Hello,
in my hunt of filtering USB i wanna try this:
a usb C to C cable with something like this on each end:
im currently halfway trough designing it, currently it looks like this:
- Data
-- Common mode choke on usb lines (i just found out ferrite beads are forbidden by usb specifications, so probably not much more i can do here?)
- VBUS
-- Pi filter with ferrite bead and 1uF on either side, only 1uF because usb specifications dont allow more than 10uF
- GND
-- i kinda wanna try the same ferrite bead here as in the vbus line, while its usually not recommended there are some scenarios where it could be beneficial
- Shield
--
there are some schemes, cap, ferrite bead, connected trough, disconnected....
i think ferrite bead like in the schematic doesnt make "much sense", a cap seems more reasonable (lets high frequencys trough but elimates ground loops/dc current)
tho, im just sure right now that i wanna make disconnected/connected switchable via solderpad
also there needs to be a shield soldering pad (probably on the backside)
Goal is really to throw as much filtering at USB to isolate the dirty PC, as long the eyepattern holds up there shouldnt be a problem with the data delivery, tho some could argue that ferrite bead in the power line could have some negative effects on bus powered dacs/interfaces, i wanna try this and the ferrite bead is easly bridgeable via solderpads on the backside
instead of spending hundreds of dollars on fancy usb cables that have different RFI/EMI pickup and therefore sound different i find this option much more reasonable (tho dont get me wrong, the cable will probably still matter a bit, im currently fairly happy with his one: https://www.audiophonics.fr/en/digi...pper-ptfe-double-shielding-o78mm-p-11077.html )
Do you guys have any ideas/opinions on this? (tho, please keep your "inaudible" claims to yourself)
in my hunt of filtering USB i wanna try this:
a usb C to C cable with something like this on each end:
im currently halfway trough designing it, currently it looks like this:
- Data
-- Common mode choke on usb lines (i just found out ferrite beads are forbidden by usb specifications, so probably not much more i can do here?)
- VBUS
-- Pi filter with ferrite bead and 1uF on either side, only 1uF because usb specifications dont allow more than 10uF
- GND
-- i kinda wanna try the same ferrite bead here as in the vbus line, while its usually not recommended there are some scenarios where it could be beneficial
- Shield
--
there are some schemes, cap, ferrite bead, connected trough, disconnected....
i think ferrite bead like in the schematic doesnt make "much sense", a cap seems more reasonable (lets high frequencys trough but elimates ground loops/dc current)
tho, im just sure right now that i wanna make disconnected/connected switchable via solderpad
also there needs to be a shield soldering pad (probably on the backside)
Goal is really to throw as much filtering at USB to isolate the dirty PC, as long the eyepattern holds up there shouldnt be a problem with the data delivery, tho some could argue that ferrite bead in the power line could have some negative effects on bus powered dacs/interfaces, i wanna try this and the ferrite bead is easly bridgeable via solderpads on the backside
instead of spending hundreds of dollars on fancy usb cables that have different RFI/EMI pickup and therefore sound different i find this option much more reasonable (tho dont get me wrong, the cable will probably still matter a bit, im currently fairly happy with his one: https://www.audiophonics.fr/en/digi...pper-ptfe-double-shielding-o78mm-p-11077.html )
Do you guys have any ideas/opinions on this? (tho, please keep your "inaudible" claims to yourself)
IIRC, @altor has made such a device before. Maybe one to look at to get an idea of what's involved: https://www.altor.co/products/test-equipment/usb-isolator Maybe Altor would be willing to offer some advice on how to best proceed.
well isolating is one thing, im just not concentrating on this since there are a few commercial isolaters
the point of this really is to have "additional" filtering at each cable end, to get rid of anything that gets introduced after isolating ( or by the isolator itself)
the point of this really is to have "additional" filtering at each cable end, to get rid of anything that gets introduced after isolating ( or by the isolator itself)
Semi-Final version 🙂 unless there are some ideas i will probably go with this
- The two solderpads are really for a shielding scheme like this:
shield connected at start connector - and on endconnector either not connected or trough the cap/resistor pair
- the ferrite bead on vbus can be bridged/removed
- i removed the ferrite bead on ground
If there might be a use case where you'd have an isolator plus this... I don't know, I don't think this is needed at all.
it will work just fine without, the question is if the additional filtering still makes an audible difference, my suspicion is yes but i cant confirm till i tried of course
the BOM is currently at 2 euro per piece, 1,2euro for the "quality usb connector" by molex, probably the connector alone is worth it over chinese usb c pcb for diy soldering
will see how much it cost, i hope not more than 5 euro for finished PCB
- added 5,1k resistor to CC line for 500mA negotiation
- changed the ferrite bead to a less effective but smaller ferrite bead
- improved usb signal routing + impedance calculated trace space/thickness to 90 ohm (also moved any vias out of the way as best as possible)
- better solderpad spacing
theoretically its finished now unless someone wants to chime in 🙂
Couple of comments:
- According to USB specification shield is connected to ground at both ends of USB-C cable. So no need for RC.
