Page 1 of 3

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 12:03 pm
by liz
Not the greatest news, but we're doing our very best to minimise any delay to you guys. I just want to reiterate that we are really very sorry; we know how impatiently you're all waiting.

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 12:10 pm
by teh_orph
How does the magnetic aspect work? Are all network connector magnetic? I thought the clip holds the cable in the socket?

Either way cool to see an x-ray of a part like that

(btw I can never post to the front page in any browser - it always moans about the copy/pasting of the password - any idea?)

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 12:15 pm
by steviewevie
Thanks for keeping us informed Liz. And thank goodness the problem was spotted before the boards shipped to customers !

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 12:20 pm
by mole125
Guess that means I can stop hitting the refresh button on my inbox waiting for that all important email from RS, any clues as to when I should start eagerly hitting it again?

At least the pictures and the new bit of knowledge make up for it.

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 12:20 pm
by Jongoleur
teh_orph said:


How does the magnetic aspect work? Are all network connector magnetic? I thought the clip holds the cable in the socket?

Either way cool to see an x-ray of a part like that

(btw I can never post to the front page in any browser - it always moans about the copy/pasting of the password - any idea?)



Not being a hardware bod, I don't know. 

I'm assuming that they're ferrite bead type thingies to suppress electrical interference.

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 12:23 pm
by plugwash
"Magnetics" in ethernet jargon reffers to the pulse transformers that sit between the ethernet chip and the ethernet socket. These transformers provide isolation to protect from voltage differences between ethernet devices and also block any DC bias currents from making their way onto the ethernet line.

In most PC network adaptors and switches i've seen these transformers are a seperate components but in space constrained embedded devices it is very common for them to be integrated in the network jack.

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 12:23 pm
by Burngate
Didn't you know carrier pigeons use the Earth's magnetic field to navigate?

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 12:30 pm
by RaTTuS

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 12:30 pm
by liz
teh_orph said:


How does the magnetic aspect work? Are all network connector magnetic? I thought the clip holds the cable in the socket?

Either way cool to see an x-ray of a part like that

(btw I can never post to the front page in any browser - it always moans about the copy/pasting of the password - any idea?)


Have you tried logging in with your forum name/pw? I need to look into an alternative to Spam Free WordPress when (ha!) I have a minute.

Lots of discussion in the comments thread about how the integrated magnetics work. I'll copy and paste what Gert and I have been saying there:

Me (when asked what, in that case, a jack without magnetics is useful for):





It doesn’t mean no network connection at all on all devices, but this board has been designed for a magnetic jack. The magnetic bits mean better signal integrity, better filtering and shorter transmission distances for data.






Gert:





Another very important feature a transformer adds is that you have no common ground. Long cables common ground: absolute disaster!






And Rurwin:





Magnetics refers to the presence of transformers and chokes which are used to isolate the Ethernet wires from the RaspPi’s power supply. and each other and probably to reduce high-frequency noise.
Without them you would effectively tie the RX and TX signals together and probably turn the entire network into an aerial for Radio 2 reception.





Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 12:32 pm
by Neil
For those wondering...

ALL RJ-45 Ethernet sockets are magnetically-coupled through transformers to the equipment they're connected to.  It gives galvanic isolation between the cable (and whatever is at the other end) and the device (Raspi, laptop, etc), blocks any DC that the cable might pick up from damaging the sensitive transceivers, and allows PoE to work, among other things.

Traditionally the transformers were seperate components - if you look on older network cards you can see little black boxes next to the RJ45 jack.  There are two transformers, one for Transmit and one for Receive, although they are typically in the same package.

For an example see Farnell part number 1636356.

Then the jack manufacturers got a clue and developed jack sockets with integrated transformers, or "magnetics".  A well-known brand is MagJack.

Check out Farnell part code 1572193.

Without the transformers the Ethernet just won't work.  Bit of a show-stopper.

Some have integrated passive components as well, some support Power-over-Ethernet, some have LEDs.  Lots of choices, but fundamentally they _all_ have isolating transformers.

