Got displays, looking for keymats


the rotation screen thing, for what is it necessary?

i have the impression it spends more energy and costs money that what it can makes as benefit.
As ED says theres no available displays which are landscape orientation, only portrait.

Now the question is why this makes such a big difference. I think it's because of how the LCD "scans" the display. Normally this is done pixel by pixel starting from the left edge to the right edge, then continuing on the next row bellow it.

This scheme is used to transfer the image data to the LCD. In order to make the transfer efficient and easy for the hardware, the image data is stored in the same way in memory. 

The problem is that Pyra is sort of misusing the portrait LCD by laying it on it's side. This means it scans from bottom up (or top down), column by column. This results in a mismatch between how image data is arranged in memory (because applications assume a left-to-right row by row scheme) and how the LCD expect data to be transferred.

The options to solve this are:

  • Rewrite all applications to their write image data the "correct" way
  • Use some chip/GPU to do the rotation
Android handsets use the GPU to do this (but then all applications would need to use the GPU). Pyra is using this special chip, which makes the portrait LCD appear to be a landscape LCD to the rest of the system (meaning applications requires zero adaptation).
 
Last edited by a moderator:
Since the Pyra has another layout, it should be no problem with additional shoulder buttons.

I can image two ways of realizing that. However, the buttons have to be thinner, but we could almost keep the rest of them. Their pivot shaft has to be longer, of course.

The switches:

a ) Simply place one on the top side and one on the bottom side of the mainboard.

b )Place them both on one small PCB and connect it with a wire to the mainboard. Both case parts have a notch where the PCB fits in. I have seen a similar design in some PlayStation controllers.
 
Last edited by a moderator:
There was already a discussion about the LCD, but can't a cheap FPGA do screen rotation?

If so you could later add a higher resolution LCD if the FPGA supports it.
 
There was already a discussion about the LCD, but can't a cheap FPGA do screen rotation?

If so you could later add a higher resolution LCD if the FPGA supports it.
Are there really cheap FPGAs? As far as I know they tend to be comparably expensive.If it was actually possible (I guess it is, I just don't know about the "cheap" part) and acceptably cheap, then an FPGA which could also be used for other stuff would be extremely nice.
 
There was already a discussion about the LCD, but can't a cheap FPGA do screen rotation?

If so you could later add a higher resolution LCD if the FPGA supports it.
No.We need a SRAM frame buffer of approx. 1-2 MByte with sufficient bandwidth and two MIPI compatible interfaces on the FPGA. I think no such FPGA exists (AFAIR LVDS would be available). And even if it does, it is not cheap.
 
Last edited by a moderator:
FPGA vs. a dedicated solution, the FPGA will always lose. The flexibility is an FPGA inherently makes it use more power and be more expensive. An FPGA is only really to be considered when the flexibility is really needed, or that a dedicated solution is not available.
 
There was already a discussion about the LCD, but can't a cheap FPGA do screen rotation?

If so you could later add a higher resolution LCD if the FPGA supports it.
No.We need a SRAM frame buffer of approx. 1-2 MByte with sufficient bandwidth and two MIPI compatible interfaces on the FPGA. I think no such FPGA exists (AFAIR LVDS would be available). And even if it does, it is not cheap.
Couldn't you probably do the MIPI interface in software on most FPGAs? I would have expected the RAM to be a bigger problem with the FPGA possibly requiring an additional RAM chip.But when it's expensive it doesn't matter anyway.

(if it was more expensive than the dedicated rotation chip but not by too much and could get be used for other interesting stuff I'd be interested to pay a little more for it)
 
Was just checking. Cheap FPGA's (Spartan 3A) for $4,50 would only do 640Mbps, 1080p60 would require more bandwidth (1920X1080X60X8X3 = 2985984000, +/-3Gbps). And bigger ones are more expensive and cost space/money.

Shame the scaler doesn't do 1080p.
 
Did you think of the USB and other connectors?


Horizontal Shoulder-Buttons might need a lot more space and maybe less connectors are possible then.


Would you really want to sacrifice one USB port for your Shoulder button mapping?
No need to sacrifice anything of course, this would be bad indeed. Like the Neo Geo X example, the 2. pair of buttons is smaller and so could be the main trigger buttons at the sides compared to the ones from the Pandora. :)


Again, if your vertical design works at the end, I'm fine with that but if it is choosen but doesn't work well in practise then nobody can complain that horizontal design was not discussed. Just to be on the safe side. :D


Here, the outer triggers would be pressed more with the middle part of your index finger, I already do this with the Pandora triggers, while the inner triggers are best pressed with the tip of your index finger. ^^

post-1-0-61260700-139k0uvt.jpg
 
Did you think of the USB and other connectors?


