Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > sci.electronics.design > #446644 > unrolled thread

USB cabling constraints

Back to article view | Back to sci.electronics.design

Contents

  USB cabling constraints Don Y <blockedofcourse@foo.invalid> - 2017-02-07 08:43 -0700
    Re: USB cabling constraints amdx <nojunk@knology.net> - 2017-02-07 10:28 -0600
      Re: USB cabling constraints jrwalliker@gmail.com - 2017-02-07 08:54 -0800
      Re: USB cabling constraints rickman <gnuarm@gmail.com> - 2017-02-07 20:37 -0500
      Re: USB cabling constraints Don Y <blockedofcourse@foo.invalid> - 2017-02-07 20:23 -0700
      Re: USB cabling constraints Robert Baer <robertbaer@localnet.com> - 2017-02-08 21:29 -0800
        Re: USB cabling constraints Kamen Lilov <kamenl@gmail.com> - 2017-02-09 14:01 -0800
    Re: USB cabling constraints Jeff Liebermann <jeffl@cruzio.com> - 2017-02-07 10:39 -0800
    Re: USB cabling constraints "Tim Williams" <tiwill@seventransistorlabs.com> - 2017-02-07 19:06 -0600
      Re: USB cabling constraints Don Y <blockedofcourse@foo.invalid> - 2017-02-07 20:35 -0700
        Re: USB cabling constraints "Tim Williams" <tiwill@seventransistorlabs.com> - 2017-02-08 17:49 -0600
          Re: USB cabling constraints Don Y <blockedofcourse@foo.invalid> - 2017-02-08 20:33 -0700
            Re: USB cabling constraints "Tim Williams" <tiwill@seventransistorlabs.com> - 2017-02-09 16:43 -0600
              Re: USB cabling constraints Don Y <blockedofcourse@foo.invalid> - 2017-02-09 16:18 -0700

#446644 — USB cabling constraints

Are there any rules/guidelines for manufacturing/designing USB2
cable assemblies?

Compared to the constraints on network cables and SCSI cables,
are they somewhere in-between?  Microphone cables?

Or, more like ZIP cord for speakers?  :>

[toc] | [next] | [standalone]

#446651

On 2/7/2017 9:43 AM, Don Y wrote:
> Are there any rules/guidelines for manufacturing/designing USB2
> cable assemblies?
>
> Compared to the constraints on network cables and SCSI cables,
> are they somewhere in-between?  Microphone cables?
>
> Or, more like ZIP cord for speakers?  :>

  I don't know*, but I just bought a cheap usb to apple iPod connector 
cable. I hadn't charged the iPod for years, (it was one of my kids).
  I plugged it in and the battery charged, before I got to down load
and data I dropped it and the plastic shell came off of the connector.
  I notice the white wire had been heated to the point the wire melted 
out of the insulation, of the four strands! one was broken. OK it wasn't 
touching anything so, whatever. But then I noticed two of the wires were 
shorted together at the pins. Looked up the connector and the Data+ and 
Data- were solder together. So, I cleaned it all up and put it back 
together. Now it doesn't even recognize the usb has been plugged into 
the computer.

* Well I have some clue to your question, this cheap cable has no 
shield, the wire is smaller that a #40 and there are 4 strands.
I was not able to see any twist.
   Tested with ohmmeter and it all checks correct. But no long will 
charge the battery.

   A google search show, quality cables have a shield, the data pair are 
and power pair are twisted.

> https://www.google.com/search?q=usb+cable+diagrams&source=lnms&tbm=isch&sa=X&ved=0ahUKEwjEpN2Usv7RAhXJeSYKHakKDgsQ_AUICCgB&biw=1440&bih=687#imgrc=_



---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

[toc] | [prev] | [next] | [standalone]

#446653

On Tuesday, 7 February 2017 16:28:52 UTC, amdx  wrote:
> On 2/7/2017 9:43 AM, Don Y wrote:
> > Are there any rules/guidelines for manufacturing/designing USB2
> > cable assemblies?

