ODROID W+

[mikemoy]

Common guys, when you going to release a replacement for the W? Been waiting like forever.....
To be honest I am quite shocked at how little there is in the way of something like these out there. I would figure people would really be wanting a module to plug into their own design and put all the connectors where they want them vs all these other modules out there where they are packed with connectors always in the wrong place of course.

[rooted]

It's a Broadcom issue, a conflict of interest or some such thing with the Pi Foundation. They didn't choose to quit making it.

I would love to see a replacement though.

[DanielBull]

They don't need to use the Broadcom chip though. Maybe a mini C1?

[mikemoy]

Yes, I have suggested they just take the C1 design and just start lopping off most of the hardware connectors. vs finding a new processor and basically start over. Why they are not doing something is a mystery to all of us.

[DW90]

I agree having a mini C1 would be awesome! I have been working on a embedded project that requires a small footprint, uses OpenCV, and the Raspberry camera, the the single core processor just doesn't cut it.

[mikemoy]

I agree having a mini C1 would be awesome! I have been working on a embedded project that requires a small footprint, uses OpenCV, and the Raspberry camera, the the single core processor just doesn't cut it.

I am in the same boat that i need a small footprint. It blows my mind that all these companies that make these things can seem to understand that there is large cround of people that just dont want a board with a ton of connectors on it,

[DanielBull]

Another vote for a small footprint board.
I love my W's.

[Respectech]

I also vote for such a board. I pitched the idea to HK a while back. They didn't reject the idea, so maybe something will happen soon!

[odroid]

Thank you for the valuable inputs. Time to share our idea.

We've started the C0 project yesterday.
It is based on the C1+ platform.
We removed the Ethernet port and USB 4port hub IC/connectors to reduce the PCB size.
We also removed the USB OTG port.

That's it? No.
We added a Li-Ion/Polymer battery charger circuit as well as a DCDC booster IC on the board.
So you can run the board with only a battery. Battery voltage level can be measured by ADC on the 40pin header.
A small slide switch is added to turn on/off the battery output power.

We also added 2 ports of USB host which are directly connected to the S805 CPU. It is the main reason why we removed the otg port.
2 port (stacked) host connector can be mounted.
You can also use a single port USB connector if you need a low-profile dimension system.

Below connectors will NOT be mounted for your DIY project probably.
USB host port
40pin header
IR receiver
USB-UART console connector
They will be sold as a connector pack separately at a few dollars.

But, other connectors(HDMI/DC-jack,uSD,eMMC, battery, RTC) will be mounted because they are SMD type.

Here is an engineering sample PCB layout.
eMMC and SD card slot are placed on the rear side exactly same as the C1+.

The PCB size is 65x56mm. It is bigger than our expectation.
But it is not easy to reduce the size due to the many connectors, heatsink and 2xDDR3 memory chips.
Good thing is that we can use many existing add-on boards for ODROID-C1/+ as well as RPi.

Click this image to enlarge it.

s_C0_BOARD_LAYOUT.png

(248.48 KiB) Downloaded 6728 times

How about calling this product ODROID-C0 (C_ZERO)?
The OS and software will be fully compatible with C1/C1+.
Feel free to add any comments. We need to hear your voice. It is really helpful.

[memeka]

maybe limit the CPU speed in the kernel, such that a heatsink will no longer be necessary, reducing cost and size.
people could "overclock" it to its normal speed and attach a heatsink themselves if they want to.

maybe also go back to micro-HDMI.

also, remove power jack and make it USB-only power (besides battery of course)

[Respectech]

I believe the AMLogic S805 already throttles down the CPU automatically when it gets hot, so it can run at full speed until the temperature climbs.

[stmicro]

Long waited item !
Finally I can make a powerful drone and many small sized IoT devices.
The formfactor looks like the raspberrypi A+ model.
It is reasonable for the compatibility of tons of accessories. Right?
I also want to have a smaller PCB. But when I seriously looked at the C1+ front & rear side, it seems to be almost impossible to reduce the size.
1GB RAM and 1.5Ghz Quad-core on 2.6 x 2.2 inch space is not a bad number if the height is less than 0.5inch.

Anyway, i really want to know when we can purchase it. how much is it?

