First Model A samples off the line!
We’re having a very busy, very mobile couple of days. We’ve been talking to Welsh teachers, spending time at the factory in Pencoed, doing EMC testing on the camera boards, picking up engineering samples of the Model A, visiting suppliers, and generally running up and down the south of England and Wales with our hair on fire. Sorry I wasn’t able to fit in a post here yesterday; I’ll make it up with an extra post over the weekend.
First up, here’s the very first (blurry, cameraphone) picture of a Model A board. I’ll take some more at the weekend with better detail when I’m near a proper camera.
I posted this picture on Twitter when I opened the box, and had some questions from our followers which I suspect some of you guys might be thinking about too, so I’ll deal with them now:
How much RAM does it have?
The Model A has 256MB RAM.
Updated to add: What’s the power consumption like?
Significantly lower than the Model B’s requirement. For all the gory details and exhaustive figures from our tests, visit this thread in the forums.
Isn’t the Model A kind of obsolete now the Model B has twice the RAM? What could I use it for?
We’re anticipating that those of you who buy the Model A will be using it for different applications from Model B owners. Model A has no ethernet and only one USB slot – and importantly, it consumes much less power than a Model B because the ethernet chip’s missing. We’re seeing demand for the Model A from people making industrial control modules, from roboticists, from people doing automation, for a bunch of headless operations – and, significantly, for people who want to use the Pi as a very cheap media centre. The Model A will only cost $25 (plus tax and shipping).
Wouldn’t it be cheaper for you to use a custom PCB without the existing routing for the ethernet?
Because of the volumes we’re producing, it’s actually less expensive for us to use the existing PCB we use for the Model B, and just not populate the ethernet part of the board, than to lay-out and separately manufacture a new PCB just for the Model A.
Where are you building the Model A?
We’re making the Model A at the Sony factory in Pencoed, Wales. For the foreseeable future, all Model As will be built there – whichever distributor they’re sold by.
We’re pretty excited. The Model A is something we wanted to produce months ago, but the crazy demand for the Model B has meant that we’ve not been able to build them, because to do so would mean that we have to cannibalise Model B parts – and that would mean that people who are experiencing the backlog would have to wait even longer. We’re hoping to get them off the line and into the hands of our distributors early in the new year, which will put us in a position where that strap line at the top of the page is actually accurate: you’ll be able to buy a $25 computer, which is what we’ve wanted to give you all along.
when they go on sale can you make sure that it’s during a weekday on GMT day time please…. ;-p
when they go on sale can you make sure that it’s during a weekday on GMT day time please…. ;-p
Good to know RS will also be using Sony-Wales
Actually, what’s happening (at least for the first while) is that the Foundation will be manufacturing them, and we’ll then be supplying RS and Farnell. They may take over the manufacture later on, but there were some chip-supply issues and a batch of memory that we ended up owning that meant this was the most sensible way to do it.
Does that mean once you’ve exhausted the memory you “own”, the model A will go to 512MB?
No, it doesn’t. One of the reasons the Model A is cheaper than the Model B is that 256MB RAM is cheaper than double that amount!
So what is the ‘much lower’ power consumption?
I don’t have the exact figures (I’m away from my computer) – I’ll try to get them for you this evening.
Some power measurements can be found here:
So that means the Model A uses far less than half the power the model B does. Even down to one third of the power.
Yup. Brilliant (even more brilliant) for always-on applications.
I have sometimes wondered whether the Model A really had a purpose… but thinking about it, I currently have 3 Model Bs, all fitted with USB Wifi, and only one of them has anything in its second USB port (a wireless keyboard dongle); and NONE of them has anything attached to its ethernet port. So two out of three of mine are already being treated as Model As. The lower power consumption will be great news, especially for anyone running their Pi off battery or solar power etc.
Of course, the creeping ‘gotta have them all’ mentality is inevitably going to mean Model As flying off the shelves whether needed or not… are we turning into Apple customers? Or are all customers of fruit-based technology susceptible to this? :-)
> The lower power consumption will be great news, especially
> for anyone running their Pi off battery or solar power etc.
