Tidbyt fits the bill for me. It's more expensive ($179), but it's a super polished product. It has all public transit for the main cities in the US, plus tons of other handy apps (stocks, weather, etc).
(No affiliation, other than being a happy customer. There's a promotion this week for $15 off the list price; DM me on twitter if you'd like the referral link)
Granted it's not an impressive e-ink display, but it's a complete all-in-one kit powered with a Raspberry Pi Pico and has 5 tactile buttons. So you're paying for the kit as a whole, not just the display.
The last HN article with the Lilygo T-Display (ESP32-based) that showed the news articles was cool, and screwing around with that led to finding an open source weather program for the better T-Display S3. Stupid easy, cheap and fun to throw together or mod.
you'd think they'd be useful for safety critical logging. You can guarantee some information on power loss like GPS coordinates, error states or situational Info like tidal information.
I always wanted to try overlaying physical controls for realtime readouts but use these displays for generated backgrounds like gauge scales. Or like a clock background that still had physical hands.
I am currently building an auto-updating webcomic display for a community bulletin board. It's of course a great application for an e-ink display because it only requires power to update.
Further implementation details: I'm using an ATTiny85 to feed 120 seconds of power-on time to a Raspberry Pi Zero W every 12 hours or so, via a MOSFET. That way I don't have to keep the RPi powered on. Power source is 3 D-cells.
They’ve offered this product in 5.7" and 7.3" sizes for at least a year.
I have one in the 5.7" size that I use to show realtime BART departures for the station near my house. It's useful but the 30sec refresh time means I'm limited to showing arrivals in wall-clock times (i.e. "5:38pm") rather than the more useful relative time format ("11 minutes from now").
If I was going to upgrade, I would probably ditch the color support, use a Banana Pi Pico for the controller (cheaper, has USB-C instead of MicroUSB) and go with a higher resolution monochrome display with a 1sec refresh, like this one: https://www.waveshare.com/6inch-hd-e-paper-hat.htm
The problem is the duty cycle, not screen real estate - if you show a relative time, it needs to update frequently to remain accurate.
If it's 5:27 right now and the BART comes at 5:38, I need to update once per minute if I'm showing a "+11m" for that relative time to remain accurate. Since the display takes more than 30s to refresh and isn't readable while it's refreshing, that means it's useless > 50% of the time I try to glance at it.
Whereas with my current setup, I only update it once one of the BART arrival times is no longer in the future. So, right now it's showing arrivals at 6:23, 6:29, and 6:44. Which means the display will update at 6:23, 6:29, and 6:44. There's a little extra optimization added to only show BARTs that are at least 5min in the future, since the BART is a 5min drive from the location where the display is. Depending on the BART schedule, all this works out to roughly ~8min update cycle on average, meaning that the display is only unreadable due to refreshing about 6% of the time, which I find to be very acceptable.
Clearly the solution is to have two displays refreshing on a 30 second alternating cycle! Maybe add a third black and white high frequency screen displaying an arrow pointing to the currently active screen. I jest:)
It actually really makes me want to build a Solari board for my local metro station.
(I couldn't verify this quickly, but it's something I've heard a few times)
E-Ink Carta uses an LCD layer for the color filter, so it has standard E-Ink response times. These things seem to have multiple pigments, which have different weights and are separated from each other by quickly pulsing the display back and forth right, which takes a long time.
For this kind of separation waveforms probably matter a lot more, but hobby applications seem to have been doing fine with custom waveforms. They may not perfectly respond to the current temperature or have optimal ghosting prevention, but they do (from all examples I've seen) produce good images. Getting fast updates seems to be mostly a function of daring to drive the device with high power (which can lead to stuck pigments if overdone, but this can often be corrected with some flashing).
I have a three-color, 5.3 inch display. It takes around 25 seconds for a full screen update.
But with some waveform hacking (courtesy of the seller, “GoodDisplay”) and some creative use of the controller chip interface spec which differed slightly from the display interface spec…
I’ve managed to get rapid, partial black and white updates working. So you can create a ‘background’ out of three colours, then so long as you update rectangles that only contain black/white pixels, you can do a sub-second update.
Mine is a display that sits inside a PC, it shows the time, CPU and GPU temperatures, updated once a minute, and has a black/red fractal pattern that only updates on a full refresh which I scheduled for every 15 minutes.
When are these e-ink patents expiring? I feel like there are many opportunities for low power devices using screens like this but the price is prohibitive for many applications.
I've read somewhere (a Reddit post from an insider? Not entirely sure anymore) that E-Ink Inc. isn't intentionally holding back adoption, but their yields are just really bad.
I've worked close with e-ink before in a past life getting early access to prototypes and pre-release tech, and can confirm that yields ware a huge issue back then holding back progress (I remember they had some really cool shit in the lab that I still haven't seen come to market). It's why progress is so slow. Patents are a moat, like any other tech company does to protect it's IP, investments and hard work, but it's not the main moat.
Making color/advanced e-paper displays for prototypes and concepts that only work in a trade show is not difficult, but making that at scale, with good yields to guarantee great image quality at low refresh power, low cost for your customers to stay interested and good margins for you to stay in business is the impossible part.
Very similar to the challenges of scaling down silicon nodes except you're doing it at your own expense, without the dozens of billions that governments and major companies are pouring in it.
Things will stay as is or will progress very slowly, not because e-ink is some evil corporation abusing the patent sistem to hold back progress, but because e-paper displays are used in this product niche of mostly e-readers and digital signage, which have long upgrade cycles and very low margins, meaning there's too little money to make huge leaps, while all the big money in display tech is going into OLED and micro-LED tech as that's what the consumer market wants and buys at large enough quantities and margins, so that's where the big players like Samsung, LG and BOE will be investing in, instead of in e-paper tech.
I hope they provide a larger display 14'-ish would be enough, then I can do coding with some syntax highlight. seven-color should be enough for most cases.
34 comments
[ 0.20 ms ] story [ 69.7 ms ] threadhttps://tidbyt.com/
(No affiliation, other than being a happy customer. There's a promotion this week for $15 off the list price; DM me on twitter if you'd like the referral link)
Though my next project will use the smaller 1.54 inch ones, very targeted ;)
https://github.com/alexphredorg/lilygo-weather-esphome
I always wanted to try overlaying physical controls for realtime readouts but use these displays for generated backgrounds like gauge scales. Or like a clock background that still had physical hands.
Further implementation details: I'm using an ATTiny85 to feed 120 seconds of power-on time to a Raspberry Pi Zero W every 12 hours or so, via a MOSFET. That way I don't have to keep the RPi powered on. Power source is 3 D-cells.
I have one in the 5.7" size that I use to show realtime BART departures for the station near my house. It's useful but the 30sec refresh time means I'm limited to showing arrivals in wall-clock times (i.e. "5:38pm") rather than the more useful relative time format ("11 minutes from now").
If I was going to upgrade, I would probably ditch the color support, use a Banana Pi Pico for the controller (cheaper, has USB-C instead of MicroUSB) and go with a higher resolution monochrome display with a 1sec refresh, like this one: https://www.waveshare.com/6inch-hd-e-paper-hat.htm
So they show the time for 11 minutes instead of spending 30s of every minute refreshing the display.
If it's 5:27 right now and the BART comes at 5:38, I need to update once per minute if I'm showing a "+11m" for that relative time to remain accurate. Since the display takes more than 30s to refresh and isn't readable while it's refreshing, that means it's useless > 50% of the time I try to glance at it.
Whereas with my current setup, I only update it once one of the BART arrival times is no longer in the future. So, right now it's showing arrivals at 6:23, 6:29, and 6:44. Which means the display will update at 6:23, 6:29, and 6:44. There's a little extra optimization added to only show BARTs that are at least 5min in the future, since the BART is a 5min drive from the location where the display is. Depending on the BART schedule, all this works out to roughly ~8min update cycle on average, meaning that the display is only unreadable due to refreshing about 6% of the time, which I find to be very acceptable.
It actually really makes me want to build a Solari board for my local metro station.
The problem is not inherent to the eink technology: I can read comics, browse the web in color on the Boox Nova Air C, or even watch youtube.
The problem is waveforms etc trying to provide a picture perfect rendering when rough blobs of color may do the same job.
E-Ink Carta uses an LCD layer for the color filter, so it has standard E-Ink response times. These things seem to have multiple pigments, which have different weights and are separated from each other by quickly pulsing the display back and forth right, which takes a long time.
For this kind of separation waveforms probably matter a lot more, but hobby applications seem to have been doing fine with custom waveforms. They may not perfectly respond to the current temperature or have optimal ghosting prevention, but they do (from all examples I've seen) produce good images. Getting fast updates seems to be mostly a function of daring to drive the device with high power (which can lead to stuck pigments if overdone, but this can often be corrected with some flashing).
But with some waveform hacking (courtesy of the seller, “GoodDisplay”) and some creative use of the controller chip interface spec which differed slightly from the display interface spec…
I’ve managed to get rapid, partial black and white updates working. So you can create a ‘background’ out of three colours, then so long as you update rectangles that only contain black/white pixels, you can do a sub-second update.
Mine is a display that sits inside a PC, it shows the time, CPU and GPU temperatures, updated once a minute, and has a black/red fractal pattern that only updates on a full refresh which I scheduled for every 15 minutes.
Something similar may work with yours?
Making color/advanced e-paper displays for prototypes and concepts that only work in a trade show is not difficult, but making that at scale, with good yields to guarantee great image quality at low refresh power, low cost for your customers to stay interested and good margins for you to stay in business is the impossible part.
Very similar to the challenges of scaling down silicon nodes except you're doing it at your own expense, without the dozens of billions that governments and major companies are pouring in it.
Things will stay as is or will progress very slowly, not because e-ink is some evil corporation abusing the patent sistem to hold back progress, but because e-paper displays are used in this product niche of mostly e-readers and digital signage, which have long upgrade cycles and very low margins, meaning there's too little money to make huge leaps, while all the big money in display tech is going into OLED and micro-LED tech as that's what the consumer market wants and buys at large enough quantities and margins, so that's where the big players like Samsung, LG and BOE will be investing in, instead of in e-paper tech.
Just curious because I think I've seen you repeat this on HN before. Have you read the replies to your comment?