Yup run one great units nudges out my highly modded Mac mini
Sent from my 486 IBM WIN 3.1x Desktop...
Hello,
Almost sure that i will but the aurender N100 with an external HD. Aurender ask to much for the model with a 2 TB internal HD. Usually this is the first part that will break down.
Also have to get myself an apple device to be able to use the Aurender app.
I will try to copy and paste the answer from Brian ( Belleson) below about an extra LC network for the 3 regulators sistuated on the mainboard.
Hello Eduard,
Thank you for waiting for a reply. I read your emails 3 times to be sure I understand your question(s). You want to add the choke and 100µF capacitor between 12V input and Belleson 3,3V and 5V regulators. These are my honest opinion...
Note that I am a practical man and an engineer, and I want to get the most performance for the least work and lowest cost. Only in the worst rain storm do I wear a rain coat slicker and carry an umbrella (a "bumbershoot" in UK and maybe Netherlands? ;-) ).
For a high performance power supply for a DAC, three specifications are most important:
output dynamic step response
ripple rejection
output impedance
noise
1. and 3. are two ways to see the same thing, that is-- how fast can the output respond to a rapid change of load current (a step) while the output voltage remains constant? Or in the limit, when load current changes quickly, how small is the output voltage change and how quickly does it recover back to its specified voltage?
2. is a way to measure how well the power supply blocks incoming AC and only allows DC to get to the output.
In a power supply, placing a choke and capacitor before the regulator reduces ripple at the regulator input and stores magnetic energy. Also it will (as you say) limit interaction between supply stages. In this way it acts as a pre-regulator. However, recall the physical function of an inductor -- the magnetic field acts to oppose change of current through the inductor. If an inductor slows the change of current at the regulator input, it also slows change of current at regulator output. Thus, while helping 2. , it hurts 1.
Now the task is, with choke vs. no choke, compare how much less ripple at regulator output vs. decrease in dynamic step response speed. Consider that Superpower SPM has ripple rejection greater than 100dB from DC to approx. 2kHz, and greater than 80dB past 25kHz, how much improvement can the pre-reg inductor have? I plan to make some measurements on this, but I think the loss of step response is greater than the gain to ripple rejection. For your specific system, I can't make the judgement for you and it's not available for me to measure.
Next consider the output side of the regulator only, with the assumption that the input is perfect DC. (In that case you don't need a regulator, but consider a perfect input to allow isolation of regulator error from other supply system error.) Output step response can be seen in this test as only dependent on the regulator. When you cascade two regulators, what happens to step response at the load? If you have two really fast regulators, the effect is minimum. With a fast regulator after a slow response regulator, the current can pass through the fast regulator only as quickly as the slow regulator delivers it. Same is true for a slow regulator after a fast regulator.
I originally said three specifications are important, then listed four. What about noise? Power supply noise is important in some circuits. In an amplifier, noise of the first stage is important. If the first stage is a differential pair in and out, first order noise is summed to zero at the stage output. If the first stage is differential in and single ended out, input noise is summed to zero but power supply noise is not. If the first stage uses resistive loads, power supply noise is passed directly from supply to output. But if it uses active current sources, power supply noise is blocked. These facts show why differential op amps generally have good noise performance, or at least better than the "university electrical engineering 201 class" resistor loaded diff pair amplifier.
For a DAC, the first stage is the current to voltage conversion. If this conversion uses an op amp, DAC output noise and internal op amp noise are important but power supply noise is only a secondary problem. If the first stage is a discrete transistor based design, power supply noise may be important, depending on the design as discussed in the above paragraph.
Certainly lower power supply noise is better than higher, all other performance equal 🙂.
For a regulator, output noise is mostly due to its internal voltage reference. The lowest noise references are buries zener breakdown type, but they are about 7V and not suitable for 3.3V and 5V regulators. Bandgap references are available at lower voltages (0.2, 0.6, 1.24, 2.5, et al) but have higher noise. Often the lowest voltage references have the lowest noise, but to get a higher regulated Vout, the reference voltage must be made higher with gain, which also multiplies the noise. There are ways to decrease reference noise internally that are used on discrete references such as Belleson.