Yeah, but it also opens up a lot of “cheaper” powersupplies. It’s slightly difficult to obtain adequate powersupplies for the model B. Almost every USB-power-supply will be good enough for a model A.
Great news! We all (the sensible ones anyway) appreciate the huge effort The Foundation is putting into the project. With the Model A and the Camera Module there is plenty to look forward to in 2013.
Maybe a couple of FAQ you should add:
– Is possible to upgrade model A to model B adding the ethernet chip?
– Will model A be compatible with additional shields (e.g. Gertboard) and coming camera module?
I guess the answers are no and yes, but I think someone will ask those questions.
You’re right – the answers are no and yes!
“Is possible to upgrade model A to model B adding the ethernet chip?”
I’m pretty sure it will be possible to add the ethernet chip, it won’t be easy though.
Of course you will still only have 256MB of ram
“Will model A be compatible with additional shields (e.g. Gertboard) and coming camera module?”
They will be fine.
I suppose that you could attach some USB hub+ ethernet adapter gizmo based on LAN951x chipset.
The 1st one I found is this:
but I think yuo cand find it at cheaper rates too…
Upgrading would also require upgrading the memory. I don’t expect even the people in the factory will be able to do that without a significant failure rate.
Downgrading a model B to a model A might be a more common thing. Probably still something pretty rare, but maybe nice for people who damaged their network/usb chip. I guess you can desolder the chip and you’ll have a model A with twice the memory.
But will it detect the absence of the chip and identify itself as a model A, or is the identification string hard-coded?
I think that doing so would break your USB ports, as on a model B, the SOC’s only USB port is wired to the USB hub/lan chip. You’d have to do some careful soldering to reconnect one of the physical USB sockets.
Besides adding the ethernet chip and the PROPER ethernet jack, you’ll also have to remove some resistors and add others….. NOT for the faint of heart.
Is it normal to apply solder to pads for parts that aren’t going to be populated?
I would have thought a different solder mask would be used for the Model As preventing unnecessary wastage of solder paste.
These are just engineering samples, not production boards. I anticipate that a different mask will be used when we move to production.
The lower profile might make this more attractive for some embedded projects. Although ideally I would like to see one with the analogue audio and video connectors removed and 512MB of RAM.
Doubtful that removing those connectors would save any significant cost, as the analog A/V outs are built into the SOC already.
Liz, how hard would it be to not populate the analog audio/video connectors but still have them work? I have a few embedded projects that could use this even leaner profile.
Definitely not on the cards, I’m afraid. Mind you, there’s no reason you couldn’t desolder them yourself.
Septic Peg says: after a few weeks, it will be upgraded to 512MB making the initial Model A boards instant rarities! And then it won’t be long until the Model B becomes 1GB. Well if it happens, I won’t complain! :-)
Alas, there are not enough pins between the BCM2835 and the POP memory to allow 1G. :-(
Yes, there are enough pins. The issues is that there are no PoP memories which have 1Gbyte and one chip select. Also these are LPDDR2 memories which are old fashioned and as far as I know no new models are planned. So theoretical it CAN do 1Gbyte but you need to convince a memory vendor to bring out a special model.
Call up hynix, maybe they’re crazy enough to do it
William H. Bell
Do you have any idea when they might be shipping in volume? There are probably a lot of robots in need of a brain.
Thanks and best regards,
First quarter of next year – can’t be any more precise than that right at the minute.
Hmmm … I’m surprised you didn’t quote him February “29th”, 2013 … :D
The whole Feb 29th thing does rob us of the opportunity to have a first birthday party. We’ll just have to have a cracking 4th birthday party in 2016.
Congratulations on the model A!
Great news everyone: extra mounting holes ;)
As a windmill owner I love the low energy consumption, And although I don’t have any plans for it yet, I like that fact that it can be an usb slave device (like an arduino).
I don’t know if it will become the success the model B has become. Personally, I lost faith in it. I don’t think the model A will be able to gain much momentum now almost a million model B’s exist and all currently available software, manuals, cases, etc. could safely assume there always are an ethernet port an 2 usb connections. Most software and hardware will adapt over time, but I think the Model B has secured it’s position as ‘the standard model’.
