Raspberry Pi Production Delayed By Factory's Assembly Flub 132
nk497 writes "The first shipment of Raspberry Pi devices has been delayed, after the factory manufacturing the cheap educational computer used non-magnetic jacks instead of ones with integrated magnetics. The problem is already nearly fixed, but new jacks need to be sourced for subsequent shipments, so those could be delayed slightly. 'It's inevitable, isn't it — you're freewheeling along perfectly happily and then you get a puncture,' said spokeswoman Liz Upton, apologizing for the delay."
Why the magnetics? (Score:5, Interesting)
Re:Why the magnetics? (Score:5, Informative)
I had the exact same thought and googled magnetic jacks:
Molex Magnetic Modular Jacks incorporate wire-wound components (magnetics) in standard RJ45 jacks. These integrated magnetics, resistors and/or capacitors filter common-mode noise to provide signal integrity, protect PHY chips, provide DC isolation and offer low-mode conversion.
I'm assuming that's the case here, and the magnets are providing filtering (given the cable's got a predominantly plastic and copper end it's not going to do much to hold it in place).
Re:Why the magnetics? (Score:5, Informative)
I'm assuming that's the case here, and the magnets are providing filtering (given the cable's got a predominantly plastic and copper end it's not going to do much to hold it in place).
They're not magnets. They're tiny transformers and inductors that magnetically couple the signals while providing 1.5 kV DC isolation and some filtering against common-mode disturbances.
Re: (Score:2)
Ah yes, that makes much more sense. Magnetics isn't a term I knew.
Re: (Score:3)
isolation transfomers.
had they said that, yeah, you want that in ethernet. products that omit that (popcorn hour, cough cough) have ruined NICs and bad performance for noise.
in audio (spdif) you also want pulse transformers. same idea.
Re: (Score:2)
They're not magnets. They're tiny transformers and inductors that magnetically couple the signals while providing 1.5 kV DC isolation and some filtering against common-mode disturbances.
Ahhh! ELECTRO-Magnetics. I wonder why that got left off.
Re: (Score:2)
Re: (Score:1)
Re: (Score:2)
Conditions the signal, like the little donuts on monitor cables.
Re: (Score:3)
Re: (Score:2)
You are right - I probably chose a poor example since the donuts aren't magnets. Still, like the cable ferrites, the magnets in the connectors condition the signal off-board to free up space.
Re:Why the magnetics? (Score:5, Informative)
As I stated in the other, non-annointed article and posted here [slashdot.org], the magnetics are actually tiny transformers used to convert from differential to single-ended signals and to isolate. Additionally center taps can be used for PoE.
"The magnetics in question aren't to hold the connector in like those in a Mac power cord, but rather the tiny transformers that are required for Ethernet differential signal isolation/transformation."
Re: (Score:2)
I feel a bit foolish now.
Re: (Score:2)
I have, thank you, and the transformer can be used to do exactly what I said as well as for pure isolation with the resultant signals retaining their differential status. Please consider that there are other Ethernet circuits than the one(s) you're familiar with before making general statements. Some really do need single-ended signals into the PHY.
Re: (Score:2)
And finally also to remove DC components, many PHYs have a constant DC component on their output lines because it's easier to vary the output currents while keeping it in the same direction than it is to change the direction of a current while maintaining smoothness.
Re: (Score:2)
Re: (Score:3)
Re: (Score:2)
Some controller IC's require magnetics on the PHY side in order to work at all. Others will work without magnetics, but only if short cables are used.
If the Raspberry Pi circuit was designed such that the magnetics are responsible for 'pulling up' the output stages to the positive supply, then the Ethernet function simply would not have worked at all without magnetics in the connectors.
BTW, 'magnetics' is a very common term in the field of electronics. It typically applies to inductors, transformers, and fe
Re: (Score:3, Informative)
also see http://www.molex.com/mx_upload/family/magnetic_modular_jacks/hypejak.pdf [molex.com]
Re: (Score:1)
Re: (Score:2)
Does it not need the little tab anymore (that always breaks off)?
