Well...I cant confirm that. i2Soverusb is always the same card as the USB-Cable before is the same....I think we can build lots of different hypothesis here...example:
I always thought as the USB-Chip part of the card is powered by the USB sender device ( the bluesound or Alix USB-sotm card), that the quality of this power might be a factor too...the Alix board got its own linear PSU, the sotm-card got its own linear PSU and has its own linear regs which powers the USB output...
I always thought as the USB-Chip part of the card is powered by the USB sender device ( the bluesound or Alix USB-sotm card), that the quality of this power might be a factor too...the Alix board got its own linear PSU, the sotm-card got its own linear PSU and has its own linear regs which powers the USB output...
I2SoverUSB is by default half-powered by USB, isolated, and then the clock and output side are powered by a clean 5v supply. However, there is option to run the USB side from yet another clean, fully isolated 5v supply. In severe noise cases that might help.
Regarding 320-bit perfect, so far as I can tell the concern from users is not about bit-errors, so much as it is about jitter and EMI/RFI noise that could couple into the FIFO buffer. One thing Andrea might have done differently would have been to use a dedicated ground pin for each I2S signal. Signal integrity between the USB board and FIFO board would probably be better that way. That said, its best to keep the I2S wiring very short. An inch or two should be all you need. I would also orient the ground plane side of the USB board towards the FIFO board, so as to provide a little shielding effect. Have seen radiated noise coupling between USB boards and DAC boards before.
Regarding 320-bit perfect, so far as I can tell the concern from users is not about bit-errors, so much as it is about jitter and EMI/RFI noise that could couple into the FIFO buffer. One thing Andrea might have done differently would have been to use a dedicated ground pin for each I2S signal. Signal integrity between the USB board and FIFO board would probably be better that way. That said, its best to keep the I2S wiring very short. An inch or two should be all you need. I would also orient the ground plane side of the USB board towards the FIFO board, so as to provide a little shielding effect. Have seen radiated noise coupling between USB boards and DAC boards before.
Last edited:
If the DRIXO clock board is in its aluminum case,It is preferable to place it at 10 centimeter or 1meter of FIFO.
Could be. It was a while ago. I used WaveIO which in its day seemed quite good. No doubt USB2I2S has gotten better in the past few years. I abandoned USB because it seemed fundamentally limited. Going to I2s first with a BBB and S03 isolator/reclocker was a clear improvement. Today, I am happy with the Rpi/IanCanada stack with WellAudio clock. It feeds a 30 yr old 1541a chip, so its far from SOTA. But its reached that point that I'm content to just listen to it.
I do keep the DAC and triode output stage isolated in its own little copper clad world. All the digital stuff is on the other side of copper plate connected by 2.5" ufl cables. Everything is grounded with beefy copper terminating to an 8 gauge copper ground bus to the safety ground. The clock is quite isolated in its own chassis well separated with a 30" SMA terminated coax. No doubt there are lots of opportunity for improvement. It was a fun challenge laying it all out though.
I do use rg400 coax with SMA connectors. I tried lower gauge cables with SMA and with ufl connectors and discovered the cable does impact sound. Finger tight connection vs properly torqued down SMA connection also impacts sound. That is what I meant by it is a fragile network where everything matters. I also had ufl cables that had been removed and replaced a few times and found that restoring the ground connection by crimping down the connector could make a significant difference. I concluded that once you are happy with your setup, the last thing you do is buy all new ufl cables and then never touch them again.
There is something to be said for once you like it, leave it be.
I started recording a set of baseline tracks just using my smart phone. It is interesting now to be able to go back and listen to new "improved" versions of the system vs where it was at any given stage in the past and ask yourself "is it better"?
There is something to be said for once you like it, leave it be.
I started recording a set of baseline tracks just using my smart phone. It is interesting now to be able to go back and listen to new "improved" versions of the system vs where it was at any given stage in the past and ask yourself "is it better"?
