In regard to perceptual analysis, there is much more to it than simply ABX. I have taken the time to study a lot of it, yet not some things. A fraction of my reference material:
I think I probably get it mostly right, but mistakes happen now and then too.
I think I probably get it mostly right, but mistakes happen now and then too.
May I ask if there is anyone eager to get to work building the boards of this thread, but who doesn't want to have new boards made from the Gerbers?
If there is a small number of people who, say, already built Marcel's RTZ dac and would like to take it to the next level, I have a few spare sets of the three boards available (clock, interconnect, and reclocker boards).
PM if interested.
If there is a small number of people who, say, already built Marcel's RTZ dac and would like to take it to the next level, I have a few spare sets of the three boards available (clock, interconnect, and reclocker boards).
PM if interested.
Hi marcelooms,
I see your point regarding ABX, but honestly,
- ABX is a popular stuff in the "hifi" forums... why not if someone buys the latest high-end gear and knows nothing about electronics : he wants its money back as an"enhanced" SQ. Then, psy bias can interfere... its person-dependant stuff.
- BUT here we are in diyaudio forum, thus miles away from most of "hifi" forum (ASR is the worst... sorry about that 🙄 ) because here you have guys/girls that, (I include myself)
1. have an idea : they "THINK" about a tech tweak (minor or major),
2. they spend TIME & often just a FEW Eur/USD to make their idea real as a nice PCB or home made prototype
3. they PRODUCE their nice PCB (great work Markw4 ! ) or "home-made tiny piece of PCB" (here is my tiny ball park 🙂!)
4. they TEST their "idea"
4.1. they have in mind => the current SQ of their hfi system (so far, it sounds "ok")
AND the previous iterations of their hifi system : harsher / simply less "natural" etc
4.2. when you test your "tweak" : you expect nothing because you don't know "what's best than my current gear"
4.3. you test your "idea" and... SQ is in par with my current SQ or worse => "fuc* me !" => the idea was wrong ? / I screwed up somewhere in the prod of the PCB or protoboard / etc...
To sum up : any diyer, @ 4.3, you're looking for the weakness of your tweak.
If the tweak enhances the SQ => it means I have to learn about this & improve it again 🙂
So, to be right : no need for ABX here
If you're wrong => ok > I go back to my design ... and find out what I missed previously
Interesting post on ABX at: https://www.diyaudio.com/community/threads/return-to-zero-shift-register-firdac.379406/post-7682348
hi Marw4,
thanks for this link.. but it's late & I fall asleep and do i have to read what's in the link : no !🙂
So, I fall asleep, but because you're driving this threat full out, dear Markw4, I admit I'll have to go thru the x posts I missed & give comments later on.
Great work ! cool !
thanks for this link.. but it's late & I fall asleep and do i have to read what's in the link : no !🙂
So, I fall asleep, but because you're driving this threat full out, dear Markw4, I admit I'll have to go thru the x posts I missed & give comments later on.
Great work ! cool !
Your memory of the SQ is not a reliable reference and without proper reference any comparison is pointless. This is why level-matched AB testing is needed. Making it blind (or double blind) more or less eliminates confirmation bias which is something you cannot escape in sighted listening. Claims on SQ improvements after tweaking or burn-in should be taken with a grain of salt as typically no proper AB testing is possible. Things go badly astray when these claims based on subjective sighted listening are portrayed as facts applicable to a wider audience.
This is a good read on the subject: http://seanolive.blogspot.com/2009/04/dishonesty-of-sighted-audio-product.html
i dont think it would be any fruitful to keep talking about audibility/abx things here, nor its Mark's intention to start this project thread, agree? we diyers are the sole responsible for our time/money that is spent (wasted)
Last edited:
Mark very kindly sent me a set of his pcbs to test out and prototype improvements to the clocks on my (two) implementations of Marcel's RTZ DAC + PCM2DSD + JLSounds I2SoverUSB.
I have spent the last couple of weeks trying various permutations of connections and XOs (from a limited selection).
It has been a very interesting experience and made very accessible due to the building block nature of Mark's design of the boards. Various combinations can be tried with ease. A very useful tool to have in the toolbox.
Regarding the results, I am still researching but I can say without a doubt that re-clocking the DSD lines to the RTZ definitely provides an improvement. There are many noticeable changes and I am not great at describing so I'm not the best person to comment, however some things I could say:
1. More focused sound-stage - accuracy of placement of ranges of frequencies from instruments/voices seems better
2. bass imaging improved for some reason.
3. HF improved (somehow, I am not sure I can describe)
4. Echoes and just "feeling the space around" is better
This was with Crystek CCHD-957 XOs.
