Clock jitter can do it. Enough to mess with ITD localization, sometimes enough to collapse a sound stage. Wrong power filtering for a clock oscillator can be one cause.
Hi Markw4,
Buy the entire set. It's worth it.
Normal clock jitter cannot do it. It would have to be so bad as to be non-functional as a clock. The minuet differences people talk about simply do not matter. I know, because I have measured clock stability using an HP 5372A many times with all kinds of digital systems. Any functional crystal oscillator without major problems is good enough. Ceramic resonators are most times good enough. Much of this talk is based on "could be, might be, we don't know everything" thinking. For things to be as bad as you describe, you would have major issues everywhere in that system. I've worked in service with digital audio since CDs were first introduced. We had to figure out those gen 1 systems, and we learned a lot.
If you are talking about modified CD players and similar, then I can see the modifier making errors that might contaminate common signal paths, power supplies. One issue is that people tend to change something and point to that change as the cause. Yet they changed other things with that one change and didn't control the experiment to eliminate all other factors but the one variable. Then you have inaccurate statements supporting their view or objective.
A power supply would have to be stunningly bad to cause high jitter in a competent quartz oscillator circuit. The circuits depending on clocks detect at some point on the waveform so they have something called noise immunity. It greatly reduces susceptibility to noise, which would have to be very high to affect the clock greatly.
There is one truth. Any manufacturer attempting to sell a true high end product will not allow a problem to degrade the performance. This holds for power cords, D/A converters, coupling capacitors. They may not do the best job, but they never intentionally allow substandard circuits to degrade the entire product. Made to price point, sure. But then why bother trying to upgrade that? I've worked with enough engineering departments over decades to know this as a fact. Engineers do the best job they know how, within a budget of course. But the goal is within grasp even considering the budget (which is realistic). So to use a "cheap clock" that can affect sound quality so much. Nope, not a chance.
Buy the entire set. It's worth it.
Normal clock jitter cannot do it. It would have to be so bad as to be non-functional as a clock. The minuet differences people talk about simply do not matter. I know, because I have measured clock stability using an HP 5372A many times with all kinds of digital systems. Any functional crystal oscillator without major problems is good enough. Ceramic resonators are most times good enough. Much of this talk is based on "could be, might be, we don't know everything" thinking. For things to be as bad as you describe, you would have major issues everywhere in that system. I've worked in service with digital audio since CDs were first introduced. We had to figure out those gen 1 systems, and we learned a lot.
If you are talking about modified CD players and similar, then I can see the modifier making errors that might contaminate common signal paths, power supplies. One issue is that people tend to change something and point to that change as the cause. Yet they changed other things with that one change and didn't control the experiment to eliminate all other factors but the one variable. Then you have inaccurate statements supporting their view or objective.
A power supply would have to be stunningly bad to cause high jitter in a competent quartz oscillator circuit. The circuits depending on clocks detect at some point on the waveform so they have something called noise immunity. It greatly reduces susceptibility to noise, which would have to be very high to affect the clock greatly.
There is one truth. Any manufacturer attempting to sell a true high end product will not allow a problem to degrade the performance. This holds for power cords, D/A converters, coupling capacitors. They may not do the best job, but they never intentionally allow substandard circuits to degrade the entire product. Made to price point, sure. But then why bother trying to upgrade that? I've worked with enough engineering departments over decades to know this as a fact. Engineers do the best job they know how, within a budget of course. But the goal is within grasp even considering the budget (which is realistic). So to use a "cheap clock" that can affect sound quality so much. Nope, not a chance.
All it takes is one ferrite bead and or one nonlinear bypass cap. It affects the sound, and at least some people can hear the difference blind.
Also, Crystek clocks can sound not too bad if properly babied.
Problem with engineers is that most tend to take for granted what they read in books and application notes. They don't do their own R&D to double check. They also haven't taken the time over the years to develop listening skills. They assume standard AP tests everybody uses will show them if there is a problem.
Only dac designs where I know for a fact the engineers did extensive R&D to find out what actually works for good sound (not just good measurements) was with Wadia. Probably some other high end dacs too, but I don't know the details on those. Why did Wadia engineers do the extra work? They had a good analog design consultant who showed them where they could do better, including with power supplies (including for clocks). That was enough to convince them they needed to do more of their own R&D in the digital area too.
Also, Crystek clocks can sound not too bad if properly babied.
Problem with engineers is that most tend to take for granted what they read in books and application notes. They don't do their own R&D to double check. They also haven't taken the time over the years to develop listening skills. They assume standard AP tests everybody uses will show them if there is a problem.
