Hi, I am new in DAC but i would like to give it a try during my school holiday.
My goal is to create a mini USB-key size DAC in a very low budget that every students can afford the bill and built themself a unit to play! Stepping this journey, I hope any of you willing to give me some guides.
I know nuts about this, but i would like to show you guys what i am thinking,
However, i have no idea on the upsampling parts that which should fit the bill and the quality.
There are few IC in my brain which are:
*AD1896
*SRC4192
*Or something better?
For DAC section, i have no idea AK4396 is better or CS4398 is better. The reason i choose them because both of this is the cheapest performer that i know. Any other suggestion guy?
Here is the output stage i would like to try after i read up a lot of results from google, AD8620 is used because i really need a smd size and capable to duty below 5Vdc (USB)
Is there anything wrong or any parts can be remove to minimized the parts count without degrading the output quality ? So that i can have my PCB as smaller as possible
One question, it there any output stage using discreet design that can work on USB power?
My utmost thankful to all of you who read this🙄
My goal is to create a mini USB-key size DAC in a very low budget that every students can afford the bill and built themself a unit to play! Stepping this journey, I hope any of you willing to give me some guides.
I know nuts about this, but i would like to show you guys what i am thinking,
An externally hosted image should be here but it was not working when we last tested it.
However, i have no idea on the upsampling parts that which should fit the bill and the quality.
There are few IC in my brain which are:
*AD1896
*SRC4192
*Or something better?
For DAC section, i have no idea AK4396 is better or CS4398 is better. The reason i choose them because both of this is the cheapest performer that i know. Any other suggestion guy?
Here is the output stage i would like to try after i read up a lot of results from google, AD8620 is used because i really need a smd size and capable to duty below 5Vdc (USB)
An externally hosted image should be here but it was not working when we last tested it.
Is there anything wrong or any parts can be remove to minimized the parts count without degrading the output quality ? So that i can have my PCB as smaller as possible
One question, it there any output stage using discreet design that can work on USB power?
My utmost thankful to all of you who read this🙄
I dont think AD8620 will be happy with 5vdc, let alone something discrete. if you are planning to run from the USB bus power with such small parts count and low current. I think you are being too ambitious
So your basicaly making a usb only amb gamma 2.
Might as well copy some of the design since its for your own personal use. They have all the parts you need. You just need to simplify the design down to just usb and pick your dac.
By the way usb 5v is very noisy so i hope you know your way around ferrite beads and ldo 4.75v regulators.
Also check out the grub dac and bantam dac since they use some of the parts you listed for usb.
Might as well copy some of the design since its for your own personal use. They have all the parts you need. You just need to simplify the design down to just usb and pick your dac.
By the way usb 5v is very noisy so i hope you know your way around ferrite beads and ldo 4.75v regulators.
Also check out the grub dac and bantam dac since they use some of the parts you listed for usb.
I think i better don't use the gamma 2 output stage without amb's permisson. Because i want to fabricate pcb layout myself which is not using his PCB, it's not good if i continue with his design
Any others good output stage that's is open publication that i can take a look for reference?
Any others good output stage that's is open publication that i can take a look for reference?
I think i got what you meaned, by the way, do the following circuit seem fine?
For the ferrites beads, I think it will be not a problem for me
http://users.catchnet.com.au/~rjandusimports/tut_9a.html
An externally hosted image should be here but it was not working when we last tested it.
For the ferrites beads, I think it will be not a problem for me
http://users.catchnet.com.au/~rjandusimports/tut_9a.html
Last edited:
it should go like this:
usb power = 4.9-5.1 volts
then a ferrite
then a decoupling cap (.1uf)
then the 4.75-4.5 volt regulator circuit
then you run your regulated to your opamp,analog dac section, and to your 3.3v regulator for the digital section of the dac.
You dont need 470uf on the output unless your driving headphones. if your driving something normal like a 10k pot then stick with 15uf-30uf.
Also i havent checked out the opamp yet, but if your running less then 5volts into it you need to run it below unity gain, and have a opamp thats stable below unity gain. If you dont then when the dac maxes out, the opamp will clip. Id shoot for 1-1.4rms output. If you sit safe at 4 volts of swing, thats 1.418rms.
While i havent tryed this, you could try. run the 5volts usb into the opamp via a ferrite and a decoupling cap. Opamps are very good at rejecting power supply noise (dacs are the opposite). so you would run regulated 4.5v into the analog dac, and run unregulated 5v into the opamp with unity gain.
usb power = 4.9-5.1 volts
then a ferrite
then a decoupling cap (.1uf)
then the 4.75-4.5 volt regulator circuit
then you run your regulated to your opamp,analog dac section, and to your 3.3v regulator for the digital section of the dac.
You dont need 470uf on the output unless your driving headphones. if your driving something normal like a 10k pot then stick with 15uf-30uf.
Also i havent checked out the opamp yet, but if your running less then 5volts into it you need to run it below unity gain, and have a opamp thats stable below unity gain. If you dont then when the dac maxes out, the opamp will clip. Id shoot for 1-1.4rms output. If you sit safe at 4 volts of swing, thats 1.418rms.
While i havent tryed this, you could try. run the 5volts usb into the opamp via a ferrite and a decoupling cap. Opamps are very good at rejecting power supply noise (dacs are the opposite). so you would run regulated 4.5v into the analog dac, and run unregulated 5v into the opamp with unity gain.
