The Upgradeable Allwinner Dev Board That's Laptop-Compatible Raises $50k So Far

[lkcl]

I'm not complaining about the price being too high. My suspicion wasn't that the price of this is high, it's that I'm not sure they can pull off so much complexity.

you can review the archives on how i did things, and you can see how i slowly plodded through teaching myself 3D CAD and PCB CAD design here. take a deep breath, it's a big list:

* http://lists.phcomp.co.uk/pipermail/arm-netbook/
* http://rhombus-tech.net/community_id...top_15in/news/
* http://rhombus-tech.net/allwinner_a10/news/
* http://rhombus-tech.net/community_id..._desktop/news/
* http://rhombus-tech.net/ingenic/jz4775/news/

but a short-cut is to just search "eoma68" on youtube and you'll see there's a stack of rather droning and boring not-very-high-quality videos (all of which is irrelevant) the main point of which is to *show* you that yes i did actually make the stuff and got it up-and-running... *BEFORE* saying to josh at crowdsupply, "ok ready to go".

I hope they can, and the higher price makes me MORE comfortable (not more suspicious).

yeah... i didn't feel comfortable going to the novena-level of pledge levels, i wanted to keep it within the reasonable budget... so many aspects to this project... but yeah just affordable enough so that people feel comfortable amortising the cost of the laptop housing over 8+ years, even at this early phase (bear in mind mass-volume pricing is siiignificantly lower). that's also really important to remember, even at this early strategic phase, that the laptop housing uses mass-produced parts (that can be found on ebay) and it can basically be kept going indefinitely - esp. with a 3D printed case design that's GPLv3+ licensed, just upgrade for around $50 a time every year or so, and sell the old card on ebay when you're ready to do so. live video from hope2016 explains it best https://www.crowdsupply.com/eoma68/m...ates/hope-2016

The thing is, after such a bad experience with Pi-Top (bugs, poor support, bad communication, and low quality hardware) despite the team really having the best intentions,

i know, right? like... why the hell didn't they contact me?? or, given that the rhombus-tech's project ikiwiki-based web site has been up and running for over 5 years, did they see that and go "huh we can do that, and we can do it better, too" i mean, good for them for having the energy and enthusiasm, but there's REALLY GOOD REASONS why i'm following software libre principles.

you can also see from the laptop comparisons update that i've been keeping an eye out over the past 15 years on a heck of a lot of projects, i've watched the failures and the successes and learned from them all: https://www.crowdsupply.com/eoma68/m...op-comparisons

I'm hesitant about a far MORE complex project like this going smoothly.

well... join the mailing list and help keep an eye on me then http://lists.phcomp.co.uk/mailman/listinfo/arm-netbook

The idea sounds great, but I'd suspect getting the shell of a laptop to work would be very very difficult.

it was. f*****g difficult. took me something like 15 months to develop. i *suspected* that sandwiching the plywood by PLA 3D-printed parts would work... i was extremely relieved to find that it did. plus, this is a 1.1kg laptop not a 2.5kg standard one.... achh i'll just refer you to the update it's what it's there for https://www.crowdsupply.com/eoma68/m...printing-parts

For one, the display might have undocumented proprietary connectors.

it doesn't, it's a standard IPEX connector. errr let me find its datasheet... right - page 8. 20453-040t-0x you want the extremely-common LP156WH4-TLN2. available in all those $300 walmart / best-buy / aldi's / tesco's / asda laptops you see. available in volumes 1 on ebay and 1,000,000 from random china suppliers. you should be able to find the datasheet here: http://hands.com/~lkcl/eoma/laptop_15in/

The keyboard and mouse could as well.

chicony mass-produced asus keyboard. available on ebay. please don't buy from ajparts or their aliases, they have a 75% track record of delivering the wrong f*****g keyboard. yes really that's the right way round: 75% *failure to deliver*.

firmware for the keyboard is ENTIRELY GPLv3+ RIGHT FROM DAY ONE and if you'd like to track albert's progress (and also mine) you can follow the links here https://www.crowdsupply.com/eoma68/m...ates/keyboards

the mouse track-pad, frickin hilarous, man. have you ever tried tracking down mouse trackpad suppliers? go on, have a laugh - do a google search for "trackpad supplier" and tell me how you get on

so... in the end i went "f*** this i'm going with a 4.3in CTP+LCD. i wonder if there's any SPI-based LCDs so i can wire them up to the 72mhz STM32F?" turns out that there are and you can track _that_ as well and also verify that the source code is available of the preliminary explorations that i did, and also take a look on the adafruit forum: https://forums.adafruit.com/viewtopic.php?f=47&t=83003

You could buy the laptop shell this project offers, but that ignores the point which makes the project awesome. I again don't mean anything bad about the team or about the price--it's just a very complex undertaking to get right in terms of hardware interoperability.

