Yep. I've been around for a while. And I've seen manufacturers failing and/or having their priorities wrong.
Meanwhile it's taking more then 10 years to fight USB associated issues.
All the countless USB gadgets and DACs I have around here remind of these awful 10 years.
Your new device proves it once more. USB is simply not an audio interface.
That's funny. You got it wrong again.
That's a statement that pretty much applies to the entire computer audio area. Not any computer is designed for audio purposes.
It simply the job of the manufactures who're committed to produce audio interfaces to cope with this.
And the funny thing. Even the "crappy" RPI (via I2S) allows to have audio jitter performances in the femto seconds area, we see extremely low noise and THD values. Nope the RPI is not the issue here.
Good luck.
PS: Who needs DSD512!??
We appreciate having our 59$ SMPS Nirvana and expensive 159$ LPS Shanti compared with hi quality lab supplies .
Now , we do know how to measure the transients , but we need specialized test machines (automatic load) etc to fully measure.
Also , please note that no audio PS is providing this data ( IFI)
I understand that you keep ALLO at a higher standard ? Honestly , I kinda appreciate it .
As soon as I have some time , I will find a way to measure it and post it. Expect a long rant..
Now those "nasty spikes" on Nirvana at less than 1uV, are because of the nature of switching power supply and because of our "obsession" with noise.
In any SMPS those spikes will be buried into the noise floor.Nirvana noise floor is low enough to observe them. At last , the fact that we test in "real word scenario" (since most users do not have isolation transformers) , they are rather free to express themselves .
Linear PSU has a build it isolation transformer that make it more gentle (and still you can see the 20K)
Now , we do know how to measure the transients , but we need specialized test machines (automatic load) etc to fully measure.
Also , please note that no audio PS is providing this data ( IFI)
I understand that you keep ALLO at a higher standard ? Honestly , I kinda appreciate it .
As soon as I have some time , I will find a way to measure it and post it. Expect a long rant..
Now those "nasty spikes" on Nirvana at less than 1uV, are because of the nature of switching power supply and because of our "obsession" with noise.
In any SMPS those spikes will be buried into the noise floor.Nirvana noise floor is low enough to observe them. At last , the fact that we test in "real word scenario" (since most users do not have isolation transformers) , they are rather free to express themselves .
Linear PSU has a build it isolation transformer that make it more gentle (and still you can see the 20K)
Klaus
you are wrong.
Femtosecond jitter on a RPI ? Ohh no, not even ps . Nanosecond I think(know). We have the test data.
Now you are right on the USB. Give me i2s anytime, but you must understand that we make our products for everyone and not only for you. Many people here have USB DACs and they invested money and time into them . I think they deserve a good USB source .
We have done it with Digione , and new USBridge (in my opinion) will do to USB what Digione did on Spidif. It changed the game.
I will repeat. Noise is the enemy of good SQ. Noise , include in that definition distortion. And impedance ...
USBridge Sig. Audiophile RPI , has the lowest noise(see above difinition) that out test machines ever measured . The lowest noise on the i2s AND i2c. GPIO as well
You talk about RPI4 advancing on ethernet , me I talk about how I added CMCs on the Ethernet (because I dont think isolation transformers are enough check our Nirvana and Shanti both have 15mH CMC)
Is there here anyone that learned from past ? Look at what Greg is reporting (and he has what I consider a crazy system with a real obsession of noise) an RPI with linear power sounds..is wonderful the word Greg ? (and its an old RPI)
We listened .(but we added super caps .caps to lower the impedance on the rails and changed the layers of the PCB to 10)
you are wrong.
Femtosecond jitter on a RPI ? Ohh no, not even ps . Nanosecond I think(know). We have the test data.
Now you are right on the USB. Give me i2s anytime, but you must understand that we make our products for everyone and not only for you. Many people here have USB DACs and they invested money and time into them . I think they deserve a good USB source .
We have done it with Digione , and new USBridge (in my opinion) will do to USB what Digione did on Spidif. It changed the game.
I will repeat. Noise is the enemy of good SQ. Noise , include in that definition distortion. And impedance ...
USBridge Sig. Audiophile RPI , has the lowest noise(see above difinition) that out test machines ever measured . The lowest noise on the i2s AND i2c. GPIO as well
You talk about RPI4 advancing on ethernet , me I talk about how I added CMCs on the Ethernet (because I dont think isolation transformers are enough check our Nirvana and Shanti both have 15mH CMC)
Is there here anyone that learned from past ? Look at what Greg is reporting (and he has what I consider a crazy system with a real obsession of noise) an RPI with linear power sounds..is wonderful the word Greg ? (and its an old RPI)
We listened .(but we added super caps .caps to lower the impedance on the rails and changed the layers of the PCB to 10)
Last edited:
This is a very complicated question.
Simple answer. Yes if you believe other people reports , NO if you think that test data(ap machine) shows the truth.
I think that SQ is a complex. Seems to me that even on digital streams some noise gets to the output. The thing I don't understand , is why even with galvanic isolation ICs , there are reports of increased SQ with a better source. Is it psychological bias ?
I dont know.
i2s/i2c signals on Usbridge Sig are much cleaner that RPI 3 (I dare include RPI4 but no data to back it up)
I don't think so.
Unless you tell me that your RPI audio interfaces are impacted by the RPI I2S quality.