Only dac designs where I know for a fact the engineers did extensive R&D to find out what actually works for good sound (not just good measurements) was with Wadia. Probably some other high end dacs too, but I don't know the details on those. Why did Wadia engineers do the extra work? They had a good analog design consultant who showed them where they could do better, including with power supplies (including for clocks). That was enough to convince them they needed to do more of their own R&D in the digital area too.
Last edited:
Good evening,
I said from the beginning that the value I am looking for is 10µF
Then I asked what technology is the best: chemical, MKP, MKS, MKT, etc.
I am leaning towards MKP (polypropylene) because in everything I have read, polypropylene is the best technology in audio, on the signal path.
In polypropylene, the lowest voltage I have found is 250VAC.
Of course, it is much too much for the few volts of input signal, but I have not found smaller.
It is also because polypropylene is a very good dielectric.
I said from the beginning that the value I am looking for is 10µF
Then I asked what technology is the best: chemical, MKP, MKS, MKT, etc.
I am leaning towards MKP (polypropylene) because in everything I have read, polypropylene is the best technology in audio, on the signal path.
In polypropylene, the lowest voltage I have found is 250VAC.
Of course, it is much too much for the few volts of input signal, but I have not found smaller.
It is also because polypropylene is a very good dielectric.
They used to make lower voltage polypropylene caps. The problem as I understand it is that thinner polypropylene film is no longer readily available in the low volumes needed by film cap manufacturers. At least that's my understanding of why REL stopping making lower voltage film caps. Thus, its use large body caps or find some other solution. One possibility to consider might be a smaller cap in combination with a high input impedance FET buffer. Alternatively, a now obsolete UKZ electrolytic in parallel with .01uf MKP (not FKP) can work pretty well, or it could anyway, before the obsolescence.
Last edited:
Hi Markw4,
All due respect ...
You just insulted every single audio engineer and technician, do you realise this? Your words come from a place of complete ignorance.
Every single technician and engineer listens, we do not assume anything. We measure and look for explanations for what we hear or reports of what people hear. We have undergone years of training, longer than many here have been alive. We have invested in test equipment, continuing to upgrade it and advance our knowledge. I have never met a single decent engineer or technician that doesn't have to know ... why? We are curious and investigate what we don't understand.
Your statements indicate you haven't got any understanding of how any of this works. Crystal oscillators do not have to be babied. Yes they are mechanical but in the Mhz regions they are robust. The 1 KHz oscillators were fragile because the crystal was so large. All that needs to be done is to drive the crystal within the manufacturers specs, not a massive problem or mystery. A ferrite bead affecting the sound? Where? Maybe misused, but not in a power supply feed, and a capacitor is across a supply, unchanging voltage. So unless it is breaking down all it does is pass high frequency content to the common. Either that or you used the wrong part and it has high impedance or resonance at the frequencies that exist on that supply line.
Hmm, just Wadia? Really? Denon, Nakamichi, Cyrus and Marantz off the top of my head. Plenty more. High end is full of "designs by ear" that don't perform well, and later it is admitted they didn't sound good either.
You made a lot of statements that don't hold up under even casual investigation. You and folks like you simply don't have a clue.
All due respect ...
You just insulted every single audio engineer and technician, do you realise this? Your words come from a place of complete ignorance.
Every single technician and engineer listens, we do not assume anything. We measure and look for explanations for what we hear or reports of what people hear. We have undergone years of training, longer than many here have been alive. We have invested in test equipment, continuing to upgrade it and advance our knowledge. I have never met a single decent engineer or technician that doesn't have to know ... why? We are curious and investigate what we don't understand.
Your statements indicate you haven't got any understanding of how any of this works. Crystal oscillators do not have to be babied. Yes they are mechanical but in the Mhz regions they are robust. The 1 KHz oscillators were fragile because the crystal was so large. All that needs to be done is to drive the crystal within the manufacturers specs, not a massive problem or mystery. A ferrite bead affecting the sound? Where? Maybe misused, but not in a power supply feed, and a capacitor is across a supply, unchanging voltage. So unless it is breaking down all it does is pass high frequency content to the common. Either that or you used the wrong part and it has high impedance or resonance at the frequencies that exist on that supply line.
Hmm, just Wadia? Really? Denon, Nakamichi, Cyrus and Marantz off the top of my head. Plenty more. High end is full of "designs by ear" that don't perform well, and later it is admitted they didn't sound good either.
You made a lot of statements that don't hold up under even casual investigation. You and folks like you simply don't have a clue.
Hi Alren,
Okay, you are then committed to a size of capacitor body since you've picked everything else. I'll tell you right now that an Electrolytic is fine unless you have stringent leakage requirements. In audio this isn't usually a consideration unless the voltage difference is high and the circuit very high impedance.
