I've always had this love/hate relationship with the PI where I love the amazing SW the community it putting out for it and I hate the closed source Broadcom hardware/drivers and the rookie design mistakes the RPi foundation keeps making over time giving me the feeling they don't test their designs at all before launching them out to market.
This seems like an overly dramatized comment. The issue would have been easy to avoid but on the other hand it's easy to work around so I would call it a minor issue and i can see how it was missed during testing. What other issues have you encountered?
Not OP, but the RasPi 2 resetting when exposed to bright light because they used a chip scale part in the power supply comes to mind.
In my opinion, the fact that the SOC is so sparsely documented (and in the Pi 4's case, almost entirely undocumented) is a way bigger deal than the occasional small power supply design issue though.
Your comment makes it seem like any bright light will reset a RPi2 which is not accurate. The RPi2 will only reset when exposed to specific high intensity long wave length light. Really it is only triggered by xenon camera flashes at close proximity and laser pointers targeted on the RPi2.
My favorite is not releasing MIPI DSI/CSI documentation. CSI argument being 'we use proprietary processing pipeline, Broadcom spend a lot of money tuning this particular camera sensor, cant just let it go for free to you plebs, plus users are too stupid to implement their own anyway, buy our official camera module and go away'. Once Camera module v1 started being cloned in China Foundation released v2 with DRM build in! Crypto signature of camera, full Apple mode engaged! "chip is required to make the camera work" https://www.raspberrypi.org/forums/viewtopic.php?t=178038#p1... except of course its not, its there purely for hardware DRM https://twitter.com/marcan42/status/1088506051295567872
For DSI it was 'gotta recuperate costs' by releasing one proprietary DSI screen - still $60 now while off the shelf replacement iphone/ipod DSI screens are ~$10, but cant use those since interface is behind proprietary binary blob.
Even defects follow Apple example, claiming custom programmed LDO, MxL7704-R3, is proprietary information! Why would you want to repair something we designed wrong anyway? Just buy another one and go away. You really want to repair it? ok, ill charge you $35 + VAT and postage, more than replacement unit, now go away and stop bothering me. Oh and "Don't keep shorting out your power rails" https://www.raspberrypi.org/forums/viewtopic.php?t=238779#p1...
Btw MxL7704-R3 are available in China and cost $3.
1 repair will be more than replacement
2 you are holding it wrong
3 source parts from china
Two questions that come to mind that the blog doesn't address:
1. Is the change in this new revision easily identifiable via PCB silkscreen from its deprecated variant, or will consumers be 0402 resistor hunting?
2. Are (US) authorized resellers shipping this revision yet? None of the 8 for the US appear to have any notice on their respective websites.
EDIT: The Register article[1] cited by the blog mentions the following almost-but-not-quite-relevant tidbits:
> The update, Upton told us, had also moved "the WLCSP SD card voltage switch to the top side" to protect it from damage, and also "silk screen tweaks to reduce solder bridging in manufacture".
> We asked Upton this week if the fix was out and he confirmed the update had indeed been rolled into a PCB Design for Manufacturing (DFM) process. He added that he would have expected the update "to have reached end users by now".
Someone already figured out how to tell them apart: There's a small voltage regulator component that was originally near the edge of the board. In the new revision, it has been moved to the previously empty space above the word "micro" in "micro SD card" [1]
So now you just gotta find a reseller willing to check that they're sending you the correct version.
12 comments
[ 3.3 ms ] story [ 31.0 ms ] threadIn my opinion, the fact that the SOC is so sparsely documented (and in the Pi 4's case, almost entirely undocumented) is a way bigger deal than the occasional small power supply design issue though.
While some problems should have been found already during testing the comment still is indeed overly dramatized.
For DSI it was 'gotta recuperate costs' by releasing one proprietary DSI screen - still $60 now while off the shelf replacement iphone/ipod DSI screens are ~$10, but cant use those since interface is behind proprietary binary blob.
Even defects follow Apple example, claiming custom programmed LDO, MxL7704-R3, is proprietary information! Why would you want to repair something we designed wrong anyway? Just buy another one and go away. You really want to repair it? ok, ill charge you $35 + VAT and postage, more than replacement unit, now go away and stop bothering me. Oh and "Don't keep shorting out your power rails" https://www.raspberrypi.org/forums/viewtopic.php?t=238779#p1...
Btw MxL7704-R3 are available in China and cost $3.
100% Apple service experience.1. Is the change in this new revision easily identifiable via PCB silkscreen from its deprecated variant, or will consumers be 0402 resistor hunting?
2. Are (US) authorized resellers shipping this revision yet? None of the 8 for the US appear to have any notice on their respective websites.
EDIT: The Register article[1] cited by the blog mentions the following almost-but-not-quite-relevant tidbits:
> The update, Upton told us, had also moved "the WLCSP SD card voltage switch to the top side" to protect it from damage, and also "silk screen tweaks to reduce solder bridging in manufacture".
> We asked Upton this week if the fix was out and he confirmed the update had indeed been rolled into a PCB Design for Manufacturing (DFM) process. He added that he would have expected the update "to have reached end users by now".
[1] https://www.theregister.co.uk/2020/02/21/pi_4_fixed/
So, how do I make sure I’m purchasing one of the new & improved ones rather than one of the faulty ones?
So now you just gotta find a reseller willing to check that they're sending you the correct version.
[1] https://www.cnx-software.com/2020/02/24/raspberry-pi-4-rev-1...
https://www.raspberrypi.org/forums/viewtopic.php?f=63&t=2596...
It seems like most of us will have to rely on online vendors beginning to proudly advertise “we have the new one!”.