Odroid Go Advance [NEW?] Revision

[Etx]

Greetings.
Looking at the evolution of OGA, i think it'll be a good idea to create a thread with smth. like roadmap with list of articles that may be improved or changed with minimal steps needed to the current architecture. So, we're not going to list here anything like "replace a CPU with Snapdragon 865" or "add a 7-inch screen" etc. Just a minor changes like OGA v1.1 have as a evolution from v1.0 after collecting feedbacks from customers.

1. It's time to add a HDMI port to OGA. It can be done by changing a CPU to PX30 instead of RK3326. It is identical CPU except it have two VOP's. Adding a RK618 MUX IC ($1 cost ) to the circuitry will result in cheap and easy HDMI output implementation.

2. Since we might have a HDMI, it is also a good idea to add bluetooth module, not just wi-fi as v1.1 have, to be able to add external controllers through USB and BT for multiplaying.

3. Something really SHOULD be done about all that housing and LCD stuff. Currently we have a massively perforated housing with a weird combination of non-dustproof LCD mounting method. Plus many cutomers keep cracking their LCD's while mounting them. So i have a few sub-articles about that:

a) Plastic screen protector is a weird choice for a portable device in a year 2020. Change it to a glass-made or make a one-piece LCD+glass laminated dustproof module, with adhesive tape along the edge of the glass. So the only step customers need to do is unstick a liner and put whole assembly in frame. Without all that twisting, bending, cutting etc. stuff. As a bonus we will get better view angles and image contrast.

b) Remove all those unnecessary holes as well as expansion slots. Seriously, who need those holes for UART port as well as 10-pin port at the top in a GAMING device? I see 0 projects that actually needs it, and i really don't know why i should accept those nano L2/R2 buttons with yet another hole between R2 and USB port. Okay, there always will be some users that will want to add a humidity sensor and play with it for 1-2 days, posting their success story to the forums - just leave a pre-embossed parts of the case that could be easily removed for wires in case they will be needed. Leave those test pads with standart layout for easy connector soldering and let them do their game. Change UART and 10-pin port connectors to a smth like FPC/FFC connector INSIDE the housing, and let them wire it throug the case if they'll need it.

c) I know that nobody likes a mention of "RK..." device here, but they really did a good job with a housing and L/R/L2/R2 buttons... Well, i have to admit ABXY and D-pad are nice too. I personally like their recessed analog stick a lot, since i have RG350M and found them very handy, easy to put in your pocket because of such recessed design. Instead of just screaming they stole everything from you (which, of course, is pure truth) - take from them what turned out well - housing and buttons.

4. "Demand was much higher than expected", yet again. Can anybody explain me the reasons of why a current production scheme do exist? I'm talking about that China -> Korea moment, where odroid orders battery, pcb and lcd? in Korea, while waiting for housing, screws, buttons, membranes, speaker, joystick and... everything else from China. After combining all parts together, it is being shipped from Korea to USA, Europe etc. I see no reasons for that. Time losses, logistics losses... Guess it was designed so to prevent replicas from China? Well, looks like that didn't work. Maybe it's time to optimize those production lines and start shipping those kits from China entirely ?

[npaladin2000]

b) Remove all those unnecessary holes as well as expansion slots. Seriously, who need those holes for UART port as well as 10-pin port at the top in a GAMING device?

Those holes aren't unnecessary. As you pointed out, one of them is for the UART port, which developers need for access and debugging (to get a serial console session while trying to get the screen working for example). The rest are for promoting airflow for cooling. You mentioned a certain other device, which I won't, except to say that their elimination of airflow holes has actually produced situations where it will overheat.

The 10 pin expansion port, I'm not sure I see the point of. It's traditional on the OdroidGo devices so people can create expansion devices for it, but the bandwidth is fairly low on it, so I'm not sure what it might be good for. I know someone has been working on an L2/R2 add-on board that connects to it, but that's only valid for the v1 devices. There isn't enough bandwidth in the connector for it to act as a TV-Out, for example, not to mention there's also the USB-A interface port present. I guess we'll have to see on that one.

[Etx]

Those holes aren't unnecessary. As you pointed out, one of them is for the UART...

Yeah, i know that UART is made for devs. But we currently have, let's round up, 1% of customers that are dev's and really using this port... and 99% of gaming customers asking what is this hole for and why it isn't covered. All in all, since all the devs have to solder a connector there, thus device need some extra work with UART, isn't it a good idea to make some easily-removable part of the case for devs and leave a solid housing for 99% of other customers? You're dev, you need this port - remove a fragment, solder those 4 UART-USB pins and continue your development.

Yet one word about all this ventilation through theese holes. I saw 2+ hours of continuous gameplay on RK* device at YouTube and found none overheating problems, thus i really don't believe it does exist on both consoles. Also you have to admit that tiny RAM copper heatsink on top of the CPU will dispass a heat significantly better than all those holes in the housing. Thus i vote on solid and handy case plus a note in manual that you can add a tiny copper heatsink in case you're playing at a direct sunlight and 40C outside - you might expect some overheating problems and it'll be your solution, rather than a perforated pocket sand collector.

