Way off topic but i just couldn't help centering on this. Do you realise how one-sided your description sounds? All these adjectives describe the same quality. Perhaps it is the quality most important to you, but there is so much more to sound. No mention of tonality, dynamics, soundstaging...smoothness rules all.
I would run away from this type of sound as fast as i could
I'm currently using a cap to solve the issue, I just don't like the idea of using a cap as an end-game solution. I feel like I had a better reason than that though but I can't remember off the top of my head.
The offset is stable so there's a few ways of approaching other solutions. The ones I tried didn't work without degradation. I think device linearity was the issue I think I may have a workaround but it is yet untested and I haven't gotten around to it yet.
My amp can accept arbitrary levels of input DC offset without problems so I can easily just attach the I/V converter directly to the input of the amp.
As far as the other attributes, they were all flawless as far as I could tell. I could close my eyes and actually believe I was at a live performance. Depending on the recording it can get scary realistic and my brain gets confused because it thinks something or someone is in the room with me.
That is the thing that I have strived for and the reason I believe I have arrived at my "ultimate" design. Although there are still optimizations yet to be done.
The offset is stable so there's a few ways of approaching other solutions. The ones I tried didn't work without degradation. I think device linearity was the issue I think I may have a workaround but it is yet untested and I haven't gotten around to it yet.
My amp can accept arbitrary levels of input DC offset without problems so I can easily just attach the I/V converter directly to the input of the amp.
Yes I know, audio descriptions are all pretty meaningless because it is all arbitrary and relative. I spoke of the smoothness because that aspect of the sound is what is immediately noticeable when listening compared to other things out there. The sound all the way to the highest frequencies has this clean, clear, crisp, almost tangible attribute that is indescribable. Like when a violin is playing, I feel like I can reach out and touch it. It just sounds totally real. It wasn't smooth in an unnatural way, I guess the best way to describe it is it is the pure opposite of fatiguing. Not an ounce of fatigue even in the highest pitches and no dirtiness at all with all of the subtleties in tact. Tonality would probably be the next most noticeable thing. The realistism of the tone is crazy. Things like pianos, violins, and vocals are especially impressive.Way off topic but i just couldn't help centering on this. Do you realise how one-sided your description sounds? All these adjectives describe the same quality. Perhaps it is the quality most important to you, but there is so much more to sound. No mention of tonality, dynamics, soundstaging...smoothness rules all.
I would run away from this type of sound as fast as i could
As far as the other attributes, they were all flawless as far as I could tell. I could close my eyes and actually believe I was at a live performance. Depending on the recording it can get scary realistic and my brain gets confused because it thinks something or someone is in the room with me.
That is the thing that I have strived for and the reason I believe I have arrived at my "ultimate" design. Although there are still optimizations yet to be done.
Hmmm. What volume level do you like to listen at? What kind of speakers used? Measurements for any or all of the equipment available?
Also, it's not unusual for people who know you or who are friends of friends to be enthusiastic. Also, there are lots of people looking for investment opportunities outside of the stock market and real estate, so not necessarily unusual there either. Don't know how much experience you have with those type of social effects.
Also, it's not unusual for people who know you or who are friends of friends to be enthusiastic. Also, there are lots of people looking for investment opportunities outside of the stock market and real estate, so not necessarily unusual there either. Don't know how much experience you have with those type of social effects.
Last edited:
I usually listen at medium-high volume. I don't own speakers unfortunately. I also would not use speakers for understanding how good a system is because room characteristics corrupt the result. I use HD800 headphones as my main listening medium. I didn't take any screenshots of measurement because the harmonics are below the noise floor of my QA401 for the amp and the DAC output stage. I can't remember if I measured the distortion of the dac itself yet. I don't think I did.
I don't know these people. Only one of them I kind of know from the internet. Also there is a difference between "enthusiastic" and ditching their entire system and their tube-loving ways while also offering to throw money at you. One of them has already given me thousands of dollars for R&D and he is the only reason I've gotten this far.
In any case it's pointless to discuss the sound of a system you have never heard. If things go my way you'll be able to hear it for yourself soon enough. But first I need to learn how to design a dac PCB, getting back on topic.
Unlikely. There is a lot of equipment I never have heard, even if I would like to. Probably the same for you.
It might help to understand better if you could respond to some questions about measuring:
Are you using a notch filter? Do you use averaging techniques to see below the noise floor? Where is the noise floor of QA401 at?
