Seperate USB Data and Power

[Tubetec]

Thanks again Thor ,
tons more good stuff there Im sure ,
Just doing a second readback of the other thread ,
to make sure I understand everything correctly ,

The idea of an isolator seems jamjared in peoples heads ,
This is a change to the way the grounding is arranged ,
techincially I guess you are isolating the supply , but thats different thing to isolating the data lines using opto 's .

There is extra inconvience using batteries ,
4 nicad cells goes to over 6V on charge , where supercaps are limited to 5.5v . with a fully charged battery on load it drops back to 5.4v .
Running time of the device is around 8-10 hours ,but of course that might vary depending on if the headphones are driven or not ,

What I had in mind was this ,
A battery based USB supply , if possible entirely passively filtered and unregulated ,
Then a quiet mains based charger/battery management system that takes a cue signal via midi to tell it recording(or playback)is underway , in which case the charger is shut down and disconnected at the mains side ,
If mains power for what ever reason goes down the battery seamlessly takes over without interuptions in the audio path ,
Maybe I just set the on charge voltage to never go above 5.5v , a trickle charge is the old term associated with nicads ,
the USB device would always be on load , permanently powered with a small trickle charge until playback is pressed .
Theres a few off the shelf charge controler pcbs on ebay with simple programable logic , they have a DVM and sense inputs that can be used to trigger end of charge cycle at a preset time or voltage level and dissconect the battery when it drops to preset a threshold value.

Same as yourself I realised the good in permanently powering the day to day listening system many years back , The Quad II's formed the heating system in my room in winters over the years ,
Not so much these days with the cost of electricity and tubes !

Thanks again for the info on modding the good old fashioned in line iron linear adapter as low noise powersupply , I see tons of these things in electronic recycling all the time , it easy to pull a few of the appropriate voltage when I drop off my empties down at the yard , even an non regulated DC adapter can easily have the 317 regulator added .

A 9v version of what you describe above with the supercaps sound like a great idea for use with stomp boxes ,
Nowadays the typical pedalboard has half a dozen SMPS ,each with its own unique whine/buzz and a grounding scheme like nest of vipers over back and through the pedal casings ,supply wires and signal cables , very much like the USB power problem , its a major cause of poor tone with guitarists in the modern day.

Noise gates to me are a poor way of tackling the problem ,
From a sound engineering perspective you often end up where the overall S/N ratio is determined by the noise comming off the guitarists rig and there isnt much you can do about it in a live situation .
In the studio environment the issue can be tackled at source , but its very much something you need do on a case by case basis , theres no short answer , the best grounding arrangement can be determined manually , with the help of your ears or software if you want to look .

I wonder how many people got tricked into buying fancy mains power filtering systems that cost thousands but dont end up making much difference to their buzz issue , where the proper local USB grounding/power arrangement would have cured it to a greater extent .

I'll have to keep my eye out for an Ifi Zen HP amp , like you pointed out , its easy to do better than the basic SMPS that it comes with ,
Amir over on Audio Science review did testing , results were generally very good but that the PSU running out of juice was fairly evident from the distortion readings , which someone pointed out to him ,

The other thought is a small USB interface might consume 2-3w ,
a 40Va transformer gives a very generous 10:1+ current margin , means the transformer runs nice and cool with less energy lost in the windings and core material ,as well as better handling of short term demands for current .

 

[RuudNL]

I once had the problem that a USB powered soundcard produced background noise.
You could even 'hear' the movement of the mouse!
I bought two USB connectors, linked the data lines and ground and used a separate power suppy for the +5 Volts.
Silence was the result!

 

[thor.zmt]

The idea of an isolator seems jamjared in peoples heads ,
This is a change to the way the grounding is arranged ,
techincially I guess you are isolating the supply , but
thats different thing to isolating the data lines using opto 's .

I call this style "ground loop breaker" as opposed to "isolator".

To me an isolator completely breaks any galvanic (DC coupled) connection between circuits and uses a minimal capacitance between the two grounds (which may be at very different potentials at DC or low frequency AC) just sufficient to guarantee reliable operation.

Opto-Isolation is way too slow.

I know an Isolator that uses a USB Phy that has a 12-Bit (8 Bit Data) interface at 60MHz/MbpS for which digital isolators with a surplus of bandwidth (100MbpS+) are available. This product is in effect an intelligent USB repeater with one isolated port. It sells for over 300 Euro in a beige plastic case that looks like it belongs to 1980's DIY PC peripheral.

