How about a 3.5 inch 480x320 tochscreen display?

PaulStoffregen

Well-known member
I'm considering making a PJRC product for a 3.5 inch TFT touchscreen display with 480x320 resolution. Conceptually, it would be pretty similar to this Adafruit product, with SPI interface on the bottom side and 8 bit parallel interface on the top.

So here's a thread to discuss details....

If we do this, which is still a pretty big "if" at this point, PJRC will almost certainly have a custom display made in China, and we'd assemble it onto a breakout board here in the USA. I've already talked with a few companies who would put together the display part from standard materials, but with the flex PCB part customized for us. That's important, because almost all the ones readily available support only 16 bit parallel mode. This also means we can get access to signals not commonly brought out from other displays, like the TE output or PWM backlight control pin.

A couple major decisions to make are the type of TFT and touchscreen. IPS displays are available, which offer superior color range and wide viewing angles. Normal TN types, like we have now with the common 2.8 inch side, are less expensive. Likewise, touchscreens come in cheap resistive which detects only a single touch point and requires significant pressure, or more expensive capacitive touch that works similar to cell phones and tablets. Different touch controller chips can be used, some detecting 2 touch points, others up to 5 points.

Tentatively, the controller chip is looking very likely to be ILI9488. If capacitive touch is used, the touch controller may end up being FT6206 or FT6236.

The pinout is also undecided, but my general goal is to have one side require as few signals as possible for easy DIY connection and the other side offer 8 bit parallel for higher performance. I'm debating whether to add extra circuitry like a reset chip, backlight control or other nice features, which aren't commonly found on the other breakouts.
 
I have found resistive touch screens a bit disappointing having been spoiled with capacitive screens on phones over the last several years so personally I would prefer capacitive touch. If there was an easy way to turn off the display to save power but wake it when touched that would be cool too. I like the few signals as possible side and would likely just use it. You guys have managed to stream video over SPI so for me its fast enough.
 
Is it worth having a footprint to allow a Teensy to be soldered directly onto the display? Maybe on just one edge if the pinout worked that way to avoid through holes going into the back of the display. Other choice would be a surface mount pin header kit, possibly with a PCB jig to make getting the rows lined up easier though that pushes price up. Single row might be easier, since you could either put the teensy behind the display, or flip it over and have it extending outside the display area if getting access to every single pin mattered for your project, and space mattered less.

Basic idea being that the display is already occupying a lot of project real estate, and needs decent size PCB anyway, so use that PCB to provide a motherboard for common sub units like the Micro controller. Might also make it easier for beginners to get the wiring right first time if it was a one for one line up.
 
yes but bigger screen == more pixels && more pixels == needs more speed, :) if it's fast enough, it's okay, if not, needs to be faster
 
I think this is a GREAT idea and would love to have such a display module. I'm definitely in favor of capacitive touch (two points is probably enough at that size) and reset, backlight PWM would save folks a lot of time and energy bringing such a thing up.

I'm actually waiting right now for a first spin of PCBs to come back from China to support a 3.2" display (this one from buydisplay.com). My board actually has a socket for Teensy (3.2 or 3.5/6), so acts as a baseboard with breakouts for a few GPIOs, I2C & USB host. I've also thrown in an ambient light sensor & audio amp. I would much preferred to have a larger display with more pixels and multi-point touch but sourcing LCDs is not easy for small quantities. This board, called BB0, looks like this:

Screen Shot 2017-04-01 at 10.50.43 PM.jpgScreen Shot 2017-04-01 at 10.51.01 PM.jpg

My previous builds used an existing LCD modules with Teensy on a separate (hand wired) baseboard. Combining the two boards makes the whole thing smaller and simpler. Consider having a spot on the back to drop a Teensy right on. And also consider offering multiple versions of the PCB (i.e. propshield/propshield LC) with differing sets of support circuitry if that's not possible.
 
That would be great..!

... agree with Frank - the adafruit 3.5 inch TFT touchscreen display with 480x320 has RAM buffer so it should be good from Teensy if an UNO can do it. Also the faster 8 bit interface using 12 pins would be cool if the Teensy underside pins could be used to keep edge pins and SPI free.

This display has a controller built into it with RAM buffering, so that almost no work is done by the microcontroller.*The display can be used in two modes: 8-bit or SPI.*For 8-bit mode, you'll need 8 digital data lines and 4 or 5 digital control lines to read and write to the display (12 lines total). SPI mode requires only 5 pins total (SPI data in, data out, clock, select, and d/c) but is slower than 8-bit mode. In addition, 4 pins are required for the touch screen (2 digital, 2 analog)
 
Being an old geezer and having cataracts done with my eyes 'fixed', I find these smaller displays disappointing for readability and also brightness of only 200-300 nits(cd/m2). What I'm currently looking for is a 7 or 8-inch display 800 x 600 and perhaps 1000nits for day light readability. I know they are out there for large quantity buyers. Heck if you went and integrated a display & controller, audio amp, SD Card, touch screen and a Teensy 3.6 on a board with some stick pins for attaching my own electronics behind it I'd be willing to drop 200 to 250 for it with a bright display.

Don't get me wrong as I appreciate the hard work & long hours you and the group here are putting in to support all of us. I feel my projects are benefiting from the use of a Teensy over the Arduino MEGA that I had used for a prior project.
 
7 or 8" may scare some off it it forces a big price increase. I would love one that big but not sure I could justify it for my just playing around projects.

If there is a lot of unused space on the board footprints for maybe a spi flash would be cool. You could maybe move some of the font data that is in headers now to pull from it if they populate the spot and load their fonts there.

I also like the idea of the teensy plugging right on it. Would there be room? Would it be possible to do both the end connections and a place to plug in a teensy? I mean your option which to use not use more that one interface. With stacking headers you could still plug on other boards like audio etc.
 
I wonder if a 4.3" 480x272 (original PSP screen size) would be a better alternative for size and pixel density. It appears to be a very common size glass.
 
Sounds like fun! Let me know when you have some to play with ;)

It would be great if these versions, had at least minimal back light support, such that you did not need to use resistor. Likewise would be great if the reset had the required PU or PD, so again made these optional to hookup

Kurt
 
One question, Paul: How much faster would the 8-bit parallel be relative to the speeds we're seeing with SPI on the ILI9341? 8x? 2x?
 
It won't be more than 8X faster! In some cases it may end up being about the same speed. At this point, I just don't have any solid benchmarks.
 
Sounds like fun!
What kind of speeds can we expect?

Edit, bah, missed a few comments. Sounds like fun.
I second using underside or digital pins with few peripherals. Or would a custom board with the Teensy builtin using the pins that aren't on the 3.6 be an option?
 
Last edited:
RE: Speed ​​- more pixels means more data to write - but this depends on your application. 480x320 = double the pixels of a 240*320 :) ~ 300KB (16Bit) instead of ~150KB per full screen
 
Last edited:
It won't be more than 8X faster!
Heh, of course. I don't quite understand the DMA engine and SPI vs. parallel controllers well enough to know where the bottlenecks are.

Looking over the ILI9488 datasheet briefly, it seems that the performance will be on par with the ILI9341, unless I missed something.

I've been wondering there's still some speed improvements to be made for the ILI9341 (and the ILI9488) by taking advantage of the DMA engine to do things like fast fills and blit operations from Teensy RAM that would run asynchronously from the main CPU. This wouldn't speed up drawing, per se, but means that that CPU could be busy with other stuff while the screen fills over the SPI. This could also dramatically increase the speed of opaque text drawing as one character is sent to the screen, the other could be rendered in RAM. Just a thought...
 
Yes, that's actually one of several displays I'm looking at...

In fact, the prototype PCB I designed just arrived from OSH Park and I've soldered the parts but haven't powered it up yet....

display.jpg
 
I'll share it just as soon as I can get the display to show anything. Remind me in a couple days if I forget (or if this gets set aide by other urgent stuff).

If you order one of these displays, select the option for capacitive touch. This board won't work with the resistive touch version.
 
Is it worth having a footprint to allow a Teensy to be soldered directly onto the display? Maybe on just one edge if the pinout worked that way to avoid through holes going into the back of the display. Other choice would be a surface mount pin header kit, possibly with a PCB jig to make getting the rows lined up easier though that pushes price up. Single row might be easier, since you could either put the teensy behind the display, or flip it over and have it extending outside the display area if getting access to every single pin mattered for your project, and space mattered less.

Basic idea being that the display is already occupying a lot of project real estate, and needs decent size PCB anyway, so use that PCB to provide a motherboard for common sub units like the Micro controller. Might also make it easier for beginners to get the wiring right first time if it was a one for one line up.

Mr. Wrangler is certainly a decent person, whom adores puppies, and contributes to the art of electronics. And putting the Teensy on the display board has a good rationale, but am not certain about connecting to the display, unless our hero (Sir Paul) can do this without adding more noise to the analog stuff.
 
And another thought. Having done something similar (was R, not C) for one of my employer's field test projects, have had some problems resulting from the environmental limitations of these low-end displays. And this is where our hero becomes a victim of his success. Professional designers and professional fools (myself in latter category) have found the Teensy to be a suitable solution to rapid dev of ATE, monitoring, and control systems; both embedded and stand-alone. Last month, committed larceny and took nephew's audio shield and put in field ATE for sig gen channel. The moral, or immorality, of the story is that PJRC has enabled some of us to become accustomed to consuming Stone IPA on a Bud lite budget. Understand that environmental limit specs are difficult to do, but characterization of specs for display/control part becomes mas muy importante.
 
Back
Top