However, consider again a cascade of two regulators, e.g. a 5V regulator after a 12V regulator. If the 5V regulator has high noise, that is what dominates because, unlike the case of an amplifier, the last regulator stage sets the output noise,
I don't tell people how to design their audio circuits because everyone has their favorite ways. However, I design mine using the least possible series impedance on the input side of the regulator, large capacitance from regulator input to ground and "enough" capacitance and shunt resistance at the regulator output. "Enough" depends on the regulator. With multiple regulators in series, I use the best regulators for both. For a DDDAC, that means I would use Belleson on both the main board and the series regulators on the other boards. The shunt regulator, assuming it has enough idle current to always supply the load with no drop-out, should be good "as is." This is my opinion, without the benefit of any measurements or listening tests.
I hope this helps.
Best regards,
Brian
So will try 500mH 30 ohm DCR hammonf choke soon. When i open up the dddac i can check if there is space enough for a choke like the Lundahl LL1638 which can be wired for 4000mH 36 ohm or 1000mH 9 ohm
Because the Bellesons all use a bleeder to get a minimum current they need to function properly i assume variations in current drawn because of the dddac actually getting a signal to '' handle '' are small. So i think DCR of the choke is not that important.
Any suggestions welcome about using bigger cap or not, more mH or not.
Greetings, Eduard
Almost sure that i will but the aurender N100 with an external HD. Aurender ask to much for the model with a 2 TB internal HD. Usually this is the first part that will break down.
Also have to get myself an apple device to be able to use the Aurender app.
I will try to copy and paste the answer from Brian ( Belleson) below about an extra LC network for the 3 regulators sistuated on the mainboard.
Hello Eduard,
Thank you for waiting for a reply. I read your emails 3 times to be sure I understand your question(s). You want to add the choke and 100µF capacitor between 12V input and Belleson 3,3V and 5V regulators. These are my honest opinion...
Note that I am a practical man and an engineer, and I want to get the most performance for the least work and lowest cost. Only in the worst rain storm do I wear a rain coat slicker and carry an umbrella (a "bumbershoot" in UK and maybe Netherlands? ;-) ).
For a high performance power supply for a DAC, three specifications are most important:
output dynamic step response
ripple rejection
output impedance
noise
1. and 3. are two ways to see the same thing, that is-- how fast can the output respond to a rapid change of load current (a step) while the output voltage remains constant? Or in the limit, when load current changes quickly, how small is the output voltage change and how quickly does it recover back to its specified voltage?
2. is a way to measure how well the power supply blocks incoming AC and only allows DC to get to the output.
In a power supply, placing a choke and capacitor before the regulator reduces ripple at the regulator input and stores magnetic energy. Also it will (as you say) limit interaction between supply stages. In this way it acts as a pre-regulator. However, recall the physical function of an inductor -- the magnetic field acts to oppose change of current through the inductor. If an inductor slows the change of current at the regulator input, it also slows change of current at regulator output. Thus, while helping 2. , it hurts 1.
Now the task is, with choke vs. no choke, compare how much less ripple at regulator output vs. decrease in dynamic step response speed. Consider that Superpower SPM has ripple rejection greater than 100dB from DC to approx. 2kHz, and greater than 80dB past 25kHz, how much improvement can the pre-reg inductor have? I plan to make some measurements on this, but I think the loss of step response is greater than the gain to ripple rejection. For your specific system, I can't make the judgement for you and it's not available for me to measure.
Next consider the output side of the regulator only, with the assumption that the input is perfect DC. (In that case you don't need a regulator, but consider a perfect input to allow isolation of regulator error from other supply system error.) Output step response can be seen in this test as only dependent on the regulator. When you cascade two regulators, what happens to step response at the load? If you have two really fast regulators, the effect is minimum. With a fast regulator after a slow response regulator, the current can pass through the fast regulator only as quickly as the slow regulator delivers it. Same is true for a slow regulator after a fast regulator.