- You should add CC pad as that would make the cable compatible with USB PD.
you can do this if you want to (there is a solderpad for direct connection) but be aware that the shield is actually another ground connecting, if connected on both sides it acts as half your return line, this is what the RC fixes, breaking groundloops but still having a low impedance for higher frequencys as i understand it
yea, either its a CC pad with seperate CC cable to the other side (so potential PD controllers can negogiate the right current) or a resistor on the start plug for a fixed 500mA (havent found any references to fixed 1,5A on the web, probably because it will potentially overload the usb port if its not capable of delivering 1,5A)
thing is, for reference i wanna keep using this cable: https://www.audiophonics.fr/en/digi...pper-ptfe-double-shielding-o78mm-p-11077.html or potentially a supra excalibur cable later on..... both without CC line and its probably not a good idea to route a seperate cc line outside the shield
i really just picked usb C because its a much better connector (i cant be the only one where the usb B connector stops working or you have to wiggle the cable after a few months of use, usb A is much more solid tho) and i guess more future proof, also there are alot of adapters
USB-A->USB-C
USB-C->USB-A
USB-C->USB-B
etc...
i dont think there is an easy fix, you have to treat any incoming cable into your hifi chain, atleast this gives me the greatest benefit audibly so far
i mainly made this because i never saw it done and im just curious what improvement it would bring
interesting, how are you going to make the cable? USB solder points of this type of design normally is under a lot of stress.
Its not about wanting. What the spec means is that all USB-C cables have the shield connected to ground and quite likely most USB-C receptables are wired like this as well. So having RC between shield and ground is pointless if you are using USB C.
i cant confirm yet, but i was hoping to use some aluminium shells that come with a similar board/connector
- https://www.aliexpress.us/item/1005006140757322.html
- https://www.aliexpress.us/item/1005007259866919.html (im pretty sure this one will fit as its glue injected to hold the pcb at the front)
there are some more including the viborg ones which are not too expensive: https://www.audiophonics.fr/en/usb-...-c-31-connector-gold-plated-o8mm-p-15920.html
well in the end if you wanna sell a commercial product then yes, keep to spec, but i know by testing what audible differences it can make to play around with different shielding schemes, also even not every commercial enddevice terminates shield correctly which is another thing....
usb cards (if done right) are usually terminated like this: usb shield to PC shield and RC between pc shield and usb ground, its ideally from what i gathered the same for the enddevice, but audibly and logically i dont think its a good idea to introduce the PC shield onto your audio device, tho this might be completely different when using a isolator for example, maybe shielded connected on both sides sound better then, idk yet, thats why i added solder pads for a more universal board
even some altium lectures suggest shielding only on one side in low frequency application CAN be beneficial (but i havent really found further references, maybe they were speaking about ground loop regions) but generally speaking they recommend both sides
also im using for a year now an diy ifi idefender (you can find the thread here (diy ground loop eliminator/supercap usb filter), it also had shield unconnected with no usb problems beside better SQ
things improve imo massivly if you tweak/isolate shield and ground(and vbus) with a new power supply
No, that is not the right way of doing it if the USB C connector is used.
USB-C uses double-ended grounding while what you are suggesting is single-ended grounding.
Read here more about the pros and cons of each:
https://www.fs.com/blog/grounding-cat6-shielded-cables-ensuring-safety-and-stability-13796.html
its not single ended grounding because the pc shield/chassi itself is grounded, atleast i saw this on some usb pcie cards, if they were to spec or not im not sure but thanks for the link it made some things more clear
People are arguing like no end https://electronics.stackexchange.com/questions/4515/how-to-connect-usb-connector-shield
imo there definitely isnt a definitive answer, both have pro and cons either i go by sound quality in the end or idk (well or just stay on spec, but in practice this really doesnt seem to matter that much, maybe its important for some devices or they would have a floating shield (or just one that is connected trough RC) but for me personal devices i rather go with what sounds best
There is no need to argue if the connector is USB-C. All USB-C cables have shield and ground connected so it does not matter one iota how the board connector is wired.
@bohrok2610 : IIUC @Ghoostknight wants to build a DIY cable with the filters at the USB-C receptacles. While all commercial USB-C cables have shield connected with the GND pin inside the receptacles, his one would not 🙂 At least that's how I understand it...
Yes, of course I understand that. But my point is that USB receptables are quite likely wired in the same fashion as USB-C cables. Why would any device manufacturer use RC with USB-C receptable since it is just a waste of PCB real estate and components.
Oh... i think have misunderstood bohrok, i thought he was talking about grounding it behind the female connector (on the device pcb)
but apparently spec says this:
i kinda have to think about this one as it seems kind of shitty to put PC/USB shield onto usb GND via this connection, tho i imagine it offers the lowest ground resistance which might be important for high power usb c cables
i actually saw the RC between usb shield and usb gnd on an older USB 3.0 pcie card with usb A ports
but the newer spec even suggests for USB A to USB C cables to tie shield to ground on both ends of the cable
but apparently spec says this:
i kinda have to think about this one as it seems kind of shitty to put PC/USB shield onto usb GND via this connection, tho i imagine it offers the lowest ground resistance which might be important for high power usb c cables
i actually saw the RC between usb shield and usb gnd on an older USB 3.0 pcie card with usb A ports
but the newer spec even suggests for USB A to USB C cables to tie shield to ground on both ends of the cable
Well lets just do it like this, since there is room for it...
also rearranged things abit
"pass" and "gnd" solderbridges made on both ends would be "on spec"