I’ve postponed buying a Pi for a long time. I wanted the model A and I didn’t want to order a model B because the foundation was concerned about not letting those who ordered a model B wait for their order longer because of model A’s taking up production capacity. Ordering a model B instead of a model A would not only postpone the Model A even further, it would also strengthening the false belief everybody wanted the model B.
But I’ve given in. I’ve put the Pi (thus the model B) on my Christmas list (which I sent to Santa just last week, oh the irony). But my friends already have Model B’s and I’m not going to be the odd one not being able to easily join the fun because I have a slightly different version.
Please please please don’t tell me that now I’ve finally given up on waiting for the model A, my model B will be delayed because you’re starting production of the model A or I will be stamping my feet while complaining how unfair that is as if I were a five year old kid.
Don’t worry: production of Model B isn’t going to be affected at all by Model A.
Sweet, do you have a waiting list? Or would you let us pre-order them?
W. H. Heydt
Hmmm… Model A order launch day…Feb. 28? (Or March 1, as 2013 isn’t a Leap Year?)
The pic is not all that blurry, you can still see the “Made In UK”!
One of the strongest features of the model B is its ability to be updated over its Ethernet. For the price difference compared to the lack of flexibility, I do not believe the demand is going to last. I’m glad you did not waste much time redesigning the board, and I hope you do not waste many resources making a huge pile of A boards. Make a smaller batch, and see if the interest is there.
Frankly, the B model board is so awesome for its price point, I would recommend you continue to focus your energies there. If someone wants to save 10$, they are more likely to pick a cheaper shipping option than loose hardware. The ram boost was a huge win for the platform, and if you can continue to find ways to improve the system’s power and abilities without driving up the price tag – that is where you want to focus your attentions.
And be careful about screwing with the current expansion pinouts. Add new ones, but don’t juggle pins around any more. Part of what made the Arduino such a success is its standardized footprint – allowing for a real growth of a supporting ecosystem of third party expansion hardware.
Yes, they did not even correct the fact that the various connectors are just off-grid. I hate that but at least you know they are the same on all boards.
I always thought the model A would have 512mb simply because it would work out cheaper to buy a larger quantity of that chip than to source and stock and supply two different chips.
Mind you i’m more than happy with the 256mb on my model B so i’m not complaining just wondering out loud.
The 512Mb was possible because the Foundation was willing, being a non-profit organisation, to accept a lower margin. I assume for the model A there was just not enough slack to allow that.
(Note that I am NOT officially part of the foundation, although I do a lot of work for them)
This is great! Most of the projects that I envision are “Internet of Things” type applications that will basically use the Pi as a gateway between a wi-fi network and GPIOs or serial interfaces. For these, the extra components and power consumption of the Model B are not needed or wanted.
It would actually be nice to have an even more stripped down version that omits all of the graphical components altogether. This could probably have less RAM too. Most of us who have done non-graphical embedded Linux stuff in the past have worked within 64-128M just fine.
Eric P. Scott
It would be nice if the Model A could be qualified for operation at lower temperatures; even just -10°C would open up all sorts of possibilities.
The temperature limiting factor has always been the LAN chip. The BCM2835 has been qualified to run at -40. I don’t know about the memory but there was a ‘balloon’ post on the front page a while ago where they said they had the Pi working at -40. (All other equipment failed…)
Interestingly (to someone perhaps!) when chips are signed off to run at a particular frequency is it usually the timing analysis that is performed at the low-end (-40’C) of the temperature scale that now limits the clock frequency. In the past it was the high-temperature corner. This inverted temperature dependence occurs when threshold voltage rather than mobility dominates cell delay. Well there you go….
I’ll be buying one. My squeezeplug pi only needs one usb for a wifi dongle and the lower power consumption will be great. I can free the B up for other duties!
As Brian said, “Blessed are the Pi makers”
Its Deja Vu all over again!
This time last year, we were feverishly worrying about how many Pis might be produced, what the proportion of As to Bs would be, and running a book on when this marvellous computer would go on sale, fervently hoping that we’d be able to get one for Christmas…
I’m sure we’re all far more relaxed now, and I for one can see many uses for the Model A. And yes, I’ll have some thankyewverymuch!!!!
