I have FIFOII with isolator to DUALXO board. FIFOII get signal throuh IsolatorPI v1 from either RPI4 powered by 5v PaulHynes ps or RPI2 with local regulators removed and powered separate 1.8, 3.3 and 5v from Mezzanine board (with LT3045 regulators) powered by ~6v same PaulHynes ps.
RPI2 has clear advantage in sounding more analogue, though RPI4 should be technically better..
And i should not hear that difference at all but still i hear it as noise still passes through to the dac..
RPI2 has clear advantage in sounding more analogue, though RPI4 should be technically better..
And i should not hear that difference at all but still i hear it as noise still passes through to the dac..
D
Deleted member 537459
For me the difference is clear, but I use roon, so all file is via ethernet. Now the rpi3 is ethernet via USB and rpi4 is native ethernet. For me all things with usb is worse. File read from usb or nas from ethernet is huge better from nas, i2s usb is worse than rpi or bbb native i2s. Is my opinion. Obviously I try this thinks many time and the comparisons is make in the same supply.
@dddac
I didn't do comparison between standard RPi versions. But I do make a listening test to Raspberry Zero and Zero/W last week. The result was interesting.
RPi Zero was connected to Ethernet via a USB Ethernet adapter,
While RPi Zero/W connected to network via wifi.
Both of them connected to a FifoPi Q3 through GPIO.
For me, RaspberryPi Zero/W sounds obviously better. With FifoPi installed in the system, it doesn't change the whole sound style. The only differences is the clarity. RPi Zero/w with wifi connection plays music more clear then RPi Zero with USB Ethernet connection.
The only thing I'm concerned about is the RPi Zero/W speed. It's only 1GHz one core ARM processor. However, less than $10- 😀.
I suspect wired network brings more EMI noise directly into stack which has impacts to the sound quality. (There is no prefect isolator in the real world)
I heard that there is something called EtherREGN on this planet that can clean the wired network signals for audiophiles. Just ordered one. Will be received in a month ( as a back order, no stock currently). I'll post some updates once tested.
Regards,
Ian
Last edited:
@Ian, Whatever happened to the DigitalOutputAdapter board. Did I miss that one, or was it a prototype that never saw production?
I've experienced all sorts of subtle differences with the Pi front ends of my systems. Bits still remain bits of course but that's not the end of it. I really do think that it's often local EMI created by the system itself that is the issue. Digital audio systems are super complex interactions of every electromagnetic kind. It's a mistake to make think that box A plugs into box B plugs into box C and all the sum of those box functions are blameless. To quote @GregStewart again: Everything matters!
Hello,
Just when i thought things were getting a LITTLE clearer the fog comes rolling in.
Of course among the Raspberry users , the serious audiophile is just a niche market or not? So very probably things are being added because most people like them and if that means audio quality will suffer no problem because total sales will still go up.
Some people will tell that with fifopi if things '' in front '' of it work like they should work it will sound identical.
Some people will tell wifi sucks.
They will tell you to buy '' audio grade '' network gear, of course connected with expensive cables. And to make it work like it should the power supply will be as expensive as the gear it is feeding
It is kind of hard for the hobbyist to judge who is telling the truth and who is just trying to make more money. Doede ( just retired) doesnt need our money that hard. Ian and Andrea both seem to be seriously involved and need some cash to keep things rolling.
Most of the talk, that sounds like the latest marketing tricks seems to be coming from the customers. Maybe they wanna brag about their latest arrival and if they just spend a weeks wage on a switch they can't tell us it doesnt work.
Maybe they will tell the seller it is a minimal improvement . So he will be told to get the real benefit the next 5 days he will be working for the power supply.
Greetings, Eduard
Just when i thought things were getting a LITTLE clearer the fog comes rolling in.
Of course among the Raspberry users , the serious audiophile is just a niche market or not? So very probably things are being added because most people like them and if that means audio quality will suffer no problem because total sales will still go up.
Some people will tell that with fifopi if things '' in front '' of it work like they should work it will sound identical.
Some people will tell wifi sucks.