There is extra inconvience using batteries , 4 nicad cells goes to over 6V on charge , where supercaps are limited to 5.5v . with a fully charged battery on load it drops back to 5.4v .

One solution is a commercial Phone Charger portable battery.

Modern USB-C based ones can deliver actually as much as 20V/5A and have 100Wh Lipo 18650 batteries.

The step-up switcher in these commonly switches at least at 250kHz and often higher with 750k-1Mhz currently seeming the limit. These relatively high frequencies are easily filtered.

They sell for not THAT MUCH, combine with a 100W USB-C charger brick will be << 200 USD. For that they offer a lot of power and operating time.

As these "pass through" external power when charging their internal battery, one way is to sense downstream current and to disconnect the charger from mains AC (relay) whenever the current flow indicates a load is connected.

One could even build a dedicated PSU using the gut's of two such units and a charger brick, switching over between the two so one charges and the other delivers power.

Or use the IC's used for such poweranks and roll your own system. I use these powerbank IC's as Battery controllers in portable equipment.

Thanks again for the info on modding the good old fashioned in line iron linear adapter as low noise powersupply , I see tons of these things in electronic recycling all the time , it easy to pull a few of the appropriate voltage when I drop off my empties down at the yard , even an non regulated DC adapter can easily have the 317 regulator added .

Yup. Many options in recycling, fewer in the market.

A 9v version of what you describe above with the supercaps sound like a great idea for use with stomp boxes ,

That needs two 5.5V Super Cap's in series with resistors for Voltage sharing. And yes, 9V units exist.

Nowadays the typical pedalboard has half a dozen SMPS ,each with its own unique whine/buzz and a grounding scheme like nest of vipers over back and through the pedal casings ,supply wires and signal cables , very much like the USB power problem , its a major cause of poor tone with guitarists in the modern day.

There is a sad story and LOADS of lost income here. When we did the first iFi products in 2013, I went for external 9V plug top supplies.

We started with a "low noise" version that was off the shelf and had generally decently low noise, 1A current limit.

When production stopped, we had a factory design a "very low noise" replacement with input from me, which was in 10's to 100uV noise at the switching frequency and very low general noise. There were a few hiccups and factory changes and re-designs to fix these issues.

At the end of the process around 2016 we had a SMSP with extremely low noise all across the board, including 9V, with replaceable mains connectors for worldwide use and as remarked with 100pF coupling capacitance which was strictly speaking optional and could have been systemically deleted.

This PSU was around 5 USD cost from the factory in 5kU (mixed voltage) and including all packaging, accessories and all. And it is a very good Power Supply, disregardless of operating principle.

I suggested to make a special center negative version (the 9V version always includes a gender changer for pedals, but it needs a gender changer) with an "octopus" 1-2-8 or so DC splitter cable with extra heavy grounds (for more reduced ground loop issues) especially for pedals, with a new, Musicians targeted paracking (mainly printing) and sell at 39 USD max., with the Shop cost 20 USD and local distributor or big chain direct cost 10 USD.

I think they would have sold very well. And with this one done, other voltages for other products could have been added.

I actually even had 50/60/100/120/200/240/400/480Hz AC power supply "lump on a cable" planned that would take the DC from a suitable of our SMPS and output AC, for low voltage AC powered MI or PA devices.

Instead S&M (sales and marketing - though I feel the other reading may be more appropriate) took the "HiFi" packaged version without DC splitter around big distributors and shop chains. The need to use a Gender changer and packaging with flowery hifi prose on them plus trying an inflated asking price for 59 USD retail resulted in nobody touching these power supplies, usually not even testing them.

So there went owning the worldwide aftermarket Pedalboard mains power supply market and making a killing in this AND helping a lot of guitarist to loose their noise headaches. I made a few handbuild hardwired splitters with an iFi supply for friends, always did sterling service.

Thor

 

[Newmarket]

https://www.usbgear.com/usbg-iso-m.html

Thanks

 

[Newmarket]

I once had the problem that a USB powered soundcard produced background noise.
You could even 'hear' the movement of the mouse!
I bought two USB connectors, linked the data lines and ground and used a separate power suppy for the +5 Volts.
Silence was the result!

Yes. Been there. Put eg 100R in parallel with eg 100n between the Grounds and it might be even better.
I did what you did and it was a big improvement but still a bit of "noise".
I'm not running any audio on USB atm but when I do I'll put that into the box I DIYed.