How about the USB specifications?  There is a lot of information
there and the specs are free to download.

John

[toc] | [prev] | [next] | [standalone]

#446719

On 2/7/2017 11:28 AM, amdx wrote:
> On 2/7/2017 9:43 AM, Don Y wrote:
>> Are there any rules/guidelines for manufacturing/designing USB2
>> cable assemblies?
>>
>> Compared to the constraints on network cables and SCSI cables,
>> are they somewhere in-between?  Microphone cables?
>>
>> Or, more like ZIP cord for speakers?  :>
>
>  I don't know*, but I just bought a cheap usb to apple iPod connector
> cable. I hadn't charged the iPod for years, (it was one of my kids).
>  I plugged it in and the battery charged, before I got to down load
> and data I dropped it and the plastic shell came off of the connector.
>  I notice the white wire had been heated to the point the wire melted
> out of the insulation, of the four strands! one was broken. OK it wasn't
> touching anything so, whatever. But then I noticed two of the wires were
> shorted together at the pins. Looked up the connector and the Data+ and
> Data- were solder together. So, I cleaned it all up and put it back
> together. Now it doesn't even recognize the usb has been plugged into
> the computer.
>
> * Well I have some clue to your question, this cheap cable has no
> shield, the wire is smaller that a #40 and there are 4 strands.
> I was not able to see any twist.
>   Tested with ohmmeter and it all checks correct. But no long will
> charge the battery.

I don't know the exact rules in the specs, but there are different specs 
for different speeds.  USB 1.2 Mbps I don't think uses a shield.  Check 
if your mouse has a shielded cable, I bet not... assuming it has a cable 
at all.


>   A google search show, quality cables have a shield, the data pair are
> and power pair are twisted.
>
>> https://www.google.com/search?q=usb+cable+diagrams&source=lnms&tbm=isch&sa=X&ved=0ahUKEwjEpN2Usv7RAhXJeSYKHakKDgsQ_AUICCgB&biw=1440&bih=687#imgrc=_

I expect 12 Mbps should have a shield and 480 Mbps and 4 Gbps absolutely 
need a shield as well as impedance control.

-- 

Rick C

[toc] | [prev] | [next] | [standalone]

#446738

On 2/7/2017 9:28 AM, amdx wrote:
> On 2/7/2017 9:43 AM, Don Y wrote:
>> Are there any rules/guidelines for manufacturing/designing USB2 cable
>> assemblies?
>>
>> Compared to the constraints on network cables and SCSI cables, are they
>> somewhere in-between?  Microphone cables?
>>
>> Or, more like ZIP cord for speakers?  :>
>
> I don't know*, but I just bought a cheap usb to apple iPod connector cable.

Note that there are different types of "<favorite_device>" cables.
E.g., you can purchase cables that are only designed for charging
while others can also support data exchange.

I've encountered many folks who bought cheap *charging* cables
and wondered why they couldn't "talk" to their <device>.

> I hadn't charged the iPod for years, (it was one of my kids). I plugged it
> in and the battery charged, before I got to down load and data I dropped it
> and the plastic shell came off of the connector. I notice the white wire had
> been heated to the point the wire melted out of the insulation, of the four
> strands! one was broken. OK it wasn't touching anything so, whatever. But

Yeah, store-bought cables are hard to repair.  Like the old WE cables
that had a thin strip of copper *foil* wound around a nylon (?) core...
attempts at soldering would just result in a "plastic" glob and a
piece of foil that kept "shriveling"  away from the approaching hot
tip of the iron!

> then I noticed two of the wires were shorted together at the pins. Looked up
> the connector and the Data+ and Data- were solder together. So, I cleaned it
> all up and put it back together. Now it doesn't even recognize the usb has
> been plugged into the computer.