As far as I recall you market your RPI audiointerfaces with jitter levels well below single digit picoseconds.
And that means that the actual SBC jitter is basically irrelevant.
The same logic applies to noise. What matters is what I see at the audio interface output.
I'm not commenting on Gregs wild fishing exercises.
The fact that split power supplies and additional stability on the rails can be a game-changer is known for a long long time (years).
For sure the transient capabilities of the PS and impedance conditions on the rails play a (big) role here.
Because , you know, that's what you'll actually impact if you wildly hook up Farrads of capacity to the power rails.
Unfortunately - as you outlined a couple of hours ago - you can't measure this transient behavior. That's a pity.
It might help to understand what's actually going on there.
Enjoy.
Last edited:
Hi Matt
If you want to achieve absolute best audio quality from USB streams or even for i2s , SOTM and uRendu, for example , shows that not only USB hub has to be quiet (LDOs) but the CPU itself has to be feed with linear power (ldo)
Any switching element on board , will introduce spikes (look at how long we worked on Nirvana ) that we can reduce but never fully eliminate (at noise floor)
I am attaching the test data for Boss THD+N on a RPI vs Usbridge Sig (same PSU used).
As soon as possible I will test the USbridge Sig vs RPI4.
If you want to achieve absolute best audio quality from USB streams or even for i2s , SOTM and uRendu, for example , shows that not only USB hub has to be quiet (LDOs) but the CPU itself has to be feed with linear power (ldo)
Any switching element on board , will introduce spikes (look at how long we worked on Nirvana ) that we can reduce but never fully eliminate (at noise floor)
I am attaching the test data for Boss THD+N on a RPI vs Usbridge Sig (same PSU used).
As soon as possible I will test the USbridge Sig vs RPI4.
Yeap we agree.What matters is what you see on output (ap machine test)
Look at my last test data and explain why on same i2s digital stream(master DAC) we get different THD+N (1db almost) if i2s (must be the noise) is irrelevant ?
Noise (and probably jitter ) matter on digital streams and we can observe THD+N change .
Now on USB the change is less dramatic..but still clearly observed . We get between 0.2 to 0.4db improvement in THD+N using a Topping dac.
Now , there are those of you that don't think that AP machines are able to fully capture the sound improvements of a good source.
Maybe its bias , but what I hear its an improvement well beyond what my AP is showing .
Sorry , there is 1.8db difference in THD+N on RPI3 vs USBridge Sig (Boss)
Noise is same . THD has improved by 2db
Those results are not easy to explain . Is the source noise does not matter (i2s is a digital stream and jitter is eliminated since DAC is running in master mode) why is there a difference ?
Noise is same . THD has improved by 2db
Those results are not easy to explain . Is the source noise does not matter (i2s is a digital stream and jitter is eliminated since DAC is running in master mode) why is there a difference ?
Thanks.
Looking forward to the measurements, maybe better to make these with Katana or Revolution.
What is your timetable for launch of USBridge and Revolution? I suppose USBridge comes first.
Matt
Revolution DAC , we are at the testing limits of our AP(THD version). Not ready since we are not concentrating on the THD+N but rather on SQ
Usbridge Sig, 6 weeks away max.
We got the new RPI4 on the way , next week we will test USB3 and 2 noise , THD+N of Katana on top , noise on i2s lines (to see if any improvement)
Usbridge Sig, 6 weeks away max.
We got the new RPI4 on the way , next week we will test USB3 and 2 noise , THD+N of Katana on top , noise on i2s lines (to see if any improvement)
It is well known that ES9038Q2M sounds best with DSD.
Is USBridge Sig able to send and Revolution DAC able to receive DSD native (not DoP)?
Thanks
Matt
@cdsgames
1.
When doing your measurements you shouldn't forget to also show numbers
with your isolator in place. Many people own and appreciate it.
The RPI + isolator is a very capable and affordable combo. And does make a difference.
2. THD-N
You meanwhile have the noise and the jitter very well under control on your RPIs AIFs . As I said before, and you basically agreed to it, it's the audio interface which is in charge to deliver highest quality results. No matter which source is feeding it. You've been doing extremely well on that account.
The new question if the SBC suddenly should now be in charge to improve THD-N on the audio interface is nonsense. It's again the audiointerface which is in charge.
If a new "audiophile" SBC improves the THD-N on the the audio interfaces it simply shows there's still a loophole/weak spot on the audio device that impacts THD-N when changing the input conditions.
It just shows there's still room for improvement on the audio interfaces.
Enjoy.
1.
When doing your measurements you shouldn't forget to also show numbers
with your isolator in place. Many people own and appreciate it.
The RPI + isolator is a very capable and affordable combo. And does make a difference.
2. THD-N
You meanwhile have the noise and the jitter very well under control on your RPIs AIFs . As I said before, and you basically agreed to it, it's the audio interface which is in charge to deliver highest quality results. No matter which source is feeding it. You've been doing extremely well on that account.
The new question if the SBC suddenly should now be in charge to improve THD-N on the audio interface is nonsense. It's again the audiointerface which is in charge.
If a new "audiophile" SBC improves the THD-N on the the audio interfaces it simply shows there's still a loophole/weak spot on the audio device that impacts THD-N when changing the input conditions.
It just shows there's still room for improvement on the audio interfaces.
Enjoy.