Re:Raspberry Pi: Manufacturing Hiccup
by ImprovedSilence on Thursday March 08, 2012 @09:36AM
(#3679827)
I find fascinating the "open source" view we get into manufacturing here. No "real" real company will throw out there the "oh, we mucked up, your products are coming a week late" and then go on to explain exactly what, and why.
It's nice to see the real side of how things are made, and to develop an understanding that usually nothing ever works out on the first go. It really enforces the lesson of "don't be afraid to fail, and don't get discouraged" As I know many people who have tried, do.
Re:Raspberry Pi: Manufacturing Hiccup
by strags on Thursday March 08, 2012 @01:42PM
(#3680983)
Absolutely agree. It's too bad this happened, but they've turned it into something interesting, and by clearly explaining what's going on, one would hope that customers will be understanding.
Re:Raspberry Pi: Manufacturing Hiccup
by ImprovedSilence on Thursday March 08, 2012 @05:13PM
(#3681935)
yeah, it's nice, isn't it. I've been aware of saleae for awhile, I really like the look of their products. I haven't had a practical reason to buy them yet, but someday I plan on getting their analyzer.
Re:Raspberry Pi: Manufacturing Hiccup
by strags on Thursday March 08, 2012 @07:25PM
(#3682360)
I bought a Logic-8 from them a while back - definitely worth the money. And they seem like an awesome company too. I inquired as to whether they were going add a particular feature to the Logic-16, and while the answer was essentially 'not any time soon', the founder who replied was kind enough to add "Let me know what sort of application you have in mind -- maybe I can point you to some other solutions out there that might work, and maybe even be cheaper."
Re:Raspberry Pi: Manufacturing Hiccup
by sdfjkl on Thursday March 08, 2012 @08:38AM
(#3679637)
Nice to know they care and don't just ship anyways (it would probably work anyways in many cases, but cause lots of frustration).
Since the author didn't mention it in the article, the "magnetics" are actually small transformers that filter noise, DC and make it harder to fry your board via the network port.
Alternatively these can be mounted as as separate component[1] on the board in conjunction with the "plain" type of connector that the Raspberry manufacturers used by accident.
Re:Raspberry Pi: Manufacturing Hiccup
by AlexandrB on Thursday March 08, 2012 @04:39PM
(#3681795)
Interesting. I always wondered what those little black boxes on NICs were - but I never bothered looking them up. It's pretty amazing that modern manufacturing can fit that kind of component into the ethernet port itself.
This is a pretty standard 'hiccup' in manufacturing. Sadly its more common when your factory is 'remote'[1] than it is when your factory is local. If you read Bunnie's blog [2] about the Chumby he's got a number of fascinating stories about how things can go sideways.
Take notes you budding product designers that requires manufacturing actual widgets, its a whole lot uglier than you might think :-)
Re:Raspberry Pi: Manufacturing Hiccup
by noonespecial on Thursday March 08, 2012 @06:59PM
(#3682300)
I actually doubt its an accident. One of the problems you have to deal with when manufacturing in China is "substitutions". In China, the contract you sign is just a place to start negotiations and the prototype that they first present to you is considered a loose guide to how they will build the actual product.
As time wears on and numbers increase, you'll find all sorts of tweaks to save a penny or two showing up on your latest batches. They don't try to make your product non-functional, but they might make a mistake on a substitution going after that 1/10th of a penny.
Re:Raspberry Pi: Manufacturing Hiccup
by DanBC on Thursday March 08, 2012 @11:51AM
(#3680441)
I feel sorry for the person doing the rework.
8 legs to desolder, on a 6 layer board, where the pads are going to be small (because of pin density) and where the component shape means you can't just cut the thing off and take the pins out individually. The click-through legs of the socket make it harder to apply heat to the pads and wobble the component out (which you shouldn't do anyway, but people do.)
And then the new connectors need to be soldered back on.
Re:Raspberry Pi: Manufacturing Hiccup
by joezydeco on Thursday March 08, 2012 @02:24PM
(#3681166)
The legs on those ethernet connectors are .100" centered, like the old DIP pins. Those are incredibly easy to solder and unsolder.
If this thing was surface mount, or had a hidden heat-sink pad in the center of the die, that would be way tricker. Nothing an assembly shop couldn't handle, but kind of getting beyond typical DIY work.
Re:Raspberry Pi: Manufacturing Hiccup
by DanBC on Thursday March 08, 2012 @05:41PM
(#3682045)
> Those are incredibly easy to solder and unsolder.
For sure. It gently depends on how many layers the boards are; if any of the pins is on a large ground plane (heatsinking) or if the solder side pads are flimsy.
But I'm grateful for the correction! I have some horror stories about rework, so maybe they've tinged everything since.
For a small quantity of items you can get "solder suckers" which are spring loaded pumps with a high temp nozzle. You have that in one with the iron in the other hand. You'd clean the pads with solder wick afterwards.
Re:Raspberry Pi: Manufacturing Hiccup
by femto on Thursday March 08, 2012 @05:57PM
(#3682109)
The other way to do it is to use a hot air gun or infrared to melt the solder on all pins simultaneously, and just lift the part out. Easy! Vacuum up the excess solder, with a hollow tip or hot air, and mount the new component as per normal.
Re:Raspberry Pi: Manufacturing Hiccup
by wr1472 on Thursday March 08, 2012 @09:12AM
(#3679743)
I'd be interested to know what kind of x-ray device they used to get those images? How much does one cost, and are they affordable (and safe) for a hobbyist to use?
It weighs slightly over a (metric) ton, and despite being marketed as "low cost" model, it's not something the average hobbyist can afford. Or even needs :)
And yes, it's lead-lined, completely enclosed and safe to use as long as you don't put your hamster in it.
Re:Raspberry Pi: Manufacturing Hiccup
by antirez on Thursday March 08, 2012 @08:37AM
(#3679633)
This stuff happens, it's not a problem. I just wish that the project was less in a hurry because it's a no profit organization so there are no time-to-market issues.
Re:Raspberry Pi: Manufacturing Hiccup
by celticninja on Thursday March 08, 2012 @09:25AM
(#3679787)
they were only 3 monthsd behind if you took November as their shipping date, that was pretty much the earliest it could have been for a Q4 release, whereas it could have been as late as end Dec.
Re:Raspberry Pi: Manufacturing Hiccup
by obtu on Thursday March 08, 2012 @10:48AM
(#3680106)
The design works for a certain state of the components market. As time passes some parts become harder to source, or pricier than others that provide better performance or functionality.
Re:Raspberry Pi: Manufacturing Hiccup
by gaius on Thursday March 08, 2012 @10:52AM
(#3680127)
This is totally like that time back in '83 when Acorn couldn't manufacture enough ULA's for the Electron in time for Christmas. Probably some of the same people involved too!
Re:Raspberry Pi: Manufacturing Hiccup
by joezydeco on Thursday March 08, 2012 @10:11AM
(#3679962)
I don't see it the same way. The article says they spec'ed a certain part on the BOM and the manufacturing house dropped in a different part. I see this all the time from overseas board houses that think they can get away with subbing in a cheaper part and charging the same price.
If this was my vendor and something like this happened, believe me, this would not be "Shit Happens" and a shrug all around from my management.
Re:Raspberry Pi: Manufacturing Hiccup
by elliottcarlson on Thursday March 08, 2012 @10:43AM
(#3680075)
To be fair, the "shit happens" statement is from the POV of a customer - and that's quite correct. We don't have further insight in to what the Raspberry Pi foundation is doing or communicating to their manufacturer and I doubt it's a "ship happens" attitude there as well - but that's just a guess and that's all either of us really have.
Re:Raspberry Pi: Manufacturing Hiccup
by ktizo on Thursday March 08, 2012 @08:32AM
(#3679618)
When have you ever before seen an apology from a computer supplier about delays in delivery, that actually went as far as including x-rays of the parts involved, just to illustrate to ordinary customers how exactly it all went wrong?