[Respectech]

I think it is really exciting. I think it will be a popular option for many projects that need the smallest size.

[odroid]

Thank you for the positive feedback.

@memeka
Yes, you can remove the heatsink to reduce the weight and height if your application is sensitive to the profile/weight.
But it is really hard to reduce the PCB size.
Many application needs the 2 USB host probably(two is much better than one ), so we removed the otg port.
It gave us a room to place the slide switch for power contorl too.

@stmicro
Yes, your guess is right. They have the identical formfactor of PCB.
Target MSRP is around $30 or slightly lower.
The first engineering samples will be ready around middle of December.
If its functionality and stability is acceptable, we will prepare materials.
The lead time is about 8~10 weeks normally.
If everything goes well, we can start to sell it in the middle of Feb 2016.
But we can delay the schedule if other customers give us any good ideas.

@Respectech
I really hope this C0 will be suitable for your Robot project. The dancing/speaking robots at the ARM Techcon was really impressive.
This will be helpful to make a robot kit with marketable price range.

[DanielBull]

This is great news, I love the way you included the LiPo circuits. This was a critical feature IMHO and one that other SBC's don't have.

[mikemoy]

This sure is awesome news!!

Dont know if this is possible here in the forum, but if so you should create a voting pole for things and make this a sticky post. This is a serious thing to get right and not rush it.

However I think your forgetting to keep something in mind. "When you say But, other connectors(HDMI/DC-jack,uSD,eMMC, battery, RTC) will be mounted because they are SMD type." This is not keeping with the ODROID W concept which for the most part is connector less. I understand why you want to do it, kinda crosses the bridge between the W and the Cx series. But if its to be more for a total embedded plugin module like say a Intel Edison, there is no reason for them.

To me the main goal should be to allow it to plug into a main board in which we can still have access to all pins for the removed hardware connectors. Meaning that if you remove a USB / Ethernet connector, there should be 0.1" pins somewhere to solder pins in place for it to plug into our board so we can then put the hardware connectors where we need them to be to fit our design if we need to use them. We dont want to loose features just because we are removing connectors.

uSD,eMMC is a definite must have onboard. The battery & RTC is a great perk. but for me adding a header to these is even better, I have no need for extra wires and connectors if I am plugging it into my board. The HDMI/DC-jack I don't see the point, if its for embedded who is going to be connecting a monitor to it, and if HDMI does wind up on there, for sure switch it to the micro-HDMI.

Also, I saw no mention of the the RJ45 Ethernet jack. I have no problem if its not on there, but please give access to the pins for it. If you dont, it will seriously limit the board for other possibilities. Say I want to make a small device with a DC jack and Ethernet connector only. I dont know if routing is possible for this, but maybe adding another say x numbered pin 0.1" header above J2 for all the connections for RTC, Battery, USB / Ethernet, IR, UART and so on.

If being embedded is the focus then the UART connector should be soldered on, because we know we are always going to terminal in no matter
what. If its not to be soldered but added as an option that is fine, just as long as the pin spacing is 0.1" so we can solder pins into it so that it plugs into our board.

The next big issue I see is how are you dealing with J2? Is the header going to be soldered in like the Cx devices or left open like the ODROID W ?
If it is going to be a plugable module then the pins need to stick down, not up like the Cx's are. I opt for leaving them open so that we can solder in what we need.

[DanielBull]

Mike it sounds like you are describing a different product, something like a compute module?

[mikemoy]

Hi DanielBull,

Not necessarily, though would not mind it. I just would like it to more look like the ODROID W module in the fact that most of the connector signals are brought out to headers.

[DanielBull]

On the Odroid W the HDMI, uSD, eMMC, battery, RTC headers were all fitted and are very small so they shouldn't be a problem. I think it would really limit the market if those weren't fitted as many people can't do surface mount and almost all projects will need most of those connectors.

The ethernet port isn't available AFAIK, I personally don't think this will be a problem as most people will use WiFi because the C0 would be more for portable applications (hence the built in LiPo).

[mikemoy]

I hear ya. Not to cause an argument, but what I see the market needs is more modules like the W out there that can be placed onto boards of our own design where we put the connectors where we need them to be, and if we even want to use them.
Also in doing so should not remove valuable connections like Ethernet, USB and so on. Just give points to access them via the header. Taking them out and also removing access would then limit its potential.

