LCD/Audio Expansion Board for NG100

Go To Last Post
20 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

LCD/Audio Expansion Board for NGW100
My intent is to provide a low cost, expansion functionality with kernel support similar to the STK1000.
see LCD on NGW100 for additional LCD discussions.

NGWLCD Features:

  • TFT/LCD 24bpp - OSD043TN13
  • touch screen
  • Audio Codec / drive
  • Stereo Phono plug out

I like PCB123.com for PCB prototypes. Their schematic & PCB layout tools are free and do a quick reasonable job.
See schematic attached. Parts used:

  • 40 pin LCD connector - FH19S-40S-0.55SH
  • LM2717-ADJ - provides +5V for LCD, +1.8v for Codec
  • LM2733 - current drive for LED backlight
  • TI - TSC2102 - Audio Codec/drive, touch screen input

Status / TODO

  • Order samples (done)
  • get LCD samples, feedback on price
  • Order misc parts from Digikey
  • Verify footprints
  • review port assignments with kernel
  • route board (usually 1 day)
  • order board from PCB123
  • assemble & test
  • modify kernel
  • port some streaming video/audio app
  • measure quality, power

I mentioned www.sparkfun.com to the OSD marketing guy since
LCDs and mating connectors are had to find for prototyping!

My estimate for a low volume (maybe 6) run of NGWLCD is
about $115US. Comments? Interest? Help?

Future possibilites:

  1. ) push button input via SPI (Phillips)
  2. ) IDE, HDD
    • 44 pin mini IDE
    • 2.5" or 1.8"
    • +5V power requirement
    • which port, current kernel work
    • performance/size of SD verses IDE
    • UMA performance IDE/SD dma to LCD refresh
  3. ) Car or battery portable
  4. ) Mechanical design

Frank Bennett
Mathegraphics,LLC
www.mathegraphics.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Suggestion: Instead of copying the STK1000's LM2733 circuit exactly, use the circuit shown on National's data sheet for the LM2733 (i.e. without the feedback on the LED cathode lead), and instead connect the LED cathode through a BSS138 FET (or equiv.) to ground. Then run the FET's gate to an AVR32 PWM output (using appropriate R & C), so you can adjust backplane illumination. I used PWM0 for greater lattitude in frequency selection, but you could probably use the LCDC's PWM.

Intensity control is nice to have particularly if you'd like to reduce overall current consumption, and the parts cost is trivial.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

GregoryC wrote:
Suggestion: Instead of copying the STK1000's LM2733 circuit exactly, use the circuit shown on National's data sheet for the LM2733 (i.e. without the feedback on the LED cathode lead), and instead connect the LED cathode through a BSS138 FET (or equiv.) to ground. Then run the FET's gate to an AVR32 PWM output (using appropriate R & C), so you can adjust backplane illumination. I used PWM0 for greater lattitude in frequency selection, but you could probably use the LCDC's PWM.

Intensity control is nice to have particularly if you'd like to reduce overall current consumption, and the parts cost is trivial.


Yes, this is a good idea. I can always short out the FET till I update the PWM driver/uboot splash or if I don't like it!

GregoryC wrote:
Note that the OSD 43TN13 is actually made by InnoLux. Smile I'm quite pleased with its appearance and I'm sure you will be too.

The OSD43N13 spec shows a different pinout & includes the LED backlite in the 40 pin connector (see my schematics).
The timing and interface signals look the same. I'm not sure if OSD wants their specs published but I did find the InnoLux on the web (attached).

Attachment(s): 

Frank Bennett
Mathegraphics,LLC
www.mathegraphics.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

GregoryC wrote:
Suggestion: Instead of copying the STK1000's LM2733 circuit exactly, use the circuit shown on National's data sheet for the LM2733 (i.e. without the feedback on the LED cathode lead), and instead connect the LED cathode through a BSS138 FET (or equiv.) to ground. Then run the FET's gate to an AVR32 PWM output (using appropriate R & C), so you can adjust backplane illumination. I used PWM0 for greater lattitude in frequency selection, but you could probably use the LCDC's PWM.

