VGA on header

4,242 views
Skip to first unread message

Anthony Maciel

unread,
Sep 15, 2012, 9:56:56 PM9/15/12
to cubie...@googlegroups.com
Hi, I've seen that VGA is available on the 96 pin header, and I was just wondering how much support hardware might be required to tie in to these signals? Can it be directly wired in to a VGA cable, or will additional passives (resistors etc) be required? And approximately how many pins are involved with the VGA signaling? 

Thanks! 

jons...@gmail.com

unread,
Sep 15, 2012, 11:13:10 PM9/15/12
to cubie...@googlegroups.com
This is another board, but the schematic shows how the VGA connection works.

https://github.com/OLIMEX/OLINUXINO/blob/master/HARDWARE/A10-OLinuXino/A10-OLinuXino_Rev_A.sch.pdf?raw=true
> --
>
>



--
Jon Smirl
jons...@gmail.com

Alejandro Mery

unread,
Sep 16, 2012, 3:34:32 AM9/16/12
to cubie...@googlegroups.com
hot gossip! the schematics (beware they are unannounced, so they could
be wrong) of rev 2012-08-08 are available in
http://dl.cubieboard.org/hardware/cubieboard_schematic_2012-08-08.pdf

expansion headers in the last page, 5 pins in the second header marked
as VGA related. (RGB, HSYNC, VSYNC)

cheers,
Alejandro Mery

Pallav Aggarwal

unread,
Sep 16, 2012, 11:12:41 AM9/16/12
to cubie...@googlegroups.com
Can we connect VGA connector to R,G,B signal with Hsync, Vsync and AGND so as to connect a monitor?
or we need to do level translation before taking out to VGA connector.

Thanks
Pallav

NOD

unread,
Sep 17, 2012, 6:15:16 AM9/17/12
to cubie...@googlegroups.com
IF we have access to all the video out lines, then it's just a matter of a few components and a 5volt supply.

vga_cubieboard.jpg

Henrik Nordström

unread,
Sep 17, 2012, 8:21:47 AM9/17/12
to cubie...@googlegroups.com
We have. You only need a linedriver for the sync signals and a couple of
resistors.
> --
>
>


NOD

unread,
Sep 17, 2012, 8:37:09 AM9/17/12
to cubie...@googlegroups.com
Which are all listed in this jpg file:
https://groups.google.com/group/cubieboard/attach/a0fbc0aac9591b08/vga_cubieboard.jpg?part=4&authuser=0&view=1
It's based on the schematic from another Allwinner A10 board with a VGA output.
Can easily fit on a small PCB along with VGA connector and a small 7 pin ribbon cable connector  (or 10 pin 2 rows similar to those used for internal COM ports in stationary computers) 

Floris Bos / Maxnet

unread,
Sep 17, 2012, 9:21:29 AM9/17/12
to cubie...@googlegroups.com
Hi,

With all the talk about VGA, I was wondering about DVI instead.
Could the board be made to output DVI when using a HDMI<->DVI adapter?

The A10 crane FEX guide only lists pure HDMI modes/resolutions.
Is there a way to let it output DVI VGA resolutions, and select the
maximum resolution supported by the monitor automatically? (using EDID?)


Yours sincerely,

Floris Bos

jons...@gmail.com

unread,
Sep 17, 2012, 9:39:27 AM9/17/12
to cubie...@googlegroups.com
On Mon, Sep 17, 2012 at 9:21 AM, Floris Bos / Maxnet
<b...@je-eigen-domein.nl> wrote:
> Hi,
>
> With all the talk about VGA, I was wondering about DVI instead.
> Could the board be made to output DVI when using a HDMI<->DVI adapter?

In general an adapter like that should work without doing anything
special to the board. HDMI uses EDID too. EDID is just I2C with
specially formatted data. Basic HDMI and DVI are compatible.

> The A10 crane FEX guide only lists pure HDMI modes/resolutions.
> Is there a way to let it output DVI VGA resolutions, and select the maximum
> resolution supported by the monitor automatically? (using EDID?)
>

If it doesn't work I'd suspect the Linux video device driver first and
make sure it has fully implemented EDID support. In general supporting
these other resolutions is a software problem, not a hardware one.


>
> Yours sincerely,
>
> Floris Bos

Floris Bos / Maxnet

unread,
Sep 17, 2012, 10:08:00 AM9/17/12
to cubie...@googlegroups.com
On 09/17/2012 03:39 PM, jons...@gmail.com wrote:
> In general an adapter like that should work without doing anything
> special to the board. HDMI uses EDID too. EDID is just I2C with
> specially formatted data. Basic HDMI and DVI are compatible.

Question is, is the EDID hooked up at all?

Since it seems that with A10 devices the resolution is normally
hardcoded in the configuration (fex/script.bin) file loaded at boot.

===
screen0_output_mode =xx Screen0 output mode
(used for tv/hdmi output, 0:480i 1:576i
2:480p 3:576p 4:720p50 5:720p60
6:1080i50 7:1080i60 8:1080p24
9:1080p50 10:1080p60 11:pal 14:ntsc)
===

Don't see an autodetect option.


My experience with another A10 device is that connecting a DVI monitor
results in a screen with the wrong (720p) resolution and incorrect
purple colors. So that's why I'm asking.


Yours sincerely,

Floris Bos

jons...@gmail.com

unread,
Sep 17, 2012, 10:18:27 AM9/17/12
to cubie...@googlegroups.com
On Mon, Sep 17, 2012 at 10:08 AM, Floris Bos / Maxnet
<b...@je-eigen-domein.nl> wrote:
> On 09/17/2012 03:39 PM, jons...@gmail.com wrote:
>>
>> In general an adapter like that should work without doing anything special
>> to the board. HDMI uses EDID too. EDID is just I2C with specially formatted
>> data. Basic HDMI and DVI are compatible.
>
>
> Question is, is the EDID hooked up at all?

You should be able to read it by attaching it to an I2C pin.

>
> Since it seems that with A10 devices the resolution is normally hardcoded in
> the configuration (fex/script.bin) file loaded at boot.
>
> ===
> screen0_output_mode =xx Screen0 output mode
> (used for tv/hdmi output, 0:480i 1:576i
> 2:480p 3:576p 4:720p50 5:720p60
> 6:1080i50 7:1080i60 8:1080p24
> 9:1080p50 10:1080p60 11:pal 14:ntsc)
> ===
>
> Don't see an autodetect option.
>
>
> My experience with another A10 device is that connecting a DVI monitor
> results in a screen with the wrong (720p) resolution and incorrect purple
> colors. So that's why I'm asking.
>
>
>
> Yours sincerely,
>
> Floris Bos
>

Dmitriy B.

unread,
Sep 17, 2012, 10:22:22 AM9/17/12
to cubie...@googlegroups.com
You should use this tool  http://forum.doozan.com/read.php?6,9002  to override script.bin values. Also you need to use fbset command to change your framebuffer W/H settings too.

As far as I know, EDID info is hardcoded into kernel and is not really done right, guy named techn works on this stuff right now, lets hope he will succeed. Right now, my LG monitor with DVI input just hangs when I connect A10 device with DVI-HDMI convertor, this means that monitor waits for info about what modes device want to use and, obviously, never gets this data. On good monitors there is timeout for this function, but not on mine cheap one.

2012/9/17 Floris Bos / Maxnet <b...@je-eigen-domein.nl>



Yours sincerely,

Floris Bos

--



Gabor Molnar

unread,
Sep 17, 2012, 10:46:22 AM9/17/12
to cubie...@googlegroups.com
That's a good question.
I have a Mele A2000, and that device is also produce a purple screen when I connect to a monitor with a HDMI -> DVI cable. 
Do you know somebody what is the reason? 


