USB cable shield resistance technical measurements

Discussion in 'Portable and Other Gear Measurements' started by atomicbob, Jan 15, 2018.

  1. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,447
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    USB is one of the banes of digital audio. Cheap USB cables contribute considerably to the problem. This technical measurement post will consider one specific, very important attribute of a USB cable, the shield. Shielding functions to keep electromagnetic interference (EMI) out of the USB data signals and to contain the USB data from contributing to the ambient EMI. Consider that USB logic lines have only a couple of volts of Common Mode Rejection (CMR). Light dimmers, small motors (refrigerator), nearby Switch Mode Power Supplies (SMPS), especially the cheap ones, all may contribute to EMI disturbance of USB data if the shield doesn't perform well. To perform well the shield requires a low ohm galvanic connection with the A and B end connectors.

    8 USB cables were evaluated for shield resistance to A and B end connectors.
    5 - generic, the type usually included a with USB device.
    2 - Audioquest Forest, 0.75 meter and 1.5 meters length.
    1 - Schiit Pyst.

    Data were acquired using the following:
    Keysight 34461A 6.5 digit DMM with Benchvue for data logging
    APRS World USB A breakout board
    APRS World USB B breakout board

    Each cable tested had both A and B ends inserted twice into the breakout boards to assure removal of any oxidation on the shield connection.
    Each cable was measured over a period of 10 minutes twice, first time to allow for settling of connector materials, the second for measurement of record.

    Measurement setup:

    USB cable sheild measurement setup - small.jpg

    The cables:
    USB cable 1 thru 6 - small.jpg
    1 generic clear small
    2 generic black small
    3 generic clear large
    4 generic black small
    5 Audioquest Forest 0.75 meter
    6 Audioquest Forest 1.5 meter


    USB cables 7 and 8 - small.jpg
    7 generic black stiff cable small
    8 Schiit Pyst


    Results (lower resistance in ohms is better):
    20171228 USB Cables shield R - all.png


    6.5 to 7.5 ohm range expanded y-axis:
    20171228 USB Cables shield R - 7R.png
    I wouldn't trust this cable (#2) to work reliably in printer usage.



    2.5 to 3.5 ohm range expanded y-axis:
    20171228 USB Cables shield R - 3R.png
    #1 isn't much better than #2 - note the sudden jump - explained below in mechanical disturbances

    0 to 1 ohm range expanded y-axis:
    20171228 USB Cables shield R - 1R.png
    #3 and #4 are better but still considered failures as they are much higher in resistance and continue to vary over time. #6 and #8 (Forest and Pyst respectively) are both low in R and constant over time.



    Mechanical disturbances:
    Look again at the USB cable in the 2.5 to 3.5 ohm range y-axis expansion, near the end at 9.75 min. That sudden jump was coincident with my sneezing approximately 0.5 meter distant from the cable, facing away from the cable. Galvanic resistance is easily disturbed in the cheap USB cables. Better quality USB cables are much more robust. A squeeze test consisting of gently applying pressure both A and B ends approximately 15 cm from the end demonstrates this disturbance. The fit between the USB cable and breakout board connectors is tight. Care was taken to avoid connector to connector movement during this test. Only the shield to cable connector transition is receiving stimulus. Here are some examples.

    Squeeze test examples:
    20171228 USB Cables shield R - squeeze test.png
    #7 varies considerably and unpredictably with the gentle squeeze test. #8, Pyst both varies little and is repeatable.


    Conclusion:
    Audioquest and Schiit know how to specify and build (or have built for them) cables with decent shield connections. USB cables included with USB devices that I've examined thus far appear to be mostly crap.
     
    Last edited: Feb 6, 2018
    MichaeLeroy, Sqveak, econaut and 32 others like this.
  2. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,447
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    Here are the Benchvue raw measurements.

    Please note Benchvue y-axis changes for each measurement to maximize data displayed. What appears to be a large change may be quite small relative to another USB cable raw measurement. Pay attention to y-axis values.

    01 generic clear small shield
    01 20171228-1715 generic clear small shield R.png

    02 generic black small shield
    02 20171228-1800 generic black small shield R.png

    03 generic clear large shield
    03 20171228-1730 generic clear large shield R.png

    04 generic black small shield
    04 20171228-1815 generic black small shield R.png

    05 Audioquest Forest shield - 75cm
    05 20171228-1700 Audioquest Forest shield R - 75cm.png

    06 Audioquest Forest shield - 150cm
    06 20171229-0930 Audioquest Forest shield R - 150cm.png

    07 generic black small stiff shield
    07 20180113-1205 generic black small stiff shield R.png

    08 Pyst shield
    08 20180113-1315 Pyst shield R.png


    Squeeze test raw data examples.

    Please note Benchvue y-axis changes for each measurement to maximize data displayed. What appears to be a large change may be quite small relative to another USB cable raw measurement. Pay attention to y-axis values.

    07 generic black small stiff shield - squeeze test
    07 20180113-1402 generic black small stiff shield R - squeeze test.png
    This generic cable is varying 0.3 ohms from stimulus!!!

    08 Pyst shield - squeeze test
    08 20180113-1400 Pyst shield R - squeeze test.png
    Pyst is varying only 0.008 ohms with stimulus, nearly 40x less than the generic above.
     
    Last edited: Jan 16, 2018
    Royaume, Sqveak, Josh358 and 12 others like this.
  3. Priidik

    Priidik Friend

    Friend BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    1,702
    Dislikes Received:
    3
    Trophy Points:
    93
    Location:
    Estonia
    Did you rule out the possibility that connectors caused this?
    If not connectors, it sure is interesting find. A sort of microphony captured it's pants down.
     
  4. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,447
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    Yes, look at the squeeze tests with Pyst vs generic. When I saw that jump on cable #1, I began tapping on the table during the first measurement material acclimation run. Every one of the generic cables would jump more or less. Audioquest and Schiit cables did not. Thus the squeeze test was created mid experiment. Hard to repeat sneezing in a reliable way.
     
    econaut and Dino like this.
  5. Luckbad

    Luckbad Traded in a unicorn for a Corolla

    Friend
    Joined:
    Sep 27, 2015
    Likes Received:
    3,427
    Dislikes Received:
    6
    Trophy Points:
    113
    Location:
    Laguna Hills, CA
    Home Page:
    I have some spare pepper I could send your way. Or you could get a friend with photic sneeze reflex and force them to look askew at the sun to sneeze.

    Anyway...

    This is awesome, and very interesting.

    I can't tell a difference between the stock RME cable that comes with the ADI-2 DAC and Schiit Pyst. That thing looks pretty generic, but it's probably a carefully-chosen sort of generic, and the ADI-2 DAC cleans things up in any case.
     
  6. Elnrik

    Elnrik Super Friendly

    Friend
    Joined:
    Jan 10, 2017
    Likes Received:
    8,528
    Dislikes Received:
    34
    Trophy Points:
    113
    Location:
    Denver CO
    Home Page:
    Thanks for the write-up Bob!

    You know what I'd love to see? One of those snake oil $500 USB cables put to the same test just to show that it wouldn't be any better than a Pyst.
     
  7. crazychile

    crazychile Eastern Iowa's Spiciest Pepper

    Friend BWC
    Joined:
    Jan 19, 2016
    Likes Received:
    1,763
    Dislikes Received:
    2
    Trophy Points:
    93
    Location:
    Grant Woods backyard
    I believe The Cable Company has a loaner program for a fee. If there was interest in taking the experiment a little farther, I’d be willing to chip in for part of the loaner costs. Maybe some popular models from $50 to $200. More expensive but not esotericly insane stuff.

    I’ve also got a well used Wireworld Ultraviolet that I could send to Bob as part of the experiment if interested.
     
  8. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,447
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    Wireworld Ultraviolet inbound later this week.

    Please remember all that is being tested is the shield connector resistance, though it is an important attribute. When a vendor has skimped on the fastening process of shield to USB connector shell, it doesn't instill confidence in other attributes of cable performance.
     
    Dino, bixby, cityshlacker and 3 others like this.
  9. Erikdayo

    Erikdayo Friend

    Friend
    Joined:
    Apr 29, 2017
    Likes Received:
    608
    Dislikes Received:
    10
    Trophy Points:
    93
    Location:
    Boston, MA
    I often get my cables from Monoprice or Amazon Basics. Curious whether those are as bad as the generic ones tested or a bit better made.
     
  10. Clemmaster

    Clemmaster Friend

    Friend
    Joined:
    Sep 28, 2015
    Likes Received:
    1,946
    Dislikes Received:
    4
    Trophy Points:
    113
    This is awesome work, Thanks Bob!

    I think the real conclusion from this extensive scientific experiment is that Black is bad for USB cable performance!
    |{

    Please spread the word to other forums!
     
  11. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,447
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    Second USB cable shield test run will include the following (all currently inbound):
    Wireworld Ultraviolet
    Audioquest Carbon
    Audioquest Pearl
    Amazon Basics
    Monoprice generic
     
  12. rlow

    rlow A happy woofer

    Friend
    Joined:
    Jul 18, 2017
    Likes Received:
    4,917
    Dislikes Received:
    5
    Trophy Points:
    113
    Location:
    Canada
    Loving this thread, esp since I'm evaluating USB cables right now (currently have a Curious USB "Regen Link", which I like, but it's too short for my new streamer.).

    Looking forward to the next round of tests. Thanks for this @atomicbob !
     
  13. Cspirou

    Cspirou They call me Sparky

    Friend
    Joined:
    Sep 27, 2015
    Likes Received:
    5,625
    Dislikes Received:
    5
    Trophy Points:
    113
    Location:
    Northwest France
  14. Wfojas

    Wfojas Friend

    Friend
    Joined:
    Sep 27, 2015
    Likes Received:
    944
    Dislikes Received:
    0
    Trophy Points:
    93
    Location:
    CA
    Do they sound different in at least one setup?
     
    julian67 likes this.
  15. skem

    skem Friend

    Friend
    Joined:
    Nov 1, 2017
    Likes Received:
    1,650
    Dislikes Received:
    17
    Trophy Points:
    93
    Location:
    People’s Republic of Cambridge
    @atomicbob. Thanks for these. How about measuring capacitance on the data line at 200MHz to estimate distortion?
     
    Josh358 and julian67 like this.
  16. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,447
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    Be glad to if you've got a Keysight E4982A to loan me. I can't justify buying one for the lab.

    All new cables are here. I've devised a new clamping system for immobilizing cable connectors during testing. May be able to find some time this coming week for more cable measurements. Too many tasks in the queue, so little time.
     
    Last edited: Feb 4, 2018
    Elnrik likes this.
  17. skem

    skem Friend

    Friend
    Joined:
    Nov 1, 2017
    Likes Received:
    1,650
    Dislikes Received:
    17
    Trophy Points:
    93
    Location:
    People’s Republic of Cambridge
    Does it have to be that model? ;-)

    I can give you LCR measurements to 100kHz using gear I have at home.

    I have the 1-meter Schiit Pyst. Here are the measurements at 100kHz on my cable, corrected for 0.91pf probe capacitance:
    Data-minus (pin 2) to shield = 189.47 ± 3.84 pf
    Data-plus (pin 3) to shield = 206.5 pf ± 1.5 pf
    Data-plus (pin 3) to ground (pin 4) = 70.82 ± 1.47pf
    Data-minus (pin 2) to +5V (pin 1) = 72.73 ± 1.50 pf

    Shield not connected to ground. Shield resistance = 0.07 Ω ± 0.01 (corrected for lead resistance). Note, this doesn't check against your resistance measurement all that well. I presume you are correcting for lead resistance? We care more about relative measurements than absolute -- but still.

    I may be able to do higher frequency measurements at work, but that will have to wait for a weekend and help from a friend. If you're interested in doing your other cables, PM me.
     
    Last edited: Feb 4, 2018
  18. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,447
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    Can you provide me with a mfg and model number that measures to 200 MHz (your request) that can be obtained for less than $2K? Most of the models that achieve 200 MHz I found are $16K ~ $20K. Not in my budget.

    Oh, if it is a matter of 100 KHz, then I can do that easily as can you.

    You can see from the pictures that I am capable of 4 wire measurements, however, given a first look at USB shields, I thought the relative information would be useful. No, I did not compensate for the 16 awg leadwires or the PCB breakout connectors. Also you are comparing 100 KHz measurements against DC resistance that I made. However, if you are going to be a strict stickler, I will repeat each measurement with a 4 wire system and an LCR to demonstrate the difference. But before I do, could you post a picture of your measurement setup, list of equipment used and method for consistent measurement over time? Would you also mind making DC resistance measurements for direct comparison?
     
    Grahad2, Elnrik and SSL like this.
  19. skem

    skem Friend

    Friend
    Joined:
    Nov 1, 2017
    Likes Received:
    1,650
    Dislikes Received:
    17
    Trophy Points:
    93
    Location:
    People’s Republic of Cambridge
    Whoaaa... Bob - None of that was meant as criticism! Don't take it so harshly, please. I was only trying to help out here.

    My resistance measurements were DC, which I think is good enough for the purpose of the shield. 200Mhz is only for the capacitance effect of the data line, and even then I think we can learn something at lower frequencies. Do you agree?

    I don't think you need a four-wire measurement --- the ranking is all that's important here. I pointed out the lead correction only because it might explain the factor of two difference. I don't suspect the 16awg copper as much as the PCB breakout boards as the source of additional resistance. My measurements were only 5 seconds and they are at the limit of the resolution of my Fluke 87V in high-res mode. Your Ohmmeter is 6.5 digits!, much better than what I have at home, and has better accuracy. Nevertheless, if both our meters are within spec, this alone doesn't explain the discrepancy, hence my question about the lead correction. In any case, I think a ranking is all that's really needed.

    On capacitance, I took my measurements with DER-EE DR-5000. Since you too can measure capacitance at 100Khz, you certainly don't need my help, but you can check my numbers. Also, on re-reading the spec sheet I see I mis-specified my errors on the capacitance readings, so I'll correct those now in an edit.

    Cheers
     
    Cryptowolf likes this.
  20. atomicbob

    atomicbob dScope Yoda

    Friend BWC MZR
    Joined:
    Sep 27, 2015
    Likes Received:
    13,447
    Dislikes Received:
    4
    Trophy Points:
    113
    Location:
    On planet
    My apologies, I've had several snipes in rapid succession across a few different (private) message systems and I'm a bit weary of people asking for moar (sic.) without considering that for which they are asking.

    In this case, 200 MHz measurements are a rather big deal. 200 KHz are relatively garden variety.

    Some things to contemplate. My measurements include a very important set of junctions. The USB connectors themselves. More about that in a moment.

    If you take short lead alligator clips from meter to A and B shell on the USB cable, indeed a very low resistance is found. Pyst measures 0.0576R on my 34461A. I obtain 0.079R on a BK Precision 878 @ 1KHz.

    Again, with alligator clips I measure 0.080R total R using LCR 878 and 0.0628R wiht 34461A, from A to B shell on my breakout fixture, with MDP connector shorted with 8 AWG banana shorting plug.

    But that isn't how the USB cables are used. They don't receive the benefit of alligator clips penetrating the metal surface. They are plugged into USB connectors found on the devices to be connected and thus have only a press fit. So my breakout PCBs are representative of those junctions. The junctions themselves contain a significant portion of the resistance and why @Priidik suggested immobilizing the cables during squeeze testing would better indicate the shield to shell connection inside the molded connector of the USB cables. I have built a fixture to address Pridik's concerns.
     
    Last edited: Feb 4, 2018

Share This Page