USB isolation experiment = great success!

[ggking7]

I isolated my USB-powered Wavelength Proton DAC with the DIY Paradise Ultravox:

 

http://diyparadise.com/shop/index.php?main_page=product_info&cPath=3&products_id=124

 

and applied power to the Ultravox (and thereby the Proton) with a small SLA battery. It's a huge improvement across the board. The sound is just much clearer. Even my wife is amazed.

 

The only problem is the SLA battery outputs a very bad smell when it's being recharged and it makes both my wife and I sick to our stomaches. To replace it, I have 8 AA NiMH batteries, an AA battery holder, and an AA battery recharger on the way and I'm hoping it sounds as good as the SLA. The battery holder wires the batteries in series so the voltage will be increased to 12V. Even 6 AA batteries would work because the Ultravox's stated minimum input voltage is 7V.

 

[I. G.]

A freshly charged NiMH AA can be 1.4V-1.45V and it goes down to 1.1V. So you can be fine with an 8-pack.

 

[Tipper]

I'm working on something similar at the moment but mains powered.

 

[ggking7]

Could you tell us a little more? I'm very interested in galvanic isolation lately. If it's mains-powered, how do you plan to deal with electrical noise from the mains?

 

[Tipper]

Sure, not much to tell at the moment. Too busy chasing a bit of confusion at the bass to mid crossover point in my speakers and enjoying listening to music through my new Dac.

However, I'm looking at ADuM4160 USB isolator with ferrite inductors powered by a regulated 5 Volt Power Supply. One of the problems is this does introduce some delay on the data lines and until i can get to test the actual delay and gauge its importance. I'm at a standstill.

 

@ggking7

If your still interested.

EDIT.

The reason I was interested in this chip was I’m considering buying the HRT Music Streamer Pro and IF USB power (it requires 4.5 Volts RMS which is flat out for USB power) is an issue with this Dac then the available options are basically a powered USB hub which are generally pretty noisy and not manufactured with good quality audio in mind or a power supply constructed around the ADuM4160 chip as an isolator, IF power supply noise is an issue.

For me at least, the two main attractions to the HRT are:

a) asynchronous data transfer developed in house

b) the part that seems to go largely unnoticed by many; it’s ability to isolate USB carried noise in excess of 20 M ohms.

In particular with regard to the HRT Pro, the 115 db S/N ratio is close enough to the theoretical audible limit to make this for me at least, above my hearing capabilities.

 

a) is important because HRT can refine and develop this technology without license restrictions.

b) is important because if this particular piece of technology works well, in theory you can have as much dirty power as you like on one side of the bridge without effecting the analogue side and looking at the posts on this forum concerning USB power and noise as well as all the USB cable talk which is also basically about noise (unless there are cable enthusiasts that are going to say the cable type alters the actual data) this problem/concern become largely irrelevant.

 

I thought this gave an intelligible account of how the chip works and its applications.

http://www.dataweek.co.za/news.aspx?pklNewsId=33578&pklCategoryID=44

 

Originally I had intended to buy and adapt this board

http://www.circuitsathome.com/mcu/usb/building-usb-isolator

 

but, it does more than I needed and by the time postage, tax and delay were taken into account it was starting to get expensive for a curiosity trial. A colleague of mine from my engineering days (still working in avionic acoustics and in England which is the real surprise in all this) volunteered to knock up something similar (it took a few single malts to get there) based around the ADuM4160 specific to an outboard power supply if I knocked up a regulated 5Volt power supply.

 

I managed to acquire a HRT 11+ (don’t ask) for the experiment. I did try to get hold of the Pro but to no avail. However, the + model is heading in the right direction away from my basic model and I assume the bridge is the same over the range.

 

I think it’s reasonable to assume that HRT use a similar system to the ADuM4160 to achieve their > 20 M ohm specs but without ripping one apart it is just an assumption.

I built/adapted two power supplies I had, one is very basic and pretty noisy (5volt wall wart basically)

and the other is full wave bridge rectified, smoothed and then regulated, 5volt, 1amp.