 

[Tubetec]

There is a few USB3 powered extender cables available in various lenghts , they either take power from host or automatically switch to external if its provided over the third USB plug ,located at the device end .
The USB device then has only has to drive the short run to the repeater , I think there around 30 euros to start off , might be worth a try for anyone who's having stabillity issues over distance .
The uppermost connector takes the input from the device , next down supplies the power and the last goes to the host pc .

I do and have used several battery charge adapters as seperate sources for USB devices along with the cable mod , they couldnt stand any decent sized caps at their output without misfiring ,but they basically did the job .
I now have it arranged to power the Xmos side of things with the nicads doing the audio , its a 10,000mah unit so its extends the life of the nicad if used to handle the digital 5v line .
I think the whole things runs around 12-15 hours on a single charge of the nicad , the powerbank lasts a couple of days between charges .

 

[jp8]

Hi all, my first post on this forum... To those who are interested: you will find my introduction in the New Members Introduction section.

For my DIY "Audio Analyzer", I wanted to have a USB isolator between the PC and UMC202HD Audio Interface. I decided to build my own one, based on the new TI ISOUSB211 chip. The chip supports USB 2.0 speeds up to 480 Mbps. I have never actually tested it up to this speed, but it works when using the UMC202HD full-duplex on 24bits/192kHz.
On the PC side, it is powered by the PC (of course). On the client side, power must be provided by an external power supply, for which you can use your favorite design. This power supply should be able to power the client side of the isolator (~100mA max) and the client. I have some PCBs left. The smallest parts are 0402. Is anyone interested? As these 4-layer PCBs were quite expensive ($31.16 + shipping for 5 pcs, to be exact), I hope you understand that I won't be giving them away for free. Have to decide on the price later, probably the actual cost + shipping. BTW, I live in The Netherlands, so shipping to other European countries won't cost that much.

I have also designed a USB-stick-like 3D printable housing for this isolator. I can share the .STL file of that one too if you like.

 

[Yash]

The USB Hub hacking route is definitely a good way to proceed: you get the Reclock + Regen (?: correct me if I'm wrong here) for free.

In that case, we would need to know what affordable Hub does a good job at the Reclock/Regen process.

That hacked Hub would then need to sit as close to the end device as possible.

 

[Yash]

Correction, ifi HAD this. Now they don't. Where they make new products, not things based on, or directly my designs, they no longer use either BB nor CS, which they find too challenging to implement.

Conclusion: Thor has extensive XMOS and Burr-Brown DAC knowledge

 

[Yash]

Reading the very interesting post on HeadFi reminds me that I bought a few supercaps for tests, haven't yet tested them on the output of my DIY low-noise Linear Regulated PSU but what I found made a nice difference in sound were the caps in old disposable cameras.

They might be similar to supercaps but I am not sure as they're quite old: they power the Flash.

A good portion of my custom PSU comes from discombobulated simple transformer supplies for the main transformer, some caps, inductors and wires from ATX PSUs, old disposable cameras for those special caps etc...

Did many experiments to tame the simple LM317, but nowadays you can buy much better.

 

[thor.zmt]

The USB Hub hacking route is definitely a good way to proceed: you get the Reclock + Regen (?: correct me if I'm wrong here) for free.

That's just marketing speak for "USB repeater".

Interesting story.

Many years ago AMR promoted USB equipped DAC's by offering dealers a custom PC setup based around a 24" touchscreen all in one, as all dealers were turning up their nose at PC audio and claimed they could (be bother to learn something new and) set up such a system.

As a touchscreen PC needed to be at the listening position we needed a REALLY LONG USB cable. What we got were "active" cables eith a repeater ever so many meters, more than the nominal 5m USB cable limit.

This cable for some reason sounded better than a passive cable. Inside was the early version FE1.1 USB 2 Hub Chip used as repeater.

In that case, we would need to know what affordable Hub does a good job at the Reclock/Regen process.

On my desk I use a Orico USB 3.0 based around VIA chips, which are the the same as in iFi USB 3 products.

Mind you, I also one of these "pregnant memory stick" repeater's directly plugged into the DAC (iFi Zen) and power the DAC with Linear PSU (AliExpress) with super capacitor pack.

As inside the DAC I integrated a groundloop breaker when I designed it, it lifts USB ground automagically if external power is supplied.

That hacked Hub would then need to sit as close to the end device as possible.

Actually - science applies.

480MHz is around 0.625m at light speed. Signal propagation in cables is usually at ~ 2/3 C, so actually in a cable the wavelength would be ~ 40cm.