... by now, from the above comprehensive set of links - which are a *summary* so as not to completely overwhelm you - i trust that you're getting the general feeling that i really do know... well... not that i "know in advance what i'm doing" but more that i have the attitude of "let's do it anyway, learn as we go along, and just never quit" but also that i REALLY AM running this as a transparent software-libre-inspired project.

[lkcl]

Remember One Laptop Per Child?

yeah i was encouraged by dave crosland to fire up the Sugar OS on it - there will be an update about it shortly. we encountered a couple of bugs in the Fedora 24 "spin" but overall it's pretty good.

This board can sort of revive that sort of really basic laptop but more modern and robust which can be useful as an ARM based Chromebook.

which reminds me: must put chromeos on the TODO list to put on this computer card, see if i have time to do it during the campaign.

Hopefully there be some potential laptop designs that can come out of this as well.

yeah although it'll be challenging, at least people can take a look at the *GPLv3+ licensed* parametric design and consider doing their own variant(s). https://www.youmagine.com/designs/li...-laptop-design

If this can be pulled off and there is interest in developing solid yet low cost machines using this board design then it could well revive a market in lower cost machines that run either ChromeOS or even a variant of Ubuntu.

bear in mind, the actual BOM for the laptop housing is $166 in QTY 250, so could easily be $150 or less in QTY 20,000 - the EOMA68-A20 is around $35 in QTY 250 and could easily be reduced to $16 in QTY 20,000 - and this is for a *15.6in* modular upgradeable laptop!! the very real possibility exists that we could be selling this retail for close to $200 in mass-volume. and it's 1.1kg! you *just don't get* 1.1kg 15.6in laptops anywhere else in the world.

[lkcl]

They aren't using Allwinner software and Allwinner is basically the only manufacturer where you can find low amounts of chips on sale (from third parties) too.

yyep. went over *some* of the options here (i must have evaluated a hundred SoCs over the past 5 years), incredibly the A20 by comparison on an "ethically-sourceable" criteria is genuinely at the top of the list. https://www.crowdsupply.com/eoma68/m...ng-a-processor

Support for Allwinner hardware is added (and mainlined) by team sunxi (not affiliated with Allwinner in any way) and it's relatively good.

nobody's particularly happy that allwinner don't actually pay them... but it's a complex situation.

This project is flawed on many other aspects.

go on, starshipeleven - elaborate on that: you know you want to

[lkcl]

In fact it's not exactly a dev board, more of a consumer device, even if it may end up being backed by more computer literate people. And it has a breakout board so you can kind of use it as a dev board too. But I think if you just wanted a dev board you'd get one of the many SBCs out there.

... and you'd be throwing that dev board out in a year's time as "too slow", and anything you'd created around it you'd have to redesign. on the other hand... if you design things based around an EOMA68 break-out board, if it's "too slow" in a year's time you just go out and get an upgraded Computer Card (and sell the older one on ebay so that people can buy it and put it into a co-located rack-mounted hosting space and make $EUR 3/month out of it). huh. how about that?

More interestingly, you don't have to use the laptop with Allwinner anymore. There's a passthrough card that you can use to connect an HDMI and USB cable and plug a bigger screen and keyboard to your mobile or HDMI stick or SBC or whatever. OR you can have another laptop and use the EOMA68 laptop enclosure and passthourgh card as a 1.1 Kg second screen. I think it's quite versatile.

thanks. a lot of thought went into this. full-time 5 years there had damn well better have been a lot of thought gone into it...