That's someothing that's always bugged me. Of course you're going to need mini or submini jacks in a very small device like a transistor radio or an iPod, but I could never figure out why they didn't use RCA jacks in PCs. Not that RCA jacks don't sometimes fail. Heck, they could have used 1/4 inch jacks in PCs, I've yet to see one of those fail, even in a heavily used environment like a guitar amplifier.
Re: (Score:2)
The big problem with RCA jacks is that they are not keyed. You can insert any RCA jack into any RCA Jill. Sometimes that's not an issue, others, well, good designers know that it's best not to let end users think much.
Inductors? (Score:3)
I have only dabbled in electronics, but I don't care for the term "magnetics" (nor I have ever heard it before). I would think "inductor" would be a better term. Anyone who actually knows what they're talking about care to illuminate what the difference is between "magnetics" and inductors?
Re: (Score:2)
Magnetics != Magnets. These are the Ethernet transformer coils that are missing.
Re: (Score:2)
Magnetics (Score:5, Informative)
From the Raspi forums :
"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."
"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: (Score:1)
"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."
I have no idea what this means. How is this possible?
Re: (Score:2, Troll)
I have no idea what this means. How is this possible?
I guess because you didn't study properly at school... ;-)
"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."
I disagree with the first part of this. Ethernet works off differential signals and the transformer does a good job of removing the common mode signals (non differential) coming down the wire. Without the magnetics the common mode signals (such as DC, mains interference, Radio 2 transmissions) picked up by a potentially long ethernet lead will turn up at the input to the ethernet phy (receiver) chip. This may or may not be enough to stop it from work
Re: (Score:1)
Its a joke. I think the fault makes the whole ethernet cable into, effectively, a dangling wire, and the only use of a dangling wire is as an aerial. Possibly connecting tx and rx makes it into a loop of sorts, and thus a loop antenna.
Re:Magnetics (Score:5, Informative)
Without the intervening transformer, the TX- and RX- lines would be tied together at ground on the device. In the diagram below, the differential RX+/- and TX+/- signals are turned into single-ended RX and TX by the transformers. Removing the transformers connects RX- and TX- to ground, which is a Bad Thing(tm).
RX+_____3 E_______RX
3 E
3 E
RX-_____3 E_____GND
TX+_____3 E_______TX
3 E
3 E
TX-_____3 E_____GND
Re: (Score:3, Informative)
Radio 2: UK radio station broadcasting crap music for housewives.
Re: (Score:2)
RX is recieve, TX is transmit.
Re: (Score:1)
At least the forums are holding up now with new hardware ;-p
Re: (Score:3)
They also crucially provide center taps to provide a path for DC currents from the transmitter to ground and to give the receive signal the correct common mode level. Also because of the aforementioned center taps the pinouts of a jack with integrated magnetics will almost certainly differer from a plain jack.
http://www.smsc.com/media/Downloads_Public/lan9000/9512_sch.pdf [smsc.com]
(that is a reference design for the lan chip the Pi guys are using)
So without the correct magnetics things are unlikely to work at all.
Before anyone makes any silly comments... (Score:5, Informative)
The magnetics in question aren't to hold the connector in like those in a Mac power cord, but rather the tiny transformers [molex.com] that are required for Ethernet differential signal isolation/transformation.
Re: (Score:3)
It would have been more informative to call the internal circuitry "filters" rather than "magnetics," which is conceptually more accurate.
That said, this has to be great advertising for Molex.
Re: (Score:3)
They're not just filters, they're transformers that (as a result of their inductance and capacitance characteristics) also act as filters. I chose molex simply because they were first on the search and seemed informative.
They've been called 'magnetics' in this context for quite a while - I guess it's a bit of an industry standard. They also have cute terms like PHY and MAC.
Seems reasonable enough (Score:1)
I think many people, me included, have been expecting something like this to happen. As said in the article, this is a relatively minor bump in the road that was practically inevitable and they seem to be handling it as well as could be expected.