Horizontal Shoulder-Buttons might need a lot more space and maybe less connectors are possible then.


Would you really want to sacrifice one USB port for your Shoulder button mapping?
No need to sacrifice anything of course, this would be bad indeed. Like the Neo Geo X example, the 2. pair of buttons is smaller and so could be the main trigger buttons at the sides compared to the ones from the Pandora. :)


Again, if your vertical design works at the end, I'm fine with that but if it is choosen but doesn't work well in practise then nobody can complain that horizontal design was not discussed. Just to be on the safe side. :D


Here, the outer triggers would be pressed more with the middle part of your index finger, I already do this with the Pandora triggers, while the inner triggers are best pressed with the tip of your index finger. ^^

post-1-0-61260700-139k0uvt.jpg
What if the shoulders remain stacked and only R2 and L2 are shortened lengthwise to the dimensions of the Gameboy SP triggers?
 
Last edited by a moderator:
The main PCB


Nikolaus was mostly working on the display PCBs, so the progress on the main PCB was slower, as you can imagine.


It's still nearly finished - but that 'nearly' are the last percent that always need a lot longer than the rest.
I don't want to derail this but as you said the PCB is nearly done and I feel it's the last chance to ask you to consider this. (Edit: Being able to split the two pads of the space key by desoldering a connection and soldering another one and having an unused keycode for the other pad in the driver.)
No need to worry. There has always been a 0R resistor (0402 dimension) for exactly this purpose :) It can either connect both meanders to a single matrix position or add the meanders to different matrix positions.
Sometimes we can identify through our crystal ball what the community would like to have :)
This is good to hear, but can we also get an alternate case and keymat that have the space key split in two?
 
Last edited by a moderator:
The main PCB


Nikolaus was mostly working on the display PCBs, so the progress on the main PCB was slower, as you can imagine.


It's still nearly finished - but that 'nearly' are the last percent that always need a lot longer than the rest.
I don't want to derail this but as you said the PCB is nearly done and I feel it's the last chance to ask you to consider this. (Edit: Being able to split the two pads of the space key by desoldering a connection and soldering another one and having an unused keycode for the other pad in the driver.)
No need to worry. There has always been a 0R resistor (0402 dimension) for exactly this purpose :) It can either connect both meanders to a single matrix position or add the meanders to different matrix positions.
Sometimes we can identify through our crystal ball what the community would like to have :)
This is good to hear, but can we also get an alternate case and keymat that have the space key split in two?
You can print one on your own.
 
Last edited by a moderator:
Curious to see the results of the testing. I've always been interested in electronics, I've learned a bit from this forum... can't wait to see progress.

Sent from my Z665C using Tapatalk
 
Also for the triggers can't you just have a flat little PCB with two switches on them that the shoulders attach too. Just have it be a plug in unit. You could just slide it into a slot on the case. I had a USB gamepad that used the same system for its shoulders and it worked brilliantly.. however that gamepad had tough as nails plastic. So if the plastic is brittle you have a risk of shoving the shoulder mechanism into the case itself! :unsure: Granted if that's the case for the cast we'd sure have more pressing issues!
 
Damn that A80 is beefy as hell, would love to see that as a high end model.  :D

As for the shoulder buttons having a micro pcb for them and either use a flat cable or solder leads would be the best option.

Obviously that will jump manufactoring costs though but most portables are using that design...
 
I think it's likely that by the time ED is ready for a new SoC module there will be a new "best" chip and it would be smarter to time the SoC module upgrade to the SoC release, to be as cutting edge as possible (it wouldn't be smart to delay the entire device for that, but when there is an acceptably powerful device [Pyra] available and selling, being cutting edge will get you sales from an additional target group).
 
The SoC

We're still using the OMAP5, since we don't have much more info on the A80 yet.

AllWinner will not offer Linux drivers, but I plan to get in touch with the Cubieboard designers to ask how they will solve their driver issue - but still, switching to the A80 would delay everything for at least 6 months... not an option for us right now, but maybe sometime in the future.
Hopefully near future. I'll wait for that or anything better at that time.
 
I think it's likely that by the time ED is ready for a new SoC module there will be a new "best" chip and it would be smarter to time the SoC module upgrade to the SoC release, to be as cutting edge as possible (it wouldn't be smart to delay the entire device for that, but when there is an acceptably powerful device [Pyra] available and selling, being cutting edge will get you sales from an additional target group).
Oh no doubt, I would hope to see more SoC experimentations once Pyra is sucessful. As of right now no one as seen the PowerVR G6230 in use but comparing with the OMAP5's SGX544, there is a massive GFLOPS bump that i would like to get my hands on.
 
Back
Top