2012/9/17 Floris Bos / Maxnet <b...@je-eigen-domein.nl>



Yours sincerely,

Floris Bos

--



NOD

unread,
Sep 17, 2012, 11:08:46 AM9/17/12
to cubie...@googlegroups.com
AFAIK it's because the Allwinner A10 is only designed with HDMI in mind, since it's a fixed standard.
DVI on the other hand is more flexible but operates slightly differently, which they didn't account for.
So we're "stuck" with either VGA (with some cheap components and a few of the 96 expansion pins), HDMI OR CVBS.

jons...@gmail.com

unread,
Sep 17, 2012, 12:19:17 PM9/17/12
to cubie...@googlegroups.com
On Mon, Sep 17, 2012 at 11:08 AM, NOD <ttmo...@gmail.com> wrote:
> AFAIK it's because the Allwinner A10 is only designed with HDMI in mind,
> since it's a fixed standard.
> DVI on the other hand is more flexible but operates slightly differently,
> which they didn't account for.

I haven't seen anything that makes me think that this is a fundamental
hardware problem. I think this is much more likely a problem with
missing code in the video drivers. Although the EDID input on the
HDMI plug has to be hooked up to i2c somehow for the software to have
a chance of working.

If everything is ok in the kernel the read-edid-i2c program in this
thread should be able to read the edid data.
http://stackoverflow.com/questions/5065159/reading-edid-from-eeprom

If that doesn't work then the first step is making the EDID data
appear on the i2c bus. Once you get that working then the video driver
has to act on the EDID data it sees.

Henrik Nordström

unread,
Sep 17, 2012, 2:56:13 PM9/17/12
to cubie...@googlegroups.com
mån 2012-09-17 klockan 15:21 +0200 skrev Floris Bos / Maxnet:

> With all the talk about VGA, I was wondering about DVI instead.
> Could the board be made to output DVI when using a HDMI<->DVI adapter?

The A10 apparently outputs HDMI signals in the normal HDMI YCbCr
colourspace. Most DVI-only monitors only handle RGB pixel format and
won't know what to do.

Maybe it's possible to change the pixel format used for the HDMI
connection, but I haven't found anything in some quick searches.

Regards
Henrik

jons...@gmail.com

unread,
Sep 17, 2012, 3:03:26 PM9/17/12
to cubie...@googlegroups.com
That should be setable in software. The EDID tells it which color space to use.

>
> Regards
> Henrik

jons...@gmail.com

unread,
Sep 17, 2012, 3:22:38 PM9/17/12
to cubie...@googlegroups.com
EDID spec
http://www.scribd.com/doc/72004086/103/Table-29-CEA-EDID-Timing-Extension-Version-3-new-in-861B

There is field in there for:
don't care
RGB
YCBCR 4:2:2
YCBCR 4:4:4

Video output should be programmed to match the info in the EDID block.
There should be code in the kernel doing this if the video driver has
been properly written.

I haven't played with mode setting in the kernel since they did the
big rewrite a couple of years ago, but this is the general idea of how
it is done.

--
Jon Smirl
jons...@gmail.com

jons...@gmail.com

unread,
Sep 17, 2012, 3:25:11 PM9/17/12
to cubie...@googlegroups.com
First thing to verify is if you can see the monitor's EDID on one of
the I2C buses. Then start messing with the video driver to see if it
is interpreting it correctly.

Tablets avoid all of this because they get hard coded for the LCD
panel attached.

--
Jon Smirl
jons...@gmail.com

Anthony Maciel

unread,
Sep 28, 2012, 6:12:48 PM9/28/12
to cubie...@googlegroups.com
Thanks for the info guys, really helpful.

Just curious, would anyone be interested in a sort of group buy for a VGA breakout board? Looks like I would be able to put together a full kit for about $10-$15 each with US shipping (not sure about international) assuming ~10 people would be willing to go in on it. If you already have the components you might be able to do it cheaper yourself, but should save a bit of money for those of us without a stock pile of components.

Let me know if you are interested, and if it seems like there are enough people I will pursue putting it together.

NOD

unread,
Sep 29, 2012, 5:50:42 AM9/29/12
to cubie...@googlegroups.com
AFAIK Hsync and Vsync also needs a set of 22ohm resistors right before the 74ACT08D chip.
Just in-case you missed them, because the schematic for Olimex's board is a little bit confusing.

Joao Mendonca

unread,
Sep 29, 2012, 5:33:51 PM9/29/12
to cubie...@googlegroups.com
I would be interested... Let me know if this is really going to happen!

Floris Bos / Maxnet

unread,
Sep 29, 2012, 5:37:26 PM9/29/12
to cubie...@googlegroups.com
On 09/17/2012 08:56 PM, Henrik Nordstr�m wrote:
Did some more testing.
And the good news is that at least with my monitor the color problem
only exists under Android and not under Linux.
Wonder if it was silenty fixed in the newer kernel version, or if it is
some graphics initialization that Android does that causes trouble.


Photos to illustrate it:

Android: http://s11.postimage.org/jqij7obyr/colors_android.jpg
Linux: http://s16.postimage.org/uaonmyuit/colors_linux.jpg

Now only need to find a way to get the resolution right. (specifying
resolutions+timings manually in fbset doesn't work)


Yours sincerely,

Floris Bos
<http://s16.postimage.org/uaonmyuit/colors_linux.jpg>

jons...@gmail.com

unread,
Sep 29, 2012, 6:07:21 PM9/29/12
to cubie...@googlegroups.com
On Sat, Sep 29, 2012 at 5:37 PM, Floris Bos / Maxnet
<b...@je-eigen-domein.nl> wrote:
> On 09/17/2012 08:56 PM, Henrik Nordström wrote:
>>
>> mån 2012-09-17 klockan 15:21 +0200 skrev Floris Bos / Maxnet:
>>
>>> With all the talk about VGA, I was wondering about DVI instead.
>>> Could the board be made to output DVI when using a HDMI<->DVI adapter?
>>
>> The A10 apparently outputs HDMI signals in the normal HDMI YCbCr
>> colourspace. Most DVI-only monitors only handle RGB pixel format and
>> won't know what to do.
>>
>> Maybe it's possible to change the pixel format used for the HDMI
>> connection, but I haven't found anything in some quick searches.
>
>
> Did some more testing.
> And the good news is that at least with my monitor the color problem only
> exists under Android and not under Linux.
> Wonder if it was silenty fixed in the newer kernel version, or if it is some
> graphics initialization that Android does that causes trouble.

A10/A13 chips have mainly been used in systems with fixed monitors (or
semi-fixed like TVs) . There probably hasn't been much work done on
the code that supports general purpose plug-in monitors. It not that
hard to hook the code up, but someone has to do it. The kernel already
knows how to decode the EDID. The decoded bits just need to poke into
the right places in the A10/13 video drivers to set the indicated
mode. Some of that poking code is likely missing.

In your older kernel, the kernel decodes the EDID. It then pokes the
video driver and says 'set RGB mode'. If you go look the function for
setting RGB mode it is probably empty in the source code because
nobody has written it yet. In your later kernel someone added it.

EDID also tells the resolution. The functions in the A10/13 video
driver for setting the resolution are probably do-nothing stubs and
they need to be implemented.

Things like tablet displays don't have EDID since they can't be
changed. So the programmers just hard code in the info about the
screen. That's why the tablets works and plug-in monitors don't.


>
>
> Photos to illustrate it:
>
> Android: http://s11.postimage.org/jqij7obyr/colors_android.jpg
> Linux: http://s16.postimage.org/uaonmyuit/colors_linux.jpg
>
> Now only need to find a way to get the resolution right. (specifying
> resolutions+timings manually in fbset doesn't work)
>
>
> Yours sincerely,
>
> Floris Bos
> <http://s16.postimage.org/uaonmyuit/colors_linux.jpg>
>

killwarez

unread,
Sep 30, 2012, 2:49:38 PM9/30/12
to cubie...@googlegroups.com
On Sunday, September 30, 2012 1:07:23 AM UTC+3, Jon Smirl wrote:
A10/A13 chips have mainly been used in systems with fixed monitors (or 
semi-fixed like TVs) . There probably hasn't been much work done on
the code that supports general purpose plug-in monitors.  It not that
hard to hook the code up, but someone has to do it. The kernel already
knows how to decode the EDID. The decoded bits just need to poke into
the right places in the A10/13 video drivers to set the indicated
mode. Some of that poking code is likely missing.

