No, it is distortion. Well, let's say distortion and noise together.
Then there is measured distortion vs human perceived distortion. We don't measure exactly the same way we hear.
There is a recent trend where some dacs with excellent distortion numbers have hit the market. Some sound better than others. The new Gustard which measures great is said to sound rather poor compared to others at similar prices. It means engineers have learned how to design according to measurements, but haven't learned to listen too. Maybe that will be the next thing to happen.
@cdgames
Wow. Such a post - out of nowhere - hidden in a "boring" power supply thread.
OK. I got it. You just intended to pepper the thread a tiny bit with a cute bedtime story.
You guys shuffle the market! At that pace we can hardly follow.
I guess you'll launch separate threads soon. Doesn't make sense to highjack this thread now. I mean, all these products/solutions spawn off numerous new questions/subjects.
***************************
#noise
It's not that simple. There's more.
1. most recordings and mixes ( = our audio basis ) are fundamentally flawed
2. our home listening environments are fundamentally flawed
3. any DSP is fundamentally flawed
And then comes the noise, all kind of it. We discussed it. That's the part where you put your priorities of course.
However. There are also
* transient issues
* vibration issues (clock vibration adding e.g. 30dB phase noise - vibrations = root cause)
* temperature issues
* integration issues (impedance issues between devices)
* material and parts issues
* and more
A lot of stuff needs to perform very well to end up with a nice audio setup.
It's more then just the noise - whatever noise.
And then. You get rid of one bottleneck, just to find out you're looking at 10 new ones.
Anyhow. It's amazing what you guys are coming with. Congrats to the Allo team.
Wow. Such a post - out of nowhere - hidden in a "boring" power supply thread.
OK. I got it. You just intended to pepper the thread a tiny bit with a cute bedtime story.
You guys shuffle the market! At that pace we can hardly follow.
I guess you'll launch separate threads soon. Doesn't make sense to highjack this thread now. I mean, all these products/solutions spawn off numerous new questions/subjects.
***************************
#noise
It's not that simple. There's more.
1. most recordings and mixes ( = our audio basis ) are fundamentally flawed
2. our home listening environments are fundamentally flawed
3. any DSP is fundamentally flawed
And then comes the noise, all kind of it. We discussed it. That's the part where you put your priorities of course.
However. There are also
* transient issues
* vibration issues (clock vibration adding e.g. 30dB phase noise - vibrations = root cause)
* temperature issues
* integration issues (impedance issues between devices)
* material and parts issues
* and more
A lot of stuff needs to perform very well to end up with a nice audio setup.
It's more then just the noise - whatever noise.
And then. You get rid of one bottleneck, just to find out you're looking at 10 new ones.
Anyhow. It's amazing what you guys are coming with. Congrats to the Allo team.
One of the things that I learned , is that SQ is a very complicated thing. We are good at measuring not the best in SQ testing .
However I am not measuring THD+N only. In fact , (thx you Nirvana) we have learned to lower the smps to 1uV level (before feeding to opamps with static load). Everything peachy ...no sirrie
While playing 1Khz tone , we seen noise on the rail , 1Khz transmitting through the LDO (so from output of LDO to input). Capacitance was raised and finally its gone..
Yes noise comes in many shapes. Distortion of opamp was measured and reduced not only increasing capacitance but also adding a small R on the rails..
Still like Mark will tell you right away , SQ is not only this. How about impedance on the rails..
Look , we are still testing , still improving . I have a few ideas that might change the rails impedance . Might not work ..supercaps Mark ?
Still I wont say anything about SQ beyond my machine testing . I can say that I am trying for SQ
One PSU for the whole DAC.
However I am not measuring THD+N only. In fact , (thx you Nirvana) we have learned to lower the smps to 1uV level (before feeding to opamps with static load). Everything peachy ...no sirrie
While playing 1Khz tone , we seen noise on the rail , 1Khz transmitting through the LDO (so from output of LDO to input). Capacitance was raised and finally its gone..
Yes noise comes in many shapes. Distortion of opamp was measured and reduced not only increasing capacitance but also adding a small R on the rails..
Still like Mark will tell you right away , SQ is not only this. How about impedance on the rails..
Look , we are still testing , still improving . I have a few ideas that might change the rails impedance . Might not work ..supercaps Mark ?
Still I wont say anything about SQ beyond my machine testing . I can say that I am trying for SQ
One PSU for the whole DAC.
My question:
Is it possible to power with the 3Amp rail of Shanti the new USBridge Sig and with the 1.5Amp rail of Shanti the 5V USB Vbus output to Revolution?
Thanks
Matt
Last edited:
wow.. an easter egg in shanti.. Im excited for new USbridge!
Cdsgames, will this be compatible with any future software updates of volumio, allo dietpi et al..
my usbridge is stuck with outdated software update..
Usbridge Sig aka Audiophile SBC has a RPI compute module so same CPU , any soft will run on both
However there is a driver for our network chip that has to be installed, we will talk to all players and try to update the driver in the players...if we are not able to upstream it to Rasbian (and we are trying)
In fact the main reason we chose RPI was access to the latest kernel (unlike our last USBridge that had good hardware specs and bad kernel support). In the race to hardware specs , most forget (include me) that what you are using mostly the OS and RPI is ..very good at that.
Right now they are at kernel 5.0
However there is a driver for our network chip that has to be installed, we will talk to all players and try to update the driver in the players...if we are not able to upstream it to Rasbian (and we are trying)
In fact the main reason we chose RPI was access to the latest kernel (unlike our last USBridge that had good hardware specs and bad kernel support). In the race to hardware specs , most forget (include me) that what you are using mostly the OS and RPI is ..very good at that.
Right now they are at kernel 5.0
Last edited:
I think that Katana with the right PSUs (and yes you need dual rails psu to extract the best) is unbeatable . Dont forget that Katana can be used on a USbridge Sig as well...
Revolution was made for people that don't have time or inclinations to "play"
One PSU asks for trade offs . Still the monolitic opamps are very good and consume 1/10 of what Katana analog stage does so one 5V is enough.
As for the dual rail..thats the crucial part .
We learned from Katana , the development never stopped. I have tried many ideas and found something that worked well (filter)
Same was applied to Revolution , giving some very clean rails. But its not enough (low noise) impedance is very important . Any time someone uses a LPS on Katana opamp stage they are adding "low noise" and low impedance then sound is enhanced.
So we are just fallowing on that path.
Revolution was made for people that don't have time or inclinations to "play"
One PSU asks for trade offs . Still the monolitic opamps are very good and consume 1/10 of what Katana analog stage does so one 5V is enough.
As for the dual rail..thats the crucial part .
We learned from Katana , the development never stopped. I have tried many ideas and found something that worked well (filter)
Same was applied to Revolution , giving some very clean rails. But its not enough (low noise) impedance is very important . Any time someone uses a LPS on Katana opamp stage they are adding "low noise" and low impedance then sound is enhanced.
So we are just fallowing on that path.
Thats a good question Mark
The answer is NO. Anytime you add a USB source , your GND must be shared with the source since USB are not really fully differential they need the gnd as a reference . If you share gnd with your source *be it a laptop , streamer etc) you can create a loop and introduce noise
So I did something rather (I think) clever. I am taking the USB power (and gnd) and filter it using CMC and pi filters and terminate it outside the board (power is not connected to our main board).So your source gnd is shared but filtered and power draw is about 5ma
Now if I want to take power from the source , the whole filter becomes very big and unmanageable (I tried)
So Revolution needs a power source (dc barrel)
IIUC, that does not matter because USBridge Sig / Revolution / Shanti offer a complete streaming solution.
Matt