If asynchronous UAC is properly implemented the host has very little impact (if any) on jitter. Anything prior to host (e.g. NAS or ethernet cables) has absolutely no impact on jitter. EMI/RFI noise is more or less eliminated if USB-I2S device has properly implemented digital isolation. The possible jitter generated by USB-I2S should be eliminated by the FIFO. The amount of jitter due to USB-I2S device depends on what type of DAC is used. E.g. for DS dacs the most critical MCK can be fed directly (not through USB-I2S).
In a 32-bit perfect asynchronous USB-I2S the only possible jitter from host could be due to some ground interference or noise (and even that is hypothetical). Any such issues can be eliminated if USB-I2S is powered from e.g. battery.
Last edited:
Well...not really. They got their little own transformers and linear regs...but there is no connection to earth.
As i said, we can come up with many theories...
Truth is that most if us do hear a significant difference by changing (in theory) bit perfect sources before the fifo. That has always been the case since the first days of ians first prototypes, his raspberries, the soekris 1941 and now with Andreas fifo.
And the difference of sound signature as well stays consistent btw. An alix/sotm sound signature is over all four fifos and different usb-i2s cards the same compared to a raspberry or a bluesound...and a sdcard reader sound signature i guess as well...
Just talked with the german Sotm distributor...will get the
As i said, we can come up with many theories...
Truth is that most if us do hear a significant difference by changing (in theory) bit perfect sources before the fifo. That has always been the case since the first days of ians first prototypes, his raspberries, the soekris 1941 and now with Andreas fifo.
And the difference of sound signature as well stays consistent btw. An alix/sotm sound signature is over all four fifos and different usb-i2s cards the same compared to a raspberry or a bluesound...and a sdcard reader sound signature i guess as well...
Just talked with the german Sotm distributor...will get the
SOtM sMS-200ultra and its smaller brother incl linear PSU within the next 2 weeks...(sorry for bold writing, not sure why that is...)
Last edited:
As do you, in how loopback and jitter measurements should correlate to what is actually heard.
@Blitz yep, seems that is just the way it is unfortunately.
Hopefully andrea will design his own synchronous SD source, for starters one that can operate with the 5/6MHz clocks, seeing as how the demand is there.
@Blitz yep, seems that is just the way it is unfortunately.
Hopefully andrea will design his own synchronous SD source, for starters one that can operate with the 5/6MHz clocks, seeing as how the demand is there.
I have not made any such claims anywhere. But maybe you can try to come up with a theory of how e.g. host interrupts can increase jitter in asynchronous UAC.
I agree...not everything is measurement...and lots of things have to do with how our ear perceives what happens in our individual listening room...
On the tube front we all kind of discussion as well...and in the wnd its not about being right with a hypothesis...it is about sharing observations and experience with fellow colleagues which they than can decide to ignore or try themselfs...i will now try SLC memory cards than to the input here...great crazy stuff...love it
On the tube front we all kind of discussion as well...and in the wnd its not about being right with a hypothesis...it is about sharing observations and experience with fellow colleagues which they than can decide to ignore or try themselfs...i will now try SLC memory cards than to the input here...great crazy stuff...love it
There may be some things going on that we haven't thought much about yet. Please let me suggest a couple:
1. R-core transformers have very good isolation between primary and secondary, but secondary windings are all wound on the same bobbin, so there is some capacitive coupling between secondaries. That's one possible path for common mode noise to couple between otherwise isolated power rails. For that reason it might make sense to use more than one R-core transformer and to think about keeping noisy sources on one transformer and clean circuitry on another.
2. On a FIFO board there are isolator chips and isolated ground planes. However, there is always some stray capacitive coupling between nominally isolated sections. There may be some magnetic coupling as well. The two types of stray coupling are always there. Where there is one there is the other, although the effects of one may dominate over the other in a given situation.
Up to Andrea to figure out the FIFO board part.
DAC builders will have to figure out how much isolation they are really getting between the parts of the circuitry they are responsible for.
Another thing to consider is that whatever comes after a dac, say, a power amp or maybe a preamp, will be coupled to the dac clean-side ground. Such devices can introduce ground noise into the dac ground too. If that stuff is connected to the same AC power source as the computer or other noisy parts of the circuitry then an opportunity where noise ingress may also occur. Moreover, AC lines act line antennas at RF frequencies. They can couple noise in and out of all sorts of devices. Clean power in today's EMI/RFI environment should not be taken as trivial to accomplish.
