new blog entry on noise measurement (and poll)
off topic....
new thread started with a poll for my blog site
would appreciate your inputs (sorry no prizes to be won )
New Thread LINK
off topic....
new thread started with a poll for my blog site
would appreciate your inputs (sorry no prizes to be won
)New Thread LINK
Hello,
Normally there are 7810 regulators that need around 12 volt to be able to supply 10 volt to the Tent shunts. There have been people here skipping these regulators and supply something close to 10 volt to get a better sound they said.
All these multi F caps contain loads of energy and i am not sure if they will " drain " completely after other parts of your digital set_up already ran out of juice so to say.
You NEED to be sure what you are doing when using these caps. Usually if something goes wrong it will go wrong terribly because there is a lot of current available.
Of course these caps need to be securely attached to the chassis and not the typical diy way using some kind of glue. Of course some planes are partially being glued but they are using exactly the right glue and use it the right way! If you go to a big diy store you can find the right glue probably. BUT i know from my former job that most glues only work with the right preparations. Some glues we used would only work with a special primer. We had to do several test for a special application and just one of the glues would stick after being summited to some force in the first month.
Just imagine that a badly glued supercap would roll around your chassis after being moved around to audiophile friends by UPS. A true firecracker.
Greetings,Eduard
So the next phase of my DDDAC journey begins! Further PSU improvements
Nice Simon, would you publish a small sketch how your going to hook up things?
It's a very good PSU for DDDAC, especially if two rails are paralleled, but I advise to use it with caution, especially in a test environment.
If your batteries somehow discharge below their critical level, an internal short can occur. I experienced this with DDDAC and this PSU. After that, I tested two A123 batteries under controlled circumstances. They do not tend to explode, but they heat to about 220-250 degrees, also they produce smoke and hot electrolyte will leak to the board. In my point of view, several actions needed to made this configuration with DDDAC safe:
1. board is completely covered by heat-resistant varnish (did it).
2. board is completely covered by high temp polyimide foil (did it)
3. battery holders are made of high temp plastic, not ABS (I printed them on my 3D printer, filament -- PC).
4. thermal sensor is placed on the top of each battery (bought all parts, not completed yet).
5. required turn-off time is carefully measured.
If your batteries somehow discharge below their critical level, an internal short can occur. I experienced this with DDDAC and this PSU. After that, I tested two A123 batteries under controlled circumstances. They do not tend to explode, but they heat to about 220-250 degrees, also they produce smoke and hot electrolyte will leak to the board. In my point of view, several actions needed to made this configuration with DDDAC safe:
1. board is completely covered by heat-resistant varnish (did it).
2. board is completely covered by high temp polyimide foil (did it)
3. battery holders are made of high temp plastic, not ABS (I printed them on my 3D printer, filament -- PC).
4. thermal sensor is placed on the top of each battery (bought all parts, not completed yet).
5. required turn-off time is carefully measured.
Hello,
A decade ago i was using a big battery with a so called intelligent charger.
It was switched on/off from time to time. I didnt install anything to check the voltage.
(Now with the DDAC i have an old fashioned panelmeter that shows the dc voltage going towards the 12 volt input terminal on the mainboard)
I thought that the intelligent charger would automatically start charger when the voltage has gone below a certain level. After some years and not using the device being fed by the batteries i noticed a sound that was kind of mediocre. I checked the voltage on the two 12 volts batteries connected in series and it was to low. The batteries also had a kind of slightly bulging appearance.
I called the dealer and they were kind of reconditioned.
How did it go wrong?
Well. the charger i was using would only react to a to low voltage after being disconnected from the batteries for a second or so.
So for a circuit taking some current like a 4 board DDAC one will need to make it kind of foolproof.
Greetings,Eduard
A decade ago i was using a big battery with a so called intelligent charger.
It was switched on/off from time to time. I didnt install anything to check the voltage.