Intensity control is nice to have particularly if you'd like to reduce overall current consumption, and the parts cost is trivial.

Cheaper yet, why not just PWM0 on the SHDN pin of the regulator in place of the LCD_CONNECTED signal? (which I'm not sure
where that comes from). With/without the inverter to ensure the LCD is off during power up depending on the PWM reset state.

Frank Bennett
Mathegraphics,LLC
www.mathegraphics.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Regarding InnoLux vs. Samsung, I meant signal wise they are quit similar, but not pinout or voltages.

I used the FET in the LED cathode because my circuit can drive several different LCDs which have different arrangements and voltages for the backplane illumination. I'm not sure exactly what you meant, but my circuit and PWM allow me continuous adjustment from dark through maximum illumination. The displays I'm using look quite acceptable at 50-60 percent full rated LED current, and my application is power sensitive.

Regarding posting the InnoLux spec., I didn't because I couldn't remember if I obtained the spec. for public disclosure.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Here the revision0 schmatic/Layout
See revision 1 below

Frank Bennett
Mathegraphics,LLC
www.mathegraphics.com

Last Edited: Fri. Jul 27, 2007 - 03:53 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Personally, I'd like to have LCD/Audio/IDE add-on's for the NGW100 available separately.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Frank,

Please can you update us on the status of your board

regards

Patrick

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Been looking at IDE but forgot to notice the EBI I/F is not brought out to the extention
connectors on the NGW100. The Atmel kernel guys have a preliminary proto which plugs into
the PCI slot on the STK1000 with SN74LVC8T245 level shifters but it looks to me like the Toshiba
drive on the 1.8" 44pin, 2mm iPOD is 3.3volts. So except for reported activity on the AD[2:0]
driving the drive crazy, the level shifters may not be needed. They have the prototype initially
running PIO but the EBI interface should eventually support DMA if the DMARQ/ACK signals are
available. So I could be right but AFAIK the NGW100 would not support DMA for say FPGAs very
well either...another future interest of mine. (FPGA support thread)

The PCB is routed & ready to send to PCB123, probably tonight after I check some footprints
using a 1:1 artwork printout. (Got a couple days before my PCB123 10% discount expires). I
changed the placement so the LCD fits better. I also added a MAX SPI part for 4 pushbutton
inputs & added EXT_IRQs for the touchpad PINT & pushbutton pushed.

Been waiting for the LCD connector & then OSD sent me 5 AT043TN13 (separate backlite conn, different
40pin pinout & no touchscreen) instead of OSD043TN13 ($65 per sample). No word yet on volume $.

So tonight I will check the footprints for the TI Codec/touchpanel I/F, the voltage converter,
backlite driver, with the samples from TI/National, send the PCB out & then get a Digikey order
figured out for the other parts.

I will post the lastest schematic/layout when I get home.

Attachment(s): 

Frank Bennett
Mathegraphics,LLC
www.mathegraphics.com

Last Edited: Fri. Jul 27, 2007 - 02:33 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I was wondering, how difficult it would be to convert the video signals coming from the AVR32 to VGA-signals so that one could plug a standard LCD to the NGW100. I'm not a HW-guru, so I not sure whether I would be able to create a proper schematic/circuit based on the STK1000 (which has a VGA-output as far as I have seen).

Has anyone tried to create such a VGA-extension board or is there something similar available in a shop?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

To get VGA-signals out you need a DAC to convert RGB => analog signals. There is a reference design you can copy on the ATSTK1000 schematics.

You are still pretty limited by resolution because of the refresh rate and that the SDRAM is shared with the CPU processes.

Hans-Christian

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Note that the Linux-OMAP tree has some tsc2102 code ... that ought to help, once the hardware exists.

Also, I think you shouldn't make the tsc2102 use the same SPI chipselect as the DataFlash chip does. :)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

mojojojo:

Yes, thanks, I think I saw the driver...one reason to select that chip.

Chip selects, DMARQ/ACK, SPI, I can change this. It would be nice if there was a table somewhere with current/proposed usage info. Maybe even a pinout definition for the AP7000...just schematics for now.

Here I started a table: (please check, contribute)
Dev Board Functional Assignments

Frank Bennett
Mathegraphics,LLC
www.mathegraphics.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I found too many mistakes on the NGWLCD1 PCB layout.

Here is NGWLCD2 I will send out Monday, so I can have a solder day
next Sat or so. I still have parts for 4 boards.

I added a 44 pin MiniIDE right angle connector(J99 backside, just grounds),
resized Capacitor Pads,fixed connector/Hole locations,turned J16 around,
moved cs0->cs2 for codec.

I have rebuilt u-boot & kernel ready to start test. I re-flashed my NGW with
avr32program, running ubuntu (apt-get but my ARCH is not x86_64)

Attachment(s): 

Frank Bennett
Mathegraphics,LLC
www.mathegraphics.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Frank,

How is the LCD adapter for the NGW100 going? I'd be interested in getting one. Also does it support the same LCD / connector as that used in the STK1000? I bought two of the LCDs off ebay.

Thanks,

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Got NGWLCD2 populated with parts but I'm fighting 2
problems at the moment:

    1) got a nasty oscillation on the +5V output I added a 68uF cap on the LM2717-ADJ input (Vin), helped
    some but I need to look at it some more.

    +1v8 looks reasonable - some ripple

    2) still no pixel, hsync or vsync signals (no /dev/fb0)
    So I'm also playing with the kernel. Today busybox/console
    prompt keeps coming & going on the serial port.

Oops, Update before Hans jumps in. I got LCD signals
and the TSC2102 driver reporting in. I see the stk2000
uses "fbv image.image" which needs to be added to the
/etc/init.d/rcS for a splash screen...I could post my
.config file but that's not much help without my kernel
patches, which the kernel dudes said were too big, so I
will wait till NGWLCD is a flashing and a sqawking!

I would recommend you check the LCD pinout (J1) on my
schematic. There is no standard out there and I'm pretty
sure it's different from the STK2000. The 40 pin pinout is
for the OSD043TN13. OSD finally send me 5 correct LCDs and
I RMAed the 5 AT043TN13...same part#, go figure. The
OSD043TN13 $65 have a single "unified single" cable and a
touchscreen. The AT043TN13 had 2 cables, one little one
for the Backlite and would have had a third if it had a
touchscreen.

Frank Bennett
Mathegraphics,LLC
www.mathegraphics.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I fixed the oscillation problem - the LM2733 is a DC-DC
Buck oscillator and the output capacitor(1uf) was missing. I
also changed the +5V to Vin on the input and now both
DC-DC converters are working & running a whole lot cooler.

The LCD/backlite now come up white then all black. The PCLK,
HSYNC, VSYNC timing I set is close to the LCD spec but the MODE/DVAL
signals are missing....back to reading the AT32AP700 (doc32003.pdf)

I see if you are not logged in you can not see the NGWLCD
design docs...how about over at avr32linux.org:

NGWLCD2 schematic

NGWLCD2 layout

Frank Bennett
Mathegraphics,LLC
www.mathegraphics.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi Frank,

I am watching this thread with eager breath as it is just the sort of thing I'm looking for. Are you anywhere closer to being able to supply this?

Many thanks
rob.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Status update:
Sorry, I havn't been able to spend alot of time on this but:

    o I can display a single image on the LCD (fbv bpp=16) o mplayer produces random video. Might be a DMA issue.
    o ripple on my power supply input shows up on the LCD backlite.
    o got gclk0 running to the TSC2102 codec (7 Mhz)
    o spi.2 registers, chip select fires, some SPI signals exist
    o yet, kernel is unable to read the version ID via SPI
I need to do more oscoping & kernel debugging.

Frank Bennett
Mathegraphics,LLC
www.mathegraphics.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Status of the project ?