I dislike allwinner like anyone, I've heard the SOC itself is about 10% of the price of the CPU card, so you could think it's just too little to base your decision on. You could also think anything else you buy may be produced by companies that colaborate in weapons trade or mass surveilance or something. Maybe the RAM manufacturer is more evil that Allwinner, and we just don't know. Yet once you know, since you always have the option of not buying anything, it's hard to go and buy Allwinner. With the passthrough card at least you may be able to not buy another CPU.

we need to be able to send a "financial incentive message" to allwinner. buying *GPL-compliant* processors like the A20 and boycotting *only* those that are not will do that. if we boycott the *entire company* then we lose the power to influence them.

i will be going to speak to allwinner when i go over to S.E. Asia. yes i will be mentioning quite how pissed-off everyone is.

It's obviously not the best thing in any of the features. It's more a promising basis for a different evolution of computer design and commercialization.

now why the hell didn't i think of saying that?

But then a mainstream device is also a promise.

... and how's that working out for people? motorola *promising* to commit to providing security updates... and then reneging on that. google pulling the plug on those home-IoT boxes, bricking them with an OTA firmware update and shutting down the "cloud" services i mean WTF?? hard lesson for people to learn, huh?

this is primarily one of the reasons why i'm running this project along transparent libre-software principles, so that you KNOW that not only the promises that are made are kept, but also if for some reason i *CAN'T* keep them, someone else can consider forking the entire [hardware *and* software] project. if they do that, they'll need to respect the EOMA68 standard though: that's inviolate, not up for debate, and not up for sale.

I'd rather take the risk on a project that is trying to do the right thing that in a project that is only profit driven. Even if only for the consolation, when this project and others after it fail, and computing stops being a tool for people and ends up being a way to control everyone, and younger ones ask you what you did to stop it, to be able to say "not enough, but at least some of us tried".

YAAAAA! someone who gets it, YIPEEEE!! oh btw that's also why i release the full CAD files under software libre licenses, so that people *could* carry on if i got hit by a bus.

P.S.: I didn't mean the EOMA68 campaign is likely to fail. Just that even if that was the case it might be reasonable to back it, or even more so. In fact the last update says they're at 80% of the units they need for the EOMA68 cards at arroudn 50% of days left.

92.4% (231 out of 250) which is awesome.

[lkcl]

In fact it's not exactly a dev board, more of a consumer device, even if it may end up being backed by more computer literate people. And it has a breakout board so you can kind of use it as a dev board too. But I think if you just wanted a dev board you'd get one of the many SBCs out there.

... and you'd be throwing that dev board out in a year's time as "too slow", and anything you'd created around it you'd have to redesign. on the other hand... if you design things based around an EOMA68 break-out board, if it's "too slow" in a year's time you just go out and get an upgraded Computer Card (and sell the older one on ebay so that people can buy it and put it into a co-located rack-mounted hosting space and make $EUR 3/month out of it). huh. how about that?

More interestingly, you don't have to use the laptop with Allwinner anymore. There's a passthrough card that you can use to connect an HDMI and USB cable and plug a bigger screen and keyboard to your mobile or HDMI stick or SBC or whatever. OR you can have another laptop and use the EOMA68 laptop enclosure and passthourgh card as a 1.1 Kg second screen. I think it's quite versatile.

thanks. a lot of thought went into this. full-time 5 years there had damn well better have been a lot of thought gone into it...

I dislike allwinner like anyone, I've heard the SOC itself is about 10% of the price of the CPU card, so you could think it's just too little to base your decision on. You could also think anything else you buy may be produced by companies that colaborate in weapons trade or mass surveilance or something. Maybe the RAM manufacturer is more evil that Allwinner, and we just don't know. Yet once you know, since you always have the option of not buying anything, it's hard to go and buy Allwinner. With the passthrough card at least you may be able to not buy another CPU.

we need to be able to send a "financial incentive message" to allwinner. buying *GPL-compliant* processors like the A20 and boycotting *only* those that are not will do that. if we boycott the *entire company* then we lose the power to influence them.

i will be going to speak to allwinner when i go over to S.E. Asia. yes i will be mentioning quite how pissed-off everyone is.

It's obviously not the best thing in any of the features. It's more a promising basis for a different evolution of computer design and commercialization.

now why the hell didn't i think of saying that?

