USB data only cable

[Tubetec]

Usb audio interfaces generally come in two varities , USB bus powered or with their own external power unit .
Ive modified my SSL2 and provide a battery supplied 5v rail via a 4pin xlr socket I installed on the back .

So I no longer need USB bus power from the pc , I took a standard short usb-c to usb-c cable ,cut the red power wire towards one end of the cable , patched it back up and it works ,
But I was thinking , now my data wires still have unwanted incidental capacitance to both the external supplied power and the USB bus ,
I havent tested the idea yet ,but I was thinking , cutting the red wire at both ends of the cable might improve matters a bit .

Various and sundry snake oil salesmen all over the internet will sell you a dedicated audio data only USB cable for anything upto 300 dollars. FFS

[AnalogJoe]

If the cable was designed by someone competent, he probably took into account the effect of the adjacent wires on the characteristic impedance of the data lines. Your really want to keep the rated characteristic impedance as close to its nominal value as possible (I am unaware of the value for the USB nominal characteristic impedance in the data lines) in order to avoid signal end reflections, which would translate into transmission errors. Grounding or removing adjacent conductors could affect the characteristic impedance of the data lines, how much or how relevant this is is something left to be seen.

Perhaps, rather than simply cutting or removing the unused wire, it would be a better idea to ground one end and add a resistor to ground at the other end to emulate normal operation.

My 2 cents.

[Tubetec]

The usual arrangement in a usb2 cable is the 4 wires are twisted , Data+ ,5V+ ,Data- ,5v-, cable screen is connected to the metal on the plug at both ends of the cable .
Connecting the isolated 5v+ wire to 0v at the computer end sounds like a good plan , at least then we balance input capacity from each data leg to ground , we also eliminate any mutual capacity between the two 5v power sources and ground.
From what I can gather USB C plugs contain a 10k resistor already , between +5v and 0v , but I need to dismantle one to verify this .

The USBC uses seperate screened pairs for the extra data lines , but also remains pin compatible with USB2 via the usual green/white wires ,

Maybe an improvement in CMRR over the standard USB arrangement is possible with a kind of starquad arrangement / data only connection ,

[Tubetec]

I did surgery on a USBc to USBa cable , found there was no dedicated black 0V wire inside , instead an extra heavy screen carried the 0V .
I seperated the red wire at both ends , heatshrinked it .
made no real difference .

What I did discover though was ,when I powered both the audio and USB from a common battery it injected quite a lot of noise down the supply wires , a cheap dynamic mic worked quite effectively as a search coil ,
With the battery just powering the audio supplies the wires didnt radiate anymore.
It seems the noise is coming from the area where 5v is turned into +/- 15V , its around 1.2khz fundamental with a splatter of harmonics extending up much further. The noise source seems to be coupled via stray capacitance as when I switch from mic to line to highZ its more prone to picking up .
I might try copper foil screening the area just to see if that helps ,

[AnalogJoe]

Boost converters are notorious for creating all sorts of artifacts due to fast switching and transition times.

In general, what matters most are the transition times (rise and fall of the square wave) more than the frequency of oscillation itself. You could have a circuit switching at 100 MHz with slow rise/fall times that radiates less than a 10 kHz circuit with ultra fast rise/fall times. Hence, interference and radiation can be reduced by slowing down the transition times of a switching circuit, but that tends to increase power dissipation and reduces power efficiency.

Do try using copper foil. In many cases, good-old simple shielding can do wonders; a piece of aluminum or copper foil in the right places can be very effective.

[Tubetec]

Heres a photo of the area the noise emanates from ,
This is the underside of the board ,
Even through the aluminium top panel this noise is detectable with a search coil ,
with two distinct noise peaks in the vicinty of the inductors .

More detailed photos from the SSL2 here ,
https://www.ifixit.com/Teardown/Solid+S ... own/135354

[mediatechnology]

Enjoyed reading the teardown.

In the comments there is a link to a sketch of the "4K" circuit. The author writes "something like this:" https://www.dropbox.com/scl/fi/peegqr9h ... g&e=1&dl=0

[Tubetec]

The ifixit site is quite useful for getting a look under the hood before you buy .
The guy went to quite a bit of trouble and correctly identifies all major parts of the circuit with quality photos that allow you zoom in .
I hadnt spotted the hidden reference to the 4K circuit .

[AnalogJoe]

Enjoyed reading the teardown.

In the comments there is a link to a sketch of the "4K" circuit. The author writes "something like this:" https://www.dropbox.com/scl/fi/peegqr9h ... g&e=1&dl=0

So is the distortion some sort of combination between crossover distortion and clipping? Not a fan.

[Tubetec]

I posted an RTA graph in the SSL2 topic which shows what the 4K button does to a single sine wave ,
The unit itself has increasing lower order harmonics (mainly 2nd.3rd) as you go from around -20dbfs to 0dbfs ,
Engaging the 4k mode caused bristling harmonics upto 9th or maybe more to appear at all levels ,
as well as a few db's boost above 1khz .
Id happily trade the 4K mode for a mute switch per channel , I cant envisage any reason Id want to 'crowbar' my signal with zener diodes either