429k through one distributor, about the same through the other. This suggests that the next pool is when will the 1 millionth raspberry pi ship?
Model A board: $25
Camera module: $25
USB WiFi-adapter: $10
This thing is going to put security-camera producers out of buisiness!
You forgot the power supply :-)
It might not put security-camera producers out of business if they switch to using the Pi! It might be a problem for supply if industries also start buying them in bulk.
No, you can get wireless IP cameras on ebay for about £30 delivered (with power supply!)
Definitely of interest for small robots. I’m currently playing with a 256MB Pi on a battery powered mobile robot – no Ethernet and one USB connection (wi-fi). A model A would do just as well, for less money, and use less than half the power.
All this talk about how much power the ethernet port uses begs the question: is it possible to disable the ethernet module on Model Bs to save the power? Either in the driver, or some boot option?
I believe the chip that controls the ethernet is also the USB hub controller so i presume if you disabled the ethernet you would also lose all USB?
Even disabled, the 9512 uses power, directly from the 3v3 power rail. Stopping it doing so would mean lifting its power pins – not easy but do-able
Is it possible to tell in software what board type (A or B) a program is running on?
Well since they have different amounts of RAM you could possibly just do a RAM check?
Not all model B’s are 512MB, the first 9 months of Pi’s are 256MB model B’s, more of these out there than 512MB B’s. That would be a stupid test, instead look to see what is on the USB port. If it is the LAN/hub chip you know it’s a model B. Anything else and you can safely assume model A.
Looking at the foto….It seems a bit unstable….
Very nice. I really should have bought two of those ultracheap wifi dongles from china. *Must* have an A to run it headless with the EVE alpha board.
How cheap do you need? Amazon are selling the Edimax nano dongle for £9 reduced from £29, and the driver is part of the Raspian distro so it works out of the box!
You can get unbranded dongles from china for less than £3 (delivered), or from UK directly for about £4. They are essentially identical to the Edimax one, I have got a few different types of them and they all work fine out of the box.
There is no such thing as ‘Edimax’ driver – as long as the dongle is based on a popular Realtek or Ralink chip, it will work out of the box in Linux.
does this mean you’ve sold a million Pis?
It also has mounting holes – YAY!!! I figured it would use the current Model B board layout, but we’ve had the ol’ “assume” makes an a$$ out of u and me happen before.
For those who keep forgetting that the Pi was and always will be developed for education first and foremost, a 29% reduction in cost from that of the Model B is huge for educational institutions, especially in the current economic climate (which is anything but warming). It simply means that they will be able to buy 29% more units for the same limited amount of funds, on top of the fact that even the Model B is an order of magnitude less than what has traditionally been paid for computing systems (keyboard, pointing device, display, cable, network, power supply. etc., costs are equal no matter what kind of computing system is used). The school districts I advise or in which I have friends and acquaintances have been waiting to be able to order the Model A in bulk after doing evaluations of single-quantity Model B boards at the individual educator, student, and computing support levels.
This is all good news, and please feel free to continue creating the conditions that will lead to more good news!
Will model a fit in model b cases?
Yes, it’s the same board with some bits missing.
It will, and we were wondering if our Cyntech case was the only one out there that wouldn’t leave you with open gaps in the case. It has knock out blanks over the top USB and ethernet port.
i like to buy it!
I want to produce a prodcut
with raspbery pi modell A in higher quanities (100-500).
So I had to know when they go on sale ?
So after model A launches, time to design the next version of raspberry pi with a higher-clocked ARMv7 cpable CPU :)
OMG i’ve been waiting years…… https://sites.google.com/site/blackrossind/project-updates/bunnyhood-singlecore
Congratulations! This is another great step forward. Once kids start making multiple projects, they’ll come up with plenty of rationale for using the “A” or “B” models. I can see them programming and updating builds on their B model and moving the sd card over to the A for stand alone projects such as robots, rovers, weather ballons etc.
it is practical as a low cost computer and is it true that the CPU pins after opening 2/3 tabs on ethernet?And I hope they change to 512 MB RAM for the model a that would be great!
There’s a bunch of 256 devices that need to be used up so 512MB won’t happen for the Model A for some time, if at all.