IMHO if the people that cannot solder SM parts or create their own PCB then I would think they should stick to the widely available products that are already out there with all these connectors on them. I say this because if they cannot do these things then odds are they are just making something for themselves or a friend, and not to sell it. My point is that is will be very low quantity of sales from these types of people. Whereas if they were to say take the C1 and strip all the connectors off it and replace them with the 0.1" header access or even better a 2mm connector then companies can then make use of it, and high sale quantities will be there for hardkernel.

But ya, I agree it would be nice to have something small for the hobbyists, but that kind of sales is not what pays the bills

[DanielBull]

It sounds like its definitely a compute module you are after.
My only query with compute modules is if a company is going to the effort to make their own PCB to plug the compute module in, why don't they just use the reference designs from the CPU manufacturer and put the CPU and RAM straight on the PCB instead of having a separate module? If they put it straight on the PCB they would and save money and increase reliability. I'm confused as I don't see the market for what you are describing?
I also never understood it when the Raspberry Pi Foundation tried it and from what I can tell it wasn't very successful for them either.

[mikemoy]

Thought you might ask that question

Here is a few reasons why a company would not do what you are suggesting, and it is a good question you ask.

#1) When a company such as ours makes many products it's better to re-use things which not only keeps the total stock quantity of items low, but speeds up the design process. For example, say we made product A, where we took the open design and put all the parts directly onto the board. Then a few weeks later we make product B, we would need to take the time to put the same parts down again and route it out. Whereas if it was a module we would simply just add the connector for it. Some may say just copy and paste the block from one to another. In some cases you can do that and in others the design parameters prohibit it and you have to move things around.

#2) In the last 20 years all but one (a 6 layer) of our products needed to use a 2 layer board. When you go above 2 layers the PCB board cost increases dramatically. If we were to use a the open source idea we would be forced to make the whole board 4, 6 or even 8 layers. Whereas just using a module we could keep our boards at 2 layers. Typically most of out boards average 4" x 6". A 8 layer is that size would be silly expensive.

#3) If we were to use the open source idea then we would need to hire a embedded Linux engineer to create and maintain all the stuff that hardkernel spends many long hours on to have the board boot and run properly. One cannot just say well since its open source, just use the image they created. The reason is if we were to run into a bug or something strange, you cannot go to where the open source was from and say hey i got a problem with my board i designed, please fix it. They will say, wait.... you made it and you want us to fix it??
So creating you own board opens up a whole other can of worms. It's better in most cases to simply buy a known good working module from say hardkernel and use it. Its just one less piece of the puzzle in the total design to have to deal with when it comes to issues.

#4) BGA parts such as most of the parts on any hardkernel board costs extra for a board house to populate. A good board board house will x-ray the board to make sure the parts is soldered properly. This is extra to do and costs more. If we use a module we dont have to have this added cost.

"I also never understood it when the Raspberry Pi Foundation tried it and from what I can tell it wasn't very successful for them either."

One of the reasons why it because they removed the ethernet from it. You have to then add your own mac/phy chip, which they do not provide any details on how to add one and to what pins. I have tried several times requesting this info and they just dont provide any other than to say follow the design for the B+. Yet the B+ uses the onboard crystal to clock the phy chip which does not come out on the CM. So they shot themselves in the foot by not providing this information. Also when adding your own mac/phy you have to register yourself to get a block of MAC numbers to then add to the device which is another hassle. The CM was a good idea, but them not understanding what would make it good caused it to be not so good.

[DanielBull]

Ahh I understand now
OK so I guess the question is, is the C0 what you want or is it a different product?
I think maybe because the C0 has LiPo circuitry it perhaps isn't?

[mikemoy]

It does not bother me any if it has that on it. May have a need for it one day. The main thing I need is the ability to plug it into my board and stt get to the signal connections for all the hardware connectors and GPIO.

[Respectech]

I like @mikemoy's suggestions as well. As far as hobbyists that don't want to solder, ameriDroid.com could offer a service to mount the items the customer wants specifically. We have done something similar on a small scale for the ODROID-W.