Try putting ~100-200 (no more!) ohms between the data pins and see if it
will again supply charging current.

> * Well I have some clue to your question, this cheap cable has no shield,
> the wire is smaller that a #40 and there are 4 strands. I was not able to
> see any twist. Tested with ohmmeter and it all checks correct. But no long
> will charge the battery.
>
> A google search show, quality cables have a shield, the data pair are and
> power pair are twisted.

Sorting out the electrical characteristics isn't a problem -- the Standard
addresses those.

But, that leaves a wide range of "valid" cable manufacturing techniques
that range from "gold standard" through "average", "adequate", "mediocre"
and "total shit"!  :-/

E.g., it's relatively easy to make a CAT5 cable that passes DC specs.
A wee bit harder -- but easy with damn near any amount of practice -- to
make one that will also pass AC specs and "perform well" (electrically).
Quite another challenge entirely to make one that can be (ab)used normally
AND still perform well!

A microphone cable requires a bit more care -- exposing the conductors
from *inside* the shield, connecting the shield (electrically), not melting
the insulation on the inner conductors as you solder the shield, crimping
the cable "bulk" to the connector (so it doesn't fail the first time the
performer removes the microphone from the stand and walks across stage
with it), etc.

At the farthest extreme, imagine running ZIP cord to a set of speakers.About 
the only real issue is making sure you've got the polarity correct -- and
that only begins to apply if you have more than *one* speaker!  Someone could
trip over the cord RIPPING it from the connections on either end and you
can have it back in place in seconds, "operating" normally!

*Most* store-bought USB cords will meet their stated purpose (e.g., if only
spec'd as a "charging cord" then it doesn't have to also support data
transfers to be considered "good"!).  But, *many* won't be "good
investments"  :-/

[toc] | [prev] | [next] | [standalone]

#446970

amdx wrote:
> On 2/7/2017 9:43 AM, Don Y wrote:
>> Are there any rules/guidelines for manufacturing/designing USB2
>> cable assemblies?
>>
>> Compared to the constraints on network cables and SCSI cables,
>> are they somewhere in-between? Microphone cables?
>>
>> Or, more like ZIP cord for speakers? :>
>
> I don't know*, but I just bought a cheap usb to apple iPod connector
> cable. I hadn't charged the iPod for years, (it was one of my kids).
> I plugged it in and the battery charged, before I got to down load
> and data I dropped it and the plastic shell came off of the connector.
> I notice the white wire had been heated to the point the wire melted out
> of the insulation, of the four strands! one was broken. OK it wasn't
> touching anything so, whatever. But then I noticed two of the wires were
> shorted together at the pins. Looked up the connector and the Data+ and
> Data- were solder together. So, I cleaned it all up and put it back
> together. Now it doesn't even recognize the usb has been plugged into
> the computer.
>
> * Well I have some clue to your question, this cheap cable has no
> shield, the wire is smaller that a #40 and there are 4 strands.
> I was not able to see any twist.
> Tested with ohmmeter and it all checks correct. But no long will charge
> the battery.
>
> A google search show, quality cables have a shield, the data pair are
> and power pair are twisted.
>
>> https://www.google.com/search?q=usb+cable+diagrams&source=lnms&tbm=isch&sa=X&ved=0ahUKEwjEpN2Usv7RAhXJeSYKHakKDgsQ_AUICCgB&biw=1440&bih=687#imgrc=_
>>
>
>
>
> ---
> This email has been checked for viruses by Avast antivirus software.
> https://www.avast.com/antivirus
>
   Doesn't the spec state a resistor is needed? R wired one way for 
USB1, another for USB2. Don't know for USB3.

[toc] | [prev] | [next] | [standalone]

#447078

On Thursday, February 9, 2017 at 7:29:29 AM UTC+2, Robert Baer wrote:
> amdx wrote:
> > On 2/7/2017 9:43 AM, Don Y wrote:
> >> Are there any rules/guidelines for manufacturing/designing USB2

>    Doesn't the spec state a resistor is needed? R wired one way for 
> USB1, another for USB2. Don't know for USB3.

