The shell of a USB connector shouldn’t be grounded? I disagree with that, and it goes against everything I’ve read on the subject, eg:
http://www.ti.com/sc/docs/apps/msp/intr … mitest.pdf
All equipment I’ve ever seen with a USB port had the shell connected to the groundplane on the circuit board. On PCs, special effort is made to also bond the shell to the case (using conductive fingers etc) - for EMC reasons.
It’s actually in the USB spec. The shells and screen must be grounded at the PC end, of course
Leon
Interesting. It seems that no one follows the spec then!
It still surprises me that the spec would say this, as it goes against the EMI / EMC principle of bonding cable shields at both ends. The only reason I can think of not to bond the shield at both ends is to avoid ground loops (eg in some audio situations), but this seems to have become an outdated practice. The shielding effectiveness is much greater if the shield is bonded at both ends, particularly at higher frequencies.
I tried to find it again, but there are so many documents that it’s impossible. There are several references to it on the Web, here are a couple:
http://www.hardwarebook.info/Universal_ … _%28USB%29
http://www.allpinouts.org/index.php/USB … Serial_Bus
I always understood that shielding should only be connected at one end of a cable in most circumstances, BTW. It obviously doesn’t apply to RF signals and coax cable in 50 ohm systems.
Leon
ground the USB/RS232/data-cable shield at one end only… the source. :oops:
from computer to hub, grounded at computer.
from hub to device, ground at the hub. etc
To ground loop, or not to ground loop, that is the question… lol
This is my understanding as well. I've never grounded a shield either, but only because I never really thought about it. The mini-b footprints I've been using don't have a connection for the shield.leon_heller:
I always understood that shielding should only be connected at one end of a cable in most circumstances, BTW. It obviously doesn’t apply to RF signals and coax cable in 50 ohm systems.Leon
We know the shield going to the computer is grounded at the computer. That one is good.
Is the other plug shield grounded at the end device?
That is because it is an End-device plug, and assumed the shield is grounded at the other end of the cable. remember there is another ground wire inside the shield, (Signal return of sorts for the +V).TheDirty:
The mini-b footprints I’ve been using don’t have a connection for the shield.
Thanks for the links Leon - interesting…
As for ground loops, the USB protocol is quite capable of handling a bit of common-mode voltage (as it uses a differential signalling protocol). In most cases, there won’t be any ground loops anyway. The extra shielding effectiveness you get from connecting both ends can make the difference between passing EMI / EMC testing and failing - from memory it is more than a 20dB improvement in many cases.
Keith Armstrong has a lot of good stuff on this, and gives the strong impression that connecting only one end of the shield is an old-fashioned practice that has little use in today’s world of high-speed signals:
http://www.compliance-club.com/KeithArm … ?artid=453
(you have to register to view the articles, but it is well worth it & they don’t apam).
Keith Armstrong… That’s a lot of reading. What no PDF’s of the newer versions of the articles!? [EDIT; Were not dealing with ““Parallel Earth Conductor” (PEC) with a lower impedance than the shields”, perfectly balanced differential signal imp at both ends, and 50/60hz interference. Good articles, and very good points.](rathole). It was just something to think about for keeping RFI emotions at bay.
One more thing… I don’t intend to cause panic and mas hysteria, just a simple question. Are you sure the traces are going to the correct pins? I think originally the project worked with the internal clock, and not the external crystal-thing.
You guys are awesome, thank you for the tips! MichaelN, I will definitely try the smaller parts eventually. I’d love to try the skillet or hot air gun but I’m afraid I can’t right now, I’m sure my fiancee would not approve! 
I may be stretching it with even the soldering iron and the tiny electronic parts everywhere. She didnt know what she was getting into… LOL!
zarcondeegrissom, awesome tip! I will see if I can move that trace and re-post.
I read that your dealing without a solder mask, so I understand the “Floating” copper areas between stuff. I’d try to find a way to ground them with more clearance, or remove the free floating tiny antennas/tuning forks. example under the “0.1” label for C3 I think.
I think BatchPCB does provide a soldermask, but I would definitely rather be safe than sorry. You guys are the experts so I’m glad to have all the advice I can get.
The part provided in Eagle for the mini usb didnt have a connection to ground for the case, but I’m sure that’s not necessarily proof that you don’t have to ground it.