One thing I have found here is that getting a used 'Monster HTPS 7000 Mk II' from ebay or some other place helps a lot. So can other types of high quality power filtering that include well-designed common mode chokes. Many other types of power conditioners, including those with good brand names, do not perform nearly as well as the one I just mentioned. An old-style PS Audio Regenerator can also be very helpful, perhaps even better than the Monster conditioner.
Regarding radiated noise coupling, steel is your friend more than is aluminum. However its best to leave a air gap (maybe at least 2") between a steel shielding chassis and sensitive circuitry. We don't want stray magnetic coupling between the steel and the circuitry being shielded.
1. R-core transformers have very good isolation between primary and secondary, but secondary windings are all wound on the same bobbin, so there is some capacitive coupling between secondaries. That's one possible path for common mode noise to couple between otherwise isolated power rails. For that reason it might make sense to use more than one R-core transformer and to think about keeping noisy sources on one transformer and clean circuitry on another.
2. On a FIFO board there are isolator chips and isolated ground planes. However, there is always some stray capacitive coupling between nominally isolated sections. There may be some magnetic coupling as well. The two types of stray coupling are always there. Where there is one there is the other, although the effects of one may dominate over the other in a given situation.
Up to Andrea to figure out the FIFO board part.
DAC builders will have to figure out how much isolation they are really getting between the parts of the circuitry they are responsible for.
Another thing to consider is that whatever comes after a dac, say, a power amp or maybe a preamp, will be coupled to the dac clean-side ground. Such devices can introduce ground noise into the dac ground too. If that stuff is connected to the same AC power source as the computer or other noisy parts of the circuitry then an opportunity where noise ingress may also occur. Moreover, AC lines act line antennas at RF frequencies. They can couple noise in and out of all sorts of devices. Clean power in today's EMI/RFI environment should not be taken as trivial to accomplish.
One thing I have found here is that getting a used 'Monster HTPS 7000 Mk II' from ebay or some other place helps a lot. So can other types of high quality power filtering that include well-designed common mode chokes. Many other types of power conditioners, including those with good brand names, do not perform nearly as well as the one I just mentioned. An old-style PS Audio Regenerator can also be very helpful, perhaps even better than the Monster conditioner.
Regarding radiated noise coupling, steel is your friend more than is aluminum. However its best to leave a air gap (maybe at least 2") between a steel shielding chassis and sensitive circuitry. We don't want stray magnetic coupling between the steel and the circuitry being shielded.
Last edited:
Sorry, I may have understood you comment. I have strong faith in that everyone has a different sense of hearing and brains. So subjective listening results can hardly be considered as universal truths. Measurements can be used for many useful purposes but not explaining everything that is heard by everyone. One such useful purpose is troubleshooting e.g. grounding or noise issues.
...violent agreement...
That is why I always tried just to take a piece under observation and try to break it in smaller units to observe and reduce complexity...e.g. If the theory is PSU can be better, lets radically change PSU (like to battery e.g.) etc. But step by step, one change at a time and enough time to settle in new components before comin to conclusions.
I rather trust my ears without knowing always exactly why things are as they are instead trusting measurements and theories driving me to conclusions. Neverheless they may drive me to another experiment to listen to, so they are very valuable.
The funny thing is that many times stuff which could not be expained initially is later (like years later) oftenly even discovered and explained by measurements...by much better dudes than myself...nevertheless I knew what sounds good immediately...its like drinking wine...I dont need to know where it grew, what grape it was or which barrel type...I just need to know if I enjoy it (neverthess most of the time I know precise where and how its been made, but that is not important).
And Andrea's fifo and DAC are a real purist piece of (development) equipment, very important for the deveolpment of many other areas of my system now...so much easier to improve the speaker cross over now as an example...or the driver stage of my 801A/gm70 amp...as they all got much cleaner signal now.