I have two RTZ Dacs, as I've mentioned before so I can easily switch between. Both are using the same setup. ie I2SoverUSB III FIO, PCM2DSD (v3 firmware), RTZ DAC with the same DAC resistor type, opa1632 based output board and Lundahl LL1588 transformers. Both are fed from their own piCorePlayer with music synched via LMS.
I am as sure as I can be that if I make changes to one DAC I can tell the difference. Both DACs sound identical when all parts are the same. It's as good as I can do, perhaps not perfect but neither are my ears! At least I can sure that the levels will be the same and the tracks played are in synch and from the same source stream.
I will try more types of music and other XOs and report back in due course. However, many thanks to Mark for the use of the pcbs, the Kicad designs and his time and effort that will improve both my DACs once I get time.
I have spent the last couple of weeks trying various permutations of connections and XOs (from a limited selection).
It has been a very interesting experience and made very accessible due to the building block nature of Mark's design of the boards. Various combinations can be tried with ease. A very useful tool to have in the toolbox.
Regarding the results, I am still researching but I can say without a doubt that re-clocking the DSD lines to the RTZ definitely provides an improvement. There are many noticeable changes and I am not great at describing so I'm not the best person to comment, however some things I could say:
1. More focused sound-stage - accuracy of placement of ranges of frequencies from instruments/voices seems better
2. bass imaging improved for some reason.
3. HF improved (somehow, I am not sure I can describe)
4. Echoes and just "feeling the space around" is better
This was with Crystek CCHD-957 XOs.
I have two RTZ Dacs, as I've mentioned before so I can easily switch between. Both are using the same setup. ie I2SoverUSB III FIO, PCM2DSD (v3 firmware), RTZ DAC with the same DAC resistor type, opa1632 based output board and Lundahl LL1588 transformers. Both are fed from their own piCorePlayer with music synched via LMS.
I am as sure as I can be that if I make changes to one DAC I can tell the difference. Both DACs sound identical when all parts are the same. It's as good as I can do, perhaps not perfect but neither are my ears! At least I can sure that the levels will be the same and the tracks played are in synch and from the same source stream.
I will try more types of music and other XOs and report back in due course. However, many thanks to Mark for the use of the pcbs, the Kicad designs and his time and effort that will improve both my DACs once I get time.
The improvements from reclocking of DCLK were shown to be quite substantial in measurements in RTZ DAC thread. So no doubt reclocking can result in sound improvements as well. However as Marcel's RTZ dac itself reclocks DSDL and DSDR I don't see how reclocking of those would change much unless setup and hold timings are otherwise violated. Can you test reclocking of only DCLK?
@Cestrian
Hi Tony, thank you for posting about your experience so far. Maybe it will help get some of the other folks interested in trying their own projects 🙂
Mark
Hi Tony, thank you for posting about your experience so far. Maybe it will help get some of the other folks interested in trying their own projects 🙂
Mark
BTW, with the latest changes to my RTZ dac, I have switched to PCM2DSD v4 firmware. The sound is very natural, yet it is detailed enough at the lowest levels to the extent it was getting too easy (even to the point of being annoying) to hear the little bit of distortion in v3 firmware. Also with the Acko clocks and my squaring board prototype, the dynamics now are just fine with v4.
So now I am thinking using v3 firmware to get better dynamics might have been something of a bandaid fix. IOW, at this point v4 firmware does not appear to be have been the real problem.
So now I am thinking using v3 firmware to get better dynamics might have been something of a bandaid fix. IOW, at this point v4 firmware does not appear to be have been the real problem.
Last edited:
Trying to decide what to do about the squaring circuit prototype here. IMHO its probably quite a bit better than anything available now from the usual sources. The problem is that with good enough clocks it makes the dac sound so good, my pro audio designer friends would basically want to kill me if I gave it away for free. It would go over the line of helping people in the forum, yet without giving away final and especially valuable designs (not that I am in it for money, but its how those guys make their living).
So anyway, the cost of the parts to make one squarer board at this point in time (for two clocks) is roughly around $100. Add to that the cost of labor to assemble and test one, and then other costs such as warranty and technical support, to get an idea of incremental business costs.
For most consumer goods the retail markup is 3 to 5 times the incremental cost to make one unit. Say for example that its a very low production volume niche market audio device. In that case I am told the retail markup needs to be 6 times to make any kind of business sense. However some of retail cost allows for dealer profit and or distributor profits (where the distributor does marketing and first line warranty service). In that case the breakdown might be 30% to the dealer, 10% to the distributor, and 60% to the manufacturer. That way everyone can make some money for their part of the work.
