my implementation of Marcel's RTZ dac, I have been asked to explain about the interconnections and signal flows. Maybe that would be easiest starting with Cestrian's simplified and compact clock and reclocking board. If I may borrow a pic from Cestrian's thread, here is a pic of the board.
Starting at the yellow box in the upper right corner are two clocks. Dacs typically need two clocks because there are two families of digital audio sample rate frequencies. The whole right side of the board including the clocks is powered by a local voltage regulator shown in the dark blue box. In the red boxes are some control circuitry and relays to switch between the clocks based on a control signal that comes from the blue color USB board on the left side of the pic. After the correct clock signal is selected, that signal travels down into the gold-colored box area. There is a clock buffer chip there to make multiple high quality copies of the master clock signal. On the lower left corner of the gold box is a frequency doubler circuit that produces the higher clock frequencies needed by the blue USB board on the left (45/49 MHz). One of the master clock copy signals also flows down into the purple box area in the lower right corner which is where the I2S signals from USB board and FPGA board are reclocked before being sent over u.fl coax cables to the dac board.
On the left side of the pic, the blue USB board includes galvanic isolation of the USB bus from the clean power on Cestrian's board. The USB board is powered from two external mutually isolated +5v power supplies. Underneath the USB board is an FPGA board which can convert PCM audio to DSD256. The FPGA can also pass external DSD audio coming in over the USB bus. By that means HQ Player or another app can be used to do the DSD conversion, and or DSD rips from SACDs can also be played.
Thus, I2S bus signals originate at the USB board, then they are passed through the FPGA board, then they are sent over to the lower right hand area in the purple box to reclocked before being sent to the dac board.
The setup shown in the 2nd pic of the first post of this thread works about the same way except the processing is spread out over multiple boards, so it looks a little more complicated.
In addition to the I2S signals, the dac board and the clock and reclocking board require a number of external mutually isolated power supplies which are only connected together at the grounds of the load boards they supply power to. That isolation is used to help avoid ground loop problems.
There should also be an ground-breaker (aka hum-breaker) circuit, not shown here, which safely connects the dac board ground to the AC line ground in a way that minimizes the chances of problematic ground loops.
That's hopefully about enough explanation for now.
Starting at the yellow box in the upper right corner are two clocks. Dacs typically need two clocks because there are two families of digital audio sample rate frequencies. The whole right side of the board including the clocks is powered by a local voltage regulator shown in the dark blue box. In the red boxes are some control circuitry and relays to switch between the clocks based on a control signal that comes from the blue color USB board on the left side of the pic. After the correct clock signal is selected, that signal travels down into the gold-colored box area. There is a clock buffer chip there to make multiple high quality copies of the master clock signal. On the lower left corner of the gold box is a frequency doubler circuit that produces the higher clock frequencies needed by the blue USB board on the left (45/49 MHz). One of the master clock copy signals also flows down into the purple box area in the lower right corner which is where the I2S signals from USB board and FPGA board are reclocked before being sent over u.fl coax cables to the dac board.
On the left side of the pic, the blue USB board includes galvanic isolation of the USB bus from the clean power on Cestrian's board. The USB board is powered from two external mutually isolated +5v power supplies. Underneath the USB board is an FPGA board which can convert PCM audio to DSD256. The FPGA can also pass external DSD audio coming in over the USB bus. By that means HQ Player or another app can be used to do the DSD conversion, and or DSD rips from SACDs can also be played.
Thus, I2S bus signals originate at the USB board, then they are passed through the FPGA board, then they are sent over to the lower right hand area in the purple box to reclocked before being sent to the dac board.
The setup shown in the 2nd pic of the first post of this thread works about the same way except the processing is spread out over multiple boards, so it looks a little more complicated.
In addition to the I2S signals, the dac board and the clock and reclocking board require a number of external mutually isolated power supplies which are only connected together at the grounds of the load boards they supply power to. That isolation is used to help avoid ground loop problems.
There should also be an ground-breaker (aka hum-breaker) circuit, not shown here, which safely connects the dac board ground to the AC line ground in a way that minimizes the chances of problematic ground loops.
That's hopefully about enough explanation for now.
Last edited:
Indeed - and thats why I seek the best numbers available...
And then there is the spectrum of systems where DACs are used.
//
A potential trap there is you may be overly reliant on the numbers that happen to be easiest to measure, and neglect things that are real but in some cases much harder to measure and thus much less frequently, if ever, measured in a meaningful way.
Have to agree with you on that point.
Last edited:
Oh, well. In reference to #365 above, it kind of looks like trying to explain how things work is not actually of much interest.
Is it good to use mechanical relays to switch 22/24 MHz signals, instead of IC (logic, mux)?
Alex.
I believe so. Less phase noise, at least with gold contacts. Its what Andrea Mori used, and what Iancanada uses for his squaring board. Andrea has phase noise measurement equipment, and IIUC, Ian has some of that now too. I expect the choice was based on measurements. However as I have mentioned before, I'm not so sure if phase noise plots tell us all we would like to know.
Other than that, there is some other stuff going on with Cestrian's board that may not be visually obvious to everyone. MU caps are used in places that IMHO affect the sound in relation to MCLK signal processing. Also, surface CU bypass current loops are used; IOW vias are avoided for bypass current loops. In addition the voltage regulator is "tuned" a little bit with a load resistor to ground. These are things that I believe have a beneficial affect the sound. Some of it is low cost to do, although the MU caps and relays do add some cost. Also, my preference is to avoid ferrites unless they are proven to be needed. Regarding the cost, if using S-cut crystal clocks/oscillators such as SC Pure, Andrea Mori DRIXO, Acko, etc., then the project is already not going to be very low cost. If spending that kind of money on clock sources, IMHO its appropriate to use whatever MCLK circuitry the designer(s) believe is good without too much regard to cost. In the interests of disclosure: Although we do use listening tests here with trained listeners, they are not formal, publication quality tests. Sometimes blinding is used but not always.
Other than that, there is some other stuff going on with Cestrian's board that may not be visually obvious to everyone. MU caps are used in places that IMHO affect the sound in relation to MCLK signal processing. Also, surface CU bypass current loops are used; IOW vias are avoided for bypass current loops. In addition the voltage regulator is "tuned" a little bit with a load resistor to ground. These are things that I believe have a beneficial affect the sound. Some of it is low cost to do, although the MU caps and relays do add some cost. Also, my preference is to avoid ferrites unless they are proven to be needed. Regarding the cost, if using S-cut crystal clocks/oscillators such as SC Pure, Andrea Mori DRIXO, Acko, etc., then the project is already not going to be very low cost. If spending that kind of money on clock sources, IMHO its appropriate to use whatever MCLK circuitry the designer(s) believe is good without too much regard to cost. In the interests of disclosure: Although we do use listening tests here with trained listeners, they are not formal, publication quality tests. Sometimes blinding is used but not always.
Last edited:
@Markw4 , connecting to post #348 the explanation in post #365 is talking about the following (wanted to confirm) -
For the pic in post #365 I see there are three power supply inputs (a) two 15V and (b) one 5V. What is the peak current they expect? Wanted to check what power supply can be used here.
Another question is regarding the I2SoverUSB board. Its documentation has multiple configurations. How is it configured in this case?
Thanks.
For the pic in post #365 I see there are three power supply inputs (a) two 15V and (b) one 5V. What is the peak current they expect? Wanted to check what power supply can be used here.
Another question is regarding the I2SoverUSB board. Its documentation has multiple configurations. How is it configured in this case?
Thanks.
Actually, they are all +5v power supplies. I mentioned Cestrian that someone might mistake the V15 marking to think it refers to voltage. In reality, it refers to Version 15 of Cestrian's design iterations. It was only at somewhere maybe around V13 that I offered to assist. Also, one of the screw terminal strips next to the V15 marking is a status signal output for MUTE and DSD_ON. The other +5v power input is hidden under the USB board.
I use power supplies from Aliexpress or ebay. They are based on LT1083 or LT1084 because they tend to sound good for analog audio and because the load boards generally have local secondary regulators for the sensitive circuitry. What is important is that each external regulator board be isolated from all the others. That means no shared grounds at the voltage regulator boards and no shared or center-tapped power transformer secondary windings. IOW, each external regulator board needs to have its own dedicated power transformer secondary winding. Also, the FPGA board can share power with the clean +5v supply for the USB board (that's the only power sharing exception).
