Raspberry Pi Slips Out New PCB Version With USB C Power Fix

When the Raspberry Pi people release a fresh model in their line of fruity single board computers, it’s always an event of great interest. The Raspberry Pi 4 brought some significant changes to the formula: they moved to mini micro HDMI and USB-C power sockets, for instance. The early adopters who scored one of those Pi 4s were in for a shock though, if they had all but the most basic USB C power cables the device wouldn’t power up. Now the Register has news that they have slipped out with little fanfare an updated version of the board containing a fix for this problem.

Our colleague Maya Posch delved deeply into the USB C specification and delivered a pithy analysis at the time which demonstrated that the fault lay with the configuration of the sense resistors used by intelligent USB C power sources to determine what power to supply. For the addition of a single surface mount resistor the problem need never have existed, and we’re guessing that’s how they fixed it.

There’s no need to despair should you have one of the older boards, though. They will still work as they always have done with the so-called “dumb” power supplies and cables, and meanwhile we’re sure that future Pi boards will have had a lot of attention paid to their USB power circuitry.

102 thoughts on “Raspberry Pi Slips Out New PCB Version With USB C Power Fix

    1. Ugh. Revision #.

      I kind of wish they hadn’t fixed this. Knowing that we must use a dumb supply or add a resistor is one thing. Now we don’t know until we sit there and rummage through boxes looking for one with the right revision number. And that’s only for the minority of us who buy these things in a store. Order one and you get to wonder what your are going to get until it arrives.

      I think they should have just left it along until the Pi5 and then made double sure to not make that mistake again.

      1. >>Now we don’t know until we sit there and rummage through boxes looking for one with the right revision number.

        Unlikely, stores will generally only have the new revision OR the old one as they would probably want to get rid of their “inferior” models first.

      1. Both connectors are over a decade old, and being replaced in the consumer world by usb-c, so consider the war lost. Micro HDMI is common on laptops, and has been the HDMI port of choice for the BeagleBone boards since 2013, but the only application for mini HDMI I’ve seen is the Pi Zero.

        To be honest, the full size HDMI connector on previous Pi was unwieldy. A typical plug and bend radius to bring the HDMI cable out in the same direction as USB and Ethernet takes up at least as much width as the Pi itself.

        1. The full size HDMI plug is standard on TVs and monitors, and cables going from large to micro HDMI just complicates things.
          OTOH, with dual micro I can see an easier path to stereo VR for the RPi.

    1. The mini HDMI on the cool right out of the box instantly pissed me off, Ive got damn near every cable known to man and low and behold the only one id had was multi adaprted to hell with 5 or so other damn male fittings, then not only did the original power supply which idk why I got… Ahh it came with the pi 4 id gotten, but yeah their own cable didnt even power the damn thing up, tool a good hald a bottle of taquilla, some water half a hell if i remeber it was eatable and 9hours later I came up with a well similar fix then the one mentioned in this post but damn I’m kinda disappointed in myself and tbh I was starting to think theyd done it on purpose to give people a little hardware learning and what not but then again after hour 10 i was convinced the charger was at fault sept I checked it in probably 3 plugs around the house. But ultimately grabbed a hammer and nailed it to the wall and said Fuck this shit I got better things to do.. (The wife made me stop, so cruel.)

    1. Hey, mistakes occasionally happen even for the very careful sometimes. I think I’m gonna give ’em a bit more time to see if it becomes a pattern. Two events is technically a pattern but only just barely; can’t really honestly say there’s a trend with that many data points.

  1. Yeah see, the important thing isn’t that you never make a mistake, that’s impossible, it’s how you react to making the inevitable mistakes, and this is great example for all OEMs to follow in terms of product support. Ok so you can’t change the boards you shipped but you can, ya know, fix it in a reasonable time and start shipping new ones that are fixed. Good example of why they have such a solid reputation and widespread success. They don’t dump steaming hot expensive garbage on an unsuspecting public and expect them to QA it, hi Samsung hi Motorola. Aarch64 Raspbian would be nice but that ones entirely on Broadcom, the Oracle of chipmakers.

    1. And then just kinda not fix anything not deemed important enough once the public is done testing it for you, or even making things worse. This is the kind of care and attention to detail that shows they really care about their products and take pride in their work. As both an engineer who often does not, and a pretty happy customer other than with Broadcom’s drivers, that’s a tiny thing in the grand scheme of things but really pretty inspiring.

      1. Or like a certain expensive drone controller with a light version that was utter shite, but for another $130 or so you could buy the firmware upgrade to fix it. Dirty little secret was it was exactly the same firmware as on their higher priced model of the controller, proving the company could easily fix the GPS and drifting issues on the lower cost model. They just flat out refused to fix their crap for free as they ought to have in order to make the lower priced model actually operate as advertised.

    2. “Aarch64 Raspbian would be nice but that ones entirely on Broadcom, the Oracle of chipmakers.” – Raspberry Pi is using 64 bit SoC since RPi 3 (actually, A+ had it too, but version 2 was 32 bit only). 64 bit raspbian was possible since then, as there are other distributions that work on Pi3 and 4 in their 64 bit version. Ubuntu server was running on my 3B without a problem, and I believe even Debian has 64 bit versions for RPi.

      But luckily it seems to be changing, as there is an experimental aarch64 kernel for Raspbian. It’s development is focused on RPi4, so while it will work with earlier 64 bit versions it might be more buggy there. To use it just add “arm_64bit=1” to config.txt and do a rpi-update.

  2. Any news on the eventual possibility of an 8gb pi 4? I remember there was something labeled as such on a spec sheet at some point, something like that, but it’s still kind of speculative.

  3. Hardware is following the software life-cycle. Get suckers to beta test for you, for free, and fix it, and get the same suckers to pay for the fix.

    Our standards are falling.

    We used to expect high quality, now we expect “good enough”, and pay the same price as we did for high quality.

    Cell phone cameras are “good enough”
    Cell phone voice quality is “good enough”
    Cell phone data is “good enough”
    Cell phone processors are “good enough”

    Look at video games, they release a PoS, and it’s “good enough”.

    Cartridge based games, with few expections, had very good QA.

    Now, crashes, patches, fixes, are “De-rigeur”, and we even pay monthly fees for shit products, I’m looking at you, XBONE-Gold.

    Very few people take a stand against this. For example:

    -Don’t pre-order, wait for a finished product
    -Don’t pay monthly fees, you either own it, or accept that you rent it.
    -Expect to complain, expect your money back for shit products, online or otherwise

    1. All these people shitting on RPI4 are obviously forgetting that it’s a $35 computer. Sure, maybe we “used” to expect high quality, but we also used to pay tens of thousands of dollars for hardware that could do a tiny fraction of what an RPI can do. And if you went out today and spent ten grand on a piece of computer hardware I imagine you’d have similarly high expectations. But the RPI has always been so cheap it’s literally disposable. And despite this low cost, they have great support, an excellent software base, and routinely upgrade their stuff.

      Back in the 80s, $35 or it’s pre-inflation equivalent might get you one of those fake cardboard computers you could hook up a couple flashlight bulbs to with paperclips.

      1. i kind of agree about the case with the pi. you get what you pay for. there are higher quality spcs out there but they can get damn expensive and fast.

        as for other products disposable mediocrity are the primary operational parameters. but idiots still buy so they get away with it.

      2. Certainly not crapping on the Raspberry Pi foundations efforts, they make great stuff. I have at least 2 of every generation, they really are great computers.

        But it seems for far more expensive products, we are happy with “good enough”.

        iphones that don’t work left handed.
        iphones that have poor software support.

        When I was doing embedded design in university, we were told, “if you need to hit the reset button on your final product, you have failed”. Yet, here we are today, where the reset button is the cure-all.

        reboot, reset, reinstall.

      3. Have you seen the bargain bin lately? You can get used x86 computers with accessories included for $50-$100 and most of them will blow the socks off of the entire Pi series in CPU performance. And you can tap into literal decades of standardized ATX components for expandability. If you want an example you can visit American eBay right now and buy a complete quad-core HP T620 for around $40 and shipping. According to Parkytowers the T620 unit they tested ran at around 8-11 watts. Get better performance, similar TDP and prevent e-waste!

        I’m really trying to hold back my disdain here about how utterly wasteful Pi’s are. You need to jury rig the highest rated tablet power supply you have and a microSD card at minimum to run it headless. Perfectly reasonable, many of us can do that. But wait, you must realize most users opt to buy at least $20 of accessories to skip the headaches caused by the Pi Foundation’s half-baked engineering. It’s 2020 for chrissakes, using unreliable microSD cards for the OS is a sick joke. And use standardized 12v barrel connectors already. Never happening of course, pushing accessories is a profitable game. Dongles for HDMI? Pi’s are really good at signage and using flimsy display connectors makes for an oxymoron.

        And on the subject of self-sabotage none of the Broadcom SoCs in Pi’s come with AES acceleration despite the physical hardware sometimes having the capability. Nobody paid royalty fees to use the instructions to penny pinch! This makes me irrationally angry I’m sorry. Even junk phones from half a decade ago have AES acceleration (see the Snapdragon 801). Android pushed for full disk encryption to be default since around Android Lollipop, and manufacturers included AES acceleration in their SoCs to match that requirement. I know this is a false equivalency, but your personal phone is expected to support decent encryption but not your computer marketed as a desktop replacement? Look at how slow AES is on every Pi, it’s literally wasted watts that could’ve been prevented!

        Let’s get back to the point though. No, I’m not going to excuse a ‘cheap’ product for being full of idiosyncrasies because the company behind it is a sleazy salesman disguised as a non-profit and that makes me irrationally angry.

        1. MicroSDs are crazy reliable now if you buy a genuine SanDisk industrial.

          eMMC is kind of awful. You can’t swap it out except with expensive proprietary modules. And M.2 is too big, and also expensive.

          I think the Pi foundation is doing quite well, but the general industry standards for hardware are pretty dismal.

        2. Also AES is not even used all the time. TLS can use all sorts of ciphers. ChaCha is probably fast enough considering the speed of browser rendering on that kind of HW.

          Full disk encryption for this is NOT something you want to encourage. We don’t need more encrypted digital signage or kids learning kits with locked down OSes. I don’t even use it on my personal stuff, although it does have value for some there.

          12v barrel connectors are slowly becoming less standard than USB-C, and you can’t find power banks using them nearly as easily. Just get a DROK synchronous converter if you have 12v, it’s a better plan anyway since anything but direct wall warts connections usually aren’t clean 12v.

          1. Correction: I meant the Snapdragon 810 from 2015 had AES acceleration, not the 801. Off by one.

            Yes, you’re right that ChaCha has good thoroughput on most devices but it’s not a proven solution for FDE. Nor does it match the throughput of accelerated AES. Most ciphers on the Pi 4 achieve a few hundred MB/s at best compared to GB/s in most products with AES acceleration. See benchmarks for the aforementioned Snapdragon 810 from 2015.

            We take disk encryption for granted because it’s usually cheap to implement. Contrary to your opinions about it, it’s quite normal to see in everyday devices since it’s implemented transparently for ease of use. Your smartphone probably utilizes some form of filesystem encryption and most consumer laptops do the same. The performance impact in computers is usually negligible; the implementation usually encrypting the filesystem with AES and then storing the decryption ‘key escrow’ style in a Trusted Platform Module (BitLocker and whatever Apple calls their implementation).

            The tragedy of the Pi is twofold; not paying for AES instruction royalties and the common use case of storing the OS on a microSD card. Not only will the crypto be slow and inefficient, any filesystem encryption you do will complicate recovery when your storage medium inevitably glitches. Let’s be totally honest here, most microSD cards consumers use are not even close to being as reliable as other flash storage. How many consumers actually use their Pi’s with industrial microSD cards? If it’s required to operate the product reliably I reckon it should be included and not sold as an accessory. Deja vu!

            You can’t market the Pi 4 as a desktop replacement when phones with ARM SoCs from 2015 somehow do AES 10x as fast and have FDE on by default. This is just failing the consumer.

            Regarding 12v barrel connectors, anything’s better than the unfortunate 5v situation. You start with 5v on the board and it’s not a big surprise when attaching anything to it can cause enough voltage droop to turn off and reset the board. Am I being obnoxious if I expect the 4x USB ports on the board to all work properly? I can reset my board by plugging in two wireless adapters at the same time and they don’t even draw that much current. Is this somehow not a valid use case for a desktop replacement? I know I’m giving off the stench of first-world problems but quite literally no other device I own with USB ports has this problem.

            The various engineering problems in Pi’s are just dumb. Probably not due to malice, just dumb and silly.

          2. FS encryption is common, but outside the dev community not everyone really cares. I can see why some Android devices make it mandatory, but I don’t think I’d use it on a laptop by choice, and I’d be even less likely to want it on a machine that stays at home(Or just plain doesn’t have anything remotely confidential) like most Pi boards.

            Encryption is nice to have, but not essential in one-off embedded, educational, and smart TV applications.

            Android phones put the extra effort into FDE because you often can’t turn it off, and they’re a much higher theft risk that often have full access 100% of someone’s personal and business stuff, their texts to their pot dealer, and their private diaries.

            If FDE was more efficient, there’s also the possibility that it would encourage locked down devices even more, and I’d much rather give up a lot privacy than any amount of control of hardware I paid for.

            Also, the fully swappable SD cards are nearly ideal for digital signage and any other embedded systems tasks, because you can perform upgrades instantly and roll back with almost 100% confidence, and it requires zero training to swap to a newer card.

            Plus, the flash is the one thing that has a defined life, and it’s also the most useful thing to be able to expand.

            Multiple wireless adapters do seem like a perfectly valid use case(Mesh networks are great), but running on a 5v adapter is far more common.

            I expect the pi 4 partially improves that, because of the higher current limit on C, but yeah, the “real” ideal way to do things would be to use a buck-boost chip with 5-24v range, and include a software-defined battery charger, as there’s almost no application that wouldn’t benefit from a battery, and solar use is common.

        3. There’s always some utter w*nker who says “you can but a laptop on eBay for $40”. Jeez, I’ve put better PC’s in the garbage because I don’t want to spend the rest of my life downloading updates. I have 22 USB power supplies from 5 generations of family phones, plus 3 gens of iPads etc. I have more USB power cables than grandchildren and 58 USB memory sticks. Any combination of 3-4 of these turns an R Pi into another useful bit of kit and I get to use my brain to sort out Linux rather than rot it trying to fix win*doze. This stuff is actually FUN to do. Aaargh – let’s keep computing worth doing.

    2. I do wonder if perhaps you are too young to remember what we actually got for $35 back in the good old days. If you think we are paying anything close to the same price as we did back then for computer hardware, you’re dreaming. Although I don’t necessarily degree with the rest of your rant, that piece of hyperbole trashes any value the rest of your post might have had.

    3. I blame Agile and the culture that goes with it.

      It a general philosophy of a constant incremental flowing “process” rather than a solid product you buy, and use for ten years without even thinking of maintenance.

      Everything is built to serve one particular need, Unix philosophy style. Avoiding E waste means making things general purpose enough to serve reasonable future needs, be compatible with relevant standards, etc.

      I really don’t like the agile programming philosophy much at all. I like integrated packages, built to last, to handle all the edge cases, and to generally be “solved problems” not continual processes of maintenance.

      I could still use 5 year old LibreOffice, although new version is much better. That’s the kind of project I respect.

  4. Why not sell old rev board at a $10 discount and advertise new board at current price. This “secret update” is crap biz.

    I’ve been on the fence whether I need/want a RPI4, and this makes me want to wait even longer until I *need* a board, where I keep holding off ’cause I just *want* a board.

    1. I tend to stay one version behind with the PIs, not really because of issues like this. But because of retailers gouging the prices of the boards, or their single boards are always sold out, but hey we’ve got this $100 bundle of useless cables and other crap that you probably already have 50 of at home. I really wish the PI foundation would put their foot down on this practice. It pretty much means you can’t find a single alone board at MSRP for at least 6 months or more. Now i’ll probably be waiting even longer for a Pi4 for all these original revision ones to work their way out of the supply chain. These original revision ones will probably end up being the ones you find being sold alone, while the new revisions will only be purchasable in some over priced bundle for some time.

      1. I wish they’d start selling directly through Amazon/eBay/best buy/etc more. They’re big enough and those kit companies don’t add much value, except that one aluminum case.

        Pi should just do their own bundle, with a good SanDisk industrial card, HAT friendly case, and maybe even skip the USB supply.

    1. That’s a cop-out. The hardware engineer is responsible to read the necessary documents and specs as part of the job. Not having the time to do so and making a defect product does not escape the liability nor responsibility.
      I have in my profession work on stuff that requires signing off on my designs as well as any component deviations from purchasing. That’s part of the profession.

      Engineering is pretty much about reading specs, design, testing. This “bug” is traceable to a fault in the design and insufficient testing.

      It is like you failed an exam because you didn’t study or didn’t understand a 500 pages paper. Guess who’s fault is it? Ultimately it is your failure.

      Anyone can make comments and no one is getting paid for his/her time nor is it a profession. The expectation is a lot lower. I have been to a lot of design review. Most of the engineers do not went through the same details as the designer(s) and yet they are welcome to comment on a design as someone might see things that are missed.

      A simple question such as “Did you copy that USB C power delivery from the example from the specs?” “I modified it to save a resistor” “Did you sufficiently tested your version of the mod?”

      So bugger off with the fan boy weak excuses.

      1. Spot on – anyone would use a reference design from the data sheet or reference manual (these days, the DS’s can be as bulky in content as traditional reference manuals) and IF the reference example nearest your requirements isn’t close enough, then you properly develop and test a reworked version of the ref design and when it’s still fundamentally ‘compliant’ and technically stable with a standards compliant commonly available PSU or other equally compliant stabilised source (USB port on other device, if consumption demand is low enough) – then it can be signed off as a working sub assembly module and not until.

        You just don’t see this QC failing and idiocy of corner cutting in equipment that simply has to work (SBC’s that have to meet some kind of standardised TA framework) and when you talk about something that’s going to be sold in bulk quantities to educational establishments, you can’t afford to get it wrong as they don’t subscribe to the ‘bin it as a bad move’ philosophy – they continue using their investment until it’s literally too unreliable to maintain. Likewise, with the pattern of foul-ups they’re making over 4 main versions and subsequent point revisions, they’d be a laughing stock if the Foundation ever tried to launch and sell an industrial grade design (thankfully, the CM3 modules are good enough for serious use).

        If serious (non-consumer) microcontroller and high grade SBC reference dev boards designs (manufacturer designs and recognised high-quality TP ones were as full of as the Pi series, the world would be in real trouble. Thankfully, in the maker market, mc reference designs and good derivatives are over engineered enough that even if some part choices are questionable, the items are fit for purpose from the outset and will outlive any RPi design.

        If ICOM had been as shoddy as RPi Foundation over the IC-7300 SD Transceiver (which had issues, but relatively minor and firmware rectified worse case), the whole exercise would have been a bloody disaster, and given the 7300 was built for a radio ‘maker’ and educational oriented market – it’s a good indicator of why attention to detail matters.

    2. Apparently an example layout is provided to prevent issues like these and Raspberry decided to ignore it and produce what they had. If you’re not smart enough to roll your own you should be smart enough to copy the homework.

  5. Criticizing someone else’s mistakes is as easy as anything can possibly be. If you’re always and only looking for something to complain about you’ll always find it. I’ve worked with a few people who made that a habit. When they did it to me I explained that shit flows both ways and I could do it to them too. Everyone of them denied I could and I easily proved them all wrong in short time and they never did it to me again.

    That Pi 4 is the most powerful and useful computer for the price there is. If anyone thinks they can do it better, or even as good and and sell it for less, all they need to do is prove it and I’ll buy it.

  6. The USB c was/is a non issue. Use an appropriate cable and be done with it. It’s not like it doesn’t work or is crippled. Nice to see they fixed it to stop the crybabies who can’t afford a cable.

    1. My guess is your “proper cable” is probably a hack of a cable that has the resistors in the cable itself and just does not connect those lines to the Pi. The resistors that tell the power supply what power the device is capable of sinking should be on the device, not a hack of a cable. What happens when you plug that hack of a usb-c cable into some other device with different power requirements than the Pi, it’s either A) not gona work because it is being under powered or B) Fry because you are pumping too much voltage into it cause the resistors in the cable told the power supply to deliver that power.

    2. The USB-C implementation on the Pi4 is a wanna be hack to placate people who have been whining so long for USB-C. If it were done right they would have put a proper buck converter on the Pi that it could actually sink the full 20v that is capable of being delivered over USB-C and completely eliminating the power issues that have plagued the Pi since it’s inception. Instead they pretty much just swapped the micro usb plug with a USB-C plug and called it a day. It still only takes 5v in, leaving you with the same voltage dropout issues that exist with micro usb when the current demand is too high.

      Had they done it right with a buck converter that can sink 20v the brownout issues the PI has would never see the light of day again. With a proper buck converter they could run high current traces straight from the buck converter to the USB ports, and have a completely separate power rail for all the on board hardware, never again would a high current USB device plugged into the PI cause it to brown out and crash.

    3. Quite frankly USB-C is a POS The plug should have had physical locking prongs on it like micro USB. USB-C is purely a friction fit connector, once you’ve plugged and unplugged it enough times to wear it a bit the cable just comes loose. I have a dell laptop at work that uses USB-C for charging/connectivity to the dock to drive my 2 monitors, keyboard, mouse and get it’s Ethernet connection. After a year of daily plugging and unplugging it the port and or cable is now loose. Shift the laptop on the desk a few inches and theirs a better than 50/50 chance that the cable is going to come loose and drop all connectivity to the dock.

      I have older microusb cables that have been used daily to charge my phone that still keep a more positive physical connection. Like literally being able to grab the cable and have the phone just dangle there without the USB plug slipping out. Try that with USB-C, you’ll probably be taking your phone in for a screen repair.

      They could have added a prong to each side of the USB-C cable, like the two upward/downward facing prongs on a micro-usb and not made it any physically thicker where it matters in the application of something like a phone. This would have given it a more physically positive connection that would not just slip out of a device.

      1. klsjdfkl klsdjfl, USB-C does have locking prongs, they are on the inside and come in from the sides, have a look down the connector again.

        One of the many beauties of the USB-C connector is the bits that wear are on the replaceable part. Note even on your Dell Dock the cable comes off (its not a USB-A cable but look again at the doc and you will see it comes apart to change that bit).

        1. You must buy trash cables then. I have had only maybe 2-3 ever go bad and they were usually some cheap junk cable that came with something. I have multiple cables from https://www.cablestogo.com which were bought in a brick and mortar compusa before they went out of business actually during the going out of business sale (to give you an idea how old they are) that still work great to this day.

  7. the point was, the specs pages are quite a few, and I would place bets most people would just look for vcc/the/data like the old spec, and possibly miss the specifics USBc 3.2.5.3.7.2… demand for proper ID of a charger made in a southeast province of wherever. it’s alot to read. I hear all the “tech reviewers” complain randomly about devices “it’s still USB mini/micro/b/a/69…” and think it’s just a drop in part. I’m so damn tired of people calling a spec oversite on a 35 Dollar device a total POS device now, even if it still works with a dumb old last eon cable. there are so many other SBC out, migrate to them if rpi is such crap.
    guess what bothered me on the 4, HDMI powersave mode, so I took the steps to go-to GitHub and toss my voice in the bug thread and it is now fixed. what did most of you do? go blog about how much you hate the rpi for changing it’s design, or the doctor being female or how Ryan J ruined your childhood and you “can’t even”… I’ll continue enjoying one of my well supported 35$ sbc that I can setup in seconds to control many things.
    the person that read the USBc spec and actually said what the issue was is the only one that has earned the priv of saying rpi foundation screwed up, the rest of the people that only read an article about it can be silent.
    do you expect them to scream out, hey we fixed the issue from a tower? no, update, like they have been doing to many other issues and kernel bugs and just release it.
    about the USBc issue, the device draws 5 volts, I’m almost sure they had no intention of some other random device saying “Identify yourself or we won’t even give you 5 volts even of you will never need 20 some other devices would expect from me.” i hate the idea that USB can switch to that high of a voltage when all you may ever be able to handle in a project is 5 because you can’t be arsed to put in a regulator for your simple LED blinker.
    yes arm-chair was a pun.
    may a camera flash above your RPI 3 and none of your cables have a pull-up.

  8. yes there was a usbc issue, but I wonder if it was made intentionally to divert attention away from the horrible choice of HDMI connectors. (and broken cpu core)

    they sould have kept a single HDMI connector, and made it possible to mux DSI header pins to connect a 2nd HDMI !
    and they really should connect the 3rd DSI lane too (using old ipod/iphone retina screens would be so nice)

    for 2 displays they could make an Model C, with 2 USBC ports (for Power & HDMI over USBC only)
    and now they have PCIe they should make a Model D, with an PCIe switch-ic and PCIe-NVME M2-port on the backside.
    and a Model A, without USB2/3 and GBE chips/ports.

    If they put the 2 USBC Ports at the end (over the mSDC), all 4 models could use the same PCB.

    1GB Model A, 2GB Model B, 4GB Model C and 4 or 8GB Model D. is simpler than multiple Model B’s

    1. No, it was a simple mistake. And there is nothing wrong with the HDMI connectors, ana Model C as you describe would just be a horribly expensive waste of time with no backwards compatibility. And there is no third DSI lane, just two on the 2711. Although I am intrigued by the comment on broken CPU core – all my Pi4’s seem to work fine….

      1. To jump into it.

        By the nature of the concept, the crypto extensions are something that extends the base functionality. Since its never been advertised as having said crypto extensions, or claimed that it had them but them then being discovered as non-functional, “broken” is rather clearly trolling.

        (But yes, it’d be nice to have said crypto extensions, I think that the RPi people even mentioned as such.)

      2. https://developer.arm.com/support/arm-security-updates/speculative-processor-vulnerability
        just because it can be fixed in software, does not mean its not broken!
        the Pentium FDIV bug could be also be fixed in software…

        HDMI using USBC connectors would not add cost if they bit-banged USBC-PD/CEC using GPIO.

        I’m pretty sure it has at least 2x2lanes DSI (RPI1-3; had 2+3lanes)
        https://www.raspberrypi.org/documentation/hardware/raspberrypi/bcm2711/README.md

    2. You need to read Upton’s interview where he says he won’t ass $1 to the board unless it benefits the majority of users. A perfect Pi 4 would be $10-15 more but would no longer be as attractive to school kids learning to program, which is what it’s about.

  9. I’m going to need Raspberry Pi to get on the ball with engineering, WTF, I like the old standard HDMI, and what’s with the suprising c plug I’m glad I didn’t throw it away, I wonder how many people did, oh but at least they sold a bunch before they told us, holly chip fact , poe was most up two , oh and older TVs dont like the mini just an fyi , so not only do I have to buy a new cable but a new TV to,

  10. The only people who like the pi’s design and support the “simple mistakes” are the same people who don’t develop anything and think it’s reasonable to junk the computer when Mr. Computer man tells them Adobe flash player won’t work anymore.

    I don’t see the point in the pi anymore. It was ok for an inexpensive arm PC for particular hardware designs, which I myself make. But the design faults are real and to ignore them just increases the problem.

    “Target market is teaching kids to code” they can already do this in the browser, with greater support (flashy animations and the like to describe concepts) – a smaller, bare bones pc with a basic text editor doesn’t make it easier. Kids don’t need this.

    Heaven forbid you let them near the GPIO where the 5v 3v and screw holes are all close together.

    RPi is for marketing. My company mentions rpi = 200% more traffic. That’s all it is.

  11. Hi, I read about half of the comments on this page, so forgive me if I’m repeating something..

    When I decided to get a pi 3b it was because of the cost of running standard desktop computers. The power consumption of a desktop was equivalent to buying a new computer every month. So that was my reason for switching to pi and android as my primary devices for developing code for websites and Arduino..

    When I first heard about the pi4 i was looking forward to getting one.. I read everything I could about the pros and cons of the pi4.

    First my complaint about the pi3b. It freezes when you do something it doesn’t like.

    The 2 freezes I experienced most are:
    Watching full screen video on YouTube on 60 inch TV.
    Uploading sketches to the esp8266 that has cp2102 and ch340g uarts.

    I don’t have a pi4 yet, but the reviews I read had complaints about the
    USB c and micro hdmi connectors

    If I could redesign the pi from the ground up, it would have the following..

    A form factor that would fit in a extruded aluminum case similar to a hayes stack modem, although smaller..
    Barrel jack that can run from laptop power bricks
    USB c ports for hdmi
    Have a independent watchdog circuit that can reboot the system (if it freezes) via
    (Timeout or bluetooth)
    Have bug reports for each app that you can do after recovering from the system freeze!

    And do what mac failed to do with dual CPUs.. use second cpu for a full crash recovery system.. (ie: it could crash all day long and you would never know it happened)

    And most importantly! Support for more USB uarts!
    cp2102, ch340g, FTDI and others

    Get the LSUSB command to add drivers for uarts in the list…

      1. Thanks for the help, but the redesign of the pi I was hoping for. I know about cases that house the pi in it’s current formfactor. What I don’t like about the pi is having cables coming out of every side.. I did see one case that added an adapter board to the USB c, hdmi side to redirect the cables to the same side as the ethernet port. This gave the added benefit to protecting the flimsy hdmi connectors on the pi.

        The hayes stack modem case has edge rails that allow the PCB to side into the case from the end. The case is one piece tube with plastic end caps. This requires the PCB to have nothing on 2 opposing edges. This was the design I was hoping for, but I don’t know how many others would like/ appreciate it.

        The only downside of this is the pin header requires boards to be stacked. So the above case would need a header slot..

    1. “The power consumption of a desktop was equivalent to buying a new computer every month”

      Wait till you learn about the 60W incandescent bulbs everyone used. You would’ve gone bankrupt way before you even got to computers if your power bill’s that much. :P

      1. My mac g4 tower was 880 watts. If I remember correctly, I paid around $350 for it on eBay. I had it in sleep mode when ever I wasn’t using it. So it was powered 24/7. One month my power bill was $350 more than when I did a complete shutdown.

        That was what inspired me to get a pi and android. They only use 10 watts.
        I could run 88 pi for the same cost as one desktop…

          1. I was using about 8 to 16 hours a day for screen time. There might have been other factors adding to the electric bill. But that taught me a lesson in trying to find a less expensive way to get the job done.

            The only limitation of using a tablet over a desktop is drag and drop text from one window to another. On the desktop I could have 3 text windows open in one view and drag whole pages of code into new templates..

            While on the subject of multi platform code development. My major problem is invisible characters that apple/android/pi uses. I need a way to sanitize text so that it will run correctly…

        1. “My mac g4 tower was 880 watts”

          You don’t look at max PSU wattage unless you’re doing napkin math. Desktops typically idle between 50-150W so it’s more likely you had some other problem with your appliances.

          “That was what inspired me to get a pi and android. They only use 10 watts.”

          We can do some napkin math (lol) to see why comparing absolute wattages is a terrible idea for low powered computers.

          The Microsoft Model 1625 Power Supply is used for older Surface Pros, some of which are fanless. Power delivery to the laptop is spec’d for 12V 2.58A or up to ~31W.

          Pi 4B’s power supply is 5V 3A or up to ~15W.

          By golly, the Pi 4B seems to use elss wattage than Surface Pros! But wait! Look on Phoronix and you’ll notice that x86 CPUs tend to perform faster by a factor of double-digits in real-world use cases. So if you ever plan on doing any compiling whatsoever save yourself the sanity and just buy a low powered laptop. Phoronix did a piece on a $199 (more like $250) Motile M141 with a Ryzen 3 3200U and even that runs 5x faster than a Pi 4B. Caring about energy footprint is nice and all, but you’re looking at the wrong place. And if you use something that has terrible performance/watt like the Pi 4B you can actually waste more electricity than something that technically uses more wattage.

    2. “The power consumption of a desktop was equivalent to buying a new computer every month.”

      I’m going to assume that’s hyperbole, since even a gas generator fed at $2 a litre isn’t gonna cost you $300 to run a desktop that costs $300. Which probably idles around 40-50W at the wall even without turning off background updating etc… and only jumps up to 120-150W under heavy load. (Actual numbers taken from an X2 system I had running TV/PVR) and clue here, for heavy computational loads, like mine was doing transcoding videos, the Pi would need 20 times as long to finish the task. This is with no particular power saving configuration of the desktop, get it set up to sleep when idle, and it’s down to drawing a watt or so. Get a more modern $150 worth of sandybridge+ SFF desktop refurb/surplus and you can find those with 15W CPUs in, then you’ll be seeing 10W idling and 50W load, with probably 50% more ooomph thatn my X2, so now the pi is 30x slower on hard grinds. With that sort of thing properly configured, for certain tasks (Wake up, do heavy number crunching, go to sleep) it could end up more power efficient than the pi for the same task.

Leave a Reply to Migamix Cancel reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.