My favourite is LME49990, but it has been obsolete for a long time and according to TI it will not be produced anymore 
When I bought it from ebay and compared it with my genuine, the one from China was 100% fake or remarked. ADA4627 is very good, if not the best replacement for LME49990 in my opinion. Nothing wrong is with ADA4898-1 (I have a typo in BOM )
When I bought it from ebay and compared it with my genuine, the one from China was 100% fake or remarked. ADA4627 is very good, if not the best replacement for LME49990 in my opinion. Nothing wrong is with ADA4898-1 (I have a typo in BOM )
Last edited:
opa828, middle is transparent, specially low frequency, full of layers.
ad4627, balance, voice so natural.
i will try ad797 later, i dont know 823 or 826 is dual version of 828, but strongly recommend ada4627, sound likea good audio output transformer.
It is equally if not more important to use high quality power for digital part aswell IME, It is a NOS DAC so whatever comprises the ''digital'' part is very much an analogue part aswell.
I think the seperate power inputs is more a way of allowing you to isolate some noisier parts of the DAC from another and/or supply different voltages that is more optimal for some part, both are very important quality wise.
If you havent already bought a power supply I recommend a discrete reg like ultrabib or super regulator, you can power both analogue and digital parts on 12V with these and it will still sound better (and probably be cheaper) than using any seperate 12 and 5V ultra low noise LDO supply... not that those discrete regulator dont have excellent measurable performance aswell, because they do
Last edited:
Sorry, to answer your last question, yes it would work fine, just not recommended if best possible sound is the goal.
The problem is that The Raspberry need more current that The se regulators can provide...
Well I dont understand the concern about using a very low noise reg instead of an ultra low noise one for your DAC when you plan to share the supply with a very power hungry and noisy device like a Rpi.
There is more to regs than just noise, you can see miro still recommends the older regs for this DAC
The ultra low noise LDOs like to be lightly loaded, so you may end up with better sound using the older high current regs in this position anyway.
There is more to regs than just noise, you can see miro still recommends the older regs for this DAC
The ultra low noise LDOs like to be lightly loaded, so you may end up with better sound using the older high current regs in this position anyway.
So, the best practice what is?
I have a transformer that provide 2*9v 2*15v. From this one I feed the raspberry with a 3A linear regulator for example using +9v line, and as well with the same line another ultra low regulator for the +-5v DAC power.
With 2*15v i feed for example the ultra low regulator for +-12v part.
Could is it works?
But when I connect the i2s from the raspberry to the DAC, the noises it's not transmitted as well?
I have a transformer that provide 2*9v 2*15v. From this one I feed the raspberry with a 3A linear regulator for example using +9v line, and as well with the same line another ultra low regulator for the +-5v DAC power.
With 2*15v i feed for example the ultra low regulator for +-12v part.
Could is it works?
But when I connect the i2s from the raspberry to the DAC, the noises it's not transmitted as well?
Yes, that is one way to do it.
I would use the 15V line for both 5v and 12v ULN regulators for the DAC and leave the 9V line only for Rpi.
Anyway, no need to settle one way.
The best thing to do is try all these different configurations, you will either hear no difference between them or you will find one configuration that sounds best and in the process gain some experience on the influence of power supplies on sound quality.
Last edited:
Ok, the problem is that I have to buy 4 uln regulators or 2 super regulator... at least if I will not find one that provide 2 voltages value for neg/pos voltages output.
Do you know the current required for this dac?
Thank you a lot
Ps: when I connect the i2s from the raspberry to the DAC, the noises it's not transmitted as well?
Do you know the current required for this dac?
Thank you a lot
Ps: when I connect the i2s from the raspberry to the DAC, the noises it's not transmitted as well?
Last edited:
sound quality is not just low analog noise, as the consumption is not constant but at different frequencies, there may occur various distortions (which is reflected in the color of the sound and details) ... this is more of a theory than a measurement ... so you can try different power sources and write us which one you liked the most, your own research
back to you your question: digital noise is a big concern ... luckily, the AD1862 has a sophisticated separation of the analog part from the digital part ... I would like to know more details from the original authors of the chip, but I do not know their names
if the DAC chip design is perfect (AD1862 is), and the PCB is routed fine (which should be
), digital interference will not be transferred to the analog via I2S wiresYou do not need 2 super regulators if you want good sound from it.