An externally hosted image should be here but it was not working when we last tested it.
Is this correct? Both the regulator chosen has high PSRR and low noise, i think it's pretty decent to use.
You are right, since the using of IIS, volume control no longer able to control under windows anymore, it will be better to change the 470uF into 20uF with a pot afterwards.
Thanks a lot for the guide :-*
Last edited:
If your design includes the AD1896 for upsampling you will get excellent source jitter rejection as a side benefit.
if you want to make it small, good luck soldering
0402 resistors are doable by hand.
you would need a 4 layer PCB with IC's on both sides to keep it small
for price, if you make enough PCB's its not to expensive 🙂
and for the DAC's, maybe a wolfson dac?
the WM8521 is cheap (1,90 euro's at farnell for 1) kinda small, 14 pins
else the WM8501, harder to get but also not to expensive and also 14 pins
bit higher quality then the WM8521
no need for an opamp gain stage, works on 5V
downside is that the wolfson dacs have a lower SNR and higher THD
but i doubt you will notice it with the USB power supply 😉
0402 resistors are doable by hand.
you would need a 4 layer PCB with IC's on both sides to keep it small
for price, if you make enough PCB's its not to expensive 🙂
and for the DAC's, maybe a wolfson dac?
the WM8521 is cheap (1,90 euro's at farnell for 1) kinda small, 14 pins
else the WM8501, harder to get but also not to expensive and also 14 pins
bit higher quality then the WM8521
no need for an opamp gain stage, works on 5V
downside is that the wolfson dacs have a lower SNR and higher THD
but i doubt you will notice it with the USB power supply 😉
Thanks a lot for the help ^nightanole.
So this go like this right?
I think i will go for AD1896 because it's definitely cheaper compare with SRC4192.
An externally hosted image should be here but it was not working when we last tested it.
So this go like this right?
I think i will go for AD1896 because it's definitely cheaper compare with SRC4192.
it's no longer challenging 🙁
One more question, any possible to make the windows
volume control functionable if i construct like this:
PCM2707(IIS)->WM8501
And i have one question that i doubt for a very long
time, I have tried
PCM2707(IIS)->TDA1543 : the volume control in (OS)windows is not working
PCM2704(DAC) : Volume controllable in windows.
Is it mean that if i use PCM2704(IIS)->(Any DAC), then the
volume control in windows is functionable? Or..?
Thanks
One more question, any possible to make the windows
volume control functionable if i construct like this:
PCM2707(IIS)->WM8501
And i have one question that i doubt for a very long
time, I have tried
PCM2707(IIS)->TDA1543 : the volume control in (OS)windows is not working
PCM2704(DAC) : Volume controllable in windows.
Is it mean that if i use PCM2704(IIS)->(Any DAC), then the
volume control in windows is functionable? Or..?
Thanks
Last edited:
Uh you do realize that if you use the windows volume control you just lost all your resolution right?
Think about it. right now with volume disabled in windows, and no pot, the amp would be running full volume. Now you turn down the windows volume to normal listen levels. You just reduced your bit resolution from 16 bit down to 8 or even 4 bit.
PS: schematic looks good so far.
Think about it. right now with volume disabled in windows, and no pot, the amp would be running full volume. Now you turn down the windows volume to normal listen levels. You just reduced your bit resolution from 16 bit down to 8 or even 4 bit.
PS: schematic looks good so far.
chchyong,
software attenuation before DAC is bad because for each 6dB decrease in attenuation you lose 1 bit of resolution.
software attenuation before DAC is bad because for each 6dB decrease in attenuation you lose 1 bit of resolution.
Now we are going for a wolf 8501...
Pots in the right place. it doesnt have to be 10k, but it can be since your using 22uf caps, so id stick with the 10-25k mark. You will now have to buffer the output of the pot with a opamp to run headphones. You cant have a pot and run headphones directly at the same time.
i dont have much experience with running an opamp without an negative supply. You would need something that works at low voltage (5v rails) and runs at or below unity gain. Again you could run the opamp directly from usb via a ferrite bead and run it in unity gain. Amb used a AD8656 for the headphone version.
Yes your output would need to be cap coupled and be big like your original design.
This might help but its from 2000 so im not sure if you can get any of the chips.
http://courses.cit.cornell.edu/bionb440/datasheets/SingleSupply.pdf
Pots in the right place. it doesnt have to be 10k, but it can be since your using 22uf caps, so id stick with the 10-25k mark. You will now have to buffer the output of the pot with a opamp to run headphones. You cant have a pot and run headphones directly at the same time.
i dont have much experience with running an opamp without an negative supply. You would need something that works at low voltage (5v rails) and runs at or below unity gain. Again you could run the opamp directly from usb via a ferrite bead and run it in unity gain. Amb used a AD8656 for the headphone version.
Yes your output would need to be cap coupled and be big like your original design.
This might help but its from 2000 so im not sure if you can get any of the chips.
http://courses.cit.cornell.edu/bionb440/datasheets/SingleSupply.pdf
Actually in the Gamma 1 the headphone output is from the WM8051 directly so no opamps. The 8051 is capable of driving headphones directly and it runs off of a 5v supply. It's a pretty handy chip.
Then again, the PCM2707 does this too, it just sounds worse. 😀
8051 can run headphones directly or used with a pot, but not both. Thats a given with any dac. Ya gotta have an opamp right before the cans, or have the dac drive them directly with big caps.
- Status
- Not open for further replies.