[Tpimp]

I agree 100% with Mike. The next odroid should be mostly pins and module connectors. Maybe a micro hdmi. My argument for removing hdmi connector all together - if this truly is a "portable" project it will be connected to a serial display to reduce cost. Hdmi is convenient for beginners but my company would be very interested in an Amlogic module board by HardKernel! We currently, like Mike, are examining the difficulty of designing and building our own BGA/ Cortex-A5 pcb. For the cost it gets very expensive for each shop to "roll their own" but if shops like Mike's and mine work together on a community board we can share software work and improve board in interest of our company.

Current reasons C1+ does not meet my company's needs:
*Too expensive - amazing little board but to be used in an end product design there are too many features that often get "un-used"
*The I/O is much better than other odroids (3.3v, lots of it etc.) however most my my designs often still pair the odroid with a smaller chipset (arduino or PIC) to facilitating needing 3 pwm or two spi, etc. Seems like this board should never have a problem with pinout demands for the market it targets.

My ultimate C0:

Concept - Easy to drop in "module" that can make multiple designs possible (Quad-copter,Automated Home, Remote Sensors). Easy to plug into Lipo battery (I would argue lipo logic should be off-board but a cool feature) should just be a DC Barrel going into the PMIC. Lipo Charge circuits can be plugged into the barrel jack. Then you could choose to use multiple cell Lipos if you prefer. Should be easy to extend pins (need 6 pwm? sure. Need 4 uarts? we can do that - FPGA or CPLD), easy to "draw HW accelerated graphics", fast networking, and a linux filesystem. Size is important but functionality and cost are more important.

Why: Imagine the intel edison with an internal GPU to offload screen rendering. Imagine the MicroWRT or FireWRT with some actual power. Imagine the Odroid-C0 with the pinout capabilities of the Arduino-Due. Imagine USB that doesn't drop fifo through another bus. All for under $30 in low volume!

How: I vote for reduced cost first and foremost. Then the next biggest "feature" should be adding a second SOC for hardware driven I/O. Up the power IC's to handle 4A like the Xu4 (but still sip power in like a C1+). For a companion SOC - please choose AT THE MINIMUM a Cortex-m0+ (like the Kinetis KL03) but if I had a choice please include a CPLD or FPGA or FPGA2!
Then we attach the i2c and spi and/or UART from the AMLOGIC device into the fpga. The FPGA/MCU can then be broken out into the headers for the GPIO.

Again 90% of my designs that use the OdroidC1+ still need extra PWM, extra SPI, extra GPIO. Adding a second low-cost SOC solves all of these problems. Software becomes slightly more complicated but I know we could all work together to make it awesome! If there is interest I will draw up a concept image. Great discussion, I look forward to more ideas! I would even be ok with it getting a little larger if it could include these features and still meet a good price point. Remember even dropping $8 when someone is comparing this to other methods of control pcb (do it yourself, or other oem) is a big deal. in High quantity it really adds up. Also would be nice if this board could benefit from "Large inventory orders" as my company would look at 10k+ orders if it fit our needs. Long live HARDKERNEL!

[memeka]

@memeka
Yes, you can remove the heatsink to reduce the weight and height if your application is sensitive to the profile/weight.
But it is really hard to reduce the PCB size.
Many application needs the 2 USB host probably(two is much better than one ), so we removed the otg port.
It gave us a room to place the slide switch for power contorl too.

Maybe if you change the HDMI to micro-HDMI, reducing the profile a bit, you will also have room for the otg port, usb power is useful to have.
even if you don't reduce the PCB size.
also, since you added LiPO, look at ways to limit power consumption, even if it's just software - make it a bit more user friendly to run in a "low-power" mode (governor, max cpu frequency, tweak drivers?). add the heatsink to the connectors kit, you'll get the board a bit cheaper, impt for those that wanna get a lot of them.

i think the connectors you have are ok - well, except HDMI, it's kind of big. but like the prev question, what's happening with the ethernet?

[venkatbo]

It is obvious, the pov's here are governed by the specific needs of a given member. They all have their pros/cons. For example, one likes to have a full-size HDMI adapter, whereas another sees no use for a display adapter in some embedded scenarios..