In order for most (negative) cable and connector effects to become material a cable needs you be multiple wavelength long.

I would suggest that any quality USB cable shorter than 12" is probably going to be fine between hub and Device, to make the cable "mostly harmless".

So take Orico USB 3 Hub, take one output, unsolder the USB plug isolate shell if it is not already (usually is, but if not, needs cutting and scraping away ground plane on PCB) plus ground pin and +5V pin (just bend them gently flat so they no longer touch the PCB) and then wire up things.

Ground gets 100R//1nF (experiment with cap value if you get dropouts) plus back to back diodes (1N4148?) to hub ground.

External DC Supply connects direct to the Gnd/5V pin. You can add a few super capacitors as holdover and "Denoiser" at the socket and use a 5V mechanical relay as "fallback" to reconnect internal power should the external Power fail. That's easiest for DIY.

A slightly noisy great "take" is better than a lost great "take", because a plug fell out.

With TI now making a high speed isolator IC I'd buy a commercial but cheap "copied in china" version and use that between hub and PC, if needed a second linear PSU for hub.

Thor

 

[thor.zmt]

Thanks

Note, according to the page (if I read the cashed version right) USB 2.0 full speed max (that is essentially USB1.1 fallback).

So most audio devices will either not work at all, or fall back to USB1.1 which is usually 16/48 fixed.

Thor

 

[thor.zmt]

Reading the very interesting post on HeadFi reminds me that I bought a few supercaps for tests,

Make sure to buy Low ESR types. The old "coin" style memory backup types have very high ESR and are not useful in Audio. This is why I use a LOT (20pcs) in parallel, to get the ESR down to usable levels.

what I found made a nice difference in sound were the caps in old disposable cameras.

They might be similar to supercaps but I am not sure as they're quite old: they power the Flash.

They are nothing like supercap's they are "Photo Flash" high voltage types. They emphasise pulse current delivery but have very poor lifespan etc.

Did many experiments to tame the simple LM317, but nowadays you can buy much better.

Best thing to do with LM317/337, use them as "self biasing" reliably protected series pass transistors in a "super regulator".

First, with a resistor between OUT & ADJ you can power reference voltage generation (TL431 for simplicity) and the Op-Amp from the output of the regulator for improved PSRR and the resistor allows the Op-Amp to drive low current high frequency currents directly bypassing the "slow" 317/337.

Needs a little finesse to stabilise, but once done you have high precision and noise at what amounts to the Op-Amp noise if done sensibly. Use a NE5532 for a +/- audio rail supply...

Thor

 

[Tubetec]

Speaking of the camera flash capacitors ,
I made contact with a guy in a print shop , he processes about 40-50 disposable cameras a week , mostly the Fuji kind , theres a very high quality Japanese made 330v 100uf cap contained inside ,
I've used two banks of ten caps on the 5V supply Ive made , its gets me an ESR of around 25mohms per bank , not quite as good as the supercaps but still it cost me nothing , the going rate for a similar Japan made flash cap from Mauser can be 3-5 dollars each in small quantities . so well worth salvaging.

My undestanding of the spec on the flash capacitors is there not rated for higher temp like 85 or 105 c
parts ,

Can LF chokes have a place alongside supercaps and batteries in making a very quiet passive filter ?

 

[Newmarket]

Note, according to the page (if I read the cashed version right) USB 2.0 full speed max (that is essentially USB1.1 fallback).

So most audio devices will either not work at all, or fall back to USB1.1 which is usually 16/48 fixed.

Thor

Yes. I saw that this would not be a good choice for current digital audio application.
If I read it correctly the user is using this with his synths etc. So I assume it's handling the 'MIDI' control data etc that is now commonly on USB (in addition to actual MIDI standard hardware). And that this helps with keeping noise off the USB 'Ground' related to any audio function.

 

[thor.zmt]

Yes. I saw that this would not be a good choice for current digital audio application.
If I read it correctly the user is using this with his synths etc. So I assume it's handling the 'MIDI' control data etc that is now commonly on USB (in addition to actual MIDI standard hardware). And that this helps with keeping noise off the USB 'Ground' related to any audio function.

Hmmm, it only works if you have separate USB connections.

As I wrote, I do something on my Yamaha 01V/96, USB is MIDI control of the Mixer, a separate USB 2.0 high speed interface is used to convert everything to optical ADAT which is then used with the Yamaha ADAT Card.