I suspect they’ll get a bit of flack over the “4 day” thing... however they would have gotten a lot of flack if they came out with some information that turned out to be incorrect. I guess they could have come out saying “there is a minor pro
Re:Seems reasonable enough (Score:4, Interesting)
Someone in China (the same guys that did the mistake in the first place, which most mentions have assumed to be a deliberate cost-saving measure rather than a true accident) has to receive those units back, hand-unsolder 10,000 connectors and hand-solder 10,000 correct connectors back into place before then packaging them up and sending them back to the UK.
Where, still, as far as we know, there's been no tests of functionality other than networking (i.e. they haven't seen if similar issues affect the other ports like the display, etc.). And then someone has to test a good portion of them again before sending them onto the suppliers.
Meanwhile, they have to source a supply of 100,000's of the proper connectors for future runs, which they are just starting now. And hope that the network WAS the only problem.
In effect, they did no actual testing of the actual device functionality ("it'll all just work if the factory did their job") until the entire first batch was opened in the UK. The testing in the manufacturing facility was purely electronic and COMPLETELY missed this problem (surprise, surprise). And immediately upon opening them here, they spotted a problem, which took FOUR DAYS to isolate (and was isolated only because they were baffled and broke one of the connectors open and happened to spot the difference) and now it all has to be sent back for more work.
That's a mite more than a "minor bump". Not irreconcilable, but certainly not a bump. More like a hard jolt with metal grinding. I sincerely hope it doesn't turn into another OP, but given that we've gone from "No preorders" to well, pre-orders, and a full launch to, well, we'll tell you when we have a working device in the same country as our distributors, the slippery slope has certainly started. Of course they can recover the situation. The question is, what other mistakes have they made in their supply chain of making 10,000 bare PCB's with components (something that happens thousand-fold times every day).
Re:Seems reasonable enough (Score:4, Informative)
Quotesdfrom the forum
''Jamesh is right – they sent us test units which *did* have the right part on before they moved to a larger batch. "
Re: (Score:2)
Actually, no. Read the forums there. They tested ONLY the electronics (i.e. that the caps etc. were the correct value, the right current was on port X, etc.). They did not test network functionality. They NEVER plugged it into a network box.
And what sort of test regime is it if you only test with a short LAN segment without checking, e.g. signal strength, expect cable loss, etc.? That's exactly my point.
Re: (Score:2, Informative)
I hate to stereotype, but this is quite typical for China. You need to watch them closely or they will cut corners whenever possible.
Re: (Score:2)
They probably wont be sending them back. They will probably be recycled.
Re: (Score:1)
Re: (Score:2)
I'd have bet on supply problems.
Always innovating, geekphone, and other manufacturers other than pandora had similar problems.
A new company might overlook some details ending up in delays, the factories might be giving priority to big clients and go out of their way to not displease them, or maybe open hardware running open software is the #1 enemy for the modern models of marketing which rely on planned obsolescence.
We ought to look whether startup hardware companies selling cheap closed stuff go on withou
Re: (Score:2)
bah! Replying to remove mis-moderation...
Re: (Score:2)
Well, the good news with everything being manufactured in the Far East is that wait a few hundred years and the Morlocks will be so far away they won't be able to eat the Eloi.
Re: (Score:3)
Blame the extortionate tariff on importing components versus the lack of tariff on finished goods for production being outsourced. Here's a direct petition on the matter: http://epetitions.direct.gov.uk/petitions/27158 [direct.gov.uk] -- that is the direct reason manufacture is being outsourced. From what I understand of the Foundation's statements on the matter no economy of scale could overcome the cost of the tariff because the bill of materials cost would be too high to be profitable at the desired price point as a r
Re: (Score:2, Insightful)
Perhaps you would be interested to find out what the tariff actually is? Or perhaps your definition of "extortionate" isn't the same as mine. Hint: the upper limit across the various electronics tariffs, last I checked, was 14%. Broadcom does not deserve the blind trust geeks seem to be awarding it over this Raspberry Pi project: even though some of the characters may be familiar, we're a long way from the days of the BBC B where a bunch of bright uni students got together with academic entrepreneurs.