They will tell you to buy '' audio grade '' network gear, of course connected with expensive cables. And to make it work like it should the power supply will be as expensive as the gear it is feeding
It is kind of hard for the hobbyist to judge who is telling the truth and who is just trying to make more money. Doede ( just retired) doesnt need our money that hard. Ian and Andrea both seem to be seriously involved and need some cash to keep things rolling.
Most of the talk, that sounds like the latest marketing tricks seems to be coming from the customers. Maybe they wanna brag about their latest arrival and if they just spend a weeks wage on a switch they can't tell us it doesnt work.
Maybe they will tell the seller it is a minimal improvement . So he will be told to get the real benefit the next 5 days he will be working for the power supply.
Greetings, Eduard
Eduard
I am sure you mean no disrespect. Your assumption about the financial motivation of Ian and Andrea is a bit troubling. We have two really competent with serious day jobs working their spare time to share their stuff via group buys. Andrea admits he lost money running the 2019 GB but did not come back to us to top up the budget. Ian has been generous often giving me free stuff to go along with great value group buys. I think we just need to be thankful for the work of all three of these guys. I'm sure you feel the same, but I just wanted to say thanks to these guys for all of us.
I am sure you mean no disrespect. Your assumption about the financial motivation of Ian and Andrea is a bit troubling. We have two really competent with serious day jobs working their spare time to share their stuff via group buys. Andrea admits he lost money running the 2019 GB but did not come back to us to top up the budget. Ian has been generous often giving me free stuff to go along with great value group buys. I think we just need to be thankful for the work of all three of these guys. I'm sure you feel the same, but I just wanted to say thanks to these guys for all of us.
Network noise
I have no doubt that the switching noise from RJ45 is degrading our sound potential. One solution that looks interesting is to connect to the switch with optical interface. I suspect WiFi has its own EMI problems. RPi4 is supposed to be better because it puts in an ethernet chip vs sharing the USB bus.
I wonder if anyone has tried an optical ethernet connection?
I have no doubt that the switching noise from RJ45 is degrading our sound potential. One solution that looks interesting is to connect to the switch with optical interface. I suspect WiFi has its own EMI problems. RPi4 is supposed to be better because it puts in an ethernet chip vs sharing the USB bus.
I wonder if anyone has tried an optical ethernet connection?
D
Deleted member 537459
I have all in optical connections for galvanic isolation. All adapter feed with good linear supply bypassing original bad ldo, and the next step is feed with andrea 25mhz clock. in this situation with good rj45 cable and feed with good supply, is much better than WiFi.
Is my experience after 3 years work to clean all the network. I do all by my self and I don't spend all my money in that, so no worry 😁
Is my experience after 3 years work to clean all the network. I do all by my self and I don't spend all my money in that, so no worry 😁
Thanks to the flexibility of Ian's gear, I can go PC or RPi. I'm glad to have both options on the StationPiPro.
I am going to skip the RPi completely and go with a PC with a realtime kernel. Fortunately, the Combo384 works with the realtime kernel. I need to make sure the PC runs on battery tho. So a cheap NUC makes a great battery-powered alternative to an RPi. I'm not sure there will be a difference yet on the backend as I wasn't planning a frontend until I locked into a SOTA clock. NUCs are also legendary for their Audiophile SQ. I wanted to try something different with a 12V motherboard, but that's put on hold indefinitely since there is a supply shortage. So an old NUC should hold the fort for now.
To keep the PC far away from the StationPiPro, I'm using Monoprice SlimRun USB fiber optic for isolation. So low latency realtime kernel PC w/ industrial wide-temperature ram -> SlimRun optical -> Uptone USPCB -> Combo384. Later on I can scale the PC as needed if there are improvements.
I can also use a PC -> SlimRun optical -> USB Ethernet / USB WiFi / USB Storage later on if I need optical isolation from the PC. The PC will be quite barebones with just RAM and a USB boot disk as internal USB, internal Storage and internal Networking adds noise. So I offload all those services with optical isolation far away from PC. After the low latency realtime OS loads 100% in RAM, I can remove the USB boot disk.
