Raspberry Pi 4 -- is it good news for experimenters and enthusiasts?
No-one can deny that the original Raspberry Pi was a significant innovation
in single-board computing. It didn't really
break new technological ground, but the Pi made the technology hugely
more accessible.
The
ready-built, well-supported Linux operating system, and desktop-style
peripheral connections, made the cost of uptake (both in money and
time) almost negligible. Suddenly, single-board computers were mainstream --
they started turning up in schools and sheds.
It's probably fair to say that even the first Pis were over-specified for the kinds of applications they were often used for -- you don't need a 1 GHz CPU running Linux to flash an LED. Still, at £30 this wasn't a big deal for experimenters and builders of custom, one-off electronic gadgets.
At the time of writing, Over 30 million Pis have been sold, exceeding anybody's expectations. The Pi has created an entire industry for peripherals, cases, accessories, books, and kits. There's no doubt the Pi is a success. I've used the Pi in many applications since its original release. I build my own Linux for them, I put them in routers and network storage devices, distributed audio systems, radios, robots, and remote cameras. I've written on programming the Pi in assembly language and hardware interfacing. As an enthusiast, you might expect me to welcome the new Pi 4 with open, um..., pliers.
The Pi 4 was released about six months ago, and its reception among technology journalists and reviewers was universally enthusiastic. It's not difficult to see why: the new model sells for approximately the same price as the previous one (the 3B Plus), but it's significantly more powerful. It has dual HDMI connectors, USB 3 support, more RAM, and faster everything. What's not to like?
It quickly became apparent that the Pi 4 had a major weakness compared to its predecessor: it ran hot. Burn-your-fingers hot, even when not doing very much. Since then, firmware updates have tamed this problem to some extent. Still, it remains unclear whether the full potential of the Pi 4 can be reached without external active cooling. That's a fan, in plain language.
This doesn't seem a big deal, on the face of it -- fans are cheap, and don't draw huge currents. Needing to use a fan is a small price to pay, for all that extra power, right?
Well, maybe.
Using a fan, and the large heatsink it entails, significantly increases the installed size of the Pi. Even the quietest fans generate at least a little noise, and consume at least a little current. More importantly from an experimenter's perspective, most of the fans currently available block access to the GPIO pins. It is these pins that give the Pi its tinkerer's credentials. Without access to them, few if any of the most widely-user peripherals will be of any use. Quite apart from the GPIO issues, the need to use a fan demonstrates a lack of suitability for embedded applications, especially those that run for long periods of time.
I have complained, on this site and elsewhere that, as a device stated to be aimed at the enthusiast and education markets, the Pi has a remarkable dearth of detailed technical documentation. I've often wondered if the Pi is really intended for this kind of customer. I don't need to wonder any longer: the official website makes it perfectly clear:
"Your new desktop computer
The speed and performance of the new Raspberry Pi 4 is a step up from earlier models. For the first time, we've built a complete desktop experience."
To be fair, there are an awful lot more desktop computer users out there than there are experimenters. There's a lot more money to be made from casual users than there is from experimenters, who are notoriously stingy.
I have to wonder, though, who's going to run a Raspberry Pi as a desktop computer? Attempts to replace the horrible, high-cost, power-hungry, proprietary Windows desktop with something more economical have all, so-far, failed. I worked on some of those projects myself. Anybody remember the SunRay? This was Sun Microsystems' attempt to sell big businesses on the idea of low-energy desktop terminals and a central server. Technically, it was an ambitious, highly capable system. The idea that you could carry your desktop session around on a smart-card was particularly intriguing. A SunRay installation used about 1% of the energy of a building full of Windows desktops.
Did it sell? Nope.
Windows 10 has absolutely nothing to recommend it beyond the fact that everybody uses it. That, of course, is like saying cars have nothing to recommend them apart from a massive road infrastructure. It's now possible to buy a 'mini PC' running Windows 10 on Intel hardware for under £100, whose performance will trample a Raspberry Pi. If you factor in the cost of a power supply, case with cooling fan, and cables, the cost saving of a Pi 4 over a Windows mini-PC is about £40. Is anybody going to give up the familiarity of the horrible Windows 10 desktop for a saving of £40? Somehow, I doubt it. Outside the IT industry, only enthusiasts run Linux.
It's just about possible that there are some custom projects that need the Pi 4's increased power. I can see the extra CPU gumption being useful in image recognition, for example. On the whole, though, none of the new features of Pi 4 are of any interest to me. What possible use is 64-bit platform support on a single-board computer? Or a phone, for that matter -- but that's a whole other rant.
So what would an experimenter-friendly Pi 4 have looked like? Here are some new features that I think most of the community would have welcomed.
A built-in real-time clock.
More robust GPIO lines -- it shouldn't be necessary to buy a new board if your multimeter probe slips.
The ability to run a Pi on 12V or higher, rather than 5V. This would make building the Pi into other equipment much easier.
Low-power standby support, so the Pi could run unattended for long periods of time -- particularly important for remote data collection.
Software-controllable under-clocking, to reduce power consumption.
A reset button. How could a device like this not have one?
At least a couple of programmable, on-board LEDs, for status monitoring.
A real audio DAC.
A serial console connection that doesn't have to be soldered on.
Built-in battery back-up and charging, with voltage monitoring.
...and many more.
It was enthusiasts, experimenters, and educators that made the Raspberry Pi the success it is. I fear that these people are now being ignored, in the rush to make inroads into the highly lucrative desktop market, where the Pi is unlikely to be a success.
The good news is that Pi 3 boards are still available for sale. Get them while you can.