I originally said three specifications are important, then listed four. What about noise? Power supply noise is important in some circuits. In an amplifier, noise of the first stage is important. If the first stage is a differential pair in and out, first order noise is summed to zero at the stage output. If the first stage is differential in and single ended out, input noise is summed to zero but power supply noise is not. If the first stage uses resistive loads, power supply noise is passed directly from supply to output. But if it uses active current sources, power supply noise is blocked. These facts show why differential op amps generally have good noise performance, or at least better than the "university electrical engineering 201 class" resistor loaded diff pair amplifier.
For a DAC, the first stage is the current to voltage conversion. If this conversion uses an op amp, DAC output noise and internal op amp noise are important but power supply noise is only a secondary problem. If the first stage is a discrete transistor based design, power supply noise may be important, depending on the design as discussed in the above paragraph.
Certainly lower power supply noise is better than higher, all other performance equal 🙂.
For a regulator, output noise is mostly due to its internal voltage reference. The lowest noise references are buries zener breakdown type, but they are about 7V and not suitable for 3.3V and 5V regulators. Bandgap references are available at lower voltages (0.2, 0.6, 1.24, 2.5, et al) but have higher noise. Often the lowest voltage references have the lowest noise, but to get a higher regulated Vout, the reference voltage must be made higher with gain, which also multiplies the noise. There are ways to decrease reference noise internally that are used on discrete references such as Belleson.
However, consider again a cascade of two regulators, e.g. a 5V regulator after a 12V regulator. If the 5V regulator has high noise, that is what dominates because, unlike the case of an amplifier, the last regulator stage sets the output noise,
I don't tell people how to design their audio circuits because everyone has their favorite ways. However, I design mine using the least possible series impedance on the input side of the regulator, large capacitance from regulator input to ground and "enough" capacitance and shunt resistance at the regulator output. "Enough" depends on the regulator. With multiple regulators in series, I use the best regulators for both. For a DDDAC, that means I would use Belleson on both the main board and the series regulators on the other boards. The shunt regulator, assuming it has enough idle current to always supply the load with no drop-out, should be good "as is." This is my opinion, without the benefit of any measurements or listening tests.
I hope this helps.
Best regards,
Brian
So will try 500mH 30 ohm DCR hammonf choke soon. When i open up the dddac i can check if there is space enough for a choke like the Lundahl LL1638 which can be wired for 4000mH 36 ohm or 1000mH 9 ohm
Because the Bellesons all use a bleeder to get a minimum current they need to function properly i assume variations in current drawn because of the dddac actually getting a signal to '' handle '' are small. So i think DCR of the choke is not that important.
Any suggestions welcome about using bigger cap or not, more mH or not.
Greetings, Eduard
Hello,
Will invest some time in finding info about the Aurender. I guess that will be the next upgrade . Gives a higher return on investment than posting here lol.
Greetings, eduard
Will invest some time in finding info about the Aurender. I guess that will be the next upgrade . Gives a higher return on investment than posting here lol.
Greetings, eduard
Aurender
I also own a Mac Mini. This is very interesting! I will read a bit about "Aurender".
I also own a Mac Mini. This is very interesting! I will read a bit about "Aurender".
If you can build a DDDAC, you can also build an "Aurender clone", but for a fraction of the price, and without software/vendor lock in. Just search diyAudio for Volumio and it's brothers and sisters.
Hello ,
Doede who did have the guts to design his own dac didnt start making an Aurender '' clone ''. I guess it is not that easy. If it would be it would have done already. I read some threads but most of these devices make customers come up with loads of questions. More questions than can be possible with the dddac.