It is,
import tariffs? Pah! (Score:2)
No, the main reason it's not financially sensible to manufacture mass-appeal items in the UK is that the unemployment benefits are higher than chinese assembly workers' wages. You can't get native brits to take on menial work as they can get more money for being unemployed than, for example, picking vegetable or jobs that other stoop labour industries can afford to pay.
If these boards were to be assembled in the UK the costs of doing so (setting aside component costs and amortised developemnt costs) would
Re: (Score:2)
No, the main reason it's not financially sensible to manufacture mass-appeal items in the UK is that the unemployment benefits are higher than chinese assembly workers' wages.
That probably has a lot to do with the cost of keeping workers alive and healthy and fit for work here in the UK being higher than Chinese assembly workers' wages. It's kind of impossible to get workers for less than the cost of feeding and housing them.
Re: (Score:2)
I didn't realise the Daily Mail ....
Yawn! - and so completely wrong
1) What is your evidence that £65/week is "higher", cost-adjusted, than the wage at, say, Foxconn?
Didn't you think to do any research - anything at all, even a simple Google search, before spouting off.
A junior level worker in Shenzhen, China, will receive 1,800 yuan a month, according to Reuters, citing a statement by the Taiwan-based company. Monthly pay may rise to 2,200 yuan if the worker passes a technical examination. That puts the monthly pay at between $285 to $350,
That works out at about £2,600 p.a. for the top rate. Compare that with your own figure of £65/week or £3,380 for an unemployed brit.
As for all the rest of your uninformed rant: TLDR, since the first sentence was so wrong, there was no point.
Re: (Score:2)
I'd certainly pay more for these things.. but their entire goal here is to make these as cheap as possible.
They also posted a pretty damn good explanation as to why they had to outsource.
I do hope they recover from this (and I suspect they will.. it's not OpenPandora yet..) and take some hard won lessons about testing and assumptions.
I also feel forry for the poor guys down in China who have to hand unsolder/resolder 10,000 of these suckers.
Re: (Score:2)
What phone do you have that costs under $25?
Re: (Score:1)
I bought a Samsung 1080 for £9.99 at Tescos recently, SIM unlocked.
Re: (Score:2)
Once again what phone was this? I take your inability to answer such a simple question as a sign that you are lying. So good day.
Re: (Score:2)
So 40 Euro shipped plus tax. Versus 40 dollars (or 30 for the cheaper version) shipped for the pi. Get back to me when the concept of conversion rates sinks in to you head.
Also, let me know when you get your phone connected to hdmi and ethernet.
Then again you're an idiot who thinks a rasberry pi is a phone so I doubt there's hope for you.
Fucking magnets. (Score:5, Funny)
Re: (Score:1)
By attracting balls of steel.
Bye Bye Raspberry Pi (Score:5, Funny)
Bye Bye, my Raspberry Pi,
I thought that I might buy you,
but the warehouse was dry,
those good old boys say just wait one more month,
but you keep running into delays,
yeah, you keep running into delays.
Re: (Score:1)
Re: (Score:1)
Did you see the other recent post? There was a mistaken mass email by Farnell about pushing back orders.
Re: (Score:2)
That was bad. Bad rhyming, bad meter. Try this one.
Bye Bye, mister Raspberry Pi,
Ain't as heavy as a Chevy
but the warehouse was dry.
those good old boys, they just want one more try,
If I don't get one I think I will cry,
If I don't get one I think I will cry.
Maybe... (Score:2)
Eternal shame (Score:1)
Re: (Score:2)
Shame.. or having to hand desolder/resolder 10,000 of these damn things.
Re: (Score:2)
or having to hand desolder/resolder 10,000 of these damn things.
These connectors are nearly impossible to remove. Two large shield tabs + 8 (or more) through hole pins... It might be cheaper to scrap the batch.