Re:Raspberry Pi: Manufacturing Hiccup
by DanBC on Thursday March 08, 2012 @09:00AM
(#3679708)
This is an excellent post by the foundation. I'm pleased they're showing people the actual parts. (After the launch I have no doubt there are some people vigorously complaining about this too.)
It would be really nice if they now link to some good quality electronic engineering page to explain what "magnetics" are, and how they work, and why they're used in network connectors. Teaching people about computing is great. Teaching people about electronic engineering would also be great.
It's nice to see the real side of how things are made, and to develop an understanding that usually nothing ever works out on the first go. It really enforces the lesson of "don't be afraid to fail, and don't get discouraged" As I know many people who have tried, do.
It reminded me a bit of these guys:
http://blog.saleae.com/?p=143
Different problem, but same kind of "small-company" response.
But your point basically stands.
Can't wait to do cool things with mine!
Since the author didn't mention it in the article, the "magnetics" are actually small transformers that filter noise, DC and make it harder to fry your board via the network port.
Alternatively these can be mounted as as separate component[1] on the board in conjunction with the "plain" type of connector that the Raspberry manufacturers used by accident.
[1] Looks like this: http://www.alliedcomponents.com/lan.php
This is a pretty standard 'hiccup' in manufacturing. Sadly its more common when your factory is 'remote'[1] than it is when your factory is local. If you read Bunnie's blog [2] about the Chumby he's got a number of fascinating stories about how things can go sideways.
Take notes you budding product designers that requires manufacturing actual widgets, its a whole lot uglier than you might think :-)
[1] yes that is code for 'outsourced to China.'
[2] http://www.bunniestudios.com/blog/?cat=6
As time wears on and numbers increase, you'll find all sorts of tweaks to save a penny or two showing up on your latest batches. They don't try to make your product non-functional, but they might make a mistake on a substitution going after that 1/10th of a penny.
8 legs to desolder, on a 6 layer board, where the pads are going to be small (because of pin density) and where the component shape means you can't just cut the thing off and take the pins out individually. The click-through legs of the socket make it harder to apply heat to the pads and wobble the component out (which you shouldn't do anyway, but people do.)
And then the new connectors need to be soldered back on.
If this thing was surface mount, or had a hidden heat-sink pad in the center of the die, that would be way tricker. Nothing an assembly shop couldn't handle, but kind of getting beyond typical DIY work.
For sure. It gently depends on how many layers the boards are; if any of the pins is on a large ground plane (heatsinking) or if the solder side pads are flimsy.
But I'm grateful for the correction! I have some horror stories about rework, so maybe they've tinged everything since.
Would it be difficult even if they had some equipment to precisely maintain temps and suck out the solder? Does such equipment even exist?
(http://uk.rs-online.com/web/p/soldering-stations/0165396/)
For a small quantity of items you can get "solder suckers" which are spring loaded pumps with a high temp nozzle. You have that in one with the iron in the other hand. You'd clean the pads with solder wick afterwards.
It weighs slightly over a (metric) ton, and despite being marketed as "low cost" model, it's not something the average hobbyist can afford. Or even needs :)
And yes, it's lead-lined, completely enclosed and safe to use as long as you don't put your hamster in it.
This was a mistake by their manufacturers, and could have happened no matter how long they took.
Could someone explain, for a non-hardware guy, why the magnetics are necessary?
If they are necessary, why do jacks without them exist?
(http://en.wikipedia.org/wiki/Impedance_matching)
You can use a component on the PCB, or a component built into the socket.
If this was my vendor and something like this happened, believe me, this would not be "Shit Happens" and a shrug all around from my management.
It would be really nice if they now link to some good quality electronic engineering page to explain what "magnetics" are, and how they work, and why they're used in network connectors. Teaching people about computing is great. Teaching people about electronic engineering would also be great.