Hello
a few months ago the previous designer/engineer of the Ifi Idefender told me that this device applys arround 100 ohms between the usb input and usb output port, and i wanted to DIY something without complicated circuitry
actually now that i think about it im not sure if a dac would work with just the 100 ohms ground, its probably better to always use a external power supply and connect the pc ground trough the resistor (if you completely disconnect pc ground your dac usually doesnt get detected, so some gnd connection is nessecary)
i currently building a small pcb/usb dongle for this purpose
the current layout looks like this:
Has someone some ideas or sees errors with this?
Current Features:
a few months ago the previous designer/engineer of the Ifi Idefender told me that this device applys arround 100 ohms between the usb input and usb output port, and i wanted to DIY something without complicated circuitry
actually now that i think about it im not sure if a dac would work with just the 100 ohms ground, its probably better to always use a external power supply and connect the pc ground trough the resistor (if you completely disconnect pc ground your dac usually doesnt get detected, so some gnd connection is nessecary)
i currently building a small pcb/usb dongle for this purpose
the current layout looks like this:
Has someone some ideas or sees errors with this?
Current Features:
- EMI/EDV filter IC on each usb port
- variable resistance to the pc gnd (not connected, 100 ohm, 1k, 500, 333, 250, 200 (and some theoretical combinations under 100 ohm) (it might be a good idea to make a passtrough switch so A/B testing is possible)
- 4x panasonic fc 470 uF capacitor bank to improve power, which can be "disabled" for A/B purposes
- external power supply barrel connector
- usb power from the pc can be disabled or enabled, depending if you use a external power supply or wanna make sure that the power is not connected
purpose of the device is mainly that i wanna test some things related to USB in combination with a usb isolator (for example if the ground STILL matters after a isolator or if the emi ic"s do something aditionally)
Tho i see it kind of as a multi purpose device, you dont need aditional dongles to disable V+ or route external power in
Tho i see it kind of as a multi purpose device, you dont need aditional dongles to disable V+ or route external power in
probably a good idea to add some kind of reverse voltage protection for the DC jack or the caps might blow up
is there maybe a good IC out there that can do over and reverse voltage protection?
... i also might exchange the dip switch for some kind of pcb mountable 2 way switch, so the two input voltages cant hit eachother...
is there maybe a good IC out there that can do over and reverse voltage protection?
... i also might exchange the dip switch for some kind of pcb mountable 2 way switch, so the two input voltages cant hit eachother...
i picked the parts on lcsc.com, so parts might be not available on digikey or mouser
i will probably order 5 or so at first, if you (or more) are interested i can reserve one, parts cost is i think around 10-15 euro, i did not choose the cheapest connectors for example and i have to buy supercapacitors seperately and solder them myself
but i will also release all nessecary files once i tested the pcb's
its not the final version in the pics yet, some minor things i wanna improve, already did length match the data lines to +/- 1mm after the pics
let me know some improvement ideas if you have some
its after all a kind of "research" projectpretty happy with this:
one major fix was the ground plance clearance reduced to 0.127mm ... this is part of the impedance matching of the data-lines combined with length matching this should be alright... (tho the usb c port is slightly less perfect (length matching is like 1-2mm off) beside longer than the usb b input..).
one major fix was the ground plance clearance reduced to 0.127mm ... this is part of the impedance matching of the data-lines combined with length matching this should be alright... (tho the usb c port is slightly less perfect (length matching is like 1-2mm off) beside longer than the usb b input..).
some keynotes:
1. there would be probably benefits in having more supercaps, but its getting just too large, i wanna keep the dongle as small as possible, so the data lines are as short as possible
there are some 5V 10-30F double super caps in one package, in theory these can be used to upgrade the supercap performance, but its getting larger
i still have some here, i will test them against a 1F 5,5V one that would fit the silkscreen
2. i changed the layout later on so majority of data line length runs trough clean ground, not sure how much such stuff matters, its basicly my first pcb, i actually will order
3. i mainly wanna try this dongle AFTER a isolator (topping hs01 i have here) to see if it does anything
the engineer of the ifi idefender said that 100 ohm between grounds give roughly 1000x attenuation of ground loops, so in theory it should elimate the need of a isolator (specially with higher resistance values), since objectivly isolators can do just that, break ground loops
also curious if it makes a difference if i feed the power trough the isolator, or after the isolator trough the ground filter dongle... i expect that trough ground filter will be better because of the supercap and no noisy IC...