Monoprice SlimRun USB-A to USB-A Female 3.0 Extension Cable - Fiber Optic, Black, 32.8ft - Monoprice.com
I am going to skip the RPi completely and go with a PC with a realtime kernel. Fortunately, the Combo384 works with the realtime kernel. I need to make sure the PC runs on battery tho. So a cheap NUC makes a great battery-powered alternative to an RPi. I'm not sure there will be a difference yet on the backend as I wasn't planning a frontend until I locked into a SOTA clock. NUCs are also legendary for their Audiophile SQ. I wanted to try something different with a 12V motherboard, but that's put on hold indefinitely since there is a supply shortage. So an old NUC should hold the fort for now.
To keep the PC far away from the StationPiPro, I'm using Monoprice SlimRun USB fiber optic for isolation. So low latency realtime kernel PC w/ industrial wide-temperature ram -> SlimRun optical -> Uptone USPCB -> Combo384. Later on I can scale the PC as needed if there are improvements.
I can also use a PC -> SlimRun optical -> USB Ethernet / USB WiFi / USB Storage later on if I need optical isolation from the PC. The PC will be quite barebones with just RAM and a USB boot disk as internal USB, internal Storage and internal Networking adds noise. So I offload all those services with optical isolation far away from PC. After the low latency realtime OS loads 100% in RAM, I can remove the USB boot disk.
Monoprice SlimRun USB-A to USB-A Female 3.0 Extension Cable - Fiber Optic, Black, 32.8ft - Monoprice.com
Last edited:
@ilgavro,
I'm interested in the optical connections for galvanic isolation and the 25 MHz clock option. Can you post a picture of your setup?
Thanks,
Ian
My 3 cent is that this "can happen" and it could depend on EMI coming from Raspberry. (It is very system dependent so each setup will react differently: position, sensitive components, distance, reflections, etc.)
Once I had an RPI3 on a wooden board. My system would play worse after placing RPI3 in an aluminum enclosure. I verified this many times to be sure, and could not come up with a "scientific" explanation. A possible explanation came much later from a radio amateur claiming (in a forum) that he was measuring higher RF emission from RPI when placed in an aluminum case.
(But now we need a new explanation as to why this was happening 😀 )
Last edited:
For my experience, RPi 3A+ in WiFi streaming sound best. I cannot obtain good sound from RPi4B no matter in wired LAN or WiFi, maybe the onboard buck power supply of the RPi4B is noisy even at under-clocked operation.
I am looking for RPi4 compute module 8GB and wish it could be better.
I am looking for RPi4 compute module 8GB and wish it could be better.
D
Deleted member 537459
If you use a pour cable rj45 and switch or modem feed original supply I believe you. But if you make a good network chain my experience said opposite. Regards
Maybe the case acts more as an antenna rather than as a shield. It might do some of both. Maybe try a solid steel case and see if you get a different result.
Hello,
THAT is the reason i will surely go for the Stationpi design. One might say that short connections are mandatory for perfect quality but if two boards are kind of disturbing each all the time it is a bit silly to add things like ultracap boards because those probably wont diminish those disturbances.
I remember when making my DDDAC chassis i did some tests about position and orientation of the Lundahl chokes. I remember Doede telling me that the main board doesnt care to much about chokes close by. AND of course lundahl chokes dont '' spread '' a lot of garbage because the way they are constructed.
But is seems that with '' front end '' parts like clocks it is easy to do something that will diminish sound quality. Not that it is wrong but it can be done better without investing more time and money. And a few of those somethings could add up to something bigger that you cannot compensate by investing an extra 1000 euro.
I think that if executed the right way most circuits here will work better if screened properly.
Of course most people here are constantly trying to improve their diorama that came with just a single mounting plate.
Greetings, Eduard
I have some experience with RF engineering as a CAD jockey laying out pcbs and it's really tricky to nail exactly what is going on. Firstly the nature of RF gets wierder the further you go up in frequency. Geometry becomes king. Every discontinuity in the circuit path becomes electrically reflective and emissive. Ground ceases to become ground and fields develop even on ground planes. If you use a scope the very connection of the scope probe will alter the all important geometry of the circuit. Then probe capacitance will drastically interact too. So then you have to get a fantastically expensive network analyser and a near field probe to gain any meaningful insight. Ian knows all this I'm sure and will employ good design practice to minimise the compromises but I think it's really worth people reading up on this as so many assume that tens and hundreds of megahertz behave the same way as audio band signals. They do not.