[crashoverride]

1. It's time to add a HDMI port to OGA. It can be done by changing a CPU to PX30 instead of RK3326. It is identical CPU except it have two VOP's. Adding a RK618 MUX IC ($1 cost ) to the circuitry will result in cheap and easy HDMI output implementation.

There are additional design considerations. Using HDMI also means an increase in memory usage. An increase in memory usage also means an increase in memory bandwidth. More pixel also puts additional pressure on the Mali GPU.

2. Since we might have a HDMI, it is also a good idea to add bluetooth module, not just wi-fi as v1.1 have, to be able to add external controllers through USB and BT for multiplaying.

I would like to see BT added. However, one constraint is that BT uses the same 2.4Ghz spectrum that wifi does.

b) Remove all those unnecessary holes as well as expansion slots. Seriously, who need those holes for UART port as well as 10-pin port at the top in a GAMING device?

As a developer, all my devices require access to the UART port. This is critical since there is no other way to debug the device (no Ethernet). All ODROID boards target developers.

4. "Demand was much higher than expected", yet again. Can anybody explain me the reasons of why a current production scheme do exist?

COVID-19

[Etx]

There are additional design considerations. Using HDMI also means an increase in memory usage. An increase in memory usage also means an increase in memory bandwidth. More pixel also puts additional pressure on the Mali GPU.

I wonder, how all those Pi-2* based consoles with 512Mb DDR1; RG-350 with it's relic MIPS CPU and 512Mb DDR2; Retroid Pocket with 512Mb DDR2 etc. are able to output HDMI signal, while OGA with Cortex-A35 and 1024Mb DDR3 is not?

I would like to see BT added. However, one constraint is that BT uses the same 2.4Ghz spectrum that wifi does.

Your smartphone having any issues using Wi-Fi and BT at the same time? I don't think so...

COVID-19

Weeeeell, i don't think so as well. It was a problem with a 1st batch, but not with the second one. RK* was released at the same time when we had a second batch in stock.
That doesn't explain the necessity of China -> Korea -> USA/Europe logistics as well.

[ThatOneSeong]

I wonder, how all those Pi-3* based consoles with 512Mb DDR1; RG-350 with it's relic MIPS CPU and 512Mb DDR2; Retroid Pocket with 512Mb DDR2 etc. are able to output HDMI signal, while OGA with Cortex-A35 and 1024Mb DDR3 is not?

The Pi-based consoles are barely in situations where they use the GPU; and when they do, it's an extremely low resolution. (Also, wrong specs, as Pi-3's and onward are 1GB LPDDR3 at minimum) Same with the Retroid, and it doesn't look very pretty blowing up 480p onto an HDTV.
On Dingux, HDMI is extremely buggy and barely functional on half-baked firmware. It's also only 240p/480p output depending on model. Source; am from the Dingux community.

If you're fine with sub-320p output on TVs, then sure - but I'm assuming you were expecting 1080p output, which really wouldn't be practical for the reasons Crash outlined already.

Your smartphone having any issues using Wi-Fi and BT at the same time? I don't think so...

Your smartphone also runs off a sub-$5 chip which throttles at ~5mb/s? Also, your phone costs, at minimum, twice the cost of the OGA in parts (and no, we're not talking used parts - the OGA is a new device, so we're using new, on-the-market-now prices and not outdated tech being liquidated); not an apt comparison...

Tbh, a lot of the suggestions being brought up would either require a significant reworking of the hardware (which definitely wouldn't be in the same scope from the OGA to the OGA BE, as you claimed at the top), or significantly raise the price. I'd love to see some changes like bigger secondary shoulders (even though the ones we have rn are just fine) and moving the IO a bit closer to one-another, but you have to realize that most, if not all these suggested changes (with the exception of BT on-chip, which probably won't happen for power and/or price reasons) are much better suited to a sequel device which can accomodate such a design overhaul.

Also keep in mind that the device is made for the SBC and DIY community just as much as it's tailored to the gaming market; you can't just exclusively single out a chunk of the userbase like that, and what you think the general audience is doesn't really apply. It comes off as pretty selfish, ngl. :/

P.S. Voldemort's Device (as it shall not be named) has been reported to overheat in other circles - certainly significantly more than any such reports from the Odroid community. You can blame that on the chip being nearly-directly pinned against the screen.

[Etx]

The Pi-based consoles are barely in situations where they use the GPU...

To be true, i don't understand, what are you talking about here at all. Go on Youtube and look at those Pi boards playing games that they are able to handle perfectly on RetroPie builds through HDMI. Nobody is talking about an entire available ROM collection to be perfectly smooth running on the OGA. Let it handle those that could be playable! All NES roms f.e. According to your explanation HDMI is useless unless it can play Doom Eternal at 4k@60fps... What you'll say about Retroid Pocket - it have HDMI out and it is working good. Any comments to it's developers with why should they disable HDMI?