They often don't make lower voltage ratings for some film capacitors because the film is so thin, the defect rate is too high without a cost difference that makes it worthwhile. Low voltage polystyrene capacitors are an excellent example, I use 630 V units, not the 50 V types often seen shorted or open. Low voltage anything is typically more problematic, even for NP0/C0G ceramic types.
Okay, you are then committed to a size of capacitor body since you've picked everything else. I'll tell you right now that an Electrolytic is fine unless you have stringent leakage requirements. In audio this isn't usually a consideration unless the voltage difference is high and the circuit very high impedance.
They often don't make lower voltage ratings for some film capacitors because the film is so thin, the defect rate is too high without a cost difference that makes it worthwhile. Low voltage polystyrene capacitors are an excellent example, I use 630 V units, not the 50 V types often seen shorted or open. Low voltage anything is typically more problematic, even for NP0/C0G ceramic types.
With all due respect to you sir,
You are quite mistaken. There are already people starting to build @MarcelvdG RTZ DSD dacs with clocking based on my designs and or my input and advice to other designers. So far the results are mostly not public, but are quite gratifying. People in the know are never going back to crappy clocking. Moreover, it is you who is insulting designers and companies such as for example @iancanada and @acko Labs. You are the ignorant one in this case, not those guys and not me.
Last edited:
Sure, okay. A commercial interest speaking.
I'm sure those probably work well and sound good. But I am afraid you don't have the only great performing designs. So you're going to hold yourself above all other designers and T&M engineers. Oh boy!
I have many decades in the audio and test and measurement fields, along with telecom. Claims must be backed up and agree with science in my life. Your designs must do exactly what you claim and be reliable, reproducible. One of the first things we learn is that there is nothing you can do that someone else can't also do. You have to believe others can do it better than you can. It's the only way to keep your eyes and mind open and learn.
I'm sure those probably work well and sound good. But I am afraid you don't have the only great performing designs. So you're going to hold yourself above all other designers and T&M engineers. Oh boy!
I have many decades in the audio and test and measurement fields, along with telecom. Claims must be backed up and agree with science in my life. Your designs must do exactly what you claim and be reliable, reproducible. One of the first things we learn is that there is nothing you can do that someone else can't also do. You have to believe others can do it better than you can. It's the only way to keep your eyes and mind open and learn.
Thickness doesn't affect dielectric properties, nor do dimensions, dielectric properties are intrinsic, not extrinsic - they don't depend on the amount of material or its arrangement - like density or refractive index or specific heat capacity. Polypropylene is polypropylene, and all of it is hydrophobic because its an alkane. etc etc. Different formulations are going to have very minor effects if its the same bulk material, because its the bulk that is in the majority and responding to the electric field.
Have you measured different capacitors for distortion at all? I suggest that unless you are prepared to do that and find evidence you don't have any right to claim what you are claiming, its just baseless assertions. I have measured a range of capacitors and PP is definitely very linear, mylar less so, type II ceramic very very bad indeed(*). In fact I found allegedly "PP" capacitors from eBay that are type II ceramic in a plastic case, very evident from measurements, and subsequent disection.
(*) type II ceramic actually have cross-over distortion, not just odd-order harmonics, as evident from their residuals:
This actually makes sense when you realize type II materials are ferro-electric...
Here's a plot from a PP film cap, basically down at the noise floor of my QA403 analyzer:
True. However, IMHO and IME @anatech is correct about antenna effect with large body caps. Sometimes squeezing or tapping on large caps can also have some effect. Its not just about dielectric material properties. Nor is it just about easy-to-measure nonlinear distortion. This has been proven to me time and time again.
Also, as I have been criticized for saying in the past, we need to be very careful about jumping to conclusions when we don't really know what we should be trying to measure. One way to do science is to first perform an observation, then go looking for correlated and or causal factors. Doing it backwards, "this is what I know how to measure, and it doesn't show anything," isn't necessarily conclusive evidence of no audible effect. Moreover, if you don't really know what to be measuring then how do you expect someone else to know?
Last edited:
Actually the trend has been to make equivalent rated capacitors smaller, not larger, but its not dramatic differences since Cyril did his research...
Not me. My clocking designs are open source and shared in the forum. I am just trying to do a jump-start in what I consider to be a too neglected area of diy dac design. @MarcelvdG RTZ discrete DSD dac design is also fully open source. In addition, @PJotr25 "PCM2DSD" converter project is also freely available for diy use. There is also an OPA1632 output stage board schematic shared by @bohrok2610
All of these people are interested in keeping the diy world of dac building advancing.
