troubleshooting topics zooms past 1000 (now 6000)

jamodio: "...Based on the volumes we are talking, even before they realized that demand was much much higher than anticipated, the $25 target, and the credit card size, are just BS. What's the difference between a $25 product that has design and reliability issues to a $28 that works properly?..."

"...The $25/$35 is really a joke, when you factor in that you need to buy a keyboard, a mouse if you want to work in graphics mode, a power supply or several of them you need to try to make sure they work and don't blow up your TV, a collection of SD cards to figure which one works with your Pi, a powered USB hub (of which some were reported not to work properly) if you want to have some extra stuff or even power hungry USB devices connected to the Pi, the HDMI cable, and if you don't have one, a monitor. Plus if you want to provide some protection to the board, a case.     What one really can criticise now...."

I really couldn't agree more with these sentiments (and the rest of your post).

It's all a bit late now but while "cheap and flaky" might be ok for experienced hackers who are looking for a project (or for those inexplicable ebay types who seem happy to take a punt on far Eastern junk that everbody else knows is a ripoff), but most consumers would want "reasonably priced and functions properly" and the UK educational system (which can be likened to an industrial grade customer imo) wouldn't settle for less than "rock solid at a reasonable premium and with Walk On Water support".

The hardware cost of a classroom's worth of of Pi boards is pretty insignificant compared to the cost of training all those teachers in some pretty deep topics. Then there's the cost of peripherals and accessories for e.g. GPIO interfacing, study materials, SD card duplication, all those thousands of meetings...

It made no sense to cheap out on the board to the current extent, but now that it appears to be pretty much finished then it makes even less sense to cheap out on getting it working bess poss. One guy with a USB sniffer devising workarounds in his spare time? Remember that the Foundation were initially hoping to shift 10,000 units eventually to finance the educational push. It now looks like that sales target will be exceeded by two orders of magnitude.That should pay for some development time. Or a new kitchen.*

I'm sure this soap opera will run and run.

*in-joke.

It may have made some kind of marketting sense to sacrifice good design in order to get the first Model B under $35 in 10-30k volume quantities, but those original design choices no longer make any sense when half a million boards have been shipped.

The BOM cost is now far below what it was originally for the Foundation, and as a result a huge range of improvements could be made to a new model of the board while still remaining below $35.  The fact that no such wholesale redesign is being undertaken by the Foundation (as far as we know) suggests many things, all bad.

Morgaine Dinova wrote:

 

It may have made some kind of marketting sense to sacrifice good design in order to get the first Model B under $35 in 10-30k volume quantities, but those original design choices no longer make any sense when half a million boards have been shipped.

 

The BOM cost is now far below what it was originally for the Foundation, and as a result a huge range of improvements could be made to a new model of the board while still remaining below $35.  The fact that no such wholesale redesign is being undertaken by the Foundation (as far as we know) suggests many things, all bad.

I don't think they're doing wholesale re-design, since the basic combination of BCM2835 and LAN9512 still fits the goal of the Foundation: lowest-cost educational platform.  I do believe they're doing minor redesign to fix problems identified here at element14 and at the RasPi Forum, but the practice of the Foundation is to redesign in secret and the only way to find out things have improved is for alert RasPi buyers to wonder why their new RasPis doesn't look the same as the published PCB photos.  I think this is a Bad Approach, since if RasPi were to publish schematics and layouts before committing to a new rev, we'd have a chance to say things like: "Wait a tick, if you directly connect USB port A +5V to the 5V0 plane instead of using a 0 Ohm jumpers then you have no way to fix voltage dip and other problems."

I fully expect to see a future RasPi Model B rev 3.x with switching regulators, but no announcement until after it's shipped.

John Beetem wrote:

 

if RasPi were to publish schematics and layouts before committing to a new rev, we'd have a chance to say things like: "Wait a tick, if you directly connect USB port A +5V to the 5V0 plane instead of using a 0 Ohm jumpers then you have no way to fix voltage dip and other problems."