No, no and no.

The spec on chargers says nothing about resistors.

This is why ultra-chic companies who are selling glamor, vapour and fruit-branded fashion accessories are allowed to run away with your money. Example: Apple phones have circuitry that measures the resistance of D+ D- lines by injecting a small current. "Secret" resistors are placed in the $60 charger, the phone inspects if these resistors are there, and if not, refuses to charge or charges with maybe 5 milliamps.

Of course, third-party charger manufacturers are not stupid. There are now single-chip (plus a few passives) solutions out there for chargers and charging stations that actually listen for the 'tones' the chargee is singing into D+/D-, deduce what the end device is, and imitate the appropriate D+/D- resistances so the device thinks it's got the original $60 vanity jewel.

Reminds me of the eternal copy protection / anti-copy-protection game for software from years ago.

[toc] | [prev] | [next] | [standalone]

#446666

On Tue, 7 Feb 2017 08:43:08 -0700, Don Y <blockedofcourse@foo.invalid>
wrote:

>Are there any rules/guidelines for manufacturing/designing USB2
>cable assemblies?

There are rules and guidelines for everything.  See the USB 2.0
Specification at:
<http://www.usb.org/developers/docs/usb20_docs/usb_20_011317.zip>
(56MB)
More:
<http://www.usb.org/developers/docs/>
<http://www.usb.org/developers/docs/usb20_docs/>

>Compared to the constraints on network cables and SCSI cables,
>are they somewhere in-between?  Microphone cables?
>Or, more like ZIP cord for speakers?  :>

Dunno.  I've never compared the wide variety of cables available with
the written USB specification.

-- 
Jeff Liebermann     jeffl@cruzio.com
150 Felker St #D    http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann     AE6KS    831-336-2558

[toc] | [prev] | [next] | [standalone]

#446713

The most often committed offense I've seen: incomplete shielding.

Avoid opening the shield at all costs.  Needless to say, a cable like the 
kind with a pin header on one end is useless -- these are most often seen 
between PC motherboard and enclosure, and work only because (and only when!) 
the enclosure is metal.

Connector housings are often taped over with copper or aluminum tape, which 
helps, but it's pretty cheap and unreliable.  (It can be hard to tell, 
because they're usually overmolded.)

Again, needless to say -- housings need to be the metallic shielded kind.

Really, this is just reiterating what's already in the standard, as Jeff 
linked -- but it's worth emphasizing signal quality, especially in regards 
to susceptibility, because this will be the single greatest cause of 
frustratingly intermittent problems, and it can be caused by something as 
passive as a cheap cable.

Tim

-- 
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Contract Design
Website: http://seventransistorlabs.com


"Don Y" <blockedofcourse@foo.invalid> wrote in message 
news:o7cpr0$nno$1@dont-email.me...
> Are there any rules/guidelines for manufacturing/designing USB2
> cable assemblies?
>
> Compared to the constraints on network cables and SCSI cables,
> are they somewhere in-between?  Microphone cables?
>
> Or, more like ZIP cord for speakers?  :> 

[toc] | [prev] | [next] | [standalone]

#446742

On 2/7/2017 6:06 PM, Tim Williams wrote:
> The most often committed offense I've seen: incomplete shielding.
>
> Avoid opening the shield at all costs.  Needless to say, a cable like the kind
> with a pin header on one end is useless -- these are most often seen between PC
> motherboard and enclosure, and work only because (and only when!) the enclosure
> is metal.
>
> Connector housings are often taped over with copper or aluminum tape, which
> helps, but it's pretty cheap and unreliable.  (It can be hard to tell, because
> they're usually overmolded.)
>
> Again, needless to say -- housings need to be the metallic shielded kind.
>
> Really, this is just reiterating what's already in the standard, as Jeff linked
> -- but it's worth emphasizing signal quality, especially in regards to
> susceptibility, because this will be the single greatest cause of frustratingly
> intermittent problems, and it can be caused by something as passive as a cheap
> cable.