Doede is using one too.
greetings, eduard
Doede who did have the guts to design his own dac didnt start making an Aurender '' clone ''. I guess it is not that easy. If it would be it would have done already. I read some threads but most of these devices make customers come up with loads of questions. More questions than can be possible with the dddac.
Doede is using one too.
greetings, eduard
Finally got my cinemags. These will be put in maximum by next week. I am pretty excited for this change. Hopefully it will be a significant one.
Hi all, not updated here in a while but have been relatively busy trying a few bits and my dddac is really singing now 🙂
Voltage regs for vdd.
I've swapped my 3.3v tent shunts out for lt3080
I swapped out my zfoil i/v resistors for rhopoint GG102D which I was using a while ago. The zfoil were very detailed, but the sound of the rhopoint was richer while still keeping the detail clear
Vcom. It seems the vcom caps have quite an effect on the final sound. People seem to find silmic works well here and gives a rich sound but at the expense of being a little wooly. OSCON gives great detail but maybe a little clinical. I guess black gates would be the choice here if you could get them. I've opted to try the new audionote kaisei here now they are available in smaller voltages and capacitance. 100v 100uf is the closest match to the spec. I guess it's not really an option for those with multiple decks as they're so massive at 25mm tall, but no problem for me
You'll note I'm only using 1 cap per dac chip. I got the idea from the suggested layout in the latest data sheet for the pcm1794a which shows both vcom L&R joined and using a single cap. It makes sense to me in this dual mono setup where both halves of each dac chip are processing the same audio data and having the same power supply that they can share this, so I've joined the vcom at the little vias under the board and tried a single cap with certainly no negative effects.
Last change is that since I mounted my bbb in the case properly using nice standoffs I have noticed a slight underlying power supply interference noise when not going via my buffer and Cinemags. It seems that the mounting holes in the BBB are all joined to Gnd, so by mounting this I had inadvertently joined my dddac gnd to the noisy mains earth bypassing the diodes and cap
So I've drilled out the holes in the BBB and mounted it using some plastic bushings
So that's me up to date, I think the kaisei vcom and rhopoint i/v resistors have made the biggest change, I now have a sound which is detailed and precise but also wonderfully full and tonally rich. Very happy indeed. If it stayed this way for ever I would be very satisfied 🙂
Only thing left to try is salas reflektor-d mini on the analogue vcc instead of the 8v tent shunts, but that can wait for a bit. Right now I'm just enjoying the music 🙂
Voltage regs for vdd.
I've swapped my 3.3v tent shunts out for lt3080
I swapped out my zfoil i/v resistors for rhopoint GG102D which I was using a while ago. The zfoil were very detailed, but the sound of the rhopoint was richer while still keeping the detail clear
Vcom. It seems the vcom caps have quite an effect on the final sound. People seem to find silmic works well here and gives a rich sound but at the expense of being a little wooly. OSCON gives great detail but maybe a little clinical. I guess black gates would be the choice here if you could get them. I've opted to try the new audionote kaisei here now they are available in smaller voltages and capacitance. 100v 100uf is the closest match to the spec. I guess it's not really an option for those with multiple decks as they're so massive at 25mm tall, but no problem for me
You'll note I'm only using 1 cap per dac chip. I got the idea from the suggested layout in the latest data sheet for the pcm1794a which shows both vcom L&R joined and using a single cap. It makes sense to me in this dual mono setup where both halves of each dac chip are processing the same audio data and having the same power supply that they can share this, so I've joined the vcom at the little vias under the board and tried a single cap with certainly no negative effects.