4. since i implemented some filter ic"s on the datalines it will be probably hard to tell wether the audible change comes from them or the ground breaking, thats why i added a passtrough switch, so the ground property can be tested on its own
5. i placed both input connectors fairly close to eachother, it would be perfect if that blocks the ability to use two input cables.... just use one
6. i first tried to make a additional usb-c for data and power... but that would have complicated the grounding scheme, thats why i used 2 usb c connectors, one for power and one for data
1. there would be probably benefits in having more supercaps, but its getting just too large, i wanna keep the dongle as small as possible, so the data lines are as short as possible
there are some 5V 10-30F double super caps in one package, in theory these can be used to upgrade the supercap performance, but its getting larger
i still have some here, i will test them against a 1F 5,5V one that would fit the silkscreen
2. i changed the layout later on so majority of data line length runs trough clean ground, not sure how much such stuff matters, its basicly my first pcb, i actually will order
3. i mainly wanna try this dongle AFTER a isolator (topping hs01 i have here) to see if it does anything
the engineer of the ifi idefender said that 100 ohm between grounds give roughly 1000x attenuation of ground loops, so in theory it should elimate the need of a isolator (specially with higher resistance values), since objectivly isolators can do just that, break ground loops
also curious if it makes a difference if i feed the power trough the isolator, or after the isolator trough the ground filter dongle... i expect that trough ground filter will be better because of the supercap and no noisy IC...
4. since i implemented some filter ic"s on the datalines it will be probably hard to tell wether the audible change comes from them or the ground breaking, thats why i added a passtrough switch, so the ground property can be tested on its own
5. i placed both input connectors fairly close to eachother, it would be perfect if that blocks the ability to use two input cables.... just use one
6. i first tried to make a additional usb-c for data and power... but that would have complicated the grounding scheme, thats why i used 2 usb c connectors, one for power and one for data
hey guys, small update, i wanna get this done
i removed vias from the datalines... just one remains on the usb c data lines, i compensated the via too, since i was finally able to get the track tuning tool working... length match on all lines including vias is under 0,3-1mm
tho i just started placing some additional vias on the usb ground.... do we wanna do something to isolate the usb output shield from the clean ground? i could place 2 pads for a resistor to isolate the shield with resistor or directly connect it
im not sure if dacs handle the usb shield and gnd line differently, does someone know more? if the dac handle them differently in implementation it might be a good idea to place a resistor between shield and clean ground...
same question stands for the usb c external power input, we could isolated the shield from the usb cable...
i might just place 2 100 ohm resistors for the two shields on the clean side...
also i will probably place a GND pad, so this thing can be wrapped in some kind of shielding connected to the GND pad
i removed vias from the datalines... just one remains on the usb c data lines, i compensated the via too, since i was finally able to get the track tuning tool working... length match on all lines including vias is under 0,3-1mm
tho i just started placing some additional vias on the usb ground.... do we wanna do something to isolate the usb output shield from the clean ground? i could place 2 pads for a resistor to isolate the shield with resistor or directly connect it
im not sure if dacs handle the usb shield and gnd line differently, does someone know more? if the dac handle them differently in implementation it might be a good idea to place a resistor between shield and clean ground...
same question stands for the usb c external power input, we could isolated the shield from the usb cable...
i might just place 2 100 ohm resistors for the two shields on the clean side...
also i will probably place a GND pad, so this thing can be wrapped in some kind of shielding connected to the GND pad
Last edited:
OK, i think i will do it this way:
1 GND pad
and
each input connector shield (so each device side) including the usb c ext power connector is connected trough a 33k resistor to GND
the output usb shield ("host" in our case) is connected directly to GND
this seems somewhat common reading on google, even if people say ground should be connected directly, i think its a different beast for audio so i go with the resistors...
1 GND pad
and
each input connector shield (so each device side) including the usb c ext power connector is connected trough a 33k resistor to GND
the output usb shield ("host" in our case) is connected directly to GND
this seems somewhat common reading on google, even if people say ground should be connected directly, i think its a different beast for audio so i go with the resistors...
