CHANGSTAR: Audiophile Headphone Reviews and Early 90s Style BBS
Lobby => Amp and DAC Measurements => Topic started by: atomicbob on April 24, 2014, 04:35:16 AM
-
Anyone here have suggestions on best practises for breaking the USB shield galvanic connection when using one of the subject line analyzers during jitter analysis? It appears that using the same USB hub for both the analyzer and ADC or DAC under test has ground loop potential that exacerbates mains related noise on the jitter measurement. Maybe a modified USB cable with the shield connection broken at the DAC under test end?
-
USB isolators may come in handy but they will add jitter.
That is they will add jitter to the data being sent IF you measure the jitter at data level.
A DAC with a good jitter rejection (most are these days) you will not find datajitter back in the audio signal.
Also it may help to tie all equipment grounds together with (thick) wire which will become the EMI's favourite path to travel reducing noise in the data/audio path.
It all highly depends on the equipment used and HOW the PCB layout/enclosure is made.
Difficult to recommend a setup that will always work.
-
Thank-you for the suggestions. I agree that an isolator may (or may not) potentially degrade the measurement. I was thinking along the lines of modifying a USB cable at the B end, carefully disconnecting the shield from the B connector shield.
-
Your situation and setup isn't going to be exactly like mine, but I was able to clean up some "noise" related issues that I assumed were some form of ground loop by utilizing Windows 8 tablets in my tests. You can get the Atom tablets for rather cheap (new quad cores are rather zippy, especially on W8), and they usually come with a USB port (some even with a full size, so no adapter needed). I would have to imagine that on battery power, which will likely last you several hours, and with the Atom not requiring a fan, you will be able to get good results and potentially avoid some ground loop situations. I usually just use my DAC on my desktop and my ADC on a tablet, like the Surface Pro, or I will sometimes use two tablets. My ADC is USB-powered, though, so that difference could be worth keeping in mind.
If this is a possibility for you, you can often find used W8 tablets for cheap on eBay. Oh, and W8 tablets are pretty damn useful overall. ;)
-
Depending on the power supply and possibly the implementation of the isolator as well, it will make a difference in the analog side of things(sometimes improvements and sometimes degrading). In my case though, I have found the differences though measureable, to be pretty little of an impact(though it does look significant on the graphs). Sometimes, isolation can also introduce artifacts/distortion into the signal but I'm pretty sure that is due to the power supply implementation.
For jitter, as Solderdude has said, isolation makes little of a difference in DACs with high jitter rejection, for example, the ODAC even with isolation, there was barely any increase of jitter.
-
Your situation and setup isn't going to be exactly like mine, but I was able to clean up some "noise" related issues that I assumed were some form of ground loop by utilizing Windows 8 tablets in my tests. You can get the Atom tablets for rather cheap (new quad cores are rather zippy, especially on W8), and they usually come with a USB port (some even with a full size, so no adapter needed). I would have to imagine that on battery power, which will likely last you several hours, and with the Atom not requiring a fan, you will be able to get good results and potentially avoid some ground loop situations. I usually just use my DAC on my desktop and my ADC on a tablet, like the Surface Pro, or I will sometimes use two tablets. My ADC is USB-powered, though, so that difference could be worth keeping in mind.
If this is a possibility for you, you can often find used W8 tablets for cheap on eBay. Oh, and W8 tablets are pretty damn useful overall. ;)
Very good suggestion. This is exactly what I did when using rmaa with a lavry ADC and a violectric DAC for various audio electrical measurements. I noticed a considerable reduction in the time domain noise floor with the galvanic connection broken. Specifically the coupling of the digital data stream into unbalanced analog connections at approx -95 ~ -100 dBFS was reduced to less than -110 dBFS. So this led me to look at the USB cable specification and the cables I am using.
I found that many of the cables included with various devices varied in the resistance measured from A to B shield connections and could be modulated by manipulating the cable with respect to the connector. With the ground loop intact, this caused the digital data stream contamination of the analog signal to vary as well. Several of the clear jacket variety in my collection would change from 3 to 70 ohms while many of the black jacket type would vary from 1 to 10 ohms. One "audiophile" cable at the inexpensive end of the spectrum stayed at under 0.5 ohm during manipulation. Ohmmeter used was a 2 wire, not a 4 wire. Residual for the meter leads to test fixture is usually 0.3 ohm. The clear vs black jacket is only a loose correlation in my specific collection and may not be generalized to the larger population of USB cables available.
-
Appreciate the suggestions all.