(Now with the DDAC i have an old fashioned panelmeter that shows the dc voltage going towards the 12 volt input terminal on the mainboard)
I thought that the intelligent charger would automatically start charger when the voltage has gone below a certain level. After some years and not using the device being fed by the batteries i noticed a sound that was kind of mediocre. I checked the voltage on the two 12 volts batteries connected in series and it was to low. The batteries also had a kind of slightly bulging appearance.
I called the dealer and they were kind of reconditioned.
How did it go wrong?
Well. the charger i was using would only react to a to low voltage after being disconnected from the batteries for a second or so.
So for a circuit taking some current like a 4 board DDAC one will need to make it kind of foolproof.
Greetings,Eduard
Hello Eduard.
Batteries are not safe, Ultracaps are even worse. Moreover, ultracaps widely used there do not have an even datasheet with a frequency response chart. Besides, batteries tend to change their ESR depending on the charge level, this is hearable. All battery-based PSUs here is a far away from real reference. Also, they are far-far away from zero-ohm ESR. It's my point of view.
But...
My personal story began almost 25 years ago, I was 19, and I heard one high-end system, it was so close to reality that I could not believe it. Perhaps at that time I did not have the money to buy 1/100 of this system. I don't remember the details after 25 years, it may have been a purely analog audio system. Thus began my personal dream that one day I could buy such a magical audio system. After many years, I realized that what I cannot buy, it ALL sounded worse than my internal reference was.
I bought the DDDAC kit 3 years ago, soldered it, and heavily upgraded it. It sounded really cool, but I decided to take the test for my DDDAC. I went to organ live concert, at this concert professional studio made recording. And ... sorry, Doede, DDDAC did not survive in this test. It was probably the best DAC, I liked it so much, and it was so close to my dream. I unsoldered the DDDAC stack and didn't listen to music for a year after that. Many there are switching to other solutions in the same case, but this forum looks like a new version of the audio market, 10 years ago people connected transport and a DAC, nowadays it looks like connecting Jan's reclocker with a resistive ladder DAC, and so on. There are a lot of fans around any solution. There is almost no collaboration between developers here. Only Lorien adapted his solution for DDDAC! This despite the fact that Doede is very open minded, and participating in almost all significant threads here.
The real problem for me was the lack of knowledge. I spent a year with books, university publications and LTSpice. I did two unsuitable upgrades (a lot of people here did the same) and haven't done many others (even Tent shunt implementation becomes a bottleneck at a certain level). Also, and it's most significant conclusion for me, switching to another solution must be based on understanding what you are really doing. It's the only way to achieve results in the terms of my dream. Anything else gives results like switching from Philips CD-player AAA to Sony CD-player BBB. = NOTHING.
Returning to the topic, DDDAC with a good clock is extremely sensitive to low-frequency noise on digital power rails. I agree that batteries are only a temporary solution, but regulators must achieve booth high PSRR, low 1/f noise and precise voltage levels. It's really hard to achieve in the real world (not a digit on paper), and some clean PSU is needed. As a stack of boards gives you SQRT, some sort of problems can be partially solved with increasing boards stack. But this method is not a panacea either.
As the final solution it's better to eliminate batteries, it's not a future|best and so on... and it's clearly a wrong way, but it means that too many goals have to be achieved at the same time.
Batteries are not safe, Ultracaps are even worse. Moreover, ultracaps widely used there do not have an even datasheet with a frequency response chart. Besides, batteries tend to change their ESR depending on the charge level, this is hearable. All battery-based PSUs here is a far away from real reference. Also, they are far-far away from zero-ohm ESR. It's my point of view.
But...
My personal story began almost 25 years ago, I was 19, and I heard one high-end system, it was so close to reality that I could not believe it. Perhaps at that time I did not have the money to buy 1/100 of this system. I don't remember the details after 25 years, it may have been a purely analog audio system. Thus began my personal dream that one day I could buy such a magical audio system. After many years, I realized that what I cannot buy, it ALL sounded worse than my internal reference was.