My concern lies beyond the "electrical specs".  An "ethical" manufacturer
(i.e., someone who actually wants to make a cable that does what he claims:
"USB2 compliant") can easily meet those.

But, he can produce something that is total shit.  That falls apart during
"normal" (what's that?) use.  Or, whose performance degrades quickly
(shields opening up from flexure).  Or, outright failures without being
unduly stressed (connections opening/wires fraying from handling).

You can also disguise a crappy cable to make it appear to be well made
("obvious" strain reliefs don't say anything about the underlying
quality of the cable; flexibility approaching finely stranded Litz wire
doesn't mean there's actually a metallic conductor contained within; etc!)

I'm looking for guidelines to present to cable manufacturers to ensure
getting assemblies that "aren't shit" (yet aren't "gold standard")
once the "form" issues are sorted out.

E.g., cables intended to see frequent mating/demating would tend to
have more stringent tensile strength requirements than those that
remain /in situ/ for most of their functional lives.   Cables mated
by teenagers probably seeing more abuse than those mated by "professionals"
(upon which their livelihood depends)

[toc] | [prev] | [next] | [standalone]

#446912

"Don Y" <blockedofcourse@foo.invalid> wrote in message 
news:o7e3iu$3k3$1@dont-email.me...
> But, he can produce something that is total shit.  That falls apart during
> "normal" (what's that?) use.  Or, whose performance degrades quickly
> (shields opening up from flexure).  Or, outright failures without being
> unduly stressed (connections opening/wires fraying from handling).

Standards for all these things exist, along with the means to perform the 
tests.  Example: flexure test, consisting of wobbling the cable back and 
forth (do a test at the connector end, too, to see the efficacy of the 
strain relief and braid-to-housing join), for a suitable bend radius and 
number of cycles.  Qualification consists of DC checks (resistance per 
conductor, no shorts between wires), and perhaps a nice RF spot-check too 
(e.g., impedance within tolerance, shield attenuation >60dB, etc.).

I don't know offhand what is recommended, but I expect the USB standard has 
some preferences; if not (or, if not sufficient for your purposes), you can 
surely find better specs, all the way up to mil spec (which is free, also!), 
if you like. :)

Keep in mind that stacking specs will likely give you a higher cost.  You 
can just as well pass that off as in-house quality checks (even better! --  
but more expensive over time), or as qualification testing from a few 
different suppliers and sticking with one (presumably doing spot checks 
every so often to determine if they've cheaped their product on you, and 
it's time to search again).

Tim

-- 
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Contract Design
Website: http://seventransistorlabs.com 

[toc] | [prev] | [next] | [standalone]

#446955

On 2/8/2017 4:49 PM, Tim Williams wrote:
> "Don Y" <blockedofcourse@foo.invalid> wrote in message
> news:o7e3iu$3k3$1@dont-email.me...
>> But, he can produce something that is total shit.  That falls apart during
>> "normal" (what's that?) use.  Or, whose performance degrades quickly
>> (shields opening up from flexure).  Or, outright failures without being
>> unduly stressed (connections opening/wires fraying from handling).
>
> Standards for all these things exist, along with the means to perform the
> tests.  Example: flexure test, consisting of wobbling the cable back and forth
> (do a test at the connector end, too, to see the efficacy of the strain relief
> and braid-to-housing join), for a suitable bend radius and number of cycles.
> Qualification consists of DC checks (resistance per conductor, no shorts
> between wires), and perhaps a nice RF spot-check too (e.g., impedance within
> tolerance, shield attenuation >60dB, etc.).
>
> I don't know offhand what is recommended, but I expect the USB standard has
> some preferences; if not (or, if not sufficient for your purposes), you can
> surely find better specs, all the way up to mil spec (which is free, also!), if
> you like. :)

The standard spends a lot of time on electrical and mechanical details
(dimensionality) and a *wee* bit on suggested materials, etc.  Of course,
invariably qualified with "or better".