In your older kernel, the kernel decodes the EDID. It then pokes the
video driver and says 'set RGB mode'. If you go look the function for
setting RGB mode it is probably empty in the source code because
nobody has written it yet. In your later kernel someone added it.

EDID also tells the resolution. The functions in the A10/13 video
driver for setting the resolution are probably do-nothing stubs and
they need to be implemented.

I have a task to set custom resolution on vga port of A10. There is predefined set of display modes in kernel. At doozan's forum there is a tool called a10_display that passes a param and sets on of predefined modes http://forum.doozan.com/read.php?6,9002,9002,quote=1.


PS: i don't have exp in embdd development 

Floris Bos / Maxnet

unread,
Sep 30, 2012, 3:09:47 PM9/30/12
to cubie...@googlegroups.com
On 09/30/2012 08:49 PM, killwarez wrote:
> I have a task to set custom resolution on vga port of A10. There is
> predefined set of display modes in kernel. At doozan's forum there is
> a tool called a10_display that passes a param and sets on of
> predefined modes http://forum.doozan.com/read.php?6,9002,9002,quote=1.
>

Problem is that when using DVI, one is not using the VGA port, but HDMI
instead.
Can only get a10_display to do either 720p/1080 on a DVI monitor, and
not the other modes.
Hard coded timings here as well:
https://github.com/amery/linux-allwinner/blob/allwinner-v3.0-android-v2/drivers/video/sunxi/hdmi/aw/hdmi_core.c


Yours sincerely,

Floris Bos

jons...@gmail.com

unread,
Sep 30, 2012, 3:16:01 PM9/30/12
to cubie...@googlegroups.com
There are two pieces to mode setting.

1) What is the A10's capablity. What resolutions does it support, what
display formats, etc.

2) What does the monitor support.

These two pieces of info get AND'd together. If the A10 support the
mode and if the monitor supports the mode, then the mode is legal. You
can then use command line tools to pick any of the legal modes. Or the
Linux desktop has a GUI for choosing from the legal modes. The
general kernel support code for doing this has been around for over a
decade so it is well supported.

From what I see posted all (or most) of the info needed to correctly
implement mode setting in the kernel is available. So the first thing
to do is make sure you are using the most up to date kernel possible.
Linux and Andriod kernels are not merged yet so there are two
independent kernel branches to check.

If neither of these work, then you have to poke around in the A10
video driver code and implement the missing functions.

-------------------------

Also, I don't know what display system you are using. The X server has
not yet been converted all of the way over to using KMS. You may need
to implement mode setting in an X video driver. That's simply an
exercise in copying the code out of the kernel and reimplementing it
in X (which is why it is being eliminated).

If you're really into video see if you can get Wayland running.

--
Jon Smirl
jons...@gmail.com

Anthony Maciel

unread,
Sep 30, 2012, 6:13:42 PM9/30/12
to cubie...@googlegroups.com
Joao and Ettore - Thanks, you've been added to the list. That brings us up to 4 boards, so only a few more and it'll hit the target price. For anyone else interested in the breakout board you can fill out this quick form: https://docs.google.com/spreadsheet/viewform?formkey=dEtuQUJHdkZycEpmckpwcjN1VWVmYXc6MQ . Should make it a little easier to keep track of everyone who is interested.

NOD - Good catch, totally missed that. That could have caused some headaches...

Does anyone know of another A10 board with VGA that has schematics available? AFAIK, the Olimex A10 board is still fairly early on in its development process correct? Does anyone know if the VGA circuit is known to be correct? I'm not saying anything against the Olimex devs, I'm sure they know what they're doing (way more than I do), but it would suck if there were a bug with this breakout board that could have been avoided by double checking against a second reference. I looked around a little and wasn't able to find another A10 based board with VGA that had schematics readily available, but if anyone knows of any I'd really appreciate it.


Thanks,

Anthony

jons...@gmail.com

unread,
Sep 30, 2012, 6:57:24 PM9/30/12
to cubie...@googlegroups.com
On Sun, Sep 30, 2012 at 6:13 PM, Anthony Maciel <maci...@gmail.com> wrote:
> Joao and Ettore - Thanks, you've been added to the list. That brings us up
> to 4 boards, so only a few more and it'll hit the target price. For anyone
> else interested in the breakout board you can fill out this quick form:
> https://docs.google.com/spreadsheet/viewform?formkey=dEtuQUJHdkZycEpmckpwcjN1VWVmYXc6MQ
> . Should make it a little easier to keep track of everyone who is
> interested.
>
> NOD - Good catch, totally missed that. That could have caused some
> headaches...
>
> Does anyone know of another A10 board with VGA that has schematics
> available? AFAIK, the Olimex A10 board is still fairly early on in its
> development process correct? Does anyone know if the VGA circuit is known to
> be correct? I'm not saying anything against the Olimex devs, I'm sure they
> know what they're doing (way more than I do), but it would suck if there
> were a bug with this breakout board that could have been avoided by double
> checking against a second reference. I looked around a little and wasn't
> able to find another A10 based board with VGA that had schematics readily
> available, but if anyone knows of any I'd really appreciate it.

Does this apply?
https://www.olimex.com/Products/OLinuXino/A13/A13-OLinuXino-WIFI-DEV/resources/A13-OLinuXino-WIFI-DEV_fix.pdf


>
>
> Thanks,
>
> Anthony
>
> On Saturday, September 29, 2012 2:33:51 PM UTC-7, Joao Mendonca wrote:
>>
>> I would be interested... Let me know if this is really going to happen!
>

Alejandro Mery

unread,
Sep 30, 2012, 7:02:06 PM9/30/12
to cubie...@googlegroups.com
On 1 October 2012 00:57, jons...@gmail.com <jons...@gmail.com> wrote:
> On Sun, Sep 30, 2012 at 6:13 PM, Anthony Maciel <maci...@gmail.com> wrote:
>> Joao and Ettore - Thanks, you've been added to the list. That brings us up
>> to 4 boards, so only a few more and it'll hit the target price. For anyone
>> else interested in the breakout board you can fill out this quick form:
>> https://docs.google.com/spreadsheet/viewform?formkey=dEtuQUJHdkZycEpmckpwcjN1VWVmYXc6MQ
>> . Should make it a little easier to keep track of everyone who is
>> interested.
>>
>> NOD - Good catch, totally missed that. That could have caused some
>> headaches...
>>
>> Does anyone know of another A10 board with VGA that has schematics
>> available? AFAIK, the Olimex A10 board is still fairly early on in its
>> development process correct? Does anyone know if the VGA circuit is known to
>> be correct? I'm not saying anything against the Olimex devs, I'm sure they
>> know what they're doing (way more than I do), but it would suck if there
>> were a bug with this breakout board that could have been avoided by double
>> checking against a second reference. I looked around a little and wasn't
>> able to find another A10 based board with VGA that had schematics readily
>> available, but if anyone knows of any I'd really appreciate it.
>
> Does this apply?
> https://www.olimex.com/Products/OLinuXino/A13/A13-OLinuXino-WIFI-DEV/resources/A13-OLinuXino-WIFI-DEV_fix.pdf

no relation at all. for the A13-olinuxino olimex made a VGA output out
of the LCD using some buffers and black magic. the A10 has native RGB.

