Hi All,
Returning after years and having forgotten most of what I knew! I was running the originl Raspberry Pi with a Hifiberry Dac hat. I have kept this going with sellotape and sheer will power for too long. I've given in and purchased a Raspberry Pi 4. I have not bought a Dac hat for it (Yet)
I would like to see if I can repurpose another project - A chinese AK4399 based Dac that had an Amanero board connected via I2S for USB input from a Linux PC
I have no PC now and would like to run Volumio on my Pi 4 as music source. My question is -
How will I connect raspberry Pi4 to my AK4399? Directly by I2S or via USB and Amanero? I have some parts lying around to make a smooth 5V supply and intend to place the Pi in the box with the Dac as shown in pic. Any suggestions welcome!
Returning after years and having forgotten most of what I knew! I was running the originl Raspberry Pi with a Hifiberry Dac hat. I have kept this going with sellotape and sheer will power for too long. I've given in and purchased a Raspberry Pi 4. I have not bought a Dac hat for it (Yet)
I would like to see if I can repurpose another project - A chinese AK4399 based Dac that had an Amanero board connected via I2S for USB input from a Linux PC
I have no PC now and would like to run Volumio on my Pi 4 as music source. My question is -
How will I connect raspberry Pi4 to my AK4399? Directly by I2S or via USB and Amanero? I have some parts lying around to make a smooth 5V supply and intend to place the Pi in the box with the Dac as shown in pic. Any suggestions welcome!
Catching up with my knowledge. It's seems the AK43xx chips are only capable of running in slave mode. Most people on this forum state the issue with the Pi clock as not being divisible into 44.1khz and therefore introducing lots of unwanted jitter.
So do I run a USB from Pi to Amamero to feed the AK4399 or just save myself time and buy a master clock Dac hat such as hifiberry dac Pro or Allo boss?
Thanks Stu
So do I run a USB from Pi to Amamero to feed the AK4399 or just save myself time and buy a master clock Dac hat such as hifiberry dac Pro or Allo boss?
Thanks Stu
If you use the Amanero to interface the dac to RPi then you won't have to deal with RPi GPIO bus jitter. Instead the Amanero will provide crystal clocking. However, Amanero boards produce the cleanest I2S and MCLK signals when run from clean +5v power, rather than using USB power. There are a couple of ways to get clean power into Amanero. One way is to modify a USB cable by cutting the +5v line and splicing in an external +5v power supply for the Amanero end of the cable. The other common way is to remove the component labeled L1 from the Amanero PCB. L1 is located very close the USB connector. One end of L1 connects to USB power, and the other end connects to the local 3.3v regulators on the Amanero board. The external +5v supply can be connected to the load side L1 pad and to the ground plane on the back of the board.
Also, the way you have I2S signals connected between Amanero and the dac board is not ideal. The signals are RF and so signal integrity will better maintained if proper RF wiring techniques are used. Each of the three I2S signals and the MCLK signal should have a dedicated ground wire. So that would be a total of 4 ground wires. Each I2S signal should be paired with its own ground. The pairs can be loosely twisted and each pair separated from the other pairs by as much space as is practical. The other way is to use ribbon cable and make every other conductor a ground. In any case the I2S and MCLK wiring should be kept a short as possible. Also, it is advisable to use gold flashed or gold plated pin header connectors which are obtained from a reliable source (in particular gold color metal used on most Chinese connectors is not actually gold). Tin corrodes and the connection quality can become unreliable after as little as several hours. When that happens jitter will likely get worse. So, if you don't like GPIO bus jitter you probably shouldn't like wires stuffed into pin header sockets either 🙂
Thank you so much Mark - I was actually coming to the same conclusion about PSU myself and I was already considering making a short USB cable for inside the box from Pi to Amanero. Questions came up for me though. Surely I need to share the supply between Pi USB and Amanero USB or can the data run independent of common 5V as long as both bits of kit are powered? i.e. does there need to be a common ground?
I have 3 x TPS7A4700 LDO regulators with rectifyers built in. I know they can only handle 1A so I ruled out using them on the Pi but thought I could go the way you suggested by powering the Amanero. Then I got stuck on the common ground and didn't know how the 5V from the Pi would 'leak' into the Amanero supply. With a dropout voltage of 0.3V could I just cut the 5V and ground of USB and splice in the TPS7A4700 (bypassing the diode recitification) is series?
