Seperate USB Data and Power

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

john12ax7

Well-known member
Joined
Oct 15, 2010
Messages
2,507
Location
California, US
I found @Tubetec experiments with the SSL-2 interesting, thread here
https://groupdiy.com/threads/ssl-2-audio-interface.84031/
Was thinking on a more general level is there something available that separates the data and power on a USB connection? The only thing I've found are powered USB hubs. Not sure how they are wired internally but could you use your own separate clean supply with them?

The overall goal would be to improve performance of USB powered audio devices without having to hack them internally.
 
A normal USB supply is limited in what size filter capacitor it can feed , typically 10uf is whats quoted as max
I did try adding a 100uf and the USB 5v line collapses completely unless the capacitor is charged up already .
The nicad supply I used has no problem charging up 1000's of uF very quickly , that and the other small mods gets the noise floor of the SSL2 down by almost 14db , which by anyones measure is a worthwhile improvement .

Theres lots and lots of chatter online about this and that USB isolator or power filter cleaning up things , very often particular interfaces wont work with them and they dont do what they say they'll do , but it can cost a few hundred to find this out for yourself .
No matter what you do your audio and USB sections of the interface share the same source so are subject to interactions ,saying that having a large capacitance in that position helps a lot .

Once I had the audio rails supplied from the nicad , I then tried standard USB power from the laptop , USB power from a battery bank , a shared connection from the nicad supply and finally another entirely seperate nicad supply for the USB power , there was really very little difference between any of these setups .

So how I see it is this any decent 5v supply that can handle a few 1000uf of smoothing and a drop of around 250mV across the inductors will help get better noise performance out of USB powered interfaces , theres no need to hack up the PCB , but you will need to wire your USB cable with seperate power and data , in other words cut the red line in the USB cable and connect what ever 5v supply your using there , just data +,- and ground to the PC .
 
Thats Thor.z outfit

The nicad 4.8v emergency light battery set up I use cost about 25 euros to put together , although I had the RC/hobby charger already .
 
Last edited:
Theres lots and lots of chatter online about this and that USB isolator or power filter cleaning up things , very often particular interfaces wont work with them and they dont do what they say they'll do
I have a number of analog synths here in my studio that are USB connected. The noise from the USB connection was unbearable and so I searched and found an isolator costing me a meager $50+. Noise disappeared. It runs a hub with five units connected. I liked it so much I bought another for my growing hardware synth collection.
 
Last edited:
I have a number of analog synths here in my studio that are USB connected. The noise from the USB connection was unbearable and so I searched and found an isolator costing me a meager $50+. Noise disappeared. It runs a hub with five units connected. I liked it so much I bought another for my growing hardware synth collection.

Info ? Link ?
Thanks
 
I found @Tubetec experiments with the SSL-2 interesting, thread here
https://groupdiy.com/threads/ssl-2-audio-interface.84031/
Was thinking on a more general level is there something available that separates the data and power on a USB connection? The only thing I've found are powered USB hubs. Not sure how they are wired internally but could you use your own separate clean supply with them?

The overall goal would be to improve performance of USB powered audio devices without having to hack them internally.

This thread elsewhere may help:

Easy solution to isolate power and ground on USB?

From my view there are a few issues with USB.

USB power quality affects things wherever USB power is directly used to feed circuitry, which is more often than people would think.

USB Ground/Earth loops, this is a HUGE problem if it materialises. If you use a laptop on batteries and power the AD/DA from the laptop, no issue. Use a 2-pin or 3-pin Power supply however on the same laptop and problems can make themselves felt.

Most affordable or DIY friendly USB Isolators are limited to USB1.1 speeds (aka USB 2.0 "full speed") which is not commonly used by USB Audio devices nowadays. USB 2.0 High Speed Isolators are expensive and not amenable to DIY.

Implementing "safe ground lifts" instead gives 90+% of the improvements of a full isolator.

One of the first iFi products was a USB Power Supply and Ground Loop Breaker (now 10 Years ago) that addressed this. I am attaching the schematic for that, it is very suited to DIY.

1689670435148.png
The power supply part is a bit over designed, a simple linear 5V supply as easily found on aliexpress for under 15 USD with an LM317 as regulator will probably do fine. SMPS are generally speaking not recommended, unless used with LC filtering and a linear post regulator.

USB Signal Integrity, cables, connectors etc. degrade signal integrity.

Any device using USB Audio Class uses isochronous streaming, meaning bandwidth is guaranteed, but there is no retransmission on error. This is the opposite of "bulk mode" which is what for example external hard drives use. The reaction to errors is derived from the way CD handles errors, first interpolation and then muting the stream.

USB Audio Class operates at USB1.1 Speeds and in USB 1.1 mode, so it uses 12MbpS and is very reliable and error free. But USB Audio Class (1) is limited to 2 Channels at 96kHz or 4 channels at 48kHz.

So commonly modern audio devices, domestic, semi-pro and pro use USB Audio Class 2. Now USB Audio Class 2 ALWAYS uses USB 2 High Speed (480MbpS) even when steaming just 2 channels at 48kHz. USB 2 High Speed is a lot more sensitive to signal integrity.

Cable quality matters quite a bit here, not "exotic" cables, but "spec compliant" cables. Having extra heavy gauge ground and good shielding is highly useful.

A solution is to add in a USB Hub or Repeater, specifically directly at or close to the USB Connector of the Audio device. This generally speaking is not easily to DIY as the IC's for this are at least QFN these days.

iFi has used the FE1.1 (USB2.0) in some products. That comes in an almost DIY'able case option (LQFP48 9mm X 9mm and I have some SOP cases too). Strip lines are needed for the USB Data lines, so PCB design is a little more involved, but it's doable.

Yet for DIY it is probably easier to to integrate a PCB stripped out of a cheap USB Hub and slightly modified to allow the ground loop breaker.

Power the USB Hub from the Source USB Device (computer) and implement feeding in separate power and breaking the ground line directly at the Sink USB connector. This just needs a few cut's and adding a few components plus a low noise external 5V supply adapter. That is what I would I would do for my personal use.

Thor
 
When I was experimenting around USB Isolation, I started from disassembling a common USB cable, ended up building my own USB isolator without chip. Merely preventing the DAC (it was the original iFi DSD Nano) from using the computer USB Bus and injecting my own DC Power made a massive difference.

My first test was a very unsuccessful experiment with directly filtering the Computer USB Power lines. Music became 'sluggish'. I kept that filter though and re-used it in the final connector to still filter the 'cut-off' computer USB power line. Still made a difference when it's there rather than not.

Also tested a few combos of shield configurations - also made a difference.

Thought job was done but still found additional differences by using a custom AC-side Filter in addition to the USB isolator.

All those post filters and regulators for SMPSes do nothing to reduce their pollution (by virtue of design) of the AC mains. This will affect your other components in the sound chain.

Another thing I dislike with USB: the proximity and geometric configuration of the power lines with the data lines. Here again, IFi had a solution with a cable separating the data lines and power lines which inspired all the USB cable disassembly and re-assembly experiments initially.

I injected power near the DAC, filtered the port at the source. Audirvana does the playing, my preferred sweet-spot with the DAC was DSD256 (free firmware upgrade, the ability isn't even on the box).

iFi has solid XMOS + Burr-Brown DAC knowledge.
 
Cheers Thor ,

I'd been wanting to ask in more detail about these IFI units and clean USB power ,


So even a three stage passive LC filter is appropriate ,
I think we all know well the variability USB power brings in ,
it makes the noise floor look like a porcupine , F-ing zipper noises when the mouse pointer redraws on the screen , buzzes and whistles spewing all over the place , its a total mess unless your supply can handle a few 1000uF of smoothing , computer motherboard USB power will cough and shit its pants at the same time if you do that , USB powerbank charge supplies are the same .