But then a mainstream device is also a promise.

... and how's that working out for people? motorola *promising* to commit to providing security updates... and then reneging on that. google pulling the plug on those home-IoT boxes, bricking them with an OTA firmware update and shutting down the "cloud" services i mean WTF?? hard lesson for people to learn, huh?

this is primarily one of the reasons why i'm running this project along transparent libre-software principles, so that you KNOW that not only the promises that are made are kept, but also if for some reason i *CAN'T* keep them, someone else can consider forking the entire [hardware *and* software] project. if they do that, they'll need to respect the EOMA68 standard though: that's inviolate, not up for debate, and not up for sale.

I'd rather take the risk on a project that is trying to do the right thing that in a project that is only profit driven. Even if only for the consolation, when this project and others after it fail, and computing stops being a tool for people and ends up being a way to control everyone, and younger ones ask you what you did to stop it, to be able to say "not enough, but at least some of us tried".

YAAAAA! someone who gets it, YIPEEEE!! oh btw that's also why i release the full CAD files under software libre licenses, so that people *could* carry on if i got hit by a bus.

[edit: for people who might not have heard.... http://fortune.com/2016/03/17/apple-...ok-going-dark/ - tim cook (apple's CEO) - "George Orwell could never have foreseen: we did it to ourselves"...]

P.S.: I didn't mean the EOMA68 campaign is likely to fail. Just that even if that was the case it might be reasonable to back it, or even more so. In fact the last update says they're at 80% of the units they need for the EOMA68 cards at arroudn 50% of days left.

92.4% (231 out of 250) which is awesome.

[lkcl]

What I said is that simply there isn't much choice unless you are a big OEM and you can order 100 thousands or more SoCs to the SoC manufacturer.

If you cannot pay for that, only way is looking for resellers that sell smaller quantities, and this means Allwinner and Rockchip are the only SoCs you can use.

rockchip demand $100k in one-off up-front fees for illegally-restricted NDA'd access to the linux kernel source, which is why we don't use them. i forgot to add that to the processor update when i wrote it.

[Palu Macil]

@Ikcl There is so much to read in the thread (and I've had some replies disappear in mod queue, I think) that it's hard to tell how the conversation stands at the moment. I didn't realize at first that you were even affiliated to the project but rather thought you were simply an expert with very related experience. Certainly, to me your active comments here already indicate that your project is more open than the Pi-Top. I am in the less technical category of "web developer who builds his own desktops" so I value this type of interaction a lot.

[lkcl]

In light of your points, I think the Libre Developer and community involvement is what sets this apart to be successful where Pi-Top failed. I think I will consider trying this out. However, I do disagree with your statement that the Pi-Top is good for education or a success overall. There are no available GPIO pins, the team doesn't provide information for getting the screen to work with other Linux kernels, and they don't distribute source for their Pi-Top OS. Additionally, I was an early backer and had to wait for a release which was about a year late. The delay was fine, but they never provided updates to keep backers in the loop. Finally, they supposedly provided a bunch of educational software that you can log in with your Pi-Top account and use. Do people use it? Does it work? I don't know because when I've contacted support, they always ask if I've upgraded to the latest OS and never reply when I let them know I have done so... I do think the Pi-Top was a great education for the well meaning student graduates that produced it, but I would warn anyone from buying it for themselves.

i haven't bought one (my sponsor chris however has: he's equally unimpressed)... i was trying to be... diplomatic so am glad that you wrote this (from direct experience of following the project), and yes, i think that after all's said and done it can be added to the decades-long list of "learning experiences" for people who would like to follow in the footsteps of "open hardware".

Casework: I'm not sure you understood what I tried to say:

possibly not....