Last change is that since I mounted my bbb in the case properly using nice standoffs I have noticed a slight underlying power supply interference noise when not going via my buffer and Cinemags. It seems that the mounting holes in the BBB are all joined to Gnd, so by mounting this I had inadvertently joined my dddac gnd to the noisy mains earth bypassing the diodes and cap
So I've drilled out the holes in the BBB and mounted it using some plastic bushings
So that's me up to date, I think the kaisei vcom and rhopoint i/v resistors have made the biggest change, I now have a sound which is detailed and precise but also wonderfully full and tonally rich. Very happy indeed. If it stayed this way for ever I would be very satisfied 🙂
Only thing left to try is salas reflektor-d mini on the analogue vcc instead of the 8v tent shunts, but that can wait for a bit. Right now I'm just enjoying the music 🙂
Last edited:
Interesting that you prefer the LT3080 over the Tent Shunt regs, the spec noise is a lot higher in the former.
black gate
Hello,
Did install some black gates and chokes to make 2 extra LC filters. One for the Belleson 5 volt regulator and one for the pair of Belleson 3,3 volt regulators.
Did not have enough space to mount a third choke so did go for two BG N types connected as what BG called a super E cap. For the 5 volt regulator i used just one. Not enough space to mount a second cap there.
Greetings, Eduard
Hello,
Did install some black gates and chokes to make 2 extra LC filters. One for the Belleson 5 volt regulator and one for the pair of Belleson 3,3 volt regulators.
Did not have enough space to mount a third choke so did go for two BG N types connected as what BG called a super E cap. For the 5 volt regulator i used just one. Not enough space to mount a second cap there.
Greetings, Eduard
Attachments
But at what frequency are those noise specs based? As I understand, the low noise specs of most of these voltage regulators are only really any good up to about 100khz to 150khz and for a high frequency digital device like this dac, which is operating up in the megahertz range, the choice and proximity of the local decoupling capacitors is far more significant. I know Doede did many listening tests with different regs for the VDD and found very minimal difference.
I changed these regs out of curiosity following a recommendation and the difference made was slight compared to fitting the SMD local decoupling caps.
cheers,
James
Fantastic results could be obtained just by replacing 0.1uF WIMA's (supplied with the power supply kits) with 0.1uF NPO SMD's. In addition, few extra decoupling capacitors should be arranged around the power supply boards. This will further reduce the DC rail noise. The stacked DAC boards have huge potential of delivering stunning results, if fed properly.
Furthermore, the 0.1uF SMD's do not have to be NPO's. They could be WIMA's SMD's. These will give different character to the sound that many may prefer.
I've mentioned already that Doede's power supplies gave better results compared to chokes, at least that's what happened in my attempt to compare them - and I really wanted the chokes to sound better, but they did not.
I suggest having an additional (third) power supply (12V version because it can be tuned for 5V operation as well…) that can be modified and left aside fully prepared for a quick swap over. This will provide more obvious differences, compared with an attempt to modify an in-use power supply.
It is also crucial to have a reference digital player available, to whom DDDAC can be compared. Without this reference player (it DOES NOT have to be better sounding than DDDAC), it is impossible to keep achieving the cumulative improvements in sound quality. It is also good to attend to audiophiles' meetings and compare DDDAC with other HiFi gear as much as possible.
Furthermore, the 0.1uF SMD's do not have to be NPO's. They could be WIMA's SMD's. These will give different character to the sound that many may prefer.
I've mentioned already that Doede's power supplies gave better results compared to chokes, at least that's what happened in my attempt to compare them - and I really wanted the chokes to sound better, but they did not.
I suggest having an additional (third) power supply (12V version because it can be tuned for 5V operation as well…) that can be modified and left aside fully prepared for a quick swap over. This will provide more obvious differences, compared with an attempt to modify an in-use power supply.
It is also crucial to have a reference digital player available, to whom DDDAC can be compared. Without this reference player (it DOES NOT have to be better sounding than DDDAC), it is impossible to keep achieving the cumulative improvements in sound quality. It is also good to attend to audiophiles' meetings and compare DDDAC with other HiFi gear as much as possible.
The noise specs for the Shunt reg are up on the Tent page:
• Bandwidth (all versions) > 250 kHz
Output noise • 3.3 versions: < 15 nV/SqrrtHz
• 5V versions: < 11 nV/SqrrtHz
• 15V versions: < 32 nV/SqrrtHz
"It is also crucial to have a reference digital player available"
Do you have exemples?
Do you have exemples?
I think the above is a bit of a generalisation. There's chokes and chokes and are you talking about split digital /analogue supplies?
From all my tests with split analogue and digital supplies:
1. The Tent shunts (or Bellesons) perform better fed direct from a choke capacitor network with no pre regulation.
2. The digital supply was best with an unregulated LRC network.
I agree that super regs on the digital side make little or no difference. The stock regs though still perform better with no pre regulation when split into analogue and digital.
The new Tent boards with Elan Silmics, Oscons and standard caps comfortably outperform my old boards with Black Gates, Oscons, Tent and Belleson regs and PPS capacitors, even with smaller rated transformers.
hello
Hello,
Some people say the shunts work better with an LC , CL or LCLC power supply. BUT the new dac boards have a 7810 as a pre regulator for the shunt.
If i remember well Doede told me with 2 purposes safer input voltage for the shunts and better sound. I never did remove them so i just believe Doede. AND it is a hell of a job to remove them in my situation!!.
I never did compare the standard regulators on the motherboard with the Bellesons i just start using the Belleson from the start.
NOW all the power going '' up '' into the small boards is coming from a 1000µF 25 volt Silmic on the mainboard. This caps is fed by 2 mundorf 10000 µF caps which are just a few inches away from the mainboard. My latest mod was to use these mundorf caps as a starting point to connect two small chokes followed by 100µF Black gate caps. The negative terminals of these caps are connected to the mundorf neg terminal. Adding these 2 LC filters ( one for 5 volt Belleson and the other one for the 2 3,3 Bellesons) is sure an improvement. more spacious sound and most important the sound does not get strident during dynamic passages. ooh one more voices get a vocal chord.
Maybe because i still use cd the regulators on the mainboard need extra care?
Will get an aurender when i have more time.
I already use pps caps and some oscon.
Also have a 1µF pps cap for the one near the Tent clock but did not install it yet. Someone else should try.
Greetings, Eduard
P.s Of course LCLC is also kind of pre regulation. I decided to go choke input because all my other devices use shunt and choke input too.
Hello,
Some people say the shunts work better with an LC , CL or LCLC power supply. BUT the new dac boards have a 7810 as a pre regulator for the shunt.
If i remember well Doede told me with 2 purposes safer input voltage for the shunts and better sound. I never did remove them so i just believe Doede. AND it is a hell of a job to remove them in my situation!!.
I never did compare the standard regulators on the motherboard with the Bellesons i just start using the Belleson from the start.
NOW all the power going '' up '' into the small boards is coming from a 1000µF 25 volt Silmic on the mainboard. This caps is fed by 2 mundorf 10000 µF caps which are just a few inches away from the mainboard. My latest mod was to use these mundorf caps as a starting point to connect two small chokes followed by 100µF Black gate caps. The negative terminals of these caps are connected to the mundorf neg terminal. Adding these 2 LC filters ( one for 5 volt Belleson and the other one for the 2 3,3 Bellesons) is sure an improvement. more spacious sound and most important the sound does not get strident during dynamic passages. ooh one more voices get a vocal chord.
Maybe because i still use cd the regulators on the mainboard need extra care?
Will get an aurender when i have more time.
I already use pps caps and some oscon.
Also have a 1µF pps cap for the one near the Tent clock but did not install it yet. Someone else should try.
Greetings, Eduard
P.s Of course LCLC is also kind of pre regulation. I decided to go choke input because all my other devices use shunt and choke input too.
Okey, the prototype of the first deck is working nicely! The board design is not yet perfect, so I made a few changes to the connector and parts locations.
I received a few messages indicating interest in those boards, so I attached a pictures of the revised design (both top and bottom deck). Let me know if you have any comments or suggestions for improvement (red is top layer, mostly audio signal routing; blue is bottom layer, mostly relay control).
