Additional architectural elements: Difference between revisions

From OLPC
Jump to navigation Jump to search
(merge where?)
Line 116: Line 116:


[[Category:Hardware ideas]]
[[Category:Hardware ideas]]

{{merge}}

Revision as of 15:39, 18 November 2007

Please try to limit the opinions of other people's ideas and save it for the discussion pages. If you think someone has a particularly good or bad idea, say so in the discussion page. This discussion/argument/feud can be noted in the article, but in a less familiar style.


Additional Architectural Elements

The envisioned manufacturing scale of the envisioned initiative can dramatically reduce hardware feature/sub-system costs and some important features are very inexpensive to add. Among these are IrDA (for wireless connectivity to a variety of existing/emerging devices), built-in computer tunable multi-band radio receiver with the minimal hardware required to extract also data transmissions sent using various methods (e.g. FM side band), at least minimal AD/DA and parallel I/O with suitable port enabling children to use the computer in laboratories and tinker with science/control and computer controlled toy applications/games.

Or genuine scientific instruments. I wrote a market research report, "Personal Instruments", many years ago, discussing the full range of computer interfaces, devices, and software for data acquisition and analysis. The major changes since then have been the introduction of USB, and the increasing speed and capacity of computers. Also, a wider range of instruments is available with computer data links. The built-in camera and microphone on the Laptop are powerful instruments in their own right, given appropriate software.
Compare the interfaces that were standard on the Simputer:
  • Smart card reader/writer
  • infrared
  • USB
  • Modem
  • Serial port
  • Sound in and out
  • Compact flash
There is substantial Free Software for scientific data acquisition and analysis--Mokurai 04:25, 19 October 2006 (EDT)
Much of this is covered under Hardware Ideas - Broadcasting. Note that the computer has an audio input port that can be used to connect sensors which provide varying voltage. And, of course, regular radios can be connected to this too. The Freeplay Lifeline is a hand-powered radio that could be used to provide this functionality. Note that not all OLPCs in a village need to have radio receivers since the mesh network allows easy sharing of downloaded content. That suggests that radio is best handled outside the OLPC, perhaps as a USB accessory.

If the CPU chip is custom made some of the above (e.g. AD/DA)and possibly 3D graphic acceleration and maybe even CCD etc. section (for camera) may be added without increasing cost. If operation of the augmented CPU+ chip sub-systems is optional then perhaps sub-systems would only receive power if/while they are used. Thus there may not necessarily be a power penalty for extending this aspect of system (+CPU chip) capability. Alternately perhaps these types of features would be on one custom chip.

Along same line of thinking, it would be useful if hard disc and CD/DVD controller circuitry + interfacing were included with suitable empty bays. This way children and their families could save money for add-ins and easily install them.

The OLPC has 4 USB ports therefore the wide variety of USB peripherals can be used with the OLPC. It is expected that most villages will have some kind of a library of ebooks on a USB hard drive or on a stack of CD-ROMS which can be downloaded via a USB CDROM drive.

One wonders why 10BaseT is not included,unless that would add significant cost(unlikely).

Wireless provides general networking without a fragile cable, a moderatly robust USB Crossover cable would be as fast or faster than 100BaseT. 62.252.0.11 15:50, 17 March 2006 (EST)
True. Yet, there are many "Wired LAN-s". How would the computer plug into such a network (ideally in a Plug&Play Mode") if there is no money for wireless hub? Of course wireless hubs are quite inexpensive, and maybe it can be assumed that in a few years they will be ubiquitous.
Ermm, one of these lappies could be a wireless hub with a USB <-> 100BaseT connector. Though I wouldn't be suprised if the local school gets a Ndiyo base station at some point which would probably be expensive enough to justify having a wireless link for the lappies built in. I suspect the thinking is that USB is sufficiently Universal that it means there's no compelling need for other interfaces (Edit: I mean wired ones). That is after all one of the basic design features of USB. 62.252.0.11 16:38, 17 March 2006 (EST)
Unlike "Wired LAN-s" USB has distance limitations. Also, "legacy environments" are built with wired LAN-s. If cost of 10BaseT/100BaseT on silicon +connector is trivial or converges to $0 in millions of quantities it is better to add also that interface, since it has no appreciable cost, and empower users decide how to set up their school/club networks. Of course, in some cases "thin client" (back to "time sharing") architectures may be preferred in some schools (e.g. Skolelinux: www.skolelinux.org [1]) - and, of course, the goal is to be "self contained" when the laptop is out of the school/club (in the originally envisioned "Dynabook" spirit).
It is good that the base system specification supports BOTH wireless and USB. The more choices the more opportunity for creative solutions suited to local tastes and cultures. For this reason IrDA ought to also be added since, it too, interfaces to legacy systems, provides yet another choice in some situations and is essentially a near $0 cost add-on. Also, it could be used in environments like school busses and school yards for spontaneous peer-to-peer networking - when a USB cable may not be readily available.
Regarding school busses, etc. thought should be given to how charging power could be provided for a number of units during a ride to/from school (assuming there are school busses). Also, "battery charging" could be a useful school service . Hopefully a well funded organization like UN could be stimulated to think creatively about providing grants, etc. to actually help developing nations successfully embrace such innovative initiatives. (Yet to be seen if the UN can respond to any such challenge. Would be of interest to know what if any contribution Mr Kofi Annan made to the initiative via UN channels after the Tunis demo/photo opportunity.) L Pfeffer March 18, 2006
I agree, both wired and wireless have compelling advantages so it's right to include both. But a 100BaseT connector has the disadvantage that it's not robust, you can destroy the socket by sticking a pen in it and the clip on a cable will fail without being abused. Even if the cost of an ethernet connector is zero at day one IMO it's likely to cost in some way for repairs or become unusable in short order. IrDA is certainly robust enough but it