NOD

unread,
Sep 30, 2012, 7:38:24 PM9/30/12
to cubie...@googlegroups.com
AFAIK the A13 employs a set of converter chips that takes the signals normally intended for a tablet's screen and turn them into VGA.
https://github.com/OLIMEX/OLINUXINO/blob/master/HARDWARE/A13-OLinuXino/A13-OLinuXino_Rev_B.sch.pdf

Tom Cubie

unread,
Oct 1, 2012, 1:25:01 AM10/1/12
to cubie...@googlegroups.com
On Mon, Oct 1, 2012 at 6:13 AM, Anthony Maciel <maci...@gmail.com> wrote:
> Joao and Ettore - Thanks, you've been added to the list. That brings us up
> to 4 boards, so only a few more and it'll hit the target price. For anyone
> else interested in the breakout board you can fill out this quick form:
> https://docs.google.com/spreadsheet/viewform?formkey=dEtuQUJHdkZycEpmckpwcjN1VWVmYXc6MQ
> . Should make it a little easier to keep track of everyone who is
> interested.
>
> NOD - Good catch, totally missed that. That could have caused some
> headaches...
>
> Does anyone know of another A10 board with VGA that has schematics
> available? AFAIK, the Olimex A10 board is still fairly early on in its
> development process correct? Does anyone know if the VGA circuit is known to
> be correct? I'm not saying anything against the Olimex devs, I'm sure they
> know what they're doing (way more than I do), but it would suck if there
> were a bug with this breakout board that could have been avoided by double
> checking against a second reference. I looked around a little and wasn't
> able to find another A10 based board with VGA that had schematics readily
> available, but if anyone knows of any I'd really appreciate it.
Hi,
We are making a test board to test all the gpio on cubieboard(by
driving leds) and for the analog signal like vga, we will put a vga
header to test. The schematics will be provided.
>
>
> Thanks,
>
> Anthony
>
> On Saturday, September 29, 2012 2:33:51 PM UTC-7, Joao Mendonca wrote:
>>
>> I would be interested... Let me know if this is really going to happen!
>
> --
>
>



--
Keep simple, stay foolish.

Anthony Maciel

unread,
Oct 1, 2012, 1:29:54 AM10/1/12
to cubie...@googlegroups.com
Great! Thanks, that should help out a ton.

Anthony

blackstag

unread,
Oct 1, 2012, 1:29:54 AM10/1/12
to cubie...@googlegroups.com
very cool thanks.

--



Floris Bos / Maxnet

unread,
Oct 2, 2012, 7:09:44 PM10/2/12
to cubie...@googlegroups.com
On 10/01/2012 12:13 AM, Anthony Maciel wrote:
> Does anyone know of another A10 board with VGA that has schematics
> available? AFAIK, the Olimex A10 board is still fairly early on in its
> development process correct? Does anyone know if the VGA circuit is
> known to be correct?

Only RGB and H/Vsync pins seems to be connected in the schematic?

Think the DDC pins should be hooked up as well, in order to be able to
get EDID support at some stage.
(to detect what resolutions the monitor supports, and as a means to
detect if anything is connected to the port. So that video output can be
handled dynamically in the future, instead of the current practice of
hardcoding settings in .fex files)


Yours sincerely,

Floris Bos

Anthony Maciel

unread,
Oct 2, 2012, 9:10:10 PM10/2/12
to cubie...@googlegroups.com
On Tuesday, October 2, 2012 4:09:48 PM UTC-7, Max wrote:
Think the DDC pins should be hooked up as well, in order to be able to
get EDID support at some stage.
(to detect what resolutions the monitor supports, and as a means to
detect if anything is connected to the port. So that video output can be
handled dynamically in the future, instead of the current practice of
hardcoding settings in .fex files)


Yours sincerely,

Floris Bos

I think that would be a good idea so it is future proof. Do you happen to know if the DDC pins are broken out on the cubieboard? Or even if the A10 provides this signalling natively? According to Wikipedia it looks like it is based in I2C, so theoretically we could just do it manually if it doesn't, correct?

If you could provide some additional details about DDC if you know more that'd be helpful. To be honest I don't really know anything about the VGA spec, I'm just going off of the schematics that others have posted.

Thanks for the input!

Anthony 

Floris Bos / Maxnet

unread,
Oct 2, 2012, 9:40:44 PM10/2/12
to cubie...@googlegroups.com
On 10/03/2012 03:10 AM, Anthony Maciel wrote:
On Tuesday, October 2, 2012 4:09:48 PM UTC-7, Max wrote:
Think the DDC pins should be hooked up as well, in order to be able to
get EDID support at some stage.
(to detect what resolutions the monitor supports, and as a means to
detect if anything is connected to the port. So that video output can be
handled dynamically in the future, instead of the current practice of
hardcoding settings in .fex files)


Yours sincerely,

Floris Bos

I think that would be a good idea so it is future proof. Do you happen to know if the DDC pins are broken out on the cubieboard? Or even if the A10 provides this signalling natively? According to Wikipedia it looks like it is based in I2C, so theoretically we could just do it manually if it doesn't, correct?


Don't know if the voltages used match.


If you could provide some additional details about DDC if you know more that'd be helpful. To be honest I don't really know anything about the VGA spec, I'm just going off of the schematics that others have posted.


I am afraid I'm not going to be much help with that.
More of a software development person, then into the electronics side of things.

Just know enough about it, that if the pins are not connected anywhere in the first place, DCC/EDID never has a chance of working. :-)
And I do think that functionality is essential and "not optional", if there is the intention of making devices like this a bit more mainstream in the future.
End-users expect to be able to insert the plug into either the vga or hdmi port, and have it "just work".



Yours sincerely,

Floris Bos

Anthony Maciel

unread,
Oct 3, 2012, 1:25:27 PM10/3/12
to cubie...@googlegroups.com


On Tuesday, October 2, 2012 6:40:48 PM UTC-7, Max wrote:

I am afraid I'm not going to be much help with that.
More of a software development person, then into the electronics side of things.

Just know enough about it, that if the pins are not connected anywhere in the first place, DCC/EDID never has a chance of working. :-)
And I do think that functionality is essential and "not optional", if there is the intention of making devices like this a bit more mainstream in the future.
End-users expect to be able to insert the plug into either the vga or hdmi port, and have it "just work".



Yours sincerely,

Floris Bos

Ah, ok, no problem. Same here for the most part. Software is my day job, but I dabble some with the electronics side in my spare time. 

Not certain that this is essential for this breakout board, though. I think by the fact that this is separate lends itself to only being used by those who have a little bit more experience with hardware (even if just putting together a PC), so you don't have to worry as much about an "End User" type. If  this was a part of the cubieboard directly (like it is with the olinuxino) then there might be more of an expectation for it to just work, but I don't think it'd be a deal breaker here. That said I'd love to include it if possible, but if it is too much work then I probably won't. Also keep in mind that the resolutions that we are dealing with on this VGA connector are really low comparatively, which limits the usefulness of the negotiation step since it'd likely be hard to find a monitor which didn't support the resolution.

Michal Suchanek

unread,
Jan 3, 2013, 5:59:45 AM1/3/13
to cubie...@googlegroups.com
On 29 September 2012 00:12, Anthony Maciel <maci...@gmail.com> wrote:
> Thanks for the info guys, really helpful.
>
> Just curious, would anyone be interested in a sort of group buy for a VGA
> breakout board? Looks like I would be able to put together a full kit for
> about $10-$15 each with US shipping (not sure about international) assuming
> ~10 people would be willing to go in on it. If you already have the
> components you might be able to do it cheaper yourself, but should save a
> bit of money for those of us without a stock pile of components.
>
> Let me know if you are interested, and if it seems like there are enough
> people I will pursue putting it together.
>

Hello,

how is the VGA header coming along?

Did anybody manage to connect a VGA screen?

With fixed resolution only or with DDC as well?

Thanks

Michal

George Ioakimedes

unread,
Jan 4, 2013, 9:43:59 AM1/4/13
to cubie...@googlegroups.com
Hi Everyone:

I just saw this thread about a VGA breakout and wonder if there is enough interest to add it to my Baseboard design. I think I have room for the circuit in the lower left corner of the board. You can read about it in this thread

I'm trying to finish the board layout today. I need to post updated schematics so I'll try to do that today but if someone could consolidate their findings on what's the correct circuit for VGA I'll see if I can add the circuit and a HD15 to the board.

Alejandro Mery

unread,
Jan 4, 2013, 10:18:18 AM1/4/13
to cubie...@googlegroups.com
yes please!

George Ioakimedes

unread,
Jan 4, 2013, 2:49:05 PM1/4/13
to cubie...@googlegroups.com
OK but I really would appreciate it if someone would help clarify what circuit should be designed in. Right now I'm trying to finish the rest of the board layout so it can be released soon.

Alejandro Mery

unread,
Jan 4, 2013, 2:56:00 PM1/4/13
to cubie...@googlegroups.com

killwarez

unread,
Jan 4, 2013, 3:02:18 PM1/4/13
to cubie...@googlegroups.com
I tried to make VGA output according to schematic from A10-Olimex board but nothing. I dont know maybe schematic has mistakes, maybe I made not very well, maybe breadboard not best tool to make such schemas. Today it was decided to design and make PCB, my colleagues agreed with me. So I'm going to design PCB during this weekend and start producing it and filling with the parts only on nearest Wednesday. After that I can say if schematic has mistakes.

Fadil Berisha

unread,
Jan 4, 2013, 3:18:36 PM1/4/13
to cubie...@googlegroups.com
A13-olinuxino solution for VGA is working in their board, but as I can
remember somebody on (this list)? has some criticism about that
solution. If somebody know better solution is good to show now before
board is finished.

Fadil Berisha
> --
> You received this message because you are subscribed to the Google Groups
> "Cubieboard" group.
> To unsubscribe from this group, send email to
> cubieboard+...@googlegroups.com.
> Visit this group at http://groups.google.com/group/cubieboard?hl=en.
>
>

killwarez

unread,
Jan 4, 2013, 3:53:18 PM1/4/13
to cubie...@googlegroups.com, f.ko...@gmail.com
If you're look closer to A10 and A13 differences you'll notice that A13 doesn't have VGA/RGB output. So Olimex made VGA output on A13 board by converting LCD (LVDS?) signal to VGA. We don't need to do so, because Cubieboard based on A10 which has native RGB.

But I should say that Olimex's LCD->VGA schematic is great!

Alejandro Mery

unread,
Jan 4, 2013, 3:55:00 PM1/4/13
to cubie...@googlegroups.com
On 04/01/13 21:53, killwarez wrote:
> If you're look closer to A10 and A13 differences you'll notice that A13
> doesn't have VGA/RGB output. So Olimex made VGA output on A13 board by
> converting LCD (LVDS?) signal to VGA. We don't need to do so, because
> Cubieboard based on A10 which has native RGB.
>
> But I should say that Olimex's LCD->VGA schematic is great!

the link I pasted was for the a10-olinuxino schematics, not for the
A13-olinuxino's.

regards,
Alejandro Mery

George Ioakimedes

unread,
Jan 4, 2013, 5:03:54 PM1/4/13
to cubie...@googlegroups.com
I quick look at @Alejandro's link and it looks reasonable to me. Just wondering if it's worth it to add I2C

jons...@gmail.com

unread,
Jan 4, 2013, 5:14:19 PM1/4/13
to cubieboard
On Fri, Jan 4, 2013 at 5:03 PM, George Ioakimedes <georg...@gmail.com> wrote:
> I quick look at @Alejandro's link and it looks reasonable to me. Just
> wondering if it's worth it to add I2C

The monitor's DDC signal is I2C. Without DDC the video driver will be
unable to read the monitor's capabilities and automatically adjust the
mode setting. It is hard to get a system started without having a
mode set on the monitor. You will have to ssh in and fix it up that
way since you won't be able to see what you are doing on the VGA
monitor. It is also possible to destroy CRT based VGA monitors by
setting modes into the video hardware that are not supported by the
CRT.

>
> --
> You received this message because you are subscribed to the Google Groups
> "Cubieboard" group.
> To unsubscribe from this group, send email to
> cubieboard+...@googlegroups.com.
> Visit this group at http://groups.google.com/group/cubieboard?hl=en.
>
>



--
Jon Smirl
jons...@gmail.com

George Ioakimedes

unread,
Jan 4, 2013, 5:28:32 PM1/4/13
to cubie...@googlegroups.com
So it looks like U15-6 & U15-7 (TWI0-SCK & TWI0-SDA) are available to use and already have pull-ups mounted so it's just a straight shot to VGA pin 15 and VGA pin 12.

  • RGB need 75 ohm pull downs to create the 75 ohm impedance lines
  • HSYNC/VSYNC go through buffers and protected by BAT54S
Does that work for everyone?

Alejandro Mery

unread,
Jan 4, 2013, 5:40:51 PM1/4/13
to cubie...@googlegroups.com
A good thing with that pick is that PB18/PB19 only have that function
(beside GPIO), but it might be a good idea to preserve the pins for
someone not wanting VGA but needing i2c for other purposes.

George Ioakimedes

unread,
Jan 4, 2013, 5:44:07 PM1/4/13
to cubie...@googlegroups.com
I'll be posting my latest schematic soon but all of the pins are still brought out to headers so they can be used for whatever you want. As long as a VGA monitor is not plugged in the I2C lines (PB18/19) can be used for something else.

I wonder why they didn't make the other TWI available on the header?

George Ioakimedes

unread,
Jan 4, 2013, 6:19:13 PM1/4/13
to cubie...@googlegroups.com
Anyone got time to looks for a shallow HD15 connector for the VGA? The spot I want to put it on only has about 11mm before it runs into the mounting post of the Cubieboard. I know there are some shallow HD15 connectors but the first couple I looked at were ~15mm deep.

theRat

unread,
Jan 4, 2013, 8:31:53 PM1/4/13
to cubie...@googlegroups.com

George Ioakimedes

unread,
Jan 5, 2013, 3:07:13 AM1/5/13
to cubie...@googlegroups.com
Thanks for finding that. A little on the expensive side but hopefully it has a common footprint. I went ahead and designed it in but couldn't find a 3D model to add to the design. The updated schematics, and board plots can be looked at here. If nobody finds anything wrong then I plan on sending out to Seeestudio this weekend.

George Ioakimedes

unread,
Jan 23, 2013, 7:28:49 PM1/23/13
to cubie...@googlegroups.com
OK guys, I'm posting this here as to not clutter up the Baseboard thread. I've made some progress with some other action items and decided I would try VGA today. I'm running the Linaro Ubuntu Desktop off of a SD Card because at the moment it's easier for me to modify the script.bin file.

I've tried a few different settings and am not able to get an image on 2 different monitors. My last test is only enabling 1 screen and setting it to VGA:

[disp_init]
disp_init_enable = 1
disp_mode = 0
screen0_output_type = 4
screen0_output_mode = 4
screen1_output_type = 2
screen1_output_mode = 14
fb0_framebuffer_num = 2
fb0_format = 10
fb0_pixel_sequence = 0
fb0_scaler_mode_enable = 1
fb1_framebuffer_num = 2
fb1_format = 10
fb1_pixel_sequence = 0
fb1_scaler_mode_enable = 1


I tried to add comments to the FEX file but it seems that fex2bin utility strips them out. So you don't have to hunt for the meanings here's what the setting mean:


;----------------------------------------------------------------------------------
;disp init configuration
;
;disp_mode (0:screen0<screen0,fb0> 1:screen1<screen1,fb0> 2:two_diff_screen_diff_contents<screen0,screen1,fb0,fb1>
; 3:two_same_screen_diff_contets<screen0,screen1,fb0> 4:two_diff_screen_same_contents<screen0,screen1,fb0>)
;screenx_output_type (0:none; 1:lcd; 2:tv; 3:hdmi; 4:vga)
;screenx_output_mode (used for tv/hdmi output, 0:480i 1:576i 2:480p 3:576p 4:720p50 5:720p60 6:1080i50 7:1080i60 8:1080p24 9:1080p50 10:1080p60 11:pal 14:ntsc)
;screenx_output_mode (used for vga output, 0:1680*1050 1:1440*900 2:1360*768 3:1280*1024 4:1024*768 5:800*600 6:640*480 10:1920*1080 11:1280*720)
;fbx format (4:RGB655 5:RGB565 6:RGB556 7:ARGB1555 8:RGBA5551 9:RGB888 10:ARGB8888 12:ARGB4444)
;fbx pixel sequence (0:ARGB 1:BGRA 2:ABGR 3:RGBA) --- 0 for linux, 2 for android
;---------------------------------------------------------------------------------


Anyone got any ideas on what else I can try?

Floris Bos

unread,
Jan 23, 2013, 10:35:31 PM1/23/13
to cubie...@googlegroups.com
On 01/24/2013 01:28 AM, George Ioakimedes wrote:
OK guys, I'm posting this here as to not clutter up the Baseboard thread. I've made some progress with some other action items and decided I would try VGA today. I'm running the Linaro Ubuntu Desktop off of a SD Card because at the moment it's easier for me to modify the script.bin file.

Anyone got any ideas on what else I can try?


Came accross the following cubieboard.fex file commit labeled "enable vga":

https://github.com/matson-hall/allwinner-pack-tools/commit/bf84adf33e5bee3583c777fd7db61c208ea976ed

You might want to give those exact settings a try.

-- 
Yours sincerely,

Floris Bos

George Ioakimedes

unread,
Jan 23, 2013, 11:53:40 PM1/23/13
to cubie...@googlegroups.com
Thanks for pointing to that but unfortunately it did not work either. It does help as there is a signal recognized but no picture. I tried a couple of different settings but not get VGA to work. When I get to the Linaro desktop and check displays only 1 display is found but I would expect to see 2 displays.

I wonder if I have to configure more than just the script.bin file??

Roman Mamedov

unread,
Jan 24, 2013, 12:03:05 AM1/24/13
to cubie...@googlegroups.com, georg...@gmail.com
On Wed, 23 Jan 2013 20:53:40 -0800 (PST)
George Ioakimedes <georg...@gmail.com> wrote:

> I wonder if I have to configure more than just the script.bin file??

Did you try a10_display?

http://forum.doozan.com/read.php?6,9002,9002

e.g. "a10_display vga on 3"


> When I get to the Linaro desktop and check displays only 1 display is found
> but I would expect to see 2 displays.

AFAIK there is always one display, as in, you can't show a different picture
on HDMI/VGA/LCD.

--
With respect,
Roman

~~~~~~~~~~~~~~~~~~~~~~~~~~~
"Stallman had a printer,
with code he could not see.
So he began to tinker,
and set the software free."
signature.asc

George Ioakimedes

unread,
Jan 24, 2013, 12:20:23 AM1/24/13
to cubie...@googlegroups.com, georg...@gmail.com
No, I forgot to try that. I did find this note :
 
To use VGA or composite output, download mele-a1000-vga_hwpack_2012.06.13.7z instead of mele-a1000_hwpack_2012.06.13.7z.
 
 From CNX Software so perhaps your image has to built differently as well as configuring script.bin??

Floris Bos

unread,
Jan 24, 2013, 1:10:10 AM1/24/13
to cubie...@googlegroups.com
No, he only uses a different kernel configuration for the headless mele-a1000-server.
All others use the default sun4i_defconfig kernel, and just a different .fex file.

See: https://github.com/cnxsoft/a10-config
https://github.com/cnxsoft/a10-tools/blob/master/a10-hwpack-bld.sh#L136


Yours sincerely,

Floris Bos

killwarez

unread,
Jan 24, 2013, 4:15:57 AM1/24/13
to cubie...@googlegroups.com, georg...@gmail.com
George, your fex params should work, i tried them on original mele a1000 device many times and they are work fine. Maybe some other such as bad kernel config or incorrect schematic. I'll try to make this schematic on breadboard with DIP components and android.

jons...@gmail.com

unread,
Jan 24, 2013, 11:17:36 AM1/24/13
to cubieboard, georg...@gmail.com
On Thu, Jan 24, 2013 at 12:03 AM, Roman Mamedov <r...@romanrm.ru> wrote:
> On Wed, 23 Jan 2013 20:53:40 -0800 (PST)
> George Ioakimedes <georg...@gmail.com> wrote:
>
>> I wonder if I have to configure more than just the script.bin file??
>
> Did you try a10_display?
>
> http://forum.doozan.com/read.php?6,9002,9002
>
> e.g. "a10_display vga on 3"
>
>
>> When I get to the Linaro desktop and check displays only 1 display is found
>> but I would expect to see 2 displays.
>
> AFAIK there is always one display, as in, you can't show a different picture
> on HDMI/VGA/LCD.


I don't know what the state of the X display driver is, but it
probably needs code added to teach it about multi-head support. Normal
way of doing that is to make one framebuffer that has two view ports
onto it. You can probably get X to make the big framebuffer but it
doesn't know how to set multiple simultaneous viewports. If it already
supports single monitor HDMI and VGA it shouldn't be too much work to
make it support simultaneous HDMI and VGA.

I'm assuming that the A10 supports having HDMI, VGA, LCD
simultaneously active. With the right driver software you should be
able to get as many simultaneous displays as the hardware supports.

>
> --
> With respect,
> Roman
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~
> "Stallman had a printer,
> with code he could not see.
> So he began to tinker,
> and set the software free."



--
Jon Smirl
jons...@gmail.com

George Ioakimedes

unread,
Jan 24, 2013, 11:49:24 AM1/24/13
to cubie...@googlegroups.com, georg...@gmail.com
I'm not good at all with Linux and never developed for Android so I'm at this group's mercy for help on bringing some of these things up. I have to switch gears and do something else this morning but I wanted to get a couple of quick scope plots to see what is going on with the signals. Don't put too much into the spikes as I'm pretty much worst case in hooking up the scope probes. For reference I inserted paper clips into the VGA connector (they are a perfect fit BTW) and am using the clips on the scope probes. For scope ground it's even worse, alligator clip to a jumper wire shoved (not soldered into a ground hole on the baseboard).

Anyway, for your viewing pleasure...


So we are getting signals out to the VGA connector, there's just something not right. Note that the RED signal is at a lower level than the syncs which I believe is correct, syncs should be at 5V and RGB at 1.8V? 


 

Floris Bos

unread,
Jan 24, 2013, 12:09:54 PM1/24/13
to cubie...@googlegroups.com
On 01/24/2013 05:49 PM, George Ioakimedes wrote:
So we are getting signals out to the VGA connector, there's just something not right. Note that the RED signal is at a lower level than the syncs which I believe is correct, syncs should be at 5V and RGB at 1.8V?

George Ioakimedes

unread,
Jan 24, 2013, 1:41:07 PM1/24/13
to cubie...@googlegroups.com
Looks like we're a little more than that and that is coming straight from the A10 if memory serves me right.

George Ioakimedes

unread,
Jan 24, 2013, 3:23:02 PM1/24/13
to cubie...@googlegroups.com
The level is actually at 1.4V so double what it's supposed to be. I wonder if the levels need to be reduced and that's the problem? There wasn't any on the reference schematic and it would be a bit odd if the A10 has dedicated RGB pins yet puts out the wrong level, or on 2nd hand that could just be yet another undocumented item we're uncovering...

Floris Bos

unread,
Jan 24, 2013, 4:27:03 PM1/24/13
to cubie...@googlegroups.com
On 01/24/2013 09:23 PM, George Ioakimedes wrote:
The level is actually at 1.4V so double what it's supposed to be. I wonder if the levels need to be reduced and that's the problem? There wasn't any on the reference schematic and it would be a bit odd if the A10 has dedicated RGB pins yet puts out the wrong level, or on 2nd hand that could just be yet another undocumented item we're uncovering... -- 

If you don't mind questionable Chinese sites.
Google terms "A10 development board schematics" gives a reference schematic.

http://www.docin.com/p-478600010.html


Not sure how actual the schematic is.
One thing I did notice is that in the "power tree" it shows DCDC2 voltage being 1.25v with that schematic, while it is configued as 1.4v in all .fex files, which happens to be the voltage you are seeing.
Maybe voltage was raised later, and you now need to add a resistor or something to bring it down to an acceptable level for VGA.



Yours sincerely,

Floris Bos

Lawrence Sheed

unread,
Jan 24, 2013, 8:21:58 PM1/24/13
to cubie...@googlegroups.com
Or if you don't mind datasheets previously downloaded from not so questionable Chinese sites (eg Baidu) -

http://dl.cubieforums.com/pdf/


Lawrence Sheed

COMPUTER SOLUTIONS
Room 401-402 Han Wen Xuan Building 2
No.14, 955 Yan An Middle Road
Shanghai China 200040

Tel: +86 400 820 0919
Fax: +86 21 62890700
Mobile: +86 13901 802 269
Skype: computersolutions.cn
Web: www.computersolutions.cn
E-mail: sup...@computersolutions.cn

George Ioakimedes

unread,
Jan 24, 2013, 9:35:04 PM1/24/13
to cubie...@googlegroups.com
Thanks for finding that EVM schematic, I see that the AXP209 DCDC2 is set at 1.25V (on page 7) but SY8021L DCDC1 is set at 3.3V and is labeled that it used for TVOUT.

I don't see how those would impact the voltage levels coming out of the A10 though. In the EVM schematic they also bring RGB straight out to the HD connector so perhaps there's some register setting that changes the voltage level of the RGB signals.

Probably have to go back to the A10 datasheet and see how those pins get configured, oh yeah, we don't have access to that! What I do see is that those pins are defined as TVOUT1, TVOUT2, TVOUT3. If they're only used for VGA-RGB and that is defined at 0.7V levels then why is 1.4V coming out?

George Ioakimedes

unread,
Jan 24, 2013, 9:39:09 PM1/24/13
to cubie...@googlegroups.com
Holy crap Lawrence, THANKS!!! Looks like you've collected a ton of useful information but I was planning on killing some brain cells tonight :)

Lawrence Sheed

unread,
Jan 24, 2013, 11:10:11 PM1/24/13
to cubie...@googlegroups.com
I've emailed George off-list with what should be sufficient info for him to get VGA out working on his baseboard.



Lawrence Sheed

COMPUTER SOLUTIONS
Room 401-402 Han Wen Xuan Building 2
No.14, 955 Yan An Middle Road 
Shanghai China 200040


Tel:
       +86 400 820 0919
Fax:      +86 21 62890700
Mobile:  +86 13901 802 269
Skype:  computersolutions.cn

--
You received this message because you are subscribed to the Google Groups "Cubieboard" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cubieboard+...@googlegroups.com.

georg...@gmail.com

unread,
Jan 25, 2013, 12:34:01 AM1/25/13
to cubie...@googlegroups.com

I do have a ton of new information to digest but what still isn’t making sense is that the TVOUT circuit in the A10 is referenced to 3.3V and all schematics I have seen only have the 75 ohm pull-downs on the RGB signals so if I’m measuring 1.4V there must be something I/we are missing?

Lawrence Sheed

unread,
Jan 25, 2013, 12:44:17 AM1/25/13
to cubie...@googlegroups.com
TV out isn't via the VGA RGB, there is a separate CVBS out for that.


Lawrence Sheed

COMPUTER SOLUTIONS
Room 401-402 Han Wen Xuan Building 2
No.14, 955 Yan An Middle Road 
Shanghai China 200040


Tel:
       +86 400 820 0919
Fax:      +86 21 62890700
Mobile:  +86 13901 802 269
Skype:  computersolutions.cn

On 25 Jan 2013, at 1:34 PM, georg...@gmail.com wrote:


I do have a ton of new information to digest but what still isn’t making sense is that the TVOUT circuit in the A10 is referenced to 3.3V and all schematics I have seen only have the 75 ohm pull-downs on the RGB signals so if I’m measuring 1.4V there must be something I/we are missing?

-- 
You received this message because you are subscribed to the Google Groups "Cubieboard" group.
To unsubscribe from this group, send email to cubieboard+...@googlegroups.com.

georg...@gmail.com

unread,
Jan 25, 2013, 12:59:46 AM1/25/13
to cubie...@googlegroups.com
The actual pin names on the A10 are called TVOUT0, 1, 2 which are then directly connected to VGA-RGB. TVOUT has a reference voltage pin which is connected to 3.3V (the pin is even referenced with TVOUT_3.3V_VCC (something like this name).
 
Sent from Windows Mail
 
To unsubscribe from this group and stop receiving emails from it, send an email to cubieboard+...@googlegroups.com.

George Ioakimedes

unread,
Jan 25, 2013, 7:54:00 PM1/25/13
to cubie...@googlegroups.com
OK guys, I need you help please. I'm not getting a signal on the VGA and I'm thinking it must be something to do with the kernel not set properly. I started thinking about the 1.4V levels on the RGB signals and decided that the level was double because the signal was being single-ended terminated by my 75 ohm pull-downs. Ideally the Cubie would also have these 75 ohm pull-downs at the source creating a 75 ohm impedance signal.

Anyway, I changed the pull-down resistors and am now seeing 0.7V signal levels. I haven't checked the actual timing but I'm thinking that shouldn't be a problem.

I've tried to mess with different kernel configs and did not have much luck as I don't have a lot of experience doing it. I thought that I read that the factory was going to be testing all the functions of the board so can't we get some details on that?

I was hoping for a productive day but it's not looking like that unfortunately...

Simos Xenitellis

unread,