I'm amazed that this was allowed to happen. Mightily unprofessional. You never assemble a large production run of anything until the same people assemble a hundred boards from the same kit and then you personally test those 100 boards. Then they are returned as "approved" samples, and any deviati
Serves them right (Score:1)
They advertise the hell out of their product, then they predictably can't deliver, then they silence any and all criticism on their forum (because not being all positive is "bad attitude", and they don't allow that), then another disaster strikes. I guess it's called karma.
If you order now, you're going to get a delivery estimate about 3 months from now. Their mailing list had more than 100000 subscribers, and Liz Banhammer has the audacity to claim surprise when demand exceeds the initial 10000 batch.
For once, this fits! (Score:1)
How's that cheap Chinese production looking now? (Score:3, Insightful)
Still good value?
What about when the returns start flooding in because a 1 cent component failed when a 2 cent one might have soldiered on? Budgeted for handling that?
I know these guys are amateurs, but do they really need to keep demonstrating it?
Re: (Score:2)
Why do you assume that the same mistake - or one like it - would NOT have been made if the boards were assembled in the west? Since western wage rates are so much higher than chinese ones if this error had been made in a british or american plant it would probably be cheaper to simply crush the whole batch and start again,
Then instead of a 1 month delay, you'd be waiting 6 months - or never, since the RPi foundation would have gone bust as it was banking on the sales of these units.
Re: (Score:3)
This sort of thing is precisely why the standard Slashdot rant of "all they did was put x, y and z together, this isn't innovating!" is so much silliness. It's not easy to mass produce things. It takes planning and more planning. It takes money and more money than you planned on because some small aspect of Murphy's law is going to pop up and rip your balls off.
It's why the Motorola Xooms of the world come with stupid little missing bits and even why our fearless denizen of perfection, Apple, still screw
Re: (Score:1)
I wonder... (Score:2)
How many people are killing themselves trying to keep up production with the demand for Raspberry Pi.
Re: (Score:2)
Hard to tell? (Score:2)
From TFA:
It's actually very hard to tell unless you look at the insides of the part,
Ohmmeter?
Re: (Score:2)
Yep, that should do the job. Certainly a lot easier to get hold of than an X-Ray machine...
UK Manufacturing (Score:2)
"Firstly, the schedule for manufacture for every UK business we approached was between 12 and 14 weeks (compared to a 3-4 week turnaround in the Far East)."
http://www.raspberrypi.org/archives/509 [raspberrypi.org]
[Posted on January 10th, 8.5 weeks ago - and manufacturing had already started at that point].
I guess UK manufacturing wouldn't have been much slower, after all.
RS
Re: (Score:2)
Re:How wrong I was! (Score:4, Insightful)
I can understand this critisism as I've had it many times over many different periods of "omg, not another <whatever> story".
But I'm way too damned excited over the Rasperry Pi to care! Kinda fun being on the other side of things for a change :D
Re:So much for opensource hardware taking off... (Score:4, Informative)
Actually, no. This is what happens when you outsource manufacturing to vendors in countries like China where it is common practice for them to quietly substitute parts between the reference design stage and when the device hits production. Sometimes, you get lucky and they even tell you in advance that they're doing this and you have a chance to evaluate impact on the design. Most times, they simply do it and pocket the difference in cost while hoping not to get caught.
Re: (Score:2)
Consumer electronics widgets are constantly having their ship dates quietly revised, usually with a terse announcement from some PR flack that 'Release of Widget Foo has been moved from late Q1 to mid Q2'. Or they just go up for pre-order and take longer than initially promised to ship. Annoying; but hardly unusual.
In this specific
Re: (Score:2)
In this specific case, I'm a bit surprised that they didn't have somebody plug one into their laptop and then wonder why the NIC wasn't working slightly earlier in the process; but so it goes.
Reading through the forums, it sounds like they were sent early test boards which contained the correct connector before the full run. Sounds to me like the "accident" happened between the test run boards being signed off on and the full run.