It addresses number of mating cycles (at a maximum frequency of occurrence),
connector retension force, etc.

But, it's also abundantly clear that "practice" deviates considerably from
"compliance".  E.g., many of the things EXPRESSLY FORBIDDEN by the standard
are easy to buy off-the-shelf.  Ooops!

Note the number of drek products out there that give the impression of being
"compatible" yet clearly aren't fulfilling even the most basic of the criteria
laid out (wire gauge, etc.)

And (more below), have you ever seen a cable that CLAIMED to embody the
standard -- AND NOTHING MORE?  (i.e., how else would you know the standard
was "sufficient" for your particular needs?)

> Keep in mind that stacking specs will likely give you a higher cost.  You can
> just as well pass that off as in-house quality checks (even better! --  but
> more expensive over time), or as qualification testing from a few different
> suppliers and sticking with one (presumably doing spot checks every so often to
> determine if they've cheaped their product on you, and it's time to search again).

A reply to the post in another forum suggested approaching cable manufacturers
for "samples" -- at a variety of quality levels/price points.  Then, testing
these for suitability.  Unfortunately, it would tend to require larger
populations than just "samples" to account for expected variation between
units.

So, I'll tackle it in two steps:
- samples to get a feel for the range of offerings (between vendors
   and within a vendor's capabilities matrix)
- sample *lots* for select vendors that can be more explicitly measured
- codify requirements based on an assessment of the characteristics of
   the "most appropriate" sample lots purchased
- develop a procedure to test/inspect production units to verify that
   compliance

<frown>  I guess its sort of like trying to evaluate the quality of a
*hammer*...

[toc] | [prev] | [next] | [standalone]

#447087

"Don Y" <blockedofcourse@foo.invalid> wrote in message 
news:o7gnq8$fg2$1@dont-email.me...
> <frown>  I guess its sort of like trying to evaluate the quality of a
> *hammer*...
>

Your only choice is to bang it around and see how it feels.  :^)

Tim

-- 
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Contract Design
Website: http://seventransistorlabs.com 

[toc] | [prev] | [next] | [standalone]

#447103

On 2/9/2017 3:43 PM, Tim Williams wrote:
> "Don Y" <blockedofcourse@foo.invalid> wrote in message
> news:o7gnq8$fg2$1@dont-email.me...
>> <frown>  I guess its sort of like trying to evaluate the quality of a
>> *hammer*...
>
> Your only choice is to bang it around and see how it feels.  :^)

For *consumers*, that's typically the only choice you have!
There are no "specifications" that you can claim the hammer
fails to meet.  Unless the failure is spectacular (or, justifies
litigation), your only recorse is to avoid that same *brand*
when purchasing your NEXT hammer!

OTOH, a hammer *manufacturer* can impose standards on *his* products
and test to verify "current production" is meeting those standards.

And, compare *his* products to those of his competitors (worst case:
by purchasing samples OTC and subjecting them to the same tests).

But, you'd want to have some idea as to the pertinent criteria that
are of concern to your consumers (as well as "Legal") so you could
gauge where you sit in the potential "quality" spectrum.

E.g., I suspect mikek (amdx) will be loath to purchase another cable
from that same vendor!  (*if* he recalls the vendor's name)

OTOH, if it had been prominently branded -- or otherwise memorably
associated with a particular vendor -- it becomes a reflection on
THAT vendor, thereafter.

(As I'm not interested in being remembered as a "bottom feeding" vendor
of schlock products, I'm concerned with sorting out where other "good"
products sit in the specification continuum)

[toc] | [prev] | [standalone]

Back to top | Article view | sci.electronics.design

csiph-web