Jan 25, 2013, 8:46:59 PM1/25/13
to cubie...@googlegroups.com
On Linux can you run

xrandr

and post the output here?
It should show what it recognises in terms of display connectors,
resolutions, etc. You can then change parameters.

Simos

George Ioakimedes

unread,
Jan 25, 2013, 8:57:12 PM1/25/13
to cubie...@googlegroups.com
OK, I need to fix my SD Card but I'll do that shortly and post the results. I forgot all about xrandr!

Bastiaan van den Berg

unread,
Jan 25, 2013, 9:24:31 PM1/25/13
to cubie...@googlegroups.com
On Sat, Jan 26, 2013 at 2:57 AM, George Ioakimedes <georg...@gmail.com> wrote:
OK, I need to fix my SD Card but I'll do that shortly and post the results. I forgot all about xrandr!

Sad to say, current drivers lack in xrandr extension. This won't get you going.

First of all, you need to enable VGA in script.fex. Check out https://github.com/cnxsoft/a10-config/blob/master/script.fex/mele-a1000-vga.fex for an example of a different device with VGA.

Then, you might still need to make kernel adjustments, but I think by default it follows what you set in script.fex...

--
buZz

George Ioakimedes

unread,
Jan 25, 2013, 9:26:47 PM1/25/13
to cubie...@googlegroups.com
I've already tried just about every FEX setting I can think of so I'm beginning to think/hope that there must be something in the kernel.

Floris Bos

unread,
Jan 25, 2013, 9:47:12 PM1/25/13
to cubie...@googlegroups.com, linux...@googlegroups.com
On 01/26/2013 01:54 AM, George Ioakimedes wrote:
OK guys, I need you help please. I'm not getting a signal on the VGA and I'm thinking it must be something to do with the kernel not set properly. I started thinking about the 1.4V levels on the RGB signals and decided that the level was double because the signal was being single-ended terminated by my 75 ohm pull-downs. Ideally the Cubie would also have these 75 ohm pull-downs at the source creating a 75 ohm impedance signal.

Hmm, there seems to be a register of interest.


And a quick look in the code of the TVE_set_vga_mode() method shows that a previous kernel branch had a different setting for it than now:

https://github.com/linux-sunxi/linux-sunxi/blob/sunxi-2.6.36/drivers/video/sun4i/disp/de_bsp/de/ebios/de_tvec.c#L334

==
TVE_WUINT32(sel,TVE_008, 0x403e1ac7);
==

https://github.com/linux-sunxi/linux-sunxi/blob/sunxi-3.0/drivers/video/sunxi/disp/de_tvec.c#L347

==
TVE_WUINT32(sel, TVE_008, 0x403f1ac7);
==


Register Name: TVE_008_REG
[...]
Bits 17:16 R/W 11 Internal_DAC_Mode_Sel 
- 0: 150ohms terminal mode
- 2: 75 ohms terminal mode
- 3: 37.5 ohms terminal mode


Does anyone know why it was changed from 75 to 37.5 ohm?
If not, you might want to change it back to 0x403e1ac7 to see if that works better.

BTW is there anyone with a Mele that can confirm that VGA output still works properly with recent kernel versions?

George Ioakimedes

unread,
Jan 25, 2013, 9:56:48 PM1/25/13
to cubie...@googlegroups.com, linux...@googlegroups.com
Floris:

Thanks for finding that. This confirms my suspicion of the terminating impedance being part of the problem. On one of my boards I have the standard 75 ohm pull-down and today I changed one board by adding another 75 ohm resistor on top of each of the RGB 75 ohm resistors giving  37.5 ohms.

I just tried the A10_Display Utility and using that utility I am able to get a very scrambled picture (on both boards). I haven't looked at the source of that code to see if it does anything different than editing the FEX file but I have not seen this yet so at least it is a little progress!

The screen is "rolling" so I think there is also some timing issue.

georg...@gmail.com

unread,
Jan 28, 2013, 11:35:11 AM1/28/13
to linux...@googlegroups.com, Floris Bos, cubie...@googlegroups.com
Hi Hans:

Would you mind letting me know what your FEX file looks like and what value resistors are on the RGB signals?

Best Regards,
George

killwarez

unread,
Jan 30, 2013, 2:52:58 PM1/30/13
to cubie...@googlegroups.com, linux...@googlegroups.com, Floris Bos
I got Mele A1000 that can be flashed with anything you want. So if you guys need to check something on another hardware just let me know.

George Ioakimedes

unread,
Jan 30, 2013, 3:14:33 PM1/30/13
to cubie...@googlegroups.com
jus need to know working FEX file for VGA output and also what termination resistors are on the RGB signals

On Wed, Jan 30, 2013 at 11:52 AM, killwarez <cat...@gmail.com> wrote:
I got Mele A1000 that can be flashed with anything you want. So if you guys need to check something on another hardware just let me know.

--
You received this message because you are subscribed to the Google Groups "Cubieboard" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cubieboard+...@googlegroups.com.
Visit this group at http://groups.google.com/group/cubieboard?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

Lawrence Sheed

unread,
Jan 30, 2013, 3:17:45 PM1/30/13
to cubie...@googlegroups.com
from the bits people emailed you before on list re patches , can be set in software so suggest add the 75ohm then set single terminator + appropriate resolution.


Lawrence Sheed

COMPUTER SOLUTIONS
Room 401-402 Han Wen Xuan Building 2
No.14, 955 Yan An Middle Road 
Shanghai China 200040


Tel:
       +86 400 820 0919
Fax:      +86 21 62890700
Mobile:  +86 13901 802 269
Skype:  computersolutions.cn

George Ioakimedes

unread,
Jan 30, 2013, 4:49:16 PM1/30/13
to cubie...@googlegroups.com

I just feel better putting out a board that I confirmed worked but I just don't have the time right now for VGA. Need to start shipping the baseboard too, I see others are already starting to design them for some reason.

killwarez

unread,
Jan 31, 2013, 3:38:45 AM1/31/13
to cubie...@googlegroups.com
Checked on Mele A1000



George Ioakimedes

unread,
Jan 31, 2013, 10:28:59 AM1/31/13
to cubie...@googlegroups.com
OK good, that's what I have too. Sure they don't have another set closer to the processor?

On Thu, Jan 31, 2013 at 12:38 AM, killwarez <cat...@gmail.com> wrote:
Checked on Mele A1000

killwarez

unread,
Jan 31, 2013, 11:15:18 AM1/31/13
to cubie...@googlegroups.com
Near to CPU there are a lot of routing and test point, you easily google for Mele's photos or I can make detailed one.

George Ioakimedes

unread,
Jan 31, 2013, 11:26:20 AM1/31/13
to cubie...@googlegroups.com
is there a schematic available for the board you sent a picture of?

killwarez

unread,
Feb 1, 2013, 12:57:03 PM2/1/13
to cubie...@googlegroups.com
Mele didnt published A1000's schematic. You can get A10 dev board manual and look into. On 34th page there is zoomed photo of VGA output. I see only resistors are there.

killwarez

unread,
Feb 1, 2013, 12:59:00 PM2/1/13
to cubie...@googlegroups.com

George Ioakimedes

unread,
Feb 1, 2013, 3:50:12 PM2/1/13
to cubie...@googlegroups.com
yes, that does appear to be straight from the A10 to the HD15 connector. What's a little odd is that the reference schematic we found had HSYNC/VSYNC going through a "buffer". I haven't tried to bypass the buffer but I can try that once I solve Mali drivers not loading...

On Fri, Feb 1, 2013 at 9:59 AM, killwarez <cat...@gmail.com> wrote:

Floris Bos

unread,
Feb 1, 2013, 8:24:16 PM2/1/13
to cubie...@googlegroups.com
On 02/01/2013 09:50 PM, George Ioakimedes wrote:
> yes, that does appear to be straight from the A10 to the HD15
> connector. What's a little odd is that the reference schematic we
> found had HSYNC/VSYNC going through a "buffer".

Think that schematic does belong to that board.
You can see a chip labeled U10 on the back picture, right bottom of the
center.


Yours sincerly,

Floris Bos

Message has been deleted

George Ioakimedes

unread,
Feb 15, 2013, 10:26:10 AM2/15/13
to cubie...@googlegroups.com
Shashank:

Good point about the speed of the buffer chips used, I'm using SN74ACT08D only because that is what was in the reference design and I still haven't had the time to debug this issue but maybe soon. 

I believe all of the initial testing was done with lower resolutions but when I get back to testing I can certainly try different resolutions and check the results.

-George

On Thu, Feb 14, 2013 at 11:16 PM, शशांक परब <shasha...@gmail.com> wrote:
HI, 

Do you use buffers for Hsync and Vsync? What is the buffer chip you use? and whats screen resolution is set? I guess 640X480 resolution clocks the pixel information at 25Mhz. and for something like 1024X768 it is far more than that, causing the buffer IC to drop some pulses and to cause the issue you mentioned.

You will have to set the resolution to minimum and see. 

- Shashank

--
You received this message because you are subscribed to the Google Groups "Cubieboard" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cubieboard+...@googlegroups.com.

George Ioakimedes

unread,
Feb 25, 2013, 8:46:13 PM2/25/13
to linux...@googlegroups.com, cubie...@googlegroups.com, Floris Bos, cat...@gmail.com
It was discovered that the Wiki had incorrectly defined the HSYNC/VSYNC pins on the Cubieboard. Using HSYNC/VSYNC from LCD0 (PD26/PD27) and not from LCD1(PH26/PH27) allowed VGA to work properly.

How is the A10 configured through the FEX file for VGA output? I have a bunch of boards built up with PH26/PH27 wired to the VGA circuit and rather than cutting traces and adding jumper wires it seems that you should be able to set the A10 to output HSYNC/VSYNC on PH26/27

Henrik Nordström

unread,
Feb 26, 2013, 3:05:13 AM2/26/13
to cubie...@googlegroups.com
mån 2013-02-25 klockan 17:46 -0800 skrev George Ioakimedes:
> It was discovered that the Wiki had incorrectly defined the
> HSYNC/VSYNC pins on the Cubieboard. Using HSYNC/VSYNC from LCD0
> (PD26/PD27) and not from LCD1(PH26/PH27) allowed VGA to work properly.

Sorry about that.




> How is the A10 configured through the FEX file for VGA output? I have
> a bunch of boards built up with PH26/PH27 wired to the VGA circuit and
> rather than cutting traces and adding jumper wires it seems that you
> should be able to set the A10 to output HSYNC/VSYNC on PH26/27

No.

PH26 has the following modes: GPIO, LCD1_HSYNC, ATAIOR, KP_OUT6,
SDC1_D2, CSI1_HSYNC

Regards
Henrik


Michal Suchanek

unread,
Feb 26, 2013, 5:32:07 AM2/26/13
to cubie...@googlegroups.com
On 26 February 2013 09:05, Henrik Nordström <hen...@henriknordstrom.net> wrote:
> mån 2013-02-25 klockan 17:46 -0800 skrev George Ioakimedes:
>> It was discovered that the Wiki had incorrectly defined the
>> HSYNC/VSYNC pins on the Cubieboard. Using HSYNC/VSYNC from LCD0
>> (PD26/PD27) and not from LCD1(PH26/PH27) allowed VGA to work properly.
>
> Sorry about that.
>

So VGA is wired to LCD0 inside A10 and to use VGA+LCD you need to use
LCD0 for VGA and LCD1 for LCD?

What is CSI?

I am really confused by these fex parameters. We have LCD0, LCD1,
CSI0, CSI1, TVOUT, TVIN. Out of that it's supposedly possible to pull
out HDMI, LVDS, VGA, and TVOUT.

Thanks

Michal
It is loading more messages.
0 new messages