Share your name in shining technical prowess.
Already have an account? Log in.
To make the experience fit your profile, pick a username and tell us what interests you.
7" is big, almost too big to wear on a shirt, so people should see it. I can make a switching power supply to run on batteries. There are some options for storing graphics in memory and supplying processing power. I'm starting with a Raspberry Pi-based prototype and I hope to quickly move to something more fun.
I keep getting distracted from this. Plus work.
And I'm at the difficult point. It was looking more complex than I expected and my KiCAD skills aren't what they should be. I haven't read as much of datasheets as I need to to understand. Maybe I should ask someone? Dunno who.
I think I'll stick with the prototype for this thing. Sometime I'll take up a smaller project to figure out KiCAD. And then maybe come back to it?
Some other neat video stuff on Hackaday but not sure if anyone is directly going to a 40ish pin screen the way I was thinking.
Here's an incomplete schematic:
A couple tips for producivity:
MPLAB Harmony (and the Configurator for some smaller PICs) have a pin diagram that you can use to figure out where modules are and line up what goes on your PIC without absolutely scouring the datasheet. Looks neat so far. I'll use it later for code too.
KiCAD parts can be slow to generate, so I use an online "quicklib" tool here http://kicad.rohrbacher.net/quicklib.php. Not as good as getting free Eagle parts, but it fits into the KiCAD design regimen well and I get to customize things at every step.
Now, the ram chip has a clock pin which I hadn't figured on. The PIC PMP might be able to use a chip select pin sort of as a clock, not sure, need to read more. An annoying placement too, I was figuring to use PORTD as most of the LCD color specifying bits.
Also, 24 bits of connection is way more than I want to actually do now that I look at it in the project. The PIC port pins don't line up and it's a struggle to use only 3 different ports to make these 16 bits work. I think this will be plenty.
It's frustrating but it's interesting! But I'm getting burned out for this weekend.
So, I've made a bit of a prototype. It was a rush job putting together stuff and the lack of planning shows. Here is an overall image:
I have 4x 3.7v batteries as the main supply, where everything but the RPI can run anywhere from 24V to 7V. So the RPI gets a 5V buck converter. No problems with this switcher: http://www.newegg.com/Product/Product.aspx?Item=9SIA6742FN3277.
Did I mention that these 50 pin, 7" TFT LCD displays are called "AT070TN90"? So far working very well and available for cheap on the internet. They go into car monitors for backup cameras and also are intended to work with the Raspberry Pi. So I bought a RPI driver for it.
The screen looks amazing (to me anyway). No sign of a flicker like with the RCA driver. Very readable, lots of color, nothing like lag or flickering.
These things would be really neat for a little display box like everyone is making, but I wanted these things to fit in my pocket. The HDMI cable isn't flexible enough to bend into there, and the two boards and extra batteries just won't fit. I've managed to do the stuff that everyone else is doing, but not the stuff that I wanted to do! I might be able to shift the batteries into another pocket or something. There are RPI model A boards that might work better on the RPI side.
We could play with this more, but I'd like to go back to the microcontroller side of the project. There's about a month before some important stuff in August where I'd like to have this thing to handle introductions for me and I'd rather have the greatly reduced bulk. Plus I'm bored of buying other people's electronics and I want to make my own. The wallet protests too.
More on a schematic maybe this weekend? I'm thinking of using a PIC and the deadline for the sponsor contest is Monday, so I might burn some midnight oil for that and the August deadline. I've used some PIC32 in the past, so I expect that to be faster than STM32. Good for trying a board.
Got an LCD! The plastic cover looked nice, had great mounting options, but just a little more bulky than I'd like. It wasn't made to go on my shirt.
It was intended as a car video screen for a rear view camera, so some RCA electronics come with the package. I hooked up my Raspberry Pi as a test along with a 12v power supply. Had to change some config.txt settings, need to tinker with boundaries for full visibility. Boot up and displays!
There might be a slight flicker or some other visual distortion... Doesn't look great yet. I haven't used RCA in so long I'm unsure if that output is the cause of the problem. Possibly the RCA converting electronics (ever hear of MStar?) are holding it back. Maybe it's worse because of the small screen? Anyway, it's not the Pi because HDMI looks fine from it.
Now the thing to consider is whether to build a prototype out of this with some code and try showing it to people. I was thinking of the Pi as a "backup plan", but perhaps a "product verification" approach would be better. Main questions about the project: Does anyone else in the world want this? Would you rather have a tablet computer on your shirt? Would you pay 50$ for it? Is the large screen too gaudy? How many hours do you want it to run? Do you need wireless access? Is animation important? Varied colors? Scrolling background?
So the way this could work is with a few batteries up to around 12v, a buck for the Pi power, and some code for displaying things. I'd like to learn more about the Pi so I like this option for prototyping, but I'm still concerned about the RCA connector keeping my prototype ugly.
The other option for a quick prototype is to use some Adafruit parts, like here: https://www.adafruit.com/product/2353
Ordering parts on Adafruit is warming up to me. It's like they work on part of your projects for you. Anyway, they have a video chip to go with it called RA8875. I can't find on http://findchips.com, but it'd be OK for a prototype and closer to the kind of parallel HSYNC graphics I want to make.
But this works with a different 40 pin FFC connector LCD in the Adafruit store. The difference is that the Adafruit LCD has its own charge pump/boost/inverting TFT supplies, I'm guessing those are the components on the FFC cable itself. That Adafruit 40 pin style looks like more a convenience for hackers than for manufacturing money. I'd rather have the parts on a PCB.
Do I want to develop this project as a cheaply manufactured thing? It'd be neat for the purpose of the contest. So far I've been approaching from that angle. But it'd be good to take a step back and make something with this Pi + LCD to test it as a product. If it's a complete flop, I might leave it here. People can say to fix the flickery part if they see potential in it.
The thing I could use most right now is a good camera. I have an old smartphone camera, but LCDs and LEDs glowing don't look very good on it. Next update will feature pictures of some hardware I've got already, but for now I'm talking about the things I've found on google like specifications and the calculations I'm figuring. I'm not a graphics processor guru, so correct me if I'm wrong because I'm only thinking out loud.
Naturally there are some interesting problems with this project. Otherwise, more people would be using 7" LCDs or selling them for small projects with 32 bit microcontrollers.
First, 800 x 480 pixels is five times as many pixels as say 320 x 240, and also there are more bits per pixel in a 24 bit RGB format. I think I can cut this down to 16 bits without losing much for the purposes of this project but I'd prefer not to. Why have a 7" screen if it's terrible resolution and color? But if I spring for 16 bits, it would cut the amount of data I need to store and shuffle around by a third.
And I think the big thing for this project will be storing and accessing that data as quick as it needs. 800 x 480 x 24 means 9 million bits for the whole screen. I'd rather not use a processor with a large amount of memory like that, so we'd have to pull it in from somewhere. If we let the display run at 10fps (slow!), that takes 7MB per second to pipe into ram (overwriting some other section of the screen) and out. So we're getting into computer-sized numbers.
Naturally, you can fix this with money. Or, if you have large quantities in manufacture and an amazing relationship with some people then you can get more for less, like a 9$ computer. I'd just like to see what I can do with chips from vendors like Mouser and how much it costs.
An SD card can be accessed by an SPI connection at maybe 25MHz, but that's only 3MB per second. The STM32 I was looking at had a 85MHz SDIO module for a more tolerable 10MBps. That's still not much; I'd hate to be stuck around ~10fps in the best of worlds. Sounds like I need to venture into RAM, which I'm not too familiar with. If you can read one byte in 20ns, parallel, then that's 50MBps. Sounds like something more like 60fps might be possible there if the data acrobatics work.
I like the RAM idea and I'd like to reinvent the whole graphics thing (it'd be fun) but I'm wondering if someone has already made a cheap chip for it. Better look around more.
For occasional fun, I'll cruise around ebay and amazon and newegg and similar sites for some inspiration and to see how cheap stuff from China is. Electronics shouldn't get designed in a vacuum, eh? A few ideas for this project came together:
Cheap development boards have been around for a while, search for "TFT LCD Development Board". Right now I see some good options for 30-40$. I suspect it takes some effort with suppliers to sell for that low, even from China. The ARM boards have some neat features like super fast SD card read and the processors are cheap on DigiKey. They compile with GNU tools. I should try one in a project, to figure out how they work. There's probably lots of tutorials out there.
The bigger LCD screens are not expensive either. The 7" one I'm looking at cost 20-30$. I want to tear off the plastic to see what lies underneath. Looks like they use composite video, so I expect some electronics for a conversion to parallel port from RCA in addition to a switcher. Some people stick that directly into a Raspberry Pi (separate power), which doesn't sound like a bad deal. I like the STM32 idea better because a Pi would be a little bulky for my taste. May be a good backup plan though.
So I'm going to buy some things and figure out how they work. There's lots of datasheets to read. The overall plan is to have a boost from battery voltage to whatever the LCD needs, store some visual data in a microSD card, read it from the SDIO on an ARM M3 STM32F103 chip, maybe add additional visuals, and use some GPIO as parallel output for the LCD, with the HSYNC and VSYNC (or whatever it's called) according to a spec sheet that looks like it might be what's inside.
I knew there were some libraries. I'll take this as a recommendation and look into it, thanks!
Depot
Jacob David C Cunningham
Michael R Colton
kelvinA
Beth
May I suggest using the uGFX library? (http://ugfx.org)
It is optimized for low resource systems such as your STM32F103 and it runs out of the box on said platform (and many others).