 

I fully expect to see a future RasPi Model B rev 3.x with switching regulators, but no announcement until after it's shipped.

Well they are inept organizationally, we know that.  But what's worse than their organizational ineptness is that their secretive approach results in a far worse product than if they did their work openly, as has already been demonstrated through the community identifying faults and other issues that could easily have been caught at design time.

They really don't get it that hundreds or thousands of engineering eyes are better than the few that they have in-house.  I'm not sure if it's because of unwarranted confidence in their own abilities, or simply a matter of ego, but it's present and it's very strong.

Everyone in the Pi ecosystem is suffering because of the Foundation's myopia, including Element 14 and RS who would benefit greatly from a product without faults.  But the most ridiculous part of the situation is that even RPF themselves would benefit, but are too short-sighted to realize.

Morgaine.

Hello all!  I'm new to the forum here.  I also frequent the RPi forum, but from the way you describe their mods I'm sure you'll forgive me if I don't disclouse my username I use over there! 

@coder: On the troubleshooting thing, I looked at the forum and to be fair, very little of it actually has to do with the pi itself.  It's full of things like "I can't figure out Samba" or "my cheap-o passive PS/2 to usb convertor isn't working" which is to say newbie linux questions.  You make it sound as if every post was representing some new issue tied solely to some issue with the pi, or at least a very large group set, when in reality it would appear 99% of the posts are about issues you would see no matter what peice of hardware.  Just from reading through that section of the forum it would seem the main problem is the USB issue (which having followed the elephant thread closely seems to be tied more to the LAN9512 chip more than the broadcom chip, but that's not to say the broadcom chip is completely innocent) and power supply issues, which manifests in a multitude of ways.

@Mor I'm all for open source as well (who doesn't like free stuff, right?), but it does seem odd to act as if they're being more secretive than any other industry out there.  Shoot, at least they're trying to fix the usb issues, Apple wouldn't even admit that they had a significant design flaw with their phone that created bad signal reception.  Their solution was to use your other hand to hold the phone!  I wonder about the two options you put forth unwarranted confidence or ego (which to me sounds like ego or ego) and wonder if there's not more to it than that.  I mean the world of business is a large and tricky place and I'd be fairly shocked to find out that there wasn't some non-disclosure agreements going around.  That's not to discount the two options you've suggested, it just seems like you're turning a blind eye to some very large and probable possibilities that are known to exist for a propisition that by your own admission doesn't make sense for the RPF to take.

"Jim"

that's is exactly the problem with that section of the forum, the total lack of formatting and processing of issues so the information can be aggregated in an inteligent and useful way.

Right now there is no process, no system, no way, to effectively track what issues are out there, what fixes are the correct ones for some of those issues, what is critical and what is not.

You show some lack of knowledge about the USB matter since the biggest issue is associated directly with the piece of IP that Broadcom licensed from Synopsys to implemente the USB features on the chip, the LAN9512 is not the culprit, the thing goes so deep that somebody has been reviewing the Verilog files for that part.

About the secrecy which is related to the lack of information about some particular subjects like the electrical characteristics and pinout of the Broadcom SoC, schematics for the new board Rev, etc, etc, there is no reasonable argument for not releasing that information and the argument about being afraid of "clonning" is 100% BS. Even if you are able to clone the PCB design unless you also clone the SoC there is no way anybody will be able to make clone of the board as is.

BTW Open Source does not mean Free...

-J

jamodio wrote:

 

"Jim"

 

that's is exactly the problem with that section of the forum, the total lack of formatting and processing of issues so the information can be aggregated in an inteligent and useful way.

 

Right now there is no process, no system, no way, to effectively track what issues are out there, what fixes are the correct ones for some of those issues, what is critical and what is not.

 

You show some lack of knowledge about the USB matter since the biggest issue is associated directly with the piece of IP that Broadcom licensed from Synopsys to implemente the USB features on the chip, the LAN9512 is not the culprit, the thing goes so deep that somebody has been reviewing the Verilog files for that part.

 

About the secrecy which is related to the lack of information about some particular subjects like the electrical characteristics and pinout of the Broadcom SoC, schematics for the new board Rev, etc, etc, there is no reasonable argument for not releasing that information and the argument about being afraid of "clonning" is 100% BS. Even if you are able to clone the PCB design unless you also clone the SoC there is no way anybody will be able to make clone of the board as is.

 

BTW Open Source does not mean Free...

 

-J

I wonder about the troubleshooting section.  On the one hand the pi forum has grown increasingly large and further segmenting will just add to the difficulty of navigating it.  But as you say it might indeed help to split it into two sections a "Problems with the hardware" Troubleshooting section and a "Linux Troubleshooting" section.  It seems some of you have had one on one dealings with the mods, which one would be the most receptive to such a PM?

As for the USB issue, the second USB thread discussing packets loss http://www.raspberrypi.org/phpBB3/viewtopic.php?f=28&t=5249 indicates that the problem is at least partially related to the LAN9512 as removing it or using one of the few model As in existance solved some of the USB  issues.  But as I said, the culprit seems to be mostly the LAN9512 (or perhaps I should be more specific and say the linux drivers interfacing with it) rather than the broadcom chip.  But that may be a matter of splitting hairs as ultimately no matter how you slice it, it's still a problem that exists between the broadcom chip and the LAN9512 chip.

I wouldn't be surprised if the line about preventing clonning isn't entirely true.  I also wouldn't be surprised based on the level of candor the "RPF" provided at the beginning of their posts if one of the distributors are the ones holding things up.  There are at least three companies involved here, it seems awefully shortsighted to lump all the blame into one basket.  That of course meanse that I am by no means assuming the pi foundation is innocent, just that I'm aware of the fact that there are at least 3 places where there could be a break down.

That's a semantic game on the open source definition ;-)

jamodio wrote:

 

"Jim"

 

that's is exactly the problem with that section of the forum, the total lack of formatting and processing of issues so the information can be aggregated in an inteligent and useful way.

 

Right now there is no process, no system, no way, to effectively track what issues are out there, what fixes are the correct ones for some of those issues, what is critical and what is not.

 

You show some lack of knowledge about the USB matter since the biggest issue is associated directly with the piece of IP that Broadcom licensed from Synopsys to implemente the USB features on the chip, the LAN9512 is not the culprit, the thing goes so deep that somebody has been reviewing the Verilog files for that part.

 

About the secrecy which is related to the lack of information about some particular subjects like the electrical characteristics and pinout of the Broadcom SoC, schematics for the new board Rev, etc, etc, there is no reasonable argument for not releasing that information and the argument about being afraid of "clonning" is 100% BS. Even if you are able to clone the PCB design unless you also clone the SoC there is no way anybody will be able to make clone of the board as is.

 

BTW Open Source does not mean Free...

 

-J

I wonder about the troubleshooting section.  On the one hand the pi forum has grown increasingly large and further segmenting will just add to the difficulty of navigating it.  But as you say it might indeed help to split it into two sections a "Problems with the hardware" Troubleshooting section and a "Linux Troubleshooting" section.  It seems some of you have had one on one dealings with the mods, which one would be the most receptive to such a PM?

As for the USB issue, the second USB thread discussing packets loss http://www.raspberrypi.org/phpBB3/viewtopic.php?f=28&t=5249 indicates that the problem is at least partially related to the LAN9512 as removing it or using one of the few model As in existance solved some of the USB  issues.  But as I said, the culprit seems to be mostly the LAN9512 (or perhaps I should be more specific and say the linux drivers interfacing with it) rather than the broadcom chip.  But that may be a matter of splitting hairs as ultimately no matter how you slice it, it's still a problem that exists between the broadcom chip and the LAN9512 chip.

I wouldn't be surprised if the line about preventing clonning isn't entirely true.  I also wouldn't be surprised based on the level of candor the "RPF" provided at the beginning of their posts if one of the distributors are the ones holding things up.  There are at least three companies involved here, it seems awefully shortsighted to lump all the blame into one basket.  That of course meanse that I am by no means assuming the pi foundation is innocent, just that I'm aware of the fact that there are at least 3 places where there could be a break down.

That's a semantic game on the open source definition ;-)

Elementary my dear Watson, if you remove the ethernet controller the network packet loss goes away ...

BTW, the problem is with the driver and the architecture associated with it which includes the piece of silicon baked on the Broadcom SoC.

What are the three companies you are talking about ?

-J

It was USB packet loss, some of which included network stuff (which yeah obviously went away when you take the ethernet chip away) since the ethernet on the pi uses the USB bus, but those yoctopuce guys were talking much broader than ethernet

The three companies I was talking about are the Raspberry Pi Foundation + the two distributors (Farnell and RS Components).  I discount Broadcom as they're not associated with the Pi Foundation, not beyond selling the chip at any rate so they wouldn't have any controll of the day to day operations of the Pi beyond what the Pi Foundation is allowed to release on the B-chip itself.  But Farnell and RS... well they would have a lot of sway in what the Pi Foundation was and was not allowed to mention in up coming releases.  They've put a lot of capital in the Pi and it would be devestating if word got out that company X was releasing some new version of the pi while company Y had just made a new huge batch of the old model.  So company Y works it out such that no one can say anything until they have a chance to get their inventory on the old stuff down.  It seems silly (read stupid), I know, but I don't make the rules, I just observe the way things tend to roll, lol! 

The Raspeberry Pi Foundation as the name says it is a Foundation (Charitable under UK law) not a company, besides whoever is involved, they are the primary responsible entity for the design of the Raspberry Pi board.

There are more than three companies involved and it is not very clear what level of involvment each one has and what are the contract terms between them, just to mention a few:

- Broadcom, Inc. (You need to read more about where the RPF people came from and who designed and produced the first alpha boards, and how actually

   Broadcom is involved in several activities through their Broadcom Foundation)

- Nortcott Technologies, Ltd

- Premier Farnel Group

- RS Components, Ltd

- Sony UK

- Unknown Assembly house in China

Companies that sell products for a profit and are succesful know how to handle releases, we are not talking about information of "future" products, we are talking about a device that is in the hands of hundreds of thousand people who were told that the Rpi can do what any other computer do, well it doesn't.

Without going in too much detail, the USB module IP licensed from Synoptycs has some limitations, which are consistent with the original design and target market for the Broadcom SoC, that module was never designed/intended to be use as it is being used on the Rpi, so besides some inherent limitations on the hardware there are other issues associated with the "software," ie drivers. Removing the ethernet controller with the integrated two port USB hub obviously reduces the number of interrupts and packets the USB module and driver has to handle on the SoC, but it does not eliminate the inherent limitations of the original design.

-J

Foundation/company, that's semantics again but you knew what I meant

I was aware of the Foundations history, but that's quite the conspiracy you're suggesting there.

Aside from that, you are correct that I had not considered all the other companies involved, which further empasises my point that to throw blame for secrecy entirely at the Pi Foundation's feet is impractical and unrealistic.

Jim,

  I think you're the one suggesting a conspiracy.  Broadcom has very openly

designed and built the RPi alpha board, even putting their own logo on it, several

of the most active RPi.org mods are openly employees of Broadcom, Broadcom

supplies the main chip on the RPi board, Broadcom employs the RPi CEO,

Broadcom is actively trying to fix bugs in the linux drivers, and the Broadcom

Foundation is very openly sponsoring RPi marketing events.   Broadcom

reportedly sent an inspector to the China factory to assess the working conditions.

When a journalist called for a response from RPF, it was a Broadcom employee

that responded, although it isn't always clear which hat he wears at any given time.

Sometimes Broadcom's role has been minimized for unknown reasons, such as

when their logo was taped over on the alpha board photo, or when Liz refers to them

as just a friendly chip supplier, or when Rob Bishop's employer isn't named,

but I don't think Jamodio is suggesting a conspiracy.

That is correct, I'm not suggesting any conspiracy, just stating facts being one of them the lack of transparency.

The organization at the center of the mesh is the RPF which I don't blame but clearly I believe in some aspects have done a poor job.

-J

SoCs are designed to suit the market for which they are intended.  Broadcom designed the BCM2835 for media devices such as the Roku 2 which are not generic host computers, and prior to Pi no other commercial customer other than Roku existed for this particular SoC as far as we know.

Given the intended purpose of the chip, it's no surprise that Broadcom built in a cheap cut-down USB core that has very limited ability to act as a generic USB host controller, and it shows in the faults that appear as soon as you try to use Pi's USB as if it were a generic host controller, because it's not.  It's a very limited one.

There's no conspiracy, but only an inappropriate choice of SoC for the Raspberry Pi.  It's inappropriate because the Pi is definitely intended as a general purpose computer with a fully working host USB system, which the actual system as designed does not have as a result of the designer's choice of SoC.

I don't disagree with the thoughts about the SoC, but I really can't use any other word than conspiracy for the way you refer to broadcom

Yeah, I would agree that it looks like the Pi Foundation used broadcom to manufacture the alpha board, but they did so at their own expense.  Some of the mods (2 if you don't count eben who never posts, maybe 3 if they've made dom a mod) are broadcom employees but the majority of them are independent individuals from the community and the broadcom people contribute in their spare time.  I'm unware of broadcom funding any of the makerting campaing, links?

Sorry but no, you're trying to suggest something more to than the facts would predicate.

Ah well, a difference of thought allignments?  No offense meant, but the more I read on this board the more I see a subtle (and sometimes not so stuble) tone of "The RPF is sinister." A willingness to take an extra step or two beyond what logic shows to be true to stick it to the Pi foundation.  Unforatunately, I'm just not in that boat .  To each his own though right?  You can enjoy this forum and I can find another, no sense in me pissing on your fun.

Thanks for the chats!

Jim,  

You said you were familiar with the RPF history, but you have your facts all wrong.

> Yeah, I would agree that it looks like the Pi Foundation used broadcom to manufacture the alpha board, but they did so at their own expense.

No, they have clearly stated that Broadcom designed and manufactured all 50 alpha boards at its own expense.

EDIT adding: see page 2 of:

http://www.element14.com/community/docs/DOC-46998/l/raspberry-pi-the-journey-so-far-and-a-peek-into-the-future

"Broadcom has been an early supporter of the Raspberry Pi programme.  Although not "officially" a sponsor, ...

but also supporting the design -- by paying for the PCB design & production of the first 50 (alpha) production boards."

> I'm unware of broadcom funding any of the makerting campaing, links?

http://blog.broadcom.com/broadcom-innovation/attention-raspberry-pi-oneers-broadcom-helps-spread-the-raspberry-jam/

http://blog.broadcom.com/broadcom-foundation/broadcom-masters-finalists-get-a-taste-of-raspberry-pi/

>Some of the mods (2 if you don't count eben who never posts, maybe 3 if they've made dom a mod) are broadcom employees

care to name them?

mynameisJim wrote:

 

I don't disagree with the thoughts about the SoC, but I really can't use any other word than conspiracy for the way you refer to broadcom

 

I think you need to stick to what is actually written on the page rather than swiping at imagined ghosts and mirages.  We're mostly engineers and technical enthusiasts on this forum, and we tend to say exactly what we mean as clearly and precisely as we can, because that's how technical people are built.  If it's not written on the page then it probably wasn't intended in the first place, but imagined.

Thanks for the chats!

You're welcome.  And indeed, everyone is welcome here are long as they engage in polite discussion with other members.  It can take a while though to catch up with the deluge of information which we've been examining in much depth here over the past year.  It's a complex situation, and understanding it takes some effort.