Now in this particular case, if the incremental cost of making one squaring board including labor, and if sales volume is very low, then that cost might be ~$150/unit. A 6x markup gives $900! However, if no retailers in the distribution chain, and if the manufacturer takes what would otherwise be the distributor responsibilities, then we have an estimated retail price of $900 x .7 = $630/unit, (less dealer markup) for one squaring board! So if sold direct, $630 might a rough estimate sale price for one unit.
Of course, if people can afford a pair of Acko Lab clocks at $680/each, then they could probably afford another $630 for the clocks to actually work up to their potential in terms of producing good dac sound.
OTOH, most diy'ers are not using exotic sine wave clocks and most are on fairly limited budgets.
In view of all the above, I am wondering if anyone else here is interested in designing a squaring circuit suitable for the best clocks available and sharing it? If there is such interest, then I could drop some hints as to what I know about doing it. OTOH, my only other options are probably to keep what I have as a secret, or to sell boards based on realistic business numbers.
I am in a quandary as to what to do.
So anyway, the cost of the parts to make one squarer board at this point in time (for two clocks) is roughly around $100. Add to that the cost of labor to assemble and test one, and then other costs such as warranty and technical support, to get an idea of incremental business costs.
For most consumer goods the retail markup is 3 to 5 times the incremental cost to make one unit. Say for example that its a very low production volume niche market audio device. In that case I am told the retail markup needs to be 6 times to make any kind of business sense. However some of retail cost allows for dealer profit and or distributor profits (where the distributor does marketing and first line warranty service). In that case the breakdown might be 30% to the dealer, 10% to the distributor, and 60% to the manufacturer. That way everyone can make some money for their part of the work.
Now in this particular case, if the incremental cost of making one squaring board including labor, and if sales volume is very low, then that cost might be ~$150/unit. A 6x markup gives $900! However, if no retailers in the distribution chain, and if the manufacturer takes what would otherwise be the distributor responsibilities, then we have an estimated retail price of $900 x .7 = $630/unit, (less dealer markup) for one squaring board! So if sold direct, $630 might a rough estimate sale price for one unit.
Of course, if people can afford a pair of Acko Lab clocks at $680/each, then they could probably afford another $630 for the clocks to actually work up to their potential in terms of producing good dac sound.
OTOH, most diy'ers are not using exotic sine wave clocks and most are on fairly limited budgets.
In view of all the above, I am wondering if anyone else here is interested in designing a squaring circuit suitable for the best clocks available and sharing it? If there is such interest, then I could drop some hints as to what I know about doing it. OTOH, my only other options are probably to keep what I have as a secret, or to sell boards based on realistic business numbers.
I am in a quandary as to what to do.
$630 is quite a steep price and diy people aren’t usually the target audience for “high-end” hi-fi. To stay true to the forum’s spirit, sharing a schematic would be the right approach (and you could also offer the complete boards for sale).
Thank you for the comment.
Having thought about it some more, at this point there appears to be no practical way to evaluate the potential demand for such a device. Thus, no justification for spending more time on it.
Also, there are existing squaring designs that measure well enough in terms of phase noise. IMHO the problem is that phase noise measurements do not completely correlate with SQ. They don't show peak time jitter and its effect on the operation of a particular dac architecture to produce a given SQ. IOW, I'm thinking low phase noise may turn out to be a "necessary but not sufficient condition."
Having thought about it some more, at this point there appears to be no practical way to evaluate the potential demand for such a device. Thus, no justification for spending more time on it.
Also, there are existing squaring designs that measure well enough in terms of phase noise. IMHO the problem is that phase noise measurements do not completely correlate with SQ. They don't show peak time jitter and its effect on the operation of a particular dac architecture to produce a given SQ. IOW, I'm thinking low phase noise may turn out to be a "necessary but not sufficient condition."
Okay. Here is the central concept of the prototype:
Notes: One should take component values with a grain of salt. For instance, a resistor marked 805 is not 805R, rather it means 805 SMD size. Also, solder jumpers allow experimentation with signal routing options.
BTW, someone mentioned to me about this High Speed PCB Design place: https://speedingedge.com/
They have published some books and newsletters. Here is a little snippet from one book:
Turns out they don't like ferrites.
They have published some books and newsletters. Here is a little snippet from one book:
Turns out they don't like ferrites.
For EMI reasons as I read it…. But aren’t we audiophiles who do a lot of stuff “ which sounds better”, where “true engineers“ are claiming it cannot make any difference or even will make things worse. Still ears tell something different, sometimes…
For what it is worth, I did a comparison between two of my dddacmk3 boards, one with and one without the ferrites… I could not seriously make out any real sonical difference.
but that was not clock supply, but DAC chip supply.
For what it is worth, I did a comparison between two of my dddacmk3 boards, one with and one without the ferrites… I could not seriously make out any real sonical difference.
but that was not clock supply, but DAC chip supply.