Possible Hardware fixes

looks like a good list.  Too bad the PoP ram is so hard to replace,

as that would be another nice improvement.  I've been thinking of

modifying one for high-altitude balooning, or similar uses, where

size and/or weight matter.  So here's my list:

  1) remove csi connector,

  2) remove dsi connector,

  3) remove hdmi connector,

  4) remove RCA connector,

  5) remove audio connector,

  6) remove ethernet connector,

  7) remove micro usb connector (directly solder power wires),

  8) remove GPIO pins,

  9) remove LDO voltage regulators, replace with switch mode,

10) remove LAN chip,

11) remove dual-usb connector, directly solder one usb device,

12) remove polyfuses, replace with 1 Ohm resistor,

13) remove the capacitor that frequently falls off,

14) as you suggest, replace SD with micro SD.

I'll be replacying the polyfuses with 0 ohm resistors, no need ot waste power.

I'll probably keep ethernet and HDMI.

Agreed with the two FPC connectors, good move by RPF to offer a camera that will be more expensive than the board.

The RCA jack takes too much space, same as audio, perhaps something done on similar designs like Roku2, replace both by a single jack that can carry all signals, but resolution and image quality out of the composite video output is not very good, perhaps better use something that can support VGA instead.

the 1 Ohm resistor is to avoid voltage drops from

power surges on the USB device.

For that you should have a pair of good tantal bulk caps as recommended by the USB spec sort of 150µF, and I'd add USB power switches.

I'd be nice if Broadcom started selling the "alpha" boards :-)

-J

and I'd add USB power switches.

I probably didn't make it clear enough in the list, the AP1212 I mentioned is a usb power switch.

Making the obvious comparison, the cheapest 140mA polyfuse from farnell is 0.113 gbp in qty 5, and the AP1212 is 0.33 gbp in qty 1. So I'm not at all sure the case for using polyfuses is a good one. Don't know what costs @ 10K, 100K etc would be like though.

Anyway, remove D10, D11 and both polyfuses and it should be simple to put the power switch in, the datasheet suggests a 33uF cap on the usb connector side and it has a soft-start feature to avoid the power dip when connecting a device.

The RCA jack takes too much space, same as audio,

I didn't add those to my list, almost goes without saying that I  won't use them and will likely remove them. For add-on boards that plug onto the gpio on top of the board that RCA connector is just too high.

Ohh you right I missed that one.

Cost and the fixation on the credit card size form factor are very lame design guidelines, It is not bad to have an approximate target size, but they didn't have any mechanical restrictions or requirements, so component and connector layout could have been much better.

-J

Dom said today:

      I agree that a small resistor is a better solution.  ...    

     The problem was that by the time the first report of the small resistor         

     fixing the reset problem was posted, the rev2 PCB had already been laid out.

http://www.raspberrypi.org/phpBB3/viewtopic.php?f=28&t=5249&start=166

I wonder exactly when the rev2 PCB was laid out, especially since PeteL

was apparently getting up to speed on the 1.8V issue quite recently.

Dom said today:

      I agree that a small resistor is a better solution.  ...    

     The problem was that by the time the first report of the small resistor         

     fixing the reset problem was posted, the rev2 PCB had already been laid out.

http://www.raspberrypi.org/phpBB3/viewtopic.php?f=28&t=5249&start=166

I wonder exactly when the rev2 PCB was laid out, especially since PeteL

was apparently getting up to speed on the 1.8V issue quite recently.

coder27 wrote:

 

Dom said today:

 

      I agree that a small resistor is a better solution.  ...   

It's really not. The better solution is to use a proper USB Power Switch IC like a AP1212 or similar that has a slow-start feature and big enough reservoir capacitors.  As Burngate said in the post after the one you linked, v2 is a backward step in this regard as it removes the possibility to do something better yourself.

I wonder exactly when the rev2 PCB was laid out, especially since PeteL

was apparently getting up to speed on the 1.8V issue quite recently.

It was my thought that v2 was probably done in conjunction with the move to UK manufacturing. The chinese factory has built enough without needing holes for the ATE fixture, but a new manufacturer might well demand some improvements to help their process.

I could obviously be wrong in that assumption, but if I'm right it would likely put it's layout at around the time of Pete's last visit here.  That ties in reasonably well with the 1.8v fix, but I'm also fairly sure the low ohm resistor ideas were being talked about much earlier. 

But was anyone listening ?  One of the big problems I see with the level of interest in the Pi is that discussions inevitably get drowned out by the noise, especially in the RPF forums, and solutions with solid engineering reasoning often get blown out of the water or simply ignored by the clueless masses flapping around in the dark. Just look at jamoido's post on the RPF forum on the 1.8v thing, as far as I can tell it was completely ignored at the time.

Not sure there's an easy solution though, the RPF is somewhat a victim of it's success and the way it's forums are run doesn't engourage proper discussion on issues or the engineering reasons why they're an issue.

I somtimes have the impression they really want to create confusion on the RPF forum.

Like Jamesh agreeing there is a problem with the usb stack, and a bit later claiming that he fixed his stuck keys by replacing his supply...

You don't need a 5000 euro usb analyser to test that theory. A simple lab adjustable power supply can do the trick.

When I told them Element 14 was selling usb hubs for the Pi that apparently provided power on their upstream port, they said it wasn't their problem.

Every technical question about the current progress on the issues is simply ignored....

A lot of people have no problems. That sure helps if you go to a forum with a problem.

But as Jamesh pointed out very accurate yesterday...It's just a waste of time...

And even better I found ...how much more open can the foundation be?

I reported the problem as soon as the schematics were released, I've no way of knowing when they realized there was a problem with it since nobody acknowledged or commented about it.

I agree with selsinork that the resistors or even the complete elimination of the polyfuses/resistors on Rev 2.0 are not a solution but a quick "fix."

It will be a solution when they finally accept an understand that the entire power design needs to be redone.

Also promoting this fix saying that it has side benefits like powering the Rpi from a hub via a USB port opens another can of worms. While that approach is interesting for an experiment it is not a very good idea since the Rpi does not behave like a USB device and people with limited knowledge could end damaging the device provide power or even the Rpi.

I really don't know who is leading over there the engineering side of the show, some claim that they work hard and put a lot of effort thinking about how to fix problems and improve the design, but for example the placement of the new P5 tells a different story, something not worth a "knighthood"

BTW did anybody seen the schematics for the Rev 2 board ?

-J

They are on the elinux wiki - so I'm told.

Fergus Byrne wrote:

 

They are on the elinux wiki - so I'm told.

The schematics on the Wiki are for the R1.0 board. Nothing yet for R2.0.

-J

I was told that someone changed them the day of the release of rev2.  They must have been wrong.  Oh well.

Fergus Byrne wrote:

 

I was told that someone changed them the day of the release of rev2.  They must have been wrong.  Oh well.

IIRC, someone said the PCB Gerbers were posted at the same place as the Rev 2 schematics.  This was a gamesome way to say that neither was available at the time of the comment.  As far as I know it's still the case that neither is available.