To ground loop, or not to ground loop, that is the question… lol
Forgive my ignorance, but what do you guys mean by “ground loop” used as a verb? :oops: The one time I encountered that term was when I installed an aux input in my car stereo system. When I try to charge it at the same time as I play it through the aux in, I get noise. From what I understood their grounds are at different potentials, causing extra noise in the system. You could use an isolation transformer to eliminate this problem. I think this was a 1:1 transformer that decoupled the aux input.
I looked it up here:
http://en.wikipedia.org/wiki/Ground_loop_(electricity)
This isolation transformer also mentions ground loops:
http://www.jaycar.com.au/productView.asp?ID=AA3085
Sorry if this sounds like a stupid question, just trying to learn.
Also: I partially populated the second copy of the original circuit board i had created, but this time added a 33uF decoupling capacitor between Vdd and Vss, and the USB seems to work fine now. As someone had said earlier, even though it was suboptimal it should still work… well it does seem to work. Haven’t added back in the FT232RL yet though; hopefully that will not cause any problems.
I’ve never heard “ground loop” used as a verb before, but I think what he’s saying is that there are trade-offs to be made when deciding whether to connect the shield at both ends.
I have noticed that most modern AV gear (DVD players, amplifiers etc) seems to connect the shield of audio and video cables to the chassis of the equipment, which gives much better shielding than only connecting it at one end. To remove the ground loop, this equipment is normally “double insulated” - ie the metal chassis is NOT connected to mains earth. Seems like a safety issue to me, but most companies seem to do it this way…
Differential signals (USB, Ethernet etc), are tolerant to the common mode voltage on the signal, and the benefit of improved EMC performance will often outweigh the disadvantages of common mode voltage / currents in the shield.
To ground loop, or not to ground loop, that is the question… lol
Yes MichaelN, and as a little bit of a Shakespearean like joke.
Ground loops are just something to keep in the back of your mind, when dealing with higher then audio frequency stuff. Wide loops, in the shape of open center rectangles/squares, ovals, or circles, make ring/loop antennas. think of that round metal thing sticking out of the back of some of the old TV’s between the bunny-ear antennas.
Thanks again, all… I got most of the original circuit board running, except for the FT232RL-- which I [wrongfully] assumed was a no-brainer. I thought it should be just as easy to use than a MAX232, which is what I was using on my breadboard. TX->RX, and RX->TX… Not even a level conversion. How much easier could you get? I’m confounded. Basically followed the data sheet.
The computer recognized it but for some reason it kept freezing the microcontroller upon transmit/receive.
It may have very well been the firmware-- I’m not sure-- but I think before I send in this design, I might home etch a breakout board for the FT232RL to work out my issues.
Thanks again, you guys are geniuses! You’ve helped me identify some good things to look for when I do send in my design
Update: it was my firmware after all. I should have spent more time with the software… Instead I destroyed the prototype boards I had from over working them. :lol:
In any case, I’ve sent in the designs for the new board you guys helped me with. Hopefully I’ll have the boards shortly.
Thanks again for all your help.
Steve
–
HI russo, its 2012 long since this post but now I am having your problem with the oscillator, did you solve your problem? did your circuit work good with your new PCB design?
thanks.
lol Move the crystal closer. CLOSER!
I would us an SMD Crystal w/caps, much easier, and much nicer looking. Plus you dont need to add the caps.
Add two FIDUCIALs on kitty corners.
Library: SparkFun
Package: FIDUCIAL-1X2.5
Fix all “top left” of the tracks leading to acute angles.
Add another .1uF to the bottom/right of the micro, VCC.
@dap20 - I think the problems were due to my firmware. The new design has been working well for the last 2 1/2 years… just used it last night.
bamboosamurai:
lol Move the crystal closer. CLOSER!
I would us an SMD Crystal w/caps, much easier, and much nicer looking. Plus you dont need to add the caps.
Add two FIDUCIALs on kitty corners.
Library: SparkFun
Package: FIDUCIAL-1X2.5
Fix all “top left” of the tracks leading to acute angles.
Add another .1uF to the bottom/right of the micro, VCC.
I haven’t had time for any of this stuff for probably a year and a half now… thanks for the tips though! If I ever get time to get back to electronics I will definitely take this into account.