As a side note: The Chinese power supply boards output stability can be improved by replacing the original trim pots used for voltage adjustment with quality Bourns trim pots.
In addition to the three power supplies used by the clock and reclocking board, the dac board needs 3 power supplies of its own. Also, if the output stage is a powered one then it will also need its own external power supplies.
The USB board is configured for external clocking, and to output a 22/24MHz MCLK for the FPGA board. This is accomplished by installing two zero ohm resistors on the bottom of the USB board. The USB board is also configured not to run on USB power for the dirty side. That option requires cutting a solder jumper trace on the bottom of the USB board.
Last edited:
Forgot to mention something new: Cestrian is working on a V16 clock and reclocker board with basically a minor change that won't affect most users. The change adds u.fl connectors on the bottom of the board to bring in external clocks. That's to allow for people using Andrea Mori DRIXO or Acko clocks that require external squaring. In that case clock switching would probably be done at the squaring board and the output of that then brought into one of the new added u.fl connectors.
For anyone already using existing V15 clock and reclocking boards, its easy enough to hack on one or two u.fl connectors if needed. I can provide instructions for how I have done it before if anyone would like to see one way of doing it.
Also, an external squaring board would require an additional +5v (or maybe +3.3v) power supply.
For anyone already using existing V15 clock and reclocking boards, its easy enough to hack on one or two u.fl connectors if needed. I can provide instructions for how I have done it before if anyone would like to see one way of doing it.
Also, an external squaring board would require an additional +5v (or maybe +3.3v) power supply.
So, is anyone interested in a brief overview of the dac board itself? If, so maybe it would be possible to write a simplified theory of operation for it. But before designing this dac @MarcelvdG published an article in Linear Audio magazine about a tube dac project. The article is full of information and references about sigma-delta DSD dacs, so its intended to be a semi-tutorial to learn from. Attached is a copy if anyone would like to see it.
Looks like there have been a few downloads of @MarcelvdG Linear Audio article. Before we go on to look at the RTZ dac design from Marcel, are there any questions about the above article? I will be happy to take a stab at answering the easy ones. Marcel may chime in to correct me or to further explain some point. He's a good guy and a very good engineer.
You would be welcome to cover the solid state version if you wish. Have not built the solid-state version myself. Can't speak to how that one sounds with different supporting circuitry. However, if Marcel reckons the solid state version is better, I would likely be inclined to build it and see what I can find out.
Last edited:
I'm also interested in additional general information about the DSD-only version, albeit I don't have any specific questions at this point. I'm still wrapping my small-brained head around the general RTZ concept, as well as how a FIRDAC does what it does 🙂
Edit: I'm planning on building a decent number of DSD-only solid state versions, for a multichannel application (doing the PCM->DSD conversion in the box, as they say). So I'm interested in those specifics and observations later on down the line.
Just as an aside: @Markw4 thank you for all of the detail here and in other threads. I'm learning a lot here, especially from your posts. I don't at all understand why certain (now ignored) members here insist on being a-holes, and thanks much for handing it with grace.
Edit: I'm planning on building a decent number of DSD-only solid state versions, for a multichannel application (doing the PCM->DSD conversion in the box, as they say). So I'm interested in those specifics and observations later on down the line.
Just as an aside: @Markw4 thank you for all of the detail here and in other threads. I'm learning a lot here, especially from your posts. I don't at all understand why certain (now ignored) members here insist on being a-holes, and thanks much for handing it with grace.
Last edited:
As there is a thread dedicated to Marcel's RTZ dac it is more appropriate to discuss about it in that thread. Unless the purpose is to censor the discussion.
If Marcel feels that way then I would be happy to move it over there. If he is okay with keeping a simpler and more introductory version over here, then I think that may make it easier to navigate through for people who have been up until very recently plugging together modules the details of which they don't really understand.
However, if you would like to help teach less experienced folks more about basic electronics in your own way then maybe better if you open your own thread to do so.
However, if you would like to help teach less experienced folks more about basic electronics in your own way then maybe better if you open your own thread to do so.