What if we could follow the path taken by 3 highly successful kickstarter projects (in no particular order):
. https://onion.io/
. http://domino.io/
. http://digistump.com/
They all share a common approach:

• There is a core pluggable module that has the absolute essentials of the system - CPU, memory, and such. This cannot be used by itself, but will plug into a varity of "base boards". In our case, this pluggable module would be the C0re! (that's a '0' btw , not an 'o'
• Each "base board" that the C0re plugs into, can serve a different need

C0re + base-board = customizable C0 !

For example, the base boards can be like so:

• wifi/BT/BLE base board: will have no RJ45 connector, but can be used as wifi/BT/BLE sensor hub, wearable compute-brain, etc.
• ethernet base board: with a USB port, maybe we could use it to create an AP or some such networking device
• Octa-core based board: what if we can mount two C0res to create an octa-core for more oomph ?
• Atheros AR9331 based board: to create a smart-home hub
• breadboard base board, so tinkerers can have a field day
• display base board: lcd, OLED, etc.

and so on... Hope the point is made... 'am not suggesting one base board per use-case, but as a group we come up with base boards incorporating commonly clubbed ports/IO and such...

Then there is this potential - suppose there is the nextgen S905 soc in the horizon, what if the C0re is designed such that it can be made pin-compatible with existing base boards - viola, we don't have to throw away old base boards - all we need is get the new C0re !!! imho, a well-thought out set of base boards will definitely make for a great product suite with a long life.

Also, it would be immensely helpful if the eMMC and SD card connectors are actually placed on the top of the board, it would help with easy access during repeat image burn/tests cycles, especially if we wish to mount the board in a case or some such... And the attach-once-and-forget type connects (UART, rtc, hdmi, etc) can go to the bottom of the board. This will also help hide wires/cable at the bottom. One can use horizontal connectors (even 4-pin UART) to minimize height requirements.

Dream, can't I ;?) - It's for free !

[odroid]

Thank you guys for the tons of inputs.

Let's clarify the Ethernet interface first.
The S805 SoC has a Gbit Ethernet MAC. But there is no PHY.
So we added a PHY RTL8211F between the SoC and RJ45 on the C1/C1+ board.
Please look into the schematics page-12. http://dn.odroid.com/S805/Schematics/od ... 150615.pdf
C0 will not have the PHY definitely.
There are 16 pins at least if you really want to utilize the MAC.
If we palce 0.1" pitch throughole soldering points x 16, the board size must be much bigger.
I believe most people doesn't like it. So let's give up the Ethernet connectivity.
As Venkat mentioned a USB-Ethernet bridge IC can be a alternative solution too.


BTW, it is very hard to make a common target application between "DIY/Maker friendly market" and "Industrial/Commercial projects market".
So we may need to consider two different models the C0re and the C-zero.

[mikemoy]

Let me ask this. how many USB connections will we have.
You might also consider a SM 0.1" header to help keep down the board size without using through hole headers.

[odroid]

There are two root USB hosts in the SoC.

Can you estimate space for the 2.54mm of 8x2 pin pads?
It seems to be 22x12mm for the SMD type header footprint. It is a similar space for the RJ45 connector. It is too big.

[Tpimp]

Im slightly sad to see the ethernet go.

To clarify my point of no hdmi. - We have so many boards with hdmi/display ports. The C1+ is the better choice for beginners over the C0 (from my understanding). Also the situation where a display exists, almost always there is a separate power supply. Making the case of a Lipo battery + HDMI display pretty rare. However, I believe RJ-45 is high priority and personally would prefer RJ45 connect in place of the hdmi out. Because this isnt about what I want personally, the best solution might be to bring HDMI and MAC pins (both) out to a single surface mount connector. Then using the kernel device tree and configurations you could enable which hardware you want through a "daughter" board.

Odroid can you comment on the possibilities of adding another SOC on board? (Maybe? Or completely out of question?).
If not my concern is, how can we improve hardware driven peripherals on board? I think the next board should have more pwm. For instance, 2 PWMs on the c1+ (assuming you can sacrifice spi) is not enough to drive 3-axis servos for camera. I understand that the AMLOGIC s805 has 5 pwm drivers. I did not dive deep into the docs but I assume these were tied up by the MAC/RJ45 and may be available now?

[fkpwolf]

Can you use "memory on chip" as you have done on Samsung soc?

Maybe s905 has no advantage compared with samsung soc. Perofrmance isn't important in this application.

[odroid]

@Tpimp,
I think the high-speed analog switches for selecting the HDMI-MAC singals will be really expensive and space consuming by complex routing.
So it might not be a good idea since the SoC has not any internal switch nor pin-muxer either.

@fkpwolf
Amlogic has no plan to make any PoP(Package on Package) option for their SoCs.

I think we can't meet all the requirements from fully different two usage models. One side must blame us.

As I mentioned, we want to make a DIY/Maker friendly C0 first.
And we will make a more serious/professional version something similar to our Exynos-4412 module which was retired almost two years ago.
http://www.hardkernel.com/main/products ... 5270682824
It will have all the signals from the S805 SoC to the connectors/pads as much as we can.

[DanielBull]

Yes that sounds sensible, it does sound like a dedicated compute module style unit and a DIY unit (C0) are two different things.
That way you can add more of the features these guys have described such as (Ethernet) and drop ones which are not useful (LiPo) on the compute module.

[odroid]

@DanielBull
Yes. your short description is much more understandable than my long words.
It is still hard to speak English.

[mikemoy]

Im slightly sad to see the ethernet go.

To clarify my point of no hdmi. - We have so many boards with hdmi/display ports. The C1+ is the better choice for beginners over the C0 (from my understanding). Also the situation where a display exists, almost always there is a separate power supply. Making the case of a Lipo battery + HDMI display pretty rare. However, I believe RJ-45 is high priority and personally would prefer RJ45 connect in place of the hdmi out. Because this isnt about what I want personally, the best solution might be to bring HDMI and MAC pins (both) out to a single surface mount connector. Then using the kernel device tree and configurations you could enable which hardware you want through a "daughter" board.

Odroid can you comment on the possibilities of adding another SOC on board? (Maybe? Or completely out of question?).
If not my concern is, how can we improve hardware driven peripherals on board? I think the next board should have more pwm. For instance, 2 PWMs on the c1+ (assuming you can sacrifice spi) is not enough to drive 3-axis servos for camera. I understand that the AMLOGIC s805 has 5 pwm drivers. I did not dive deep into the docs but I assume these were tied up by the MAC/RJ45 and may be available now?

Tpimp, Not to get off topic, but have you checked out the SX1509 I/O Expander IC. This may help you with your PWM. I have this and it is a very cool IC. Each IO[X] pin has 4 registers, to control the rise, on time, fall time and off time. You may be able to use this to help with any device not having enough PWM pins for you.
https://learn.sparkfun.com/tutorials/sx ... okup-guide

[Respectech]

I agree that it sounds like we need two different modules. One for DIY and one for industrial.

If you don't mind me asking, why was the prime module discontinued?

[manuti]

Maybe a 96Board layout could be interesting.

[Tpimp]

Can we clarify DIY?

Do we mean do-it-yourself soldering of the connectors?
or do we mean for do-it-yourself projects?

To start discussing for a potential "compute module" I will begin a new thread such to not derail the C0.

Odroid a few comments and questions:

Comment - if the new board is a stripped down C1, why not name C1-.
So then we have -
C1- DIY C1
C1 (assumed to be discontinued when stock runs out)
C1+ - Feature model

Question:
What do you believe hardkernel could sell packet of connectors for? If C1- and the packet of connectors is is $35 total (or even $33) then most would benefit from just purchasing C1+ (do you plan to raise price on C1+?).

Also - How many cells can the Lipo circuit handle? Can you link to part documentation?
I fear the IC only handles 1 Cell Lipo and systems I build often utilize 3 or 4 cell for greater capacity and provide power for other systems i.e motors. A quadcopter utilizing C1- will then need either two battery types (motor and control) or DC/DC converter and bypass Lipo circuit all together and use barrel jack.

[DanielBull]

I personally would class DIY as DIY for projects and also that it can include solder on connectors but not surface mount. Although I'm happy with surface mount its beyond the skillset of a lot of people.

[DanielBull]

By the way, on a related note to the C0 its worth pointing out the new Raspberry Pi Zero announcement this morning. I've covered the details here but basically it's not a competitor to the C0 or even the W IMHO as it doesn't include LiPo circuitry:
https://plus.google.com/+DanielBull/posts/e7zCxMpKZVY

It is however groundbreakingly cheap.