Hi Markw4,
Well, in normal threads, it isn't the place to jump start anything. Please remain on topic. "Commercial" also includes reputation as a driver. A motivation beyond what is being talked about.
As for measurements, when we see a phenomenon we can't explain, we test via different means. We aren't trying to prove something, we are discovering what causes a certain effect. Then we make sure what we think we are measuring is what in fact is causal. Only people with a preconceived idea attempt to prove they are right. The rest of us just try to follow the evidence to find out what the truth is.
Many audiophiles don't understand what real engineers and technicians look for or how they go about it. It is a branch of physics, the same principles apply. Follow the evidence to see where it goes. Keeps us honest and on track. To my mind, trying to be "right" over what is really the truth is a massive waste of time and effort. To do that, you need all forms of evidence, weighted as to how reproducible they are of course. This normally also means some form of testing apparatus and skill to use same. The only thing that doesn't follow the rules of physics is a brain with ideas already in place.
Hi Mark T,
Yes! Absolutely!
Thickness affects dielectric strength and perhaps how fast a material may degrade over time. The properties otherwise remain the same with varying thickness.
Really, you found high K ceramics sold as PP types inside larger bodies? Wow! That's a great reason to purchase from authorized distribution only - and to check the parts!
Well, in normal threads, it isn't the place to jump start anything. Please remain on topic. "Commercial" also includes reputation as a driver. A motivation beyond what is being talked about.
As for measurements, when we see a phenomenon we can't explain, we test via different means. We aren't trying to prove something, we are discovering what causes a certain effect. Then we make sure what we think we are measuring is what in fact is causal. Only people with a preconceived idea attempt to prove they are right. The rest of us just try to follow the evidence to find out what the truth is.
Many audiophiles don't understand what real engineers and technicians look for or how they go about it. It is a branch of physics, the same principles apply. Follow the evidence to see where it goes. Keeps us honest and on track. To my mind, trying to be "right" over what is really the truth is a massive waste of time and effort. To do that, you need all forms of evidence, weighted as to how reproducible they are of course. This normally also means some form of testing apparatus and skill to use same. The only thing that doesn't follow the rules of physics is a brain with ideas already in place.
Hi Mark T,
Yes! Absolutely!
Thickness affects dielectric strength and perhaps how fast a material may degrade over time. The properties otherwise remain the same with varying thickness.
Really, you found high K ceramics sold as PP types inside larger bodies? Wow! That's a great reason to purchase from authorized distribution only - and to check the parts!
Three "100nF PP" caps I found, clearly only one is likely to be PP:
I suspect green one is mylar (MKT), but you can guess what the others are! Just taken at random from my cap storage drawers one day. I'd filled the drawers from a few different sources over time...
I suspect green one is mylar (MKT), but you can guess what the others are! Just taken at random from my cap storage drawers one day. I'd filled the drawers from a few different sources over time...
Cap 1 is ugly for sure.
I don't run THD vs level, and it depends also on the current level. So number one could be anything, a tantalum, high K ceramic. It is pretty bad!
I don't run THD vs level, and it depends also on the current level. So number one could be anything, a tantalum, high K ceramic. It is pretty bad!
What kind of crystal oscillators are you talking about ? The HP 5372A's resolution is no better than 150ps.
I use an external clock source, double oven which is extremely stable. For frequency accuracy it is locked to the GPS receiver, but in holdover when I need to check stability or noise. You can't use the GPS reference as it corrects occasionally, that produces a step shift in readings.
Looking at any audio application, 150 ps is far better than needed. I figured all this stuff out ages ago. Anyway, that is time interval. I have the 1Meg input modules as well so crystals can be measured in circuit and operating. You use normal oscilloscope probes to avoid loading the circuit under test. It has other functions to give further insight into the oscillator circuit under test. Given the testing is in situ, all other factors are taken into account for that application. In addition, yo can assess the various other timing sources, like word clocks and sampling.
Edit:
Reference frequencies are all 10MHz, your clocks are normally in the range of 4.3 MHz, maybe higher for some new things.
Looking at any audio application, 150 ps is far better than needed. I figured all this stuff out ages ago. Anyway, that is time interval. I have the 1Meg input modules as well so crystals can be measured in circuit and operating. You use normal oscilloscope probes to avoid loading the circuit under test. It has other functions to give further insight into the oscillator circuit under test. Given the testing is in situ, all other factors are taken into account for that application. In addition, yo can assess the various other timing sources, like word clocks and sampling.
Edit:
Reference frequencies are all 10MHz, your clocks are normally in the range of 4.3 MHz, maybe higher for some new things.
What exactly do you think I am "claiming"?
Think a bit before replying.
Perhaps think twice...