I had also hoped to maybe use one of the TPS modules on the op amp circuits on the DAC itself - but that is another story
And thank you for the headup about bestpractice of I2S connections - I will look to rewire this!!
bw
Stuart
I have 3 x TPS7A4700 LDO regulators with rectifyers built in. I know they can only handle 1A so I ruled out using them on the Pi but thought I could go the way you suggested by powering the Amanero. Then I got stuck on the common ground and didn't know how the 5V from the Pi would 'leak' into the Amanero supply. With a dropout voltage of 0.3V could I just cut the 5V and ground of USB and splice in the TPS7A4700 (bypassing the diode recitification) is series?
I had also hoped to maybe use one of the TPS modules on the op amp circuits on the DAC itself - but that is another story
And thank you for the headup about bestpractice of I2S connections - I will look to rewire this!!
bw
Stuart
Yes, a common USB ground is needed. Regarding a common ground, you're kind of stuck with that if using Amanero. Unless maybe you want to make your own isolator board. Easiest way to get galvanic isolation is probably to use I2SoverUSB instead of Amanero. Also I believe I2SoverUSB has somewhat better clocks.
TPS7A4700 is IME a pretty good sounding LDO. However, the most sensitive voltage on the dac would be for Vref. Maybe two of them if AK4493 has separate pins for each channel's Vref. To get the best results there I would prefer to use discrete regulators. Maybe try a shunt (or 2 of them). That might sound best. Even if using a series regulator for Vreg I often try adding a load resistor from the regulator output to ground to see if that helps SQ. It can change regulator loop gain since the pass transistor gain is current dependent.
TPS7A4700 is IME a pretty good sounding LDO. However, the most sensitive voltage on the dac would be for Vref. Maybe two of them if AK4493 has separate pins for each channel's Vref. To get the best results there I would prefer to use discrete regulators. Maybe try a shunt (or 2 of them). That might sound best. Even if using a series regulator for Vreg I often try adding a load resistor from the regulator output to ground to see if that helps SQ. It can change regulator loop gain since the pass transistor gain is current dependent.
Last edited:
I have two LDO's for Vref on Dac and will approach this after dealing with how to get data from Pi to DAC
Would rather not buy any additional kit - Is there any way I can improve power to Amanero with my spare LDO??
Stuart
Would rather not buy any additional kit - Is there any way I can improve power to Amanero with my spare LDO??
Stuart
IMHO and IME most LDOs are not good choices for Vref if SQ is the end goal. LDOs can be fine if the goal is for good looking measurements. That said, other folks may choose to disagree. In your case I might try one TPS7A4700 for each dac channel and try resistor loading of the LDO outputs (or one shared LDO could work but stereo separation may be worse). The LDOs can only dissipate so much power, which is given by the voltage drop across them times the current flowing though them. Thus you can load the output the most if the input voltage to the LDO is not too high.
This advise you keep repeating applies to differential signals but not to I2S signals that are single ended. There is no guarantee that the return current of one I2S signal wire travels along the ground wire it's twisted with so the result may actually be worse than with no twisting.
Yes, I disagree with this. IME with AKM dacs both for good sound and good measurements Vref needs to have very low noise especially at LF. TPS7A4700 does not have very low noise at LF so there are better options.
Thanks for all the advice - I've just got the AK 4399 board working with my new Pi4. Its night and day different from the old Pi with Hifiberry dac hat. The Hifiberry is much much cleaner sounding so when I first heard the AK4399 alongside it I thought it was a huge downgrade! But I have started to appreciate the AK4399 as a much fuller sound and I believe it is easier for the hifiberry to sound cleaner when it is not as full bodied! I think I would like to not give up on the AK4399 but would love to find the middle ground! The AK4399 is much more 'woolly' sounding than the hifiberry. Am i flogging a dead horse to try modding the AK4399? Are there any other decent dac module boards that are i2s and newer? I bought the AK4399 almost 8 years ago I think!
It seems the newer AK44** boards are non existent today? Is there a chip shortage?
It seems the newer AK44** boards are non existent today? Is there a chip shortage?
IIRC according to Maxwell, HF RF mirror ground currents flow very close the the signal current (even in a spacious ground plane!). At LF ground loops can be more of an issue when there are separate ground wires, but not the issue here. Besides, IME what I suggest has been tested and works quite well in practice.
Last edited:
If one is on a limited budget and can't do what is ideal, and if SQ outweighs FFT PSS measurements, then use what regulator sounds best within budget. If the dac is primarily for PSS FFT measurement use then go for best measurements within budget. Seems reasonable to me.
@mark - are you suggesting the best upgrade I could make is the I2SoverUSB v.III from JLSounds?
And then use my TPS7A4700 boards to power both the USB part and the generators with the reclock? And clean up my I2S wiring (lol)
Not sure what you mean by "try one TPS7A4700 for each dac channel and try resistor loading of the LDO outputs"? Are you referring to the Vref of the opamp circuits here?
Now that I've actually heard my AK4399 in action - I think I am prepared to spend some sensible money on it!
S
And then use my TPS7A4700 boards to power both the USB part and the generators with the reclock? And clean up my I2S wiring (lol)
Not sure what you mean by "try one TPS7A4700 for each dac channel and try resistor loading of the LDO outputs"? Are you referring to the Vref of the opamp circuits here?
Now that I've actually heard my AK4399 in action - I think I am prepared to spend some sensible money on it!
S
Basically my position would be that if you want to know the the difference between isolated ground or not, with your ears, and with your system, I2SoverUSB can be useful for that. Also, if you are trying to make your system the best it can be by fixing every issue problem you know could be a problem, maybe just so you can find out just how good you can do, then I2SoverUSB would likely prove useful too. I can't tell you if you will notice much difference right now.
So, because you are one a budget I would suggest to try clean power with Amanero first. Please let us know what you think. Then as a separate experiment try some better Vref regulators, then try loading them. Please let us know what you think at each step. If you don't change more than one thing a time then you will learn the most from each change. That is, if learning is a goal.
Thing is, in real world systems that people often have it can be very hard to say exactly whether one-off technical improvements will immediately become audible. Often there are masking factors that tend to obscure small changes in SQ (although several small improvements can add up to something more substantial). IME it doesn't take much wrong/non-ideal for such masking to occur. Thus I can only say a few things with much certainty. I can say I would recommend that you experiment to see how sound changes if you follow different people's advice, and or if you have your own ideas you want to try. Also, I can say that if you are ever in Northern California and want to stop by Auburn for a listening session, I can demonstrate effects that are audible here with this particular system.
Other than that, good to be aware that people tend to want efforts they make to improve sound be successful. Thus the immediate reaction to any audible change is often to feel that it is a change for the better. Often it turns out to be merely different, not better. Sometimes its worse. Also, some changes take time to settle in. So you should think about ways to overcome and other otherwise deal with bias and expectations. Its another subject though, so I won't try get deep into it here.
So, because you are one a budget I would suggest to try clean power with Amanero first. Please let us know what you think. Then as a separate experiment try some better Vref regulators, then try loading them. Please let us know what you think at each step. If you don't change more than one thing a time then you will learn the most from each change. That is, if learning is a goal.
Thing is, in real world systems that people often have it can be very hard to say exactly whether one-off technical improvements will immediately become audible. Often there are masking factors that tend to obscure small changes in SQ (although several small improvements can add up to something more substantial). IME it doesn't take much wrong/non-ideal for such masking to occur. Thus I can only say a few things with much certainty. I can say I would recommend that you experiment to see how sound changes if you follow different people's advice, and or if you have your own ideas you want to try. Also, I can say that if you are ever in Northern California and want to stop by Auburn for a listening session, I can demonstrate effects that are audible here with this particular system.
Other than that, good to be aware that people tend to want efforts they make to improve sound be successful. Thus the immediate reaction to any audible change is often to feel that it is a change for the better. Often it turns out to be merely different, not better. Sometimes its worse. Also, some changes take time to settle in. So you should think about ways to overcome and other otherwise deal with bias and expectations. Its another subject though, so I won't try get deep into it here.
Flat ribbon cable works just as well (or better) so no need to complicate the issue by twisting wires with uncertain results. Coaxial cables would be even better but unfortunately most USB-to-I2S boards do not have this option.
By the looks of it your regulator board has insufficient heatsinking for 1A so available current may be much less (even below 0.5A). TPS7A47 has thermal protections that kick in if the chip heats up too much but I would still be cautious with those boards as TPS7A47s are expensive and hard to come by these days.
My design goal in DACs is both good sound and measurements as IME they go hand in hand.
Thanks Both, appreciate what I'm learning here! So one thing at a time then.. I will 'try to clean up the power to the Amanero first and then I will turn my attention to a better output stage later, probably still based around the NE5534's
I'm going to 'splice' a USB cable and 'Inject with power from the TPS7A47' - I have drawn a picture for clarity because I understand that a direct parrallel splice will simply inherit 'dirty noise' from the USB surely? Or can I just keep a common ground and cut the Vcc supply from the Pi USB?
I'm going to 'splice' a USB cable and 'Inject with power from the TPS7A47' - I have drawn a picture for clarity because I understand that a direct parrallel splice will simply inherit 'dirty noise' from the USB surely? Or can I just keep a common ground and cut the Vcc supply from the Pi USB?