Im am using 1 ultrabib at 12v for both supplies at the moment,
An independant 5V supply with its own transformer using TPA7A4700/3300 ULN LDOs for 5V digital supply sounds worse than the ultrabib on its own.
Eventually a super regulator will replace the TPA7A4701 for 5V digital but as it stands power quality is of a much higher priority than optimal voltage or isolated supplies.
Im am using 1 ultrabib at 12v for both supplies at the moment,
An independant 5V supply with its own transformer using TPA7A4700/3300 ULN LDOs for 5V digital supply sounds worse than the ultrabib on its own.
Eventually a super regulator will replace the TPA7A4701 for 5V digital but as it stands power quality is of a much higher priority than optimal voltage or isolated supplies.
Last edited:
+1.
both digital and analog are important and I will split at minima the secondaries from the traffos. Strongly advised C-Core from my experience. I always use two different traffos for the analog & digital. You can not go wrong with utra low noise regs for the digital side, analog asks more care and linear regs imho. But to start, nothing wrong with ldo regs but the fact some layout gives good spec but average good sound according the source of the designed pcb. Mavrica member did cool TpA7 regs, very easy to tweak on the sound for the digital side because the main cap. I never own BiBs...
As Laserscape said : experimention is your best friend : layout, powersupply & passive parts choice can enhance or waste a design.
both digital and analog are important and I will split at minima the secondaries from the traffos. Strongly advised C-Core from my experience. I always use two different traffos for the analog & digital. You can not go wrong with utra low noise regs for the digital side, analog asks more care and linear regs imho. But to start, nothing wrong with ldo regs but the fact some layout gives good spec but average good sound according the source of the designed pcb. Mavrica member did cool TpA7 regs, very easy to tweak on the sound for the digital side because the main cap. I never own BiBs...
As Laserscape said : experimention is your best friend : layout, powersupply & passive parts choice can enhance or waste a design.
Last edited:
1 ultrabib for both positive and negative supplies?
How did you manage to get pos and neg 12v from one ultrabib?
It is and the Rpi layout is not quiet enough . How do you use the Rpi ? Some 1/3 party hats ?
Last edited:
The I2S direct hacked from the Rpi GPIO header is both noisy and not timing proof due to the Rpi architecture : you will have a great improvment by injecting a cleaner reclocked I2S with true crystals and not an I2S extrapolated from the poor standalone Rpi one that have timing issue and very noisy after passing through the embeded fgpa -dunno if right word for that- and the dna electric noise problems of the Rpi for that delicate task.
A good investment for all your futur diy I2S streaming project will profite from a the FifoPi hat for Rpi made here by IanCanada, also avaliable at Audiophonics - who I don't like especially due to the poor team I have had to talk services with- but reliable seller. the layer added basicly will get you a cleaner reclocked & fifoed I2S with noise isolation.
With NDK crystal it will mate pretty well with the AD1862 dac chip imho.
If you directly stream from USB a JLSounds lab USB to I2S is also good for less monney but will not have of course the versability streaming fro a Rpi based solution. If you stream for a computer or a NAS via USB, it's enough, the Fifo Rpi goes further though.
The In between is also the Kali reclocker Rpi hat from Allo for cheap enough : be just carefull of the frequency Xtal chosed related to the upsampling frequency of your music.
A good investment for all your futur diy I2S streaming project will profite from a the FifoPi hat for Rpi made here by IanCanada, also avaliable at Audiophonics - who I don't like especially due to the poor team I have had to talk services with- but reliable seller. the layer added basicly will get you a cleaner reclocked & fifoed I2S with noise isolation.
With NDK crystal it will mate pretty well with the AD1862 dac chip imho.
If you directly stream from USB a JLSounds lab USB to I2S is also good for less monney but will not have of course the versability streaming fro a Rpi based solution. If you stream for a computer or a NAS via USB, it's enough, the Fifo Rpi goes further though.
The In between is also the Kali reclocker Rpi hat from Allo for cheap enough : be just carefull of the frequency Xtal chosed related to the upsampling frequency of your music.
Last edited: