Which ARM processor for WVGA TFT

[Rufus]

I have a requirement for a controller providing a user interface on an  800x480 TFT display with resistive touch screen. Can probably get away with 8 bit palleted colour. Apart from managing the display processing power requirement isn't high. USB host for firmware updating would be nice. Other I/O requirements are pretty trivial serial ports, input capture, a few digital I/O.

Can anyone suggest a family to look at for lowest overall production cost for volumes of a few thousand/year?

 

[Rasz]

how about ~$30 for a whole packaged LCD module with SoC capable of USB host? 256MB ram 4GB flash etc?

if that sounds fine just look at chinese tablet suppliers

seriously, you wont be albe to beat that price point unless you move to china and design there at the lowest level

[ehughes]

LPC4357

[hagster]

Or the LPC1788

Or possibly the STM32F429

[Jeroen3]

Did you notice the request for an ARM Processor instead of a ARM Microcontroller?
I'd expect suggestions going to an Cortex A family.

Depending on the use case. What about Raspberry PI equivalent hardware/software?

Isn't the LPC1788 bundled with emWin license?

[hagster]

Did you notice the request for an ARM Processor instead of a ARM Microcontroller?
I'd expect suggestions going to an Cortex A family.

Yes, but the OP asked for serial ports, pin capture and GPIO. That sounds like a uContoller to me.

I don't know what emWin is.

[nctnico]

System on chips like the Omap from TI and Freescale's iMX series have the pheripherals (I2C, SPI, GPIO, etc) you typically find on microcontrollers. But a system based on such a device isn't exactly cheap.

[Rufus]

Yes, but the OP asked for serial ports, pin capture and GPIO. That sounds like a uContoller to me.

I don't know what emWin is.

Processor, Microprocessor, Microcontroler?

The current version of the product runs on a PIC24F with LEDs and buttons, another version on a PIC24F with mono QVGA LCD and touch screen.

The next generation the customer wants colour and overkill WVGA resolution because he says he can source the displays cheaply for long enough. It doesn't need to do more than the current PIC24s are doing apart from driving colour and much more resolution and firmware update from USB stick would be nice. The current boards cost about $15 but the mono QVGA LCD is getting old and expensive.

Thanks for the suggestions so far.

[hagster]

You might also need more RAM and flash to do much with such a display.

Also it's worth thinking about what's more important to you, part cost or development effort.

[Jeroen3]

firmware update from USB stick

USB Host capabilities... That requires some software effort too.

And for that price you'd want an applications processor, such as the new PIC32MZ, Freescale Kinetis or as mentioned STM32F4xx or LPC18xx or LPC43xx. Maybe an LPC178x but you haven't mentioned what the product is actually doing. What's on the display?

There are some more brands around, Atmel ATSAM, Infineon XMC, Renesas or Texas Instruments.
Get a chip you can buy and you can get software for. You don't want to write the USB Host stack yourself, do you?

And design in a (place for a) uSD card or SPIFI for your images.

[free_electron]

Yes, but the OP asked for serial ports, pin capture and GPIO. That sounds like a uContoller to me.

I don't know what emWin is.

Processor, Microprocessor, Microcontroler?

The current version of the product runs on a PIC24F with LEDs and buttons, another version on a PIC24F with mono QVGA LCD and touch screen.

The next generation the customer wants colour and overkill WVGA resolution because he says he can source the displays cheaply for long enough. It doesn't need to do more than the current PIC24s are doing apart from driving colour and much more resolution and firmware update from USB stick would be nice. The current boards cost about $15 but the mono QVGA LCD is getting old and expensive.

Thanks for the suggestions so far.

the question is : does the lcd panel have an on board controller ? meaning : you shove in data and it refreshes itself. or do you need a controller ?

if the display has an on board controller and uses a simple byte or word wide interface you can pretty much use anything. those displays are simple to use. if you really need a TTL display interface ( meaning you scan the display ) pyou will need to find a processor with something like a MALI in it ( MALI is arm's GPU )

lcd's with controller on board are cheap. 35$ will get you such a panel.

i;ve driven them from a STM32F102 without problems. just pick a cpu with lots of ram so you can make a frame buffer in the cpu. perform the drawing in ram first then shoot the ram to the display using a dma operation or monolithic routine.

[Rasz]

lcd's with controller on board are cheap. 35$ will get you such a panel.

ahahahah "only 35" for one component?
http://www.banggood.com/MTP28-R430C-RK2926-Dual-Core-1_0GHz-4_3-Inch-Android-4_2-Kids-Tablet-p-920975.html


> because he says he can source the displays cheaply
Key word here is "cheaply".
Rufus what your client is looking for is cutting you out of the equation. They saw how cheap chinese tablets are and now demand the same deal from you.

http://www.banggood.com/A86A13-Single-Core-12GHz-7-Inch-Android-40-Kids-Tablet-p-927932.html
Allwinner a13 is $4 in china, its a full 1GHz A9 core with powerful gpu, h264 encoder/decoder, dedicated lcd/camera controllers, usb, even sata and enthernet!

You wont be able to beat or even compete with that using western parts (like pics, avrs, or even TI ones). TI charges more for SoC alone than Chinese for whole complete boards!
http://www.digikey.com/product-search/en?v=296&mpart=AM3358ZCZD72


 Embrace it by either
-designing for manufacture in china using chinese components, you could probably do something like this at ~$20
https://www.olimex.com/Products/OLinuXino/A10/A10-OLinuXino-LIME/open-source-hardware

-use ready module
http://www.aliexpress.com/item/allwinner-a13-industry-core-board-C13/1532747657.html

-or redesign your product to be GUI agnostic, make it able to speak to any off the shelf tablet over wifi or BT. This way you will make it future proof while staying low price low power.


The only source of cheap relatively available screens with buildin controllers is cellphone replacement market.
http://andybrown.me.uk/wk/2013/10/19/vivaz-u5-lcd/
~$10 for a 3.2? 640×360, same ~$10 gets you 7" 800x480 screen with capacitive touch.

[Rufus]

Maybe an LPC178x but you haven't mentioned what the product is actually doing. What's on the display?

Basic user interface stuff - numeric displays, buttons, list boxes, touch keypad. Will probably include some small images. If the hardware can manage it some small animated images/icons might be a nice feature.

You don't want to write the USB Host stack yourself, do you?

I was expecting not to have to write USB stacks and hoping for free or at least cheap graphics library and design tools.

the question is : does the lcd panel have an on board controller ? meaning : you shove in data and it refreshes itself. or do you need a controller ?

The expectation was to use a panel needing a controller. Hard to see how a panel with controller is going to work out cheaper than just a panel with a bigger processor on the board. Can you point me at an example?

i;ve driven them from a STM32F102 without problems. just pick a cpu with lots of ram so you can make a frame buffer in the cpu. perform the drawing in ram first then shoot the ram to the display using a dma operation or monolithic routine.

WVGA need needs 384kB of RAM for an 8 bit colour frame buffer - what cheap microcontrollers have 384k + of RAM? I was expecting to need some S or SD RAM to support the display.

> because he says he can source the displays cheaply
Key word here is "cheaply".
Rufus what your client is looking for is cutting you out of the equation. They saw how cheap chinese tablets are and now demand the same deal from you.

I am not in that equation - I design they sub-contract manufacture in Asia already. They think they can get cheap WVGA panels serving video player and picture frame type markets not tablets and not have to design in a new panel every 12 months because supply dries up.

[nctnico]

One of the problems with TFT screens is that models follow eachother quickly and each screen has a different connector. That cheap screen may not turn out so well. How much time can you spend on software? Linux may be a way to have the software finished quickly if you have the knowledge or find someone who can create a stable board support package.

[mikeselectricstuff]

PIC32MZ may be worth a look, but check the current errata situation.
512K RAM and support for LCD - I think this is done using DMA plus a parallel port.
 Has USB host (hi speed I think) - not sure what state the software is in for this.
If the main reason for USB is firmware updates, SD card may be an easier option as there are fewer layers of protocol.
 

[nctnico]

NXP has a LUFA fork for their chips to create USB host applications. I takes some exploring on how to use it due to lack of documentation but it works pretty well. I have used this in several projects to interface with USB to serial converters.

[hagster]

I'll probably get beasted for suggesting something Microsoft based (it's fully open source btw), but using .net microframework really suits this sort of thing.

All the basic stuff is really easy. You get USB host(and client) with pivoted to support keyboards, mice alandmass storage, SD card(with FAT33 file system). You have a good choice of hardware prototyping boards. You have abstraction from the hardware so you can change the MCU without too much hassle. There is a whole library of controls for drop down boxes, buttons etc etc.

https://www.ghielectronics.com/catalog/product/373

The issue may be cost so this is why I ask what your priority is between dev time and BOM cost.

The other option is obviously using something Linux  or Android based. These can have there own issues but might not be a bad option for you. Often these have slow boot times and can be relatively power hungry.

Personally I think trying to create compelling user interfaces in bare metal C++ is likely to be very time consuming.

[Jeroen3]

Personally I think trying to create compelling user interfaces in bare metal C++ is likely to be very time consuming.

That why people develop software libraries such as emWin.
https://www.segger.com/emwin.html

It will require a license, but it will save you expensive development time.

[dannyf]

I don't know what emWin is.

Google is there for that.

I think there are free emwin libraries for LPC and ST chips.

[nctnico]

Emwin is available in binary form (library) for NXP's processors which is free to use. Interesting...