 

We, (the poor guy I bribed into helping me build the kit) and I listened to both supplies feeding the isolator for a number of hours and despite finishing the rest of the scotch couldn’t detect any change in sound quality with any of the combinations we tried.

 

Conclusion.

We are both deaf.

My system didn’t cost enough money to be able to hear any changes. We even tried an Ultraviolet USB cable. Not mine I hasten to add

It doesn’t matter within reason what rubbish goes in the USB end of an HRT. The system works as well as the power supply and isolator we built.

 

 

Someone is bound to say that in a really high quality system this cable or that isolator would make an audible difference; or that we are deaf.

So for the record I’m not deaf (I can hear the difference between a 24/96 file and a 16/44 file on my system; just )

With regard to the system, it doesn’t actually mean much unless you’ve heard it, however here it is with prices which may give some vague indication of its performance.

We used my colleagues Asus 73JH laptop, WASPI. (I don’t know how much he paid for it but it looks very nice)

Exposure 25RC Integrated Amp (£1150)

My hand built interconnects and a pair of Atlas Equators (£48)

Speaker Cables; CS122 HiFi Cables 4 metre airlocks (£134)

Speakers. Heavily Modified Volt Boxer.

Efficiency; 10 watts for 97db at 1 metre.

Polar response symmetrical at 180Hz to 12000Hz within 30 degrees.

Crossover at 400Hz and 3 Khz.

Volt B250 Bass. Volt ABR (tuned at 20Hz). ScanSpeak 13M8621 Midrange. ScanSpeak D2905/9300 Hi Range.

Hand built crossover (designed by Volts David Lyth) air core hand wound chokes, polypropylene capacitors.

These speakers will handle frequencies from 20 Hz to 22 Khz within an error of around 6db. Below 60Hz the performance drops off at around 1db per 1 Hz to 20 Hz. (Price for components only £1600)

 

 

[ggking7]

Hi Tipper, noise in the power applied to the HRT should reach the DAC chip no matter what. A chip like the ADuM4160 provides true galvanic isolation, which means no power can pass through it to the other side, only digital data. If the HRT is utilizing power provided by the USB connection, it is also susceptible to noise in that power. The way to fix this is to apply power on the DAC side of the isolation that is as noise-free as possible. The ADuM4160 provides for exactly that, and NiMH/SLA battery power is very low-noise.

 

[Tipper]

Hi again ggking7

Did I really make such a bad job of explaining myself (?)

 

We did apply galvanic isolation to the USB side of the HRT using the two power supplies I mentioned; the wall wart type (noisy) and the rectified one (very quiet) and the ADuM4160 chip.

 

With or without the ADuM4160, using either type of power supply didn’t make any audible difference. This was the whole point of the experiment.

 

What seems to be not understood by some people about this Dac and a few other Dacs that use some form of power rail isolation be it galvanic or otherwise is that “dirty” power doesn’t give an audible degradation in sound.

 

For example there is a thread on another forum somewhere where some contributors are discussing the audible benefits of isolating a Bel Canto Dac using the ADuM4160 chip and a battery supply.

 

To the best of my knowledge this Dac already has such an isolator built in, adding another one cant possibly make any difference be it battery powered or otherwise.

There are a number of such threads scattered around various forums concerning use of this chip when the Dacs concerned already have a similar chip in place.

 

The >20 M ohms USB to Analogue audio section isolation inbuilt into the HRT makes USB born noise irrelevant.

 

I’m quite sure there are better sounding Dacs available at a price. I’m also quite sure a more sophisticated analogue stage could improve its sound.

The two features of this Dac that particularly interest me are

the USB asynchronous data transfer which seems to be slowly catching on and getting to be accepted even by the SPDIF diehards.

The other is it’s >20 M ohm noise isolation which many other USB Dacs don’t even feature in their specs let alone achieve at this price level.

I really don’t know how better to explain this.