Also, if you listen too loud it can cause ear fatigue. Depends on the person and the listening SPL level, but for me I do best a critical listening if I keep to volume level low-ish. Turning it up sounds good but hearing looses sensitivity to details quickly. For comparing dacs I usually listen perhaps 70dB SPL. Maybe even a few dB less. It is easy to listen for a bit, then turn down the volume a little and wait for your hearing to adjust, kind of like letting your eyes adjust to the dark, and you may find after a few minutes you will start to hear more details, at which point you can turn it down just a little more. By such a downward search you can find the level at which you can hear the most details. That is the level at which you should compare amps and dacs to hear the differences most clearly.
My recommendation is that you start small and start with something easier. No offense, but you are at the stage where you don't yet realize all that you don't know. At best you are going to recreate the Buffalo DAC, but that's actually unlikely, considering this is at least their 3rd iteration to incorporate lessons learned.
Get an MCU development board for an STM32 (Discovery series boards are very cheap and there is a huge community). Learn to use the I2C and SPI peripherals to read and write to registers on other chips for configuration purposes. Find some firmware that works as a USB Audio Class Device and play with it. Attach a cheap DAC like a low end Cirrus, TI, or AKM part that is easier to work with and also doesn't have ASRC built in so you can understand the basics better and you don't have to worry about destroying a DAC IC that cost $50+ a piece with a minimum order quantity > 1.
You, my friend, are trying to run before you can walk. Hardly the first person to be guilty of this. For example, consider your FIFO requirement. You should understand WHY you think you need it before you go about implementing it. I happen to think it is pointless for a blank slate design where the master clock is at the converter, and especially for a DAC IC with a built in ASRC like the ESS parts. Falling back on only "it sounds better" is not enough and shows you don't understand what it's doing. If you can explain its purpose but still insist it sounds better, then fine, go ahead.
For what it's worth, the reason people are skeptical of your previous posts in this thread about your output stage and others about your amp design, is because you are asking basic questions on one hand and then proposing world-beating performance with no evidence on the other.
Get an MCU development board for an STM32 (Discovery series boards are very cheap and there is a huge community). Learn to use the I2C and SPI peripherals to read and write to registers on other chips for configuration purposes. Find some firmware that works as a USB Audio Class Device and play with it. Attach a cheap DAC like a low end Cirrus, TI, or AKM part that is easier to work with and also doesn't have ASRC built in so you can understand the basics better and you don't have to worry about destroying a DAC IC that cost $50+ a piece with a minimum order quantity > 1.
You, my friend, are trying to run before you can walk. Hardly the first person to be guilty of this. For example, consider your FIFO requirement. You should understand WHY you think you need it before you go about implementing it. I happen to think it is pointless for a blank slate design where the master clock is at the converter, and especially for a DAC IC with a built in ASRC like the ESS parts. Falling back on only "it sounds better" is not enough and shows you don't understand what it's doing. If you can explain its purpose but still insist it sounds better, then fine, go ahead.
For what it's worth, the reason people are skeptical of your previous posts in this thread about your output stage and others about your amp design, is because you are asking basic questions on one hand and then proposing world-beating performance with no evidence on the other.
Last edited:
I've heard almost everything on the market as far as headphone systems go. I've confirmed my contraption sounds better, usually much better, than everything on the market that I've heard. If it's at least better than most things on the market then it's at least worth being on the market.
Yesterday I verified that my amp beats the THX AAA in sound which was the only amp I was really concerned about not being able to beat.
Anyway as a user on the internet, the rule of thumb is to take everything I say with a grain of salt, especially when discussing audio. No need to get so serious.
If you get ear fatigue or loss of detail from higher SPLs then there is something wrong with your system.
I get what you are saying and I agree, but what exactly is so difficult about making the PCB I am attempting to make?
If I place the DAC footprint on the PCB, place the regulators next to the pins with the guidelines laid out in the articles linked in this thread, place a UFL connector with its own plane for the XO connection, and add some more UFL connectors for the I2S inputs I don't see the big deal. I also have a buffalo board on hand for layout guidance. What am I not getting?
I know what a FiFo does. I also made it clear twice that I don't want to DIY a FIFO. So far the fifo has improved the sound of 2 dacs that have the MCLK at the converter as you say. Can the situation be resolved through other means? Probably. But I'm an idealist when I design, I don't compromise. I work down from the top rather than up from the bottom.
None of the questions I've asked have anything to do with designing analog amplifiers which is where I specialized my learning.
If what you're saying is that it's unlikely the OP will design and build something that equals the performance of a commercial unit, you're probably right. But this is diyAudio, not Shark Tank. I'm not an EE; far from it, but I've designed and built a bunch of stuff. Is my stuff of commercial quality? I doubt it. But the design, construction, and listening has given me a great deal of pleasure. Let the OP have fun.
