Is it just me who feels that having one processing unit per display is a waste?
I mean, I get it why they did it (it’s way easier to just have one SBC per-display, both on the hardware and the software sides), but if designing such a system I would still try to come up with a single board solution if only because waste gets on my nerves.
It could just be a backlit panel that you place a semi-transparent logo in front. Could be magnetic or slid into place. More resources than a sticker but probably far less than a system-on-a-chip running an OS and displaying the same picture on a monitor all day.
But what about yet another bright light in someone’s face? Do you not want another bright light in someone’s face? Everyone loves bright lights in someone’s face!
I’d argue that a custom board is more wasteful since they are single use. Using a cheapo COTS processor that drives a single display and runs Linux is reusable in the long run.
True, such a low number of production units design would really only make sense if you could find an off the shelf solution to drive multiple displays.
If these displays are not supposed to be animated and they’re reasonably low resolution (say, 800x600 20bit RGB or less), they could be connected via SPI and pretty much every microcontroller out there has multiple SPI ports, so even a cheap SBC would work for that). However I expect that getting XWindows or Wayland in Linux to work with such displays would be a PITA.
I’ve only ever got software running under Linux to control a tiny 2-tone display via I2C - on an Orange Pi SBC - and it’s totally its own thing which happens to be running under Linux sending low-level commands via the I2C dev and not at all integrated with X-Windows or Wayland. This would also work fine if the comms was via SPI (in fact the code barelly changes since I’m using a library that does most of the low-level work for me).
To just display a static image or a sequence of static images loaded from storage in a bunch of screens low-resolution enough to support SPI (so 800x600 or less) I expect something like that would be fine.
The more I think about it, there more I expect this thing could run on a single $50 SBC as long as the connector exposes at least an SPI device and 8 independent I/O lines (given how SPI works, shared SPI bus is fine with one separate Chip Select line for each screen as long as the SPI device under Linux can run on a mode that lets your code control the CS line itself, and the other 4 I/O lines are for touch detection) assuming touch position is irrelevant.
My local gas station now has screens on the pump. Not the big unit, but the part you put in your gastank. It shows shitty ads. Also in the Netherlands you can’t lock the gas pump, so you have to manually press it to get more fuel, so you are almost forced to watch shitty ads.
Is it just me who feels that having one processing unit per display is a waste?
I mean, I get it why they did it (it’s way easier to just have one SBC per-display, both on the hardware and the software sides), but if designing such a system I would still try to come up with a single board solution if only because waste gets on my nerves.
You’d think a damn sticker would be good enough
Human replaceable printed paper labels, manual stick.
What is this? The 20th century?
It could just be a backlit panel that you place a semi-transparent logo in front. Could be magnetic or slid into place. More resources than a sticker but probably far less than a system-on-a-chip running an OS and displaying the same picture on a monitor all day.
How do you charge for a service contract on a sticker?
This guy B2Bs. Y’all think companies aim for efficiency when their client is a megacorp? Heeelllll no. Corporations bleed each other out, too.
But what about yet another bright light in someone’s face? Do you not want another bright light in someone’s face? Everyone loves bright lights in someone’s face!
I’d argue that a custom board is more wasteful since they are single use. Using a cheapo COTS processor that drives a single display and runs Linux is reusable in the long run.
True, such a low number of production units design would really only make sense if you could find an off the shelf solution to drive multiple displays.
If these displays are not supposed to be animated and they’re reasonably low resolution (say, 800x600 20bit RGB or less), they could be connected via SPI and pretty much every microcontroller out there has multiple SPI ports, so even a cheap SBC would work for that). However I expect that getting XWindows or Wayland in Linux to work with such displays would be a PITA.
I’ve only ever got software running under Linux to control a tiny 2-tone display via I2C - on an Orange Pi SBC - and it’s totally its own thing which happens to be running under Linux sending low-level commands via the I2C dev and not at all integrated with X-Windows or Wayland. This would also work fine if the comms was via SPI (in fact the code barelly changes since I’m using a library that does most of the low-level work for me).
To just display a static image or a sequence of static images loaded from storage in a bunch of screens low-resolution enough to support SPI (so 800x600 or less) I expect something like that would be fine.
The more I think about it, there more I expect this thing could run on a single $50 SBC as long as the connector exposes at least an SPI device and 8 independent I/O lines (given how SPI works, shared SPI bus is fine with one separate Chip Select line for each screen as long as the SPI device under Linux can run on a mode that lets your code control the CS line itself, and the other 4 I/O lines are for touch detection) assuming touch position is irrelevant.
My local gas station now has screens on the pump. Not the big unit, but the part you put in your gastank. It shows shitty ads. Also in the Netherlands you can’t lock the gas pump, so you have to manually press it to get more fuel, so you are almost forced to watch shitty ads.
It’s exactly like this https://www.team-bhp.com/news/ads-fuel-dispenser-nozzle-havent-seen-anything