Your smartphone also runs off a sub-$5 chip which throttles at ~5mb/s?

Yeah, my tablet runs off sub-$5 chip RTL8723AS, BT+Wi-Fi combo module, and it have no throttle problems whatsoever. If a currently used by OGA:BE module having those issues, that means poor module was chosen or bad layout made, interferencing with data lanes. Thus a better module should be chosen or rewired according to all manufacturer datasheet recommendations. Is there any obstacles to not to use RTL8822CS 802.11ac Wi-Fi/BT module in the next revision? I found them working well in many RK-based TV boxes.

Tbh, a lot of the suggestions being brought up would either require a significant reworking of the hardware...

Well, if we're talking about HDMI output - i agree, some comprehensive layout changing should be done, but everything else is a minor changes.

It comes off as pretty selfish, ngl. :/

What are we talking here about yet again? Nobody in the DIY community wants a working HDMI port? Nobody wants a better quality housing and bigger buttons? Nobody wants dust-free LCD module without THESE backlight spots? This forum is full of threads with expansion modules that uses 10-pin port? Have you ever read subreddits with buyers "success" stories full of cracking screens, desoldered buttons etc.?

aaa.png (331.01 KiB) Viewed 225 times

You can blame that on the chip being nearly-directly pinned against the screen.

It's just a complete lie. They've stolen PCB layout almost completely, thus CPU is located at the same place as OGA have it - at the opposite side of the PCB and is not being pinned against the screen. The only difference is on RK* device CPU is covered by the battery, unlike the OGA, where the battery is located at the right part.

[crashoverride]

What you'll say about Retroid Pocket - it have HDMI out and it is working good. Any comments to it's developers with why should they disable HDMI?

HDMI connectors are not designed to provide the tolerances required in a portable gaming device. During game play, a handheld gaming device experiences a lot of motion and forces. This fatigues the connector and will eventually result in failure of the part. These connectors are designed to be used in a "docked" scenario. A more suitable alternative would be to "cast" the video/audio wirelessly.

Is there any obstacles to not to use RTL8822BS 802.11ac Wi-Fi/BT module in the next revision?

Regulatory certification costs and antenna design.

P.S. Voldemort's Device (as it shall not be named)

This literally made me laugh out loud. Thanks.

I should also state that GO-Advance BE is proof that HardKernel listens to community feedback.

[Etx]

HDMI connectors are not designed to provide the tolerances required in a portable gaming device...

That's why a micro HDMI port was created. In case of "a lot of motion and forces" it will simply fall out and nothing will be breaked.
All that wireless/casting will need a massive amount of work to be done to make it work well - not a variant at all. Besides, current wi-fi perfomance is very bad also, right?
I have a gcast 2nd gen and it have many issues sometimes with forcing it to work normally.

Regulatory certification costs and antenna design.

I know nothing about certification, but antennas are being made 100% based on Realtek datasheets, can be a part of the PCB or you can supply it easily as a foil sticker with U.FL connector, very cheap.
If all that certification costs so high that RTL8822CS 802.11AC/BT module is not an option for the future revision, perhaps a PCB could be pre-made to solder this module by the buyers itself? Something like it was done with SDIO port in rev.1.0. PCB will be shipped with a poor and cheap certified module, but there will be a layout beneath it to unsolder it and replace by 802.11AC device.
Why should we accept poor Wi-Fi and lack of BT because of the current module was cheap in certification?
On the one hand we're talking about OGA as a DIY device, for DIY community and skilled buyers. And on the other - it should be as simple as piece of wood and nothing should be done by the customers except nuts tightening. Let those 2 variants co-exist. Skilled owners may do some DIY work to gain 802.11AC perfomance, customers without appropriate skills will use a device in it's basic condition.

[rooted]

A more suitable alternative would be to "cast" the video/audio wirelessly.

I've been hoping somehow we may see this in the future.

[mad_ady]

Why should we accept poor Wi-Fi and lack of BT because of the current module was cheap in certification?

See, this is exactly the reason why the original goa didn't have wifi. You could add it via usb where you have control over its type/performance...
They went with a modest wifi chip because:
* price
* sdio was available with no extra cost
* for convenience to the end user, for rom transfers.

But it was not intended for high throuhput/low latency streaming or anything...

So they knew from the start they can't please everyone. Also, not sure how many users need wifi after copying/managing their roms...

[Etx]

Well, since the console is running Linux, that was a bad idea to choose a USB adapter as a network source. Just a few adapters are actually supported, there are almost no variants to choose from. And yet you can't easily inject the drivers to the distro to make your specific adapter to work.
I still don't think all that licensing cost a lot, since all the changes made on OGA:BE cost $5.
But if there is so many expenses with RTL8822CS 802.11AC/BT module, a premade soldering layout looks like a great idea. One specific Wi-Fi/BT module will be supported (as well as those current 3? USB adapters) and will fully covers all the needs. Based on it's datasheet, it utilizes UART interface for the bluetooth, which we also have onboard. A simple solder jumper or SMD switch will be needed to switch between debug UART output and BT functionality.