Interesting to note any form of switcher supply isnt reccomended for use with the IFI filter , instead its back to the old style transformer based linear approach ,

I'd like to hear any advice you have regarding battery based supplies ,
Four nicads plus similar passive filtering to what you show has worked wonders for me, typically getting me zero excess 50hz hum whatsoever even with the laptop running on mains power ,

The inevitable voltage sag from the battery as it discharges seems manageable and its only once the voltage across the battery terminals has dropped below 4,3v do the the internal circuits start to run out of regulation and the whole thing starts motorboating

For many years I used simple USB 1.1 to optical spdif and adat input devices to kill grounding problems in computer audio ,and as you say its actually very robust but limited in bandwidth ,
most importantly it means your audio circuits are galvanically isolated from any computer garbage.

My attempts at the passive filtering are big and clumsy compared to what you describe with the SMD , neat and tidy , with perfect mirror image symetry/geometry,
I wish I was brave enough to learn a bit of CAD and try pcb design , but Im a luddite to hardwire ,

I should have paid better attention in maths in school instead of skipping off for beer spliffs n giggles with the likely lads ,
I lived in continental europe for a while in my teens , me and a few buddies cobbled together a disco ligthing rig , it was bog ass basic , a lighting controller and a few handmade wooden boxes , painted mat black , typically the house parties we'd arrive at had Daddys hifi system as FOH ,
We'd be be able to play tunes if we wanted, grab beers , get introduced to all kinds of throughbred fillies and try and get the slow dance at the end of the night with the one you fancied best .
The dj'ing side of it didnt really occur to us ,and I wasnt much of a fan of acid house music in any case ,

I was always a big fan of Marshall amps ,
its the most iconic brand in Hard Rock music to me,
I had a contact to order spares directly from Marshall years back ,
I wish now I had thrown in a cv and went over for a look at the factory back in the glory days ,

Many of my pals tried their luck early on with music , most that had sense had a day job too ,
many of the others ended up taking to the drink , in the end having either to restrain themselves from it entirely , or ending up on a non stop binge , the body count has been attrocious to be honest , the number who make a decent living from it without being consumed by the lifestyle ,very few indeed .
Its not that the talent wasnt there either , it was it all to often turned into a monster after too much liquor and dope and ended up kissing piss on the toilet floor or having bumfights down piss alleys , or making a spectacle of itself and getting bared for life from venues , or arrested etc ,
 
Defianetly agree Yash ,
the simple cable hack and additional filtering is a great start to getting lower noise from any interface . It doesnt have to be destructive to the PCB in any way , the extra filter components can be housed nearby ,
There is a kind of USB Y splitter cable , it allows power input to the device on one connector and data through to host on another , I was thinking modern on-the-go charging cables might be suitable or easily adapted .
 
Attached is a good isolator, including a HUB. The chips are still available, and they are very forgiving with the design. The 5V CHARGE is a separate 5 volt supply and ground to charge phones and power drives, USB sticks and iLoks. I put this in the Slate control, and it has never had any issues. Each unit has a PTC fuse as well. The input is connected direct ground, but no 5V. The outputs are through a 100 ohm R, but bypassed so the data has a good ground, but the current doesn't to the hub. The main issue with charging iPhones without a hub is that they have an Atom on the data pins to see if their is a hub, if not, you have to fool it with 4 resistors setting up a voltage divider, which is different for low speed charging and high speed charging.
 

Attachments

  • USB HUB.png
    USB HUB.png
    162.1 KB
  • Portable-USB-Charger-circuit-diagram-values.png
    Portable-USB-Charger-circuit-diagram-values.png
    15.6 KB
Interesting to note any form of switcher supply isnt reccomended for use with the IFI filter , instead its back to the old style transformer based linear approach ,

Not quite. I recommend the simple linear supplies for DIY injecting separatate power into USB (e.g. with a small DC Jack on a wire tail.