[odroid]

Can we clarify DIY?
Do we mean do-it-yourself soldering of the connectors?
or do we mean for do-it-yourself projects?

I have no idea how we can define the DIY.
I know a few hardcore DIY people could make their own S805 based board.
They have a nice BGA soldering station at home even they could design a 6-layers PCB.
It seems not to be a common case. But we still call it a DIY.

When I do a DIY project to build a prototype of small robot like this, I need more flexibility of connector options.
http://www.hardkernel.com/main/products ... 1089647540
If I need only one USB host and two PWM ports, I can just connect several wires.
If all the connectors are fully mounted by default, it will be very painful.

I am also considering to make a drone controller with a wifi donlge, a LiPo battery, a 3.2inch TFT and a few joysticks for my personal hobby.
In this case, I also don't like the preinstalled connectors. It will limit my idea.

I hope above two examples can be an answer for your question partially.

I still believe many people need a smaller and battery powered C1 to build something.
Is this totally wrong? In that case, we must drop this project.

Anyway, the battery management circuit can support only a single cell.
Is it a big deal for a DIY Tesla Car project? even many tiny quadcopters can fly with a single cell.
A few giant quadcopters need 70~80watt power sources.
How much power(voltage/current) do you need?

[manuti]

I think Raspberry Pi Foundation have spies in your offices. RASPBERRY PI ZERO!!!
Really ZERO!!! With a format very similar to the original ODROID-W !!!
But is interesting to reconsider the use of a mini or micro HDMI and also mini USB.

[DanielBull]

I think its just a co-incidence.

[Tpimp]

Anyway, the battery management circuit can support only a single cell.
Is it a big deal for a DIY Tesla Car project? even many tiny quadcopters can fly with a single cell.
A few giant quadcopters need 70~80watt power sources.
How much power(voltage/current) do you need?

True there are some very small quad-copters that can run on 1 cell battery. But I don't believe they can also lift Odroid-C1 and camera. Most Quads will run larger battery system with a regulator.
Here is an example from the quad I built.
https://docs.google.com/document/d/1WMx ... sp=sharing

Lipo charge circuit is still good for powering OdroidC1 alone. For many users I think is a big plus! I think to be most useful for Quads lipo circuit that support 2 cell minimum, but then PMIC becomes more complicated.

I think C0 would fit many users needs but I the s805 is too powerful to compete small/cheap market like Raspi Zero. However, at the cost of the cortex-A5 more embedded features are left to be desired. Thats why I request pairing second very cheap SOC. The USB changes in C0 are very welcome. My biggest complaint about C1+ is the USB performance.

[mikemoy]

RPI Zero, I was super excited to see the announcement today and photo, then after looking into it I was sadly disappointed yet once again.
Now that is a DYI module if I ever saw one. There is absolutely no point to have a HDMI on something that. As for a plugin to another board its useless. By not bringing the USB to the I/O pins we would have to have a cable to plug into it.
No emmc on it. SD card is old school for the OS storage and should have been dumped long ago.

Bottom line is what the hell are they thinking....

[DanielBull]

Its a totally different market to what you want Mike but its a massive market.
I spoke to 2 traders this morning both of which were selling a Pi Zero roughly every 5 seconds until they sold out and that's not including the tens of thousands which have gone out on magazines. I'm getting one (and if Odroid make the C0 I will get one of those as well).

You have to remember the Pi Zero is not a plugin board/compute module and isn't aiming to be one. Its aimed at being a super cheap DIY maker style board for educational use and those customers want connectors like HDMI and USB. Very few kids have the ability to make PCB's for a board to plug into so its understandable that the foundation aren't targeting that market (especially as they tried once before and it didn't do so well AFAIK, possibly as they got it wrong as you pointed out earlier).

[Curmudgeon]

MikeMoy: I'm glad RPi Zero has no eMMC. Odroid's cheapest eMMC module costs $25. How would that go on a $5 computer? As for microSD being "old school": microSD is still a very good solution when you need affordable, removable, bootable, scalable storage.

[DanielBull]

I like the way with Odroids you get the choice to be honest.
eMMC or SD
But its obvious the Foundation were working to a tight cost budget so the board is pretty bare, it doesnt even have any components on the rear.