Neil

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 12:33 pm
by nick.mccloud
We play a little cross-checking game in manufacturing to keep people on their toes so they don't spend half a day merrily soldering the wrong component in the wrong hole.* Fails at least once a month and takes just as long to get back on track.

That's assuming yours truly hasn't purchased a substitute part without checking the dimensions are the same.

You have my sympathy, it is very frustrating but it will get there.

How many are in this first batch that will be reworked and are they coming airfreight?

* Or is that the wrong component in the right hole or the right component in the wrong hole, we can never make up our minds.

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 12:51 pm
by teh_orph
Excellent, cheers for the info guys.

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 1:03 pm
by rurwin
And well done on noticing this before you had to decide whether to ship them back to China or do the rework yourselves.

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 1:14 pm
by texy
So did this happen to part of the initial 10k boards, or the batch after it?

Texy

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 1:17 pm
by liz
The whole first batch. And they're being airfreighted, hopefully at the start of next week.

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 1:28 pm
by SN
Well, sh*t happens.  Someone in China should be roundly b*ll*ocked and then we all move on.

Thanks for being so open and honest.

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 1:28 pm
by ukscone
liz said:


The whole first batch. And they're being airfreighted, hopefully at the start of next week.


These are going to be the most well traveled computers in history once they are back (again) They must have enough air miles by now for a free ticket to the bahamas

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 1:37 pm
by nick.mccloud
liz said:


The whole first batch. And they're being airfreighted, hopefully at the start of next week.


Once more you spoil our fun with facts

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 1:55 pm
by selectany
How long would be this potential delay?

A week or month?

I ask about that because my estimated delivery date is 16.04, but my card will be expired on 30.04 and I'm starting to worry.

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 2:15 pm
by hippy
SN said:


Well, sh*t happens.  Someone in China should be roundly b*ll*ocked and then we all move on.


Someone somewhere needs a rollocking because there ought to be processes in place to catch errors like that before 10K are built and need to be reworked. There's a moral somewhere in there about always building one exactly as the production line will and testing before committing to a full build.

But even with the best efforts in the world things can still go wrong, slip through the net. At least it is minor in the scheme of things, can be resolved, and doesn't mean binning the entire batch.

More lessons to be learned but it's still heading in the right direction.

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 2:22 pm
by jamesh
hippy said:


SN said:


Well, sh*t happens.  Someone in China should be roundly b*ll*ocked and then we all move on.


Someone somewhere needs a rollocking because there ought to be processes in place to catch errors like that before 10K are built and need to be reworked. There's a moral somewhere in there about always building one exactly as the production line will and testing before committing to a full build.

But even with the best efforts in the world things can still go wrong, slip through the net. At least it is minor in the scheme of things, can be resolved, and doesn't mean binning the entire batch.

More lessons to be learned but it's still heading in the right direction.


I believe to one off the line method was used....but Liz will probably correct me on that.

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 2:24 pm
by abishur
selectany said:


How long would be this potential delay?

A week or month?

I ask about that because my estimated delivery date is 16.04, but my card will be expired on 30.04 and I"m starting to worry.



I think that's firmly unknown.  I certainly understand your cause for getting antsy, but every time the RPF gives an estimate someone gets upset because they don't understand that an estimate is not a firm promise!

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 2:29 pm
by RaTTuS
it's just hit /. again - lets see how the forum copes now

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 2:40 pm
by liz
Jamesh is right - they sent us test units which *did* have the right part on before they moved to a larger batch.

Slashdot - well, there we go. It'll be a nice test of the new server setup.

Re: New post on main page: manufacturing hiccup. Please keep comments about it in this thread!

Posted: Thu Mar 08, 2012 3:07 pm
by crispyking
I don't get it.

The original manufacturer used the wrong part for the first batch, and now they have to be sent back for correction.

But "there may now be a slight delay in later batches if there’s a problem sourcing enough magnetic jacks".

I thought RS and Farnell were manufacturing the later batches. Have both of them messed up and sourced the wrong components as well?