I bought the DDDAC kit 3 years ago, soldered it, and heavily upgraded it. It sounded really cool, but I decided to take the test for my DDDAC. I went to organ live concert, at this concert professional studio made recording. And ... sorry, Doede, DDDAC did not survive in this test. It was probably the best DAC, I liked it so much, and it was so close to my dream. I unsoldered the DDDAC stack and didn't listen to music for a year after that. Many there are switching to other solutions in the same case, but this forum looks like a new version of the audio market, 10 years ago people connected transport and a DAC, nowadays it looks like connecting Jan's reclocker with a resistive ladder DAC, and so on. There are a lot of fans around any solution. There is almost no collaboration between developers here. Only Lorien adapted his solution for DDDAC! This despite the fact that Doede is very open minded, and participating in almost all significant threads here.
The real problem for me was the lack of knowledge. I spent a year with books, university publications and LTSpice. I did two unsuitable upgrades (a lot of people here did the same) and haven't done many others (even Tent shunt implementation becomes a bottleneck at a certain level). Also, and it's most significant conclusion for me, switching to another solution must be based on understanding what you are really doing. It's the only way to achieve results in the terms of my dream. Anything else gives results like switching from Philips CD-player AAA to Sony CD-player BBB. = NOTHING.
Returning to the topic, DDDAC with a good clock is extremely sensitive to low-frequency noise on digital power rails. I agree that batteries are only a temporary solution, but regulators must achieve booth high PSRR, low 1/f noise and precise voltage levels. It's really hard to achieve in the real world (not a digit on paper), and some clean PSU is needed. As a stack of boards gives you SQRT
, some sort of problems can be partially solved with increasing boards stack. But this method is not a panacea either.As the final solution it's better to eliminate batteries, it's not a future|best and so on... and it's clearly a wrong way, but it means that too many goals have to be achieved at the same time.
Sorry guys, I mentioned 2 of my mistakes in terms of" sound close to reality ", but did not disclose them. We all have a different vision of DAC, so follow the recommendations based on the vision of "sound close to reality" rather than "analog sound", "tube sound", "best sound", "the best DAC on the planet" etc. Maybe some of you have the same vision, in this case it could be helpful for you If your vision is different -- it's great, but ideas may be not suitable for you -- just skip this post.
While almost all people share their achievements in upgrades, I’ll share my mistakes. Exclusive
1. The 10uF capacitor next to PCM1794 analog power pins can be easily replaced with something better, it is better to use larger capacitors here.
If you are not ready to change the decoupling scheme of DAC IC or use NOS caps -- almost no chances. Any "better" capacitor is worse here. With "better" cap female voices sound brighter and younger, but it means only one thing: the main advantage of Tent shunt here (excellent phase linearity and output impedance linearity) is lost, FR on complex signals is not linear anymore too. So, it looks like improvement at first glance, but it's a huge step away from "sound close to reality". Also, DA from caps will change signal too. Is it measurable? If a sine wave signal is used for a test, and output of DDDAC is measured -- probably not. But music is not a sine wave, it is changes confirmed by spice simulation and listening test on breadboard with PCM1794 (I reproduced DDDAC board step-by-step from the scratch to find this issue) and real DDDAC boards. Newer OSCONs are the worst choice here. Lower ESR is worse, higher capacitance is worse. It's not a 7805 regulator and digital power decoupling anyway. Personally, I do not like "sweet" coloration from KZ Muzes, but almost any other choice will be even worse. If you have an old board (without simplified on-board shunt implementation) and installed standard Tentlabs shunt by yourself close to IC power pins -- no electrolytic capacitor is the best capacitor here. For new boards in my point of view a more complex solution is needed to eliminate electrolytic capacitors safely.
Initially I used slightly bigger caps (22uF) with slightly lower ESR. It’s "musical" properties was great, but shunt not forgave me this upgrade.
2. VCOM caps can be easily upgraded to something better.
It's a kind of noise reduction of internal voltage reference, noise = voltage error, error = distortion of output signal. So leakage current of capacitor means more than "musical" properties of cap here. See explanations and graphs from TI support. Leakage = noise = error = distortion here, Also this graph can be interpolated to real leakage of any caps used. They're kindly not mentioned one thing: low leakage means actual values must be below a datasheet’s promise. Manual selection is required.
AN capacitors are roulette. I measured a lot of them and think it's awful for that price range. You can win if you are ready to perform selection or you're lucky, I lost with them. The newer OSCONs are the absolute worst choice here. I measured about 70 different capacitors, only 5 outperformed Doede's choice. Even then, from my personal experience, replacement of KZ Muzes here without measurements and selection was insane.
If your vision is different -- it's great, but ideas may be not suitable for you -- just skip this post.While almost all people share their achievements in upgrades, I’ll share my mistakes. Exclusive
1. The 10uF capacitor next to PCM1794 analog power pins can be easily replaced with something better, it is better to use larger capacitors here.
If you are not ready to change the decoupling scheme of DAC IC or use NOS caps -- almost no chances. Any "better" capacitor is worse here. With "better" cap female voices sound brighter and younger, but it means only one thing: the main advantage of Tent shunt here (excellent phase linearity and output impedance linearity) is lost, FR on complex signals is not linear anymore too. So, it looks like improvement at first glance, but it's a huge step away from "sound close to reality". Also, DA from caps will change signal too. Is it measurable? If a sine wave signal is used for a test, and output of DDDAC is measured -- probably not. But music is not a sine wave, it is changes confirmed by spice simulation and listening test on breadboard with PCM1794 (I reproduced DDDAC board step-by-step from the scratch to find this issue) and real DDDAC boards. Newer OSCONs are the worst choice here. Lower ESR is worse, higher capacitance is worse. It's not a 7805 regulator and digital power decoupling anyway. Personally, I do not like "sweet" coloration from KZ Muzes, but almost any other choice will be even worse. If you have an old board (without simplified on-board shunt implementation) and installed standard Tentlabs shunt by yourself close to IC power pins -- no electrolytic capacitor is the best capacitor here. For new boards in my point of view a more complex solution is needed to eliminate electrolytic capacitors safely.
Initially I used slightly bigger caps (22uF) with slightly lower ESR. It’s "musical" properties was great, but shunt not forgave me this upgrade.
2. VCOM caps can be easily upgraded to something better.
It's a kind of noise reduction of internal voltage reference, noise = voltage error, error = distortion of output signal. So leakage current of capacitor means more than "musical" properties of cap here. See explanations and graphs from TI support. Leakage = noise = error = distortion here, Also this graph can be interpolated to real leakage of any caps used. They're kindly not mentioned one thing: low leakage means actual values must be below a datasheet’s promise. Manual selection is required.
AN capacitors are roulette. I measured a lot of them and think it's awful for that price range. You can win if you are ready to perform selection or you're lucky
, I lost with them. The newer OSCONs are the absolute worst choice here. I measured about 70 different capacitors, only 5 outperformed Doede's choice. Even then, from my personal experience, replacement of KZ Muzes here without measurements and selection was insane.
Yuri,
not sure what to say here. It is really amazing what you have been doing and I thank you for your endurance and insights! This is good practice shared and will help anyone who wants to do some more experiments.
But not listening music for one year? I could not do that )
I am working from home and in my office the whole day (almost) I play music through Roon into my "power-DDDAC" (WaveIO - 1 deck DDDAC - UCD 40 Watt) and a pair of small B&W shelf speakers. Not all very high end, but very enjoyable at work
again, fantastic you spent so much time, did it the scientific way and share it here with the community
not sure what to say here. It is really amazing what you have been doing and I thank you for your endurance and insights! This is good practice shared and will help anyone who wants to do some more experiments.
But not listening music for one year?
I could not do that )I am working from home and in my office the whole day (almost) I play music through Roon into my "power-DDDAC" (WaveIO - 1 deck DDDAC - UCD 40 Watt) and a pair of small B&W shelf speakers. Not all very high end, but very enjoyable at work
again, fantastic you spent so much time, did it the scientific way and share it here with the community
Doede,
if the dream is easily achievable, maybe it was not a dream
To be clear, I am trying to stay away from a fan zone, and trying to think about DDDAC and your decisions critically, but it’s only because staying in a fan zone your results is absolutely the maximum I'll ever achieve.
I spent a lot of time with DDDAC, and now I clearly understand how much knowledge, patience and experience you invested in this solution. Sometimes your solutions can be outperformed, but it's HARD, even if the price is not a concern. Trying to reach it with an affordable price It’s not hard, it means CHALLENGE.
I admire you so much, and thank you very much for all your works for the community in general and for realization of my dreams personally. I made the right choice with DDDAC and never regretted it.
Thank you so much again.
Well, some words about an ultra-high end. Some time before I decided to test my DDDAC, I created a small sketch -- my personal vision of DAC. Yes, it must cover my dream as the ultra version, but nowadays it's not enough. Staying only with ultra-version, new possibilities are lost. Music collection may be big or even huge, but what about live concerts, some HD recordings on Youtube not available from ultra-HQ sources and so on. Ignoring it means that I have "special occasion" DAC only, and lose some sense of presence. It can be comparable to a live concert. What's better, being part of it and listening to it live or wait a year and download the HD track? In the first case actual sound quality is almost always worse.
Unfortunately, all the latest great updates give us signal latency, so for me personally the next step is to return to the original idea, WaveIO with almost zero latency, and try to reach maximum in the live mode too. Compared to ultra-version, some level of quality will be lost, but it's a fair price for a sense of presence. It's priceless.
if the dream is easily achievable, maybe it was not a dream
To be clear, I am trying to stay away from a fan zone, and trying to think about DDDAC and your decisions critically, but it’s only because staying in a fan zone your results is absolutely the maximum I'll ever achieve.
I spent a lot of time with DDDAC, and now I clearly understand how much knowledge, patience and experience you invested in this solution. Sometimes your solutions can be outperformed, but it's HARD, even if the price is not a concern. Trying to reach it with an affordable price It’s not hard, it means CHALLENGE.
I admire you so much, and thank you very much for all your works for the community in general and for realization of my dreams personally. I made the right choice with DDDAC and never regretted it.
Thank you so much again.
Well, some words about an ultra-high end. Some time before I decided to test my DDDAC
, I created a small sketch -- my personal vision of DAC. Yes, it must cover my dream as the ultra version, but nowadays it's not enough. Staying only with ultra-version, new possibilities are lost. Music collection may be big or even huge, but what about live concerts, some HD recordings on Youtube not available from ultra-HQ sources and so on. Ignoring it means that I have "special occasion" DAC only, and lose some sense of presence. It can be comparable to a live concert. What's better, being part of it and listening to it live or wait a year and download the HD track? In the first case actual sound quality is almost always worse.Unfortunately, all the latest great updates give us signal latency, so for me personally the next step is to return to the original idea, WaveIO with almost zero latency, and try to reach maximum in the live mode too. Compared to ultra-version, some level of quality will be lost, but it's a fair price for a sense of presence. It's priceless.
Hi Yuri
Loved reading your posts and admire your dedication tremendously!
I have been thinking a lot about the pernicious effects of 1/f noise and saw you'd developed an LT3042/ ltc6655 solution - is it basically the same as the datasheet?
Would love to hear more about your specific recommendations for this and other power supply aspects.
Many thanks and best wishes
Cv
Loved reading your posts and admire your dedication tremendously!
I have been thinking a lot about the pernicious effects of 1/f noise and saw you'd developed an LT3042/ ltc6655 solution - is it basically the same as the datasheet?
Would love to hear more about your specific recommendations for this and other power supply aspects.
Many thanks and best wishes
Cv