I don't think the printable OR orderable case is the main cool factor of the project. The main thing I'd get this for is reviving an old laptop that I've gutted to try with this.

aaahh okay. right. yes. okay, i get it. okay. i actually seriously considered this approach (3-4 years ago) and on closer and comprehensive inspection decided not to run with it. at the time i was like, "omg, really?? design my own casework?? argh!" and then of course the 3d printing revolution stepped up a notch, my friend phil encouraged me to get one, pointed me at openscad and the mendel90 and i was off

The problem is that there are just a ton of proprietary things in old laptops, so working with an old laptop might be iffy. I'm not intending to complain about the case they are working with, it's just not as interesting to me as putting this in a broken old laptop. But I love the idea an hope it works!

it's not just that there's a ton of proprietary things in old laptops, it's that in many cases they had to have the connectors *custom-made*! so they would have had a good relationship with the laptop designers (IBM, HP, etc.), made up special tooling, made a run of 250,000 to 1 million... then *destroyed* the tooling to make room for the next model. now you want to buy... how many? of a connector that costs $0.20 or less in volume? so you're expecting them to jeapordise a million-dollar relationship to sell *proprietary* and probably copyrighted and trademarked goods... for what... $20 worth of business? this is why sadly the vero apparatus team are going to get a really nasty shock when they start properly investigating the approach that they're taking.

basically what i discovered (and discovered the hard way from losing a customer...) is that you really do have to take all these three things into account: (1) component sourcing (2) PCB design (3) casework design.

these three things *HAVE* to be done all at the same time, and you really really can't retro-fit one laptop's case, components or PCB into another laptop's case, components or PCB.... unless they're *designed* to be modular in the first place. and with the mass-volume focus on cost-cutting and profit-maximisation, modular is just far too risky for them. look at what happened to MXM for example.

even the LCDs, i mean it's mad: they're heavily-customised for specific make and models of laptops! as in: the first thing that a laptop manufacturer does is contact the LCD suppliers and give them a *custom* set of specifications and a whopping great cash up-front order of 250,000 units. the power requirements for the backlight? totally different from LCD-to-LCD. pinouts? totally different (so now you're looking to get a custom cable... even if the voltages and current match... which they won't).

so this is why the pi-top was a successful proposition and you can put in different SBCs, because they chose to do an HDMI-to-eDP converter board, basically. that's all the pi-top PCB really is (that and a battery charger).

now i know (because i investigated it) that there happen to be some RGB/TTL to LVDS converters out there which cover a comprehensive range of LCDs: hilariously the company that makes them keep the firmware proprietary, but somebody managed to reverse-engineer it.

honestly i'd recommend that you go down that route: PCB1 (which handles the computer card and does the RGB/TTL to LVDS conversion and backlight circuits) are *specifically* customised to suit *exactly* this *one* LCD - the LP156WH4-TLN2 and *no other LCD* except for all that LCD's "equivalent" clones, that is.

i don't want to dent your enthusiasm but i would very much like you to be fully aware of what you'd be letting yourself in for, if you get me

[lkcl]

@Ikcl There is so much to read in the thread (and I've had some replies disappear in mod queue, I think) that it's hard to tell how the conversation stands at the moment. I didn't realize at first that you were even affiliated to the project but rather thought you were simply an expert with very related experience.

i just had one reply eaten by the forum *scratches head* took me a while to write so i'm a bit peeved, let's see if it turns up

Certainly, to me your active comments here already indicate that your project is more open than the Pi-Top. I am in the less technical category of "web developer who builds his own desktops" so I value this type of interaction a lot.

yeah me too. summary of what [probably] just got lost: don't underestimate how complex it is to try to retro-fit stuff into an existing laptop case. i considered it... and quickly ruled it out after doing a comprehensive analysis. i don't want to dampen your enthusiasm, just make you aware of what you'd be letting yourself in for. the vero apparatus laptop team are going to be following this "train-wreck" approach... and drive up the 2nd-hand market and create shortages on whatever older laptop casework they choose to use...

[update: ok great it wasn't lost after all - comment #28 just above]

[kingu]

Boycotting a product from a vendor that looks to sell a modicum of Allwinners minimum order quantity is only upsetting to one of them.

Seeing as what they have on offer is the closest to attaining a laudable goal, you can practically get behind it, in the name of not only getting something, where options are limited, but also sending a message to Allwinner that this is a viable business-model.

One that Allwinner could get behind across the board, making a no-brainer of the currently some-brains-required market that is atm maybe big enough, but will grow to give back tremendously once everyone pulls in the same direction.

Yes, Allwinner is Allwinner, coincidentally, that is not nobody. Also, this project is not just about one initial board, but the modularity and ecosystem thereof.