The SMPS I designed for iFi has ~ 100pF Y-Capacitance, but it actually passed EMC without this capacitor. There is a reason it had to be fitted, the TUV lab in Kanton China refused to certify the SMPS without this despite multiple units passing EMC without.

Sadly the case on the small unit is sealed with ultrasonic welds, so it's hard to open and to clip that damn cap out. In "cooking" SMPS the Y-capacitor is typically 2200pF (so noise leakage from the mains is 22 * that if the small ifi unit) and I cannot recommend clipping it out in others.

While for a manufacturer using a SMPS is more cost efficient and gives near unlimited supply options, for DIY we do not to consider such limits and we will find that the Ali-Express linear power supplies for very little money have less noise leakage from mains than iFi's SMPS and cost less and deliver more power (using the nominally 40VA unit).

More discussion here (appy polly logie for linking to rival forum):

Super (capacitor) charging your iFi Zen Can (and other zen products and possibly other manufacturers)

I'd like to hear any advice you have regarding battery based supplies ,
Four nicads plus similar passive filtering to what you show has worked wonders for me, typically getting me zero excess 50hz hum whatsoever even with the laptop running on mains power ,

Yes, this works fine, but is unconvenient. Using off the shelf charger battery packs again brings a usually quite noise SMPS.

For many years I used simple USB 1.1 to optical spdif and adat input devices to kill grounding problems in computer audio ,and as you say its actually very robust but limited in bandwidth ,
most importantly it means your audio circuits are galvanically isolated from any computer garbage.

I agree. where usable this solution is best. I use ADAT inputs from a USB2.0 box (tricked out naturally) on my Yammie 01V/96 and a USB 1.1 isolator for the remote control.

I should have paid better attention in maths in school instead of skipping off for beer spliffs n giggles with the likely lads ,

I did both. Math Prodigy and S & D & R'n'R.

We'd be be able to play tunes if we wanted, grab beers , get introduced to all kinds of throughbred fillies and try and get the slow dance at the end of the night with the one you fancied best .

Sounds like well mis-spend youth. Except I also DJ'ed and worked as sound engineer.

I wish now I had thrown in a cv and went over for a look at the factory back in the glory days ,

I nearly worked for Marshall at Milton Keynes, when they had the first aborted attempt to get into home audio, I kinda echoed the concept then with the iFi Retro decades later. Got to meet Jim a few times, lovely old geezer. The Venture into HiFi got scotched somewhere over politics, my job offer went with it and that was that.

Thor
 
Defianetly agree Yash ,
the simple cable hack and additional filtering is a great start to getting lower noise from any interface . It doesnt have to be destructive to the PCB in any way , the extra filter components can be housed nearby ,

Take it up to the next level and hack up a 50 Quid or less USB Hub. Overall some more datapoints from elsewhere:

MEASUREMENTS: Computer USB port noise, USB hubs and the 8kHz PHY Microframe Packet Noise

MEASUREMENTS: Computer USB +5V Power Noise by Raoul Trifan & MQA's Bob Stuart Speaks Again...

Thor
 
Attached is a good isolator, including a HUB. The chips are still available, and they are very forgiving with the design.

The GL850G is NOT an isolator. It is just a hub. A USB 2.0 Hub. Nothing wrong with using it, it is comparable to the chinese FE1.1. But it is easiest to buy an existing hub and hack it, over building one's own.

There is currently no USB 2.0 high speed capable USB isolator (there was one for a while but it was EOL'ed years ago). It looked like this:

1689760335997.png

VL811 is a USB 3.0 Hub, ICE08 is the USB 2.0 Isolator.

USB 3.0 is capacitively isolated inherently and is as such easily isolated, HOWEVER as USB 3.0 needs USB2.0 high speed working to negotiate USB3.0 capabilities, USB 3.0 isolation does not work without USB 2.0 high speed isolation.

This is a fully tested and working design to isolate USB1.1, 2.0 & 3.0. The isolated power supply is a simple flyback design, 5V in 5V out.

Thor
 
Last edited:

Latest posts

Back
Top