Audio is isolated by optical ADAT and MIDI USB by a full speed isolator. The DAW PC and the Audio Mixer are completely isolated (and sit on separate Power Distributions

But it's important to remember the limitations. This is not a "universal use" design for Audio.

Thor

 

[KA-Electonics.com]

Going back to questions people have about the iFi iDefender...

I have three of them using external power injection and they do clean up most USB-induced ground loops.
Two are used with Focusrite 2i2 the third with a Roland Quad Capture. The Quad capture is the worst offender.

When they don't help - and it varies greatly with the situation - they still provide options you wouldn't otherwise have.
Thanks for bringing them to market thor.zmt.

 

[Tubetec]

I gave a buddy my old Maudio transit USB interface(with optical in/out) to use with his Yamaha 01 ,
He said it worked out flawless , he'd previously spent big money on Apogee gear but the firewire connection never worked for him , that simple bit of kit solved all his issues .

I cut my digital teeth back in the days of the Alesis Adat and Tascam Da-88 ,
The transport would let you down in the end , but the Adat optical format lived on , its something Ive used a lot and it always gave workhorse reliability , no grounding problems and it was plug and play ,

I see in SSL's newish model 12 interface they snuk in an adat optical input , its a smart move as it opens up many possible options to interface up pre-existing gear the customer has .

I also saw the review and modification of the original Behringer Speaker management EQ/RTA/measurement mic ,
Was an interesting read ,
I have to admit to having a liking for a few select bits Behringer put out back in the day ,
I'm glad I took a REV2496 before the cease and desist letters were dropped , ADA8000 ,SRC2496 are also fairly usable boxes.

Going back to USB power ,
In the Diy end where we find ourselves theres a real difficulty getting decent audio grade SMPS , theres better and worse available off the shelves ,but nothing specifically designed for low noise in an audio context ,

As Thor mentioned ,fundamentally , by design most SMPS are unsuitable for audio as they leave the factory , something to do with the approval agencies /safety etc .
Then we have the USB power problem on top ,

Killing two birds with one stone is a great opening gambit for any fledgling audio company ,
Shame the S&M department turned it into a pig in knickers with high price tag ,

 

[Yash]

they are "Photo Flash" high voltage types. They emphasise pulse current delivery

Indeed. These coupled with tiny Vishay Polymers did wonders at output. I had a set of those tiny Vishays for my DIY Analogue Modular builds. They ended up being very useful for audiophilia. The Photo Flash caps were obtained at the thrift store nearby.

Best thing to do with LM317/337, use them as "self biasing" reliably protected series pass transistors in a "super regulator".

Yes, Walt Jung did a lot of work on those. Actually, he still does.

I am more inclined to work on Shunts these days though.

 

[thor.zmt]

Going back to questions people have about the iFi iDefender...

Just a note, I feel they (small USB thingys) ended up way overpriced. Original target at design was 29 USD for the iDefender and iSilencer (defender is groundloop breaker and silencer is active USB power line noise killer) for a combined unit.

The iDefender is basically a bunch of Mosfet switches that routes power and has the groundloop defeat.

When they don't help - and it varies greatly with the situation - they still provide options you wouldn't otherwise have.
Thanks for bringing them to market thor.zmt.

One of the problems at iFi was that instead of making a single "full spectrum" product and price it fairly, such products are broken up into multiple products, each with a single function.

So the iDefender & iSilencer were meant to be one inexpensive product that would cover more use cases.

Same on the AC side.

I proposed a mains cable using fairly generic high quality shielded cables, with a DC blocker (stops toroidal transformer buzz), earth/ground lift (break earth loop) and the "ANC" active noise canceller for mains noise ~ 30kHz and up.

Now we have split up all this and the basic mains cable (without anything) costs more than the fully loaded unit was supposed to cost.

And we suddenly have a 100 USD earth lifter, a 100 USD DC blocker and the mains cables have a fancy construction that will however cause unfavorable electric parameters and costs 300 USD for a basic passive mains cable and 500 USD active noise cancelling one plus 200 USD for the rest.

This makes what was meant to be 99...149 USD a mockery and utter ripoff. It would not exactly been a cheap kettle lead. But seeing how many problems it would have fixed with full code compliance etc. (And a lack of anything remotely similar) I think the original product would have made sense and been good value for money.

So while all the iFi accessory products are "real" and work, if the problem they address are present, the extreme fragmentation of products, combined with poor pre-sale support and dealer training
makes it hard to select the right product.

And excessive (way past fair) profit margins pricing limits potential sales appeal for most customers and makes it hard for me to recommend these products.

Thor