 

 

[ggking7]

Tipper,

As long as your DAC is powered by your power supply, it will be affected by noise from that power supply. The HRT's "isolation" can't magically remove all noise from the power current that it uses. The ADuM4160 blocks the current entirely and only passes digital information. That's why a separate power supply is needed, even for a USB-powered device. You probably knew all of this already. However, you said:

 

"With or without the ADuM4160, using either type of power supply didn’t make any audible difference. This was the whole point of the experiment."

 

I'm saying you should try SLA battery power because it should be less noisy than either of the power supplies you tried. Then the ADuM4160 might make an "audible difference". Does that make sense?

 

Everyone,

I've tested 8 AA NiMH batteries now, and the sound is substantially worse than the SLA. I've switched back and forth and the sound is clearly much more full and rich with the SLA, much more bass and overall clarity. The NiMH batteries are brand Powerex, brand new, and fully charged.

 

Does this mean NiMH batteries are noisier than SLA? Is there any other possible explanation?

 

[Tipper]

I’m not getting through here am I?

Galvanic Isolation is achieved using a row of mini transformers for want of a better description.

There is I believe more than one chip available that does this.

Let’s just suppose for sake of argument that the HRT does use a galvanic Isolator to dump USB born noise. (I can see I’m going to have to dismantle mine at this rate) adding another one isn’t going to make any difference is it?

 

HRT state in their specs the USB to audio signal Isolation is greater than 20 M ohms. Thats about the performance of a galvanic bridge i beleive.

 

Next, if the chip ADuM4160 completely isolates USB born noise on the D+ and D- rails then as long as you feed it power on the correct side of the bridge, it cant make any difference whether its batteries or mains as long as the voltage and current are suitable can it? It either does isolate or it doesn’t.

 

Therefore if you have it wired up correctly using the pin config for the appropriate USB speed and you accept it does fully isolate it can’t make any difference what batteries you use can it? It isolates the noise from any power source……doesn’t it (?)

The only doubtful features I know of concerning the ADuM4160 chip are its bandwidth capability and the data delay impact.

 

Think on it.

 

[vortecjr]

Their specification says, “With its unique topology the Music Streamer II+ provides a completely isolated path between the computer and the audio system.” Their specification also says, “There are no external power supplies to worry about because the Music Streamer II+ derives its power from the USB buss while completely regenerating power for all its internal circuitry using a sophisticated set of proprietary circuits.”

 

This to me means USB data and power isolation and it sounds like the tests conducted by Tipper would confirm. I’m not sure if other power supplies will make a difference for better or for worse.

 

Jesus R

www.sonore.us

 

 

[ggking7]

I must be misunderstanding what it is you're doing. How are you applying power to your device? You said it was USB-powered, right?

 

EDIT: Wait, it's coming back to me. You were applying power via the ADUM4160, right? So you tried two different power supplies and since they sound the same you conclude the HRT is as effective as the ADUM4160? If I were you I'd try a power supply you know is low-noise. Like batteries.

 

[I. G.]

Some "gurus" recommend LiFePO4 batteries, have smallest internal resistance. 3x3.4V would be good for the 7-12V range. I use 4-pack NiMH for 5V USB supply, unfortunately 4.5-5.5V cannot be mixed from SLA (~2.1V/cell) or LiFePO4 (~3.4V/cell)...

 

[ggking7]

I did a bunch of research this morning and I came to the same conclusion. A123 Lithium Nanophosphate batteries are supposed to be great. The 26650 looks like the right choice:

 

http://www.a123systems.com/a123/products

 

The only problem is the voltage. They're 3.3v and I need 7v, although I wonder if I could get away with 6.6v since it's being regulated down to 5v on the Ultravox anyway. The capacity of 2.3Ah should be plenty for the Ultravox and Proton DAC. I just need to figure out how to wire them in series for the increased voltage. I've never soldered and I'd like to be able to buy a holder like I did with AA batteries, but there don't seem to be any available. Any ideas?