A couple of points: First, you seem enthralled by this DAC. If it sounds wonderful to you as is, no one can argue with that. But if the OP wants to use a FIFO, let him. Second, you say that the Benchmark doesn't "use one or need one." How did you come to the conclusion that it doesn't need one? Have you measured the Benchmark with and without a FIFO? If so, can you share your results? Third, if you haven't already, spend some time on the Asynchronous I2S FIFO project thread. diyAudio member iancanada seems to know his way around mixed-signal design and implementation. He also uses a FIFO in front of ESS dacs.
Sounds like good advice. As far as I can tell, that thread is concerned with hacking inexpensive Chinese DAC boards and using headphones to determine if the hack makes the board sound more like a Benchmark. That would be a good way for a neophyte to have some fun without spending a lot of money on test equipment.
I'm not going to sign an NDA to read a datasheet on a 3.3V regulator, but ESS says on its website that the ES9311's "[o]utput voltage noise is <1μVrms from 100 Hz to 100 kHz." That's not much different than the ADM7150. But the real question is, if 1/f noise is so important, why doesn't ESS brag about the noise performance below 100Hz? (Analog Devices publishes the ADM7150's noise profile down to 10Hz). Either the ES9311 isn't good down there or ESS doesn't think it matters.
I know nothing about the Sabre DACs, I use an FPGA board as interpolation chain and digital sigma-delta and E88CC tubes as DACs myself, but in general the main issues with a sigma-delta DAC are to keep the reference voltage as clean as you possibly can and to keep the DAC clock as clean as you can. Both are sensitive to interference at any frequency that's not an exact multiple of the DAC clock frequency. Some types of sigma-delta are particularly sensitive to interference around odd multiples of half the DAC clock frequency.
That means you have to look where the high-frequency current loops are, to keep them as small as you can and to make sure there is no data or unrelated clock loop coupling into the reference or the DAC clock. Usually the current drawn from the reference has some degree of data dependence, which means that the reference has to be as low impedance as possible, also at high frequencies where board trace inductance determines the impedance.
I read in the first post that the clock will come from another board. That makes the clock wiring longer and loop areas larger than when the clock generator is on the DAC board. You may need a well-shielded and properly terminated cable for the clock then.
That means you have to look where the high-frequency current loops are, to keep them as small as you can and to make sure there is no data or unrelated clock loop coupling into the reference or the DAC clock. Usually the current drawn from the reference has some degree of data dependence, which means that the reference has to be as low impedance as possible, also at high frequencies where board trace inductance determines the impedance.
I read in the first post that the clock will come from another board. That makes the clock wiring longer and loop areas larger than when the clock generator is on the DAC board. You may need a well-shielded and properly terminated cable for the clock then.
A related thing is frequency planning: try not to choose clock frequencies such that they have differences in the middle of the audible band, just in case they end up in something that acts as a mixer (in the RF sense of the word: non-linear thing that generates sum and difference frequencies). By definition, a DAC acts as a mixer between the digital signal and its reference and clock.
If you don't know what DAC-3 you could just google it. Measurements are at Stereophile, one of the best measuring dacs in the world today, but maybe a better one will come along next week, who knows. For now it is an industry standard and I use it as such. It sounds great too, it is on the Stereophile recommended equipment list for it's sound quality, but the world class measurements are nice to have too. If there is a FIFO dac that measures as well, be happy to take a look at it.
smooth...?
i have once got such smooth sound when i was hacking my dac/amp quite some years ago. it was so smooth, like silk.
it was felt like that the dac/amp/music was so eager to get my ears to listen to it all the way (sorry no better words).
it surprised me (for a while) and of course some more testing, all recordings i have tested show the same smoothness.
then i asked myself few simple questions:
- could any human sing like that? i dont think so
- could any instruments sound like that? i dont think so
- could any live performance sound like that? i dont think so
so i concluded that such sound was simply terribly wrong, a disaster, because realism was totally lost.
then i quickly undo the mod without hesitation.
unfortunately, i cant remember what mod, maybe the analog supply caps caused some ringing/instability.
and yes, those smooth sound could impress inexperienced listeners.
is it the correct way to go? you decide and think twice!
i have once got such smooth sound when i was hacking my dac/amp quite some years ago. it was so smooth, like silk.
it was felt like that the dac/amp/music was so eager to get my ears to listen to it all the way (sorry no better words).
it surprised me (for a while) and of course some more testing, all recordings i have tested show the same smoothness.
then i asked myself few simple questions:
- could any human sing like that? i dont think so
- could any instruments sound like that? i dont think so
- could any live performance sound like that? i dont think so
so i concluded that such sound was simply terribly wrong, a disaster, because realism was totally lost.
then i quickly undo the mod without hesitation.
unfortunately, i cant remember what mod, maybe the analog supply caps caused some ringing/instability.
and yes, those smooth sound could impress inexperienced listeners.
is it the correct way to go? you decide and think twice!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.