just a small update, JLCPCB wrote me about some discoleration of the solder mask but function is not altered, they will ship soon
i also already have a eaton 1,5F 5,5V supercap here and some other bigger 15F or 25F ones i would need to place on the backside of the PCB, i will test both tho i currently have not a dac that is purely usb powered so i have to see how big of a difference it might make, some dacs still use usb power to power the dirty side of the usb chip
also i got 33k 1206 resistors for the backside to tie the shield over the resistor to ground, i will test with the resistor in place and without it
its quite costly to populate both sides by JLCPCB, so i just selected top side since there is nothing beside the shield resistors on the back
excited to test the PCB, stay tuned, i will report back what it does to sound quality
here is a pic of the coloration, not sure why, maybe because of soldering temperatur, i choose purple PCB's
EDIT: just one board seems effected out of 5, so not a big deal
i also already have a eaton 1,5F 5,5V supercap here and some other bigger 15F or 25F ones i would need to place on the backside of the PCB, i will test both
tho i currently have not a dac that is purely usb powered so i have to see how big of a difference it might make, some dacs still use usb power to power the dirty side of the usb chipalso i got 33k 1206 resistors for the backside to tie the shield over the resistor to ground, i will test with the resistor in place and without it
its quite costly to populate both sides by JLCPCB, so i just selected top side since there is nothing beside the shield resistors on the back
excited to test the PCB, stay tuned, i will report back what it does to sound quality
here is a pic of the coloration, not sure why, maybe because of soldering temperatur, i choose purple PCB's
EDIT: just one board seems effected out of 5, so not a big deal
boards should tomorrow arrive, i also bought 5 of these: https://de.aliexpress.com/item/1005006172691597.html 1,5F 5V C type ones from CDA
Just a question about the data lines, it looks like they are routing over a split in the ground plane. This is worth a read https://resources.altium.com/p/splitting-planes-good-bad-and-ugly if that is the case.
im not sure if this is really applicable here, the two ground planes are connected over a resistance to even out potential between planes but essentially the "reference"-ground is for both section a different one (actually the same but from two different points, external power supply(or dac) and PC, the resistance avoids ground loops because of that
tho im not exactly sure what happens to return currents because of the split... maybe i should have moved the split to one end but the problem still is the same i think
Device was received today, it works! yay
1k ohm seems to be to high for my setup but 500 ohms seems to work fine, so it was a good thing to make different resistors switchable to get the highest "profit" from the ground loop breaking property, it may actually make isolators obsolete but i will test a bit around...
also i should notice... i have this chain under test :
2m usb extension -> 30cm usb a to usb b -> topping hs01 -> ifi isilencer -> 30 cm usb a to usb b -> my usb ground filter -> 30 cm usb a to usb b
its actually quiet long and badly routed data cables should get noticed here i think, tho i think the topping hs01 also acts as a repeater, but everything works as usual, i just noticed my aune x8 has some noise issue where sometimes if the dac gets connected it plays white noise for a few seconds.... this white noise actually changed with my ground filter which is interesting to see, tho it isnt much worse or better, just different and it was quite consistent before
tho i could test some 2m extensions i have, to see how far the dac gets detected...
connectors also feel great, apparently i never saw higher quality connectors before but its somehow pretty obvious.. also how they feel if you plug something in
currently im testing a board without supercap and without resistors on the backside, i will solder the resistors later to see what the difference sounds like..
the first 15 minutes or so i would say the sound sounds "smoothed out" like usual if you start filtering EMI/RFI and the difference was definitely bigger than with the ifi isilencer(on top of the ifi isilencer, which is quite remarkable)
1k ohm seems to be to high for my setup but 500 ohms seems to work fine, so it was a good thing to make different resistors switchable to get the highest "profit" from the ground loop breaking property, it may actually make isolators obsolete but i will test a bit around...
also i should notice... i have this chain under test :
2m usb extension -> 30cm usb a to usb b -> topping hs01 -> ifi isilencer -> 30 cm usb a to usb b -> my usb ground filter -> 30 cm usb a to usb b
its actually quiet long and badly routed data cables should get noticed here i think, tho i think the topping hs01 also acts as a repeater, but everything works as usual, i just noticed my aune x8 has some noise issue where sometimes if the dac gets connected it plays white noise for a few seconds.... this white noise actually changed with my ground filter which is interesting to see, tho it isnt much worse or better, just different and it was quite consistent before
tho i could test some 2m extensions i have, to see how far the dac gets detected...
connectors also feel great, apparently i never saw higher quality connectors before but its somehow pretty obvious.. also how they feel if you plug something in
currently im testing a board without supercap and without resistors on the backside, i will solder the resistors later to see what the difference sounds like..
the first 15 minutes or so i would say the sound sounds "smoothed out" like usual if you start filtering EMI/RFI and the difference was definitely bigger than with the ifi isilencer(on top of the ifi isilencer, which is quite remarkable)
Last edited: