Author Topic: On device's side, should I connect USB cable shielding to the black wire?  (Read 1733 times)

0 Members and 2 Guests are viewing this topic.

Offline dusanTopic starter

  • Regular Contributor
  • *
  • Posts: 78
  • Country: sk
I'm using CH340N and stm32f103 blue pill. I used USB cable and connected red to vcc, green to d+, white to d- and black to gnd. Everything works fine. But the cable also have shielding. Should I connect the breadded shielding to black wire?
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21962
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Do you have a metal or shielded enclosure? Are there any other connections to your board?

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline dusanTopic starter

  • Regular Contributor
  • *
  • Posts: 78
  • Country: sk
No metal enclosure. STM32 has it's own power. Grounds are connected between STM32 and CH340N.
 

Offline Solder_Junkie

  • Frequent Contributor
  • **
  • Posts: 364
  • Country: gb
In general, power and shielding should be kept separate. It amazes me how often the metal shell of connectors is not connected to the screen of cables, not helped by the lack of solder tag on many plugs/sockets.

SJ
 

Online wraper

  • Supporter
  • ****
  • Posts: 17253
  • Country: lv
In most of micro USB cables shield is connected to GND wire internally. The only case when you should even consider not connecting shield to GND plane is when connector shell is connected to metal enclosure that by itself has connection to GND. USB type C specification explicitly demands connector shell to be connected to GND plane.
« Last Edit: May 21, 2024, 10:25:33 am by wraper »
 
The following users thanked this post: T3sl4co1l

Offline tooki

  • Super Contributor
  • ***
  • Posts: 12021
  • Country: ch
We had this question recently, but specifically for USB-C: https://www.eevblog.com/forum/projects/why-usb-c-gnd-is-being-connected-to-the-shield-of-the-cable-after-connecting/

The answer there is unequivocal: the USB-C standard states
Quote
6. Shield and GND grounds shall be connected within the USB Type-C plug on both ends of the cable assembly.
11. The receptacle shell shall be connected to the PCB ground plane.

I haven't checked the USB 2.0 spec, but I bet it says something similar.
 

Offline radiolistener

  • Super Contributor
  • ***
  • Posts: 3567
  • Country: ua
In order to avoid EMI and common mode currents it's better to connect cable shield to GND from one side only. On second side of cable the shield should not be connected anywhere. This is because when the shield on both side of cable is connected to GND it can lead to ground loops which can receive RFI and can leads to common mode currents.

In your case, most of all, the cable shield is already connected to GND on USB connector side, so there is no need to connect cable shield anywhere on MCU side.
« Last Edit: May 21, 2024, 05:35:39 pm by radiolistener »
 

Online wraper

  • Supporter
  • ****
  • Posts: 17253
  • Country: lv
In order to avoid EMI and common mode currents it's better to connect cable shield to GND from one side only.
:palm: If your goal is to fail EMI/EMC tests as well as ESD immunity, not to say ensure unstable communication in presence of transients, then sure.
Quote
This is because when the shield on both side of cable is connected to GND it can lead to ground loops
One of the dumbest arguments I often see about this topic. There is GND wire anyway, so you screw up shielding for nothing in this regard.
 
The following users thanked this post: T3sl4co1l, tooki

Offline radiolistener

  • Super Contributor
  • ***
  • Posts: 3567
  • Country: ua
:palm: If your goal is to fail EMI/EMC tests as well as ESD immunity, not to say ensure unstable communication in presence of transients, then sure.
There is GND wire anyway, so you screw up shielding for nothing in this regard.

It don't means that there is no ground loop effects. Yes, there is GND wire everywhere and it leads to issues.

Regarding to the single point grounding scheme, I'm agree that it may not work for all cases, it depends on specific case and environment.

But it don't means that this is dumbest argument. For topic starter case I prefer to not connect the shield to GND on MCU side.

For example, here is Analog Devices Application Note, about it:
https://www.analog.com/media/en/technical-documentation/application-notes/41727248AN_347.pdf

« Last Edit: May 21, 2024, 07:11:00 pm by radiolistener »
 

Online wraper

  • Supporter
  • ****
  • Posts: 17253
  • Country: lv
For example, here is Analog Devices Application Note, about it:
https://www.analog.com/media/en/technical-documentation/application-notes/41727248AN_347.pdf
Don't you think it's a bit too outdated and does not apply for high speed digital interfaces?
« Last Edit: May 21, 2024, 07:45:35 pm by wraper »
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21962
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
In order to avoid EMI and common mode currents it's better to connect cable shield to GND from one side only. On second side of cable the shield should not be connected anywhere. This is because when the shield on both side of cable is connected to GND it can lead to ground loops which can receive RFI and can leads to common mode currents.

In your case, most of all, the cable shield is already connected to GND on USB connector side, so there is no need to connect cable shield anywhere on MCU side.

I wonder, do any of the radios you listen to, also cut grounds in the coax cables carrying RF power between sections? How about the antenna, do you ever cut the ground there?

If not, could you explain how a coax cable differs from a USB 2.0 cable?

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline radiolistener

  • Super Contributor
  • ***
  • Posts: 3567
  • Country: ua
Don't you think it's a bit too outdated and does not apply for high speed digital interfaces?

No, I don't think its outdated. And it is still applied for high speed interfaces, but high speed interface needs twisted pair instead of single wire in order to keep fixed impedance along transmission line. This is the OP case where cable has twisted pair and shield.

I wonder, do any of the radios you listen to, also cut grounds in the coax cables carrying RF power between sections? How about the antenna, do you ever cut the ground there?

Note that topic starter talking not about coax line. He is talking about twisted pair which has shield around. This is completely different thing.

In coax cable signal is carried in the insulator between center conductor and the braid. If you don't connect braid on one side it will not work and will have impedance issues.

In case of topic starter case, his cable already has a ground wire in twisted pair and it is already connected on both sides. He is asking about shield connection. Not signal ground wire.
« Last Edit: May 21, 2024, 08:19:00 pm by radiolistener »
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21962
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Note that topic starter talking not about coax line. He is talking about twisted pair which has shield around. This is completely different thing.

In coax cable signal is carried in the insulator between center conductor and the braid. If you don't connect braid on one side it will not work.

In case of topic starter it's cable already has a ground wire in twisted pair and it is already connected on both sides. He is asking about shield connection. Not signal ground wire.

Ok, cool, good.  So, about that twisted pair.  Several things:
1. What is the ground wire referenced to / paired with?
(A: usually VCC; most USB cables I've taken apart, are double twisted pair, with foil on the data pair.)
2a. Is USB fully differential? If it depends, then also, when?
2a(i). If so/when, what is USB's common-mode range? How does it compare with commercial immunity levels?
2b. Is the power fully differential? (An odd way to put it, but more to the point: would there ever be a situation where we would worry about common-mode currents on the power pair? When, generally, are we ever concerned about that on a power line?)
(A: you've already implied that GND could be used as a signal reference, so the answer is no!)
3. If the answer to at least one of (2) is no, what solution would you suggest for it? (implied: a solid shield most likely, but on what justification; what possible alternatives?)

You may have to dig deep for some of these answers; many professionals get it wrong!

Tim
« Last Edit: May 21, 2024, 08:26:02 pm by T3sl4co1l »
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline radiolistener

  • Super Contributor
  • ***
  • Posts: 3567
  • Country: ua
You may have to dig deep for some of these answers; many professionals get it wrong!

If we dig deep, both variants - RF and LF shielding (connect shield to ground on two ends and on one end) has its pros and cons.
That's why there is twisted pair, it allows to eliminate ground loop issues by using single side shield grounding and keep high speed signal protected from RFI.

Instead of twisted pair you can use coax cable inside shield, you can find such approach in sensitive instruments which uses coax cable with double braid (triaxial cable). The outer braid is used as a shield and inner braid is used as signal ground. In this case shield is connected to ground at one side (at instrument side) and is open at other side.

The RFI is received/radiated at point where characteristic impedance of transmission line is changed. But twisted pair has constant impedance along its length, so it won't receive/radiate RFI. And single side grounded shield around twisted pair allows to eliminate ground loops.


2b. Is the power fully differential? (An odd way to put it, but more to the point: would there ever be a situation where we would worry about common-mode currents on the power pair? When, generally, are we ever concerned about that on a power line?)
(A: you've already implied that GND could be used as a signal reference, so the answer is no!)
3. If the answer to at least one of (2) is no, what solution would you suggest for it? (implied: a solid shield most likely, but on what justification; what possible alternatives?)

Well, it is possible that USB cable don't have shield at all or it may use double shield for LF and RF shielding.
So it all depends on the exact case. There is no universal solution.
« Last Edit: May 21, 2024, 09:37:53 pm by radiolistener »
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21962
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
If we dig deep, both variants - RF and LF shielding (connect shield to ground on two ends and on one end) has its pros and cons.
That's why there is twisted pair, it allows to eliminate ground loop issues by using single side shield grounding and keep high speed signal protected from RFI.
(...)

No no no, not so fast -- I need you to check your assumptions, do actual research.  It's not enough to rattle off generics.  This is a specific case, in a specific implementation.  No wishy-washy needed!  I chose the list of questions very particularly, and they are objectively answerable.

The USB standard is open and available:
https://www.usb.org/document-library/usb-20-specification
crack open that ZIP and grab the "usb_20.pdf" and start perusing.

USB 3.0+ of course adds more, but since we're talking the four-wire shielded case, this covers everything.

If you would like a more concrete example, let's say we have...
- A desktop PC (metal enclosure and all that)
- 2m long USB cable to an instrument
- Instrument consists of a plastic enclosure, PCB with ground plane design, mains power (a more-or-less normal off-the-shelf isolated SMPS, with adequate filtering to meet EMC otherwise), general construction is "double insulated" (so that grounding through device connections doesn't matter), and, let's say it's a, like, USB-connected SMU or something, so there is a wired connection doing something useful.
- And let's say the wired connection is 2m long, and the device under test is grounded.
--> So, PC is earthed through its cord, and EUT is grounded through DUT (bonded) and USB (to PC).  Only mains ports are (galvanically) isolated.
- And, this is all within a semi-anechoic test chamber, so when I say "grounded", I mean the DUT is bonded to that, and the PC and instrument are elevated on an insulated table, as is standard, let's say as in CISPR 22.
- And let's say both PC and EUT are plugged into the wall via separate LISNs each.
I could go on to describe (or provide crude drawings for) cable placement and construction, relevant line filter sections (if mains circuit impedance should be relevant), USB cable construction (well heck, let's just put that in real quick anyway: say it's outer braid 90% coverage, power pair twisted, foil over data pair twisted; braid serves as drain wire for foil), etc., but at this point, you get the idea, and this should be sufficiently well constrained that anything left undefined, you can freely make assumptions about -- whether extrapolating from common practice, or to try and game a solution.

And we assess EMC in this setup by putting current clamps or CDNs on all cables, as well as measuring the LISNs, to cover conducted band; and there's an antenna in the usual configuration (variable altitude, both polarizations) to cover the radiated band.


You should be able to answer the numbered questions; if not, let it be known in this thread that your answers on this topic are not authoritative.  But so, too, you have opportunity to improve your craft, to better your radio skills, to diversify on topics (how exciting!).

It is with regret that I do this confrontationally; I realize such approach has a strong chance of simply shutting down subsequent response, rather than generating the nuanced analysis that is required on this topic, and for this I apologize.  I hope however that this leads to fruitful discussion, and learning experiences (not just you but readers too).  These really are tough topics, and one must approach them with care and humility; it is only from decades of experience that I would bring such confidence to bear here -- and even so, I would still be ready for surprises in testing real hardware of this general description.  I have a pretty good idea of what would happen here, what approximate levels such a system might pass or fail by (but I wouldn't dare stake my reputation on anything more precise than 10dB or thereabouts--), and what ways it might fail in; but also be open to surprises.  After all, it takes only one little gap to emit a couple millivolts of RF and spoil the whole thing.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: shabaz, wraper

Offline shabaz

  • Regular Contributor
  • *
  • Posts: 245
...
If you would like a more concrete example...

Hi Tim, This seems a great example for learning in-depth (I'm a novice with EMC testing). I tried to sketch the example you mention in case it helps the future discussion. Let me know if I've made a mistake or any other detail is worth adding to it, and I can edit it (the attached zip file contains the PowerPoint source of the diagram for anyone to edit too).

EDIT: Updated with the comment below.
« Last Edit: May 22, 2024, 02:18:37 am by shabaz »
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21962
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Oh thanks!  Very close; I might not've been very clear about terminology, but "DUT" would be the box on the left, and EUT the USB-connected unit.

Probably DUT would be on the ground (hence better bonding; floor is metal, and "bonding" can simply be metal [EMI] tape), and wouldn't have a LISN itself.  No other wires besides the ground connection and the +/- from the EUT.  We could also have it where there's a ground/earth wire to the negative terminal; it shouldn't be so material where the box sits along that route (between EUT and ground), as long as it's relatively small compared to wavelengths of interest.

The SMPSU would have the other LISN.

A LISN isn't really an inline block, but a network that sits on the ground (bonded in the usual way) and serves to separate the test environment from the outside mains network.  I'd be inclined to draw it as a bulkhead unit on the right-side wall, or showing the connections going down to the floor (LISN unit bonded to GND) then back up.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: shabaz

Offline shabaz

  • Regular Contributor
  • *
  • Posts: 245
Thanks for the quick feedback! I've updated the diagram in the earlier post based on your comments (I'm happy to continue refining it further since I may still have misunderstood or missed things).
 

Offline tooki

  • Super Contributor
  • ***
  • Posts: 12021
  • Country: ch
Well, it is possible that USB cable don't have shield at all or it may use double shield for LF and RF shielding.
IIRC, the only variant of USB that allows unshielded cables is USB 1.1/2.0 Low Speed, the 1.5Mbps mode used for input devices like mice and keyboards. (I don’t remember if the standards allow power-only cables at all, but I suppose those could be unshielded.) Everywhere else, shielding is mandatory, and its implementation is prescribed by the standards, which you’ve been provided the link for.

 
So it all depends on the exact case. There is no universal solution.
Which is why people were debating your absolutist proclamations about shields needing to be connected at one end only. Because that is true, except all the times when it’s not.
 

Online wraper

  • Supporter
  • ****
  • Posts: 17253
  • Country: lv
Well, it is possible that USB cable don't have shield at all or it may use double shield for LF and RF shielding.
IIRC, the only variant of USB that allows unshielded cables is USB 1.1/2.0 Low Speed, the 1.5Mbps mode used for input devices like mice and keyboards. (I don’t remember if the standards allow power-only cables at all, but I suppose those could be unshielded.) Everywhere else, shielding is mandatory, and its implementation is prescribed by the standards, which you’ve been provided the link for.
In USB 2.0 specification USB-IF made it vague stating:
Quote
6.8 USB Grounding
The shield must be terminated to the connector plug for completed assemblies. The shield and chassis are
bonded together. The user selected grounding scheme for USB devices, and cables must be consistent with
accepted industry practices and regulatory agency standards for safety and EMI/ESD/RFI.
And all shit broke loose. Some app notes adding random components between connector shell and GND without mentioning that
Quote
The shield and chassis are bonded together
USB type C spec made clear that no such wankery is allowed.
 

Offline Doctorandus_P

  • Super Contributor
  • ***
  • Posts: 3543
  • Country: nl
2a. Is USB fully differential? If it depends, then also, when?

USB is not fully differential (at least USB1 and USB2). Data is sent differential, but it has some weird things (I think they are called "T-states") that are used to signal special conditions on the USB bus. In this regard it somewhat resembles the start and stop conditions on an I2C bus.

For the rest, I have always been confused of how to handle GND versus shield in cables.
 

Offline perieanuo

  • Frequent Contributor
  • **
  • Posts: 869
  • Country: fr
no. if you have at some moment a bad power supply on pc/laptop side and the target is not floating, you're gonna burn the laptop. i changed southbridge on some laptops with this type of connection.
 

Offline m k

  • Super Contributor
  • ***
  • Posts: 2214
  • Country: fi
Maybe collection of names can help.

Earth
Neutral
Ground
Shield
Chassis

Hmm, nope, didn't do much.
Advance-Aneng-Appa-AVO-Beckman-Danbridge-Data Tech-Fluke-General Radio-H. W. Sullivan-Heathkit-HP-Kaise-Kyoritsu-Leeds & Northrup-Mastech-REO-Simpson-Sinclair-Tektronix-Tokyo Rikosha-Topward-Triplett-YFE
(plus lesser brands from the work shop of the world)
 

Online wraper

  • Supporter
  • ****
  • Posts: 17253
  • Country: lv
no. if you have at some moment a bad power supply on pc/laptop side and the target is not floating, you're gonna burn the laptop. i changed southbridge on some laptops with this type of connection.
Make it make sense. Shield making connection between the grounds first/disconnecting the last is exactly what will protect internal pins if there is potential difference between the grounds of the devices.
 

Offline radiolistener

  • Super Contributor
  • ***
  • Posts: 3567
  • Country: ua
If you would like a more concrete example, let's say we have...

Even your concrete example may have hidden details which may lead to different results and as you said you're understand it and ready for surprises despite the fact that you had good experience with it.

If you have some testing results for specific case it will may be useful to publish and analyze for better understanding what values can be expected with at specific conditions. But abstract discussion about specific cases with no real test results and measurements are useless from my point of view.

I'm not EMI QA guy, but I often catch EMI issues due to ground loops with USB, and shielded Ethernet (CAT6) cables and I understand that this question is not easy and has many hidden issues.

I'm not surprised when I see that one cable don't produce EMI and another produce it despite the fact that the second one has shielding and the first one don't have shielding at all (unshielded twisted pairs)... The difference in noise floor at 100-200 MHz is about 5-10 dB. I have example of such cables and can reproduce it. And it can be reproduced on certified equipment.
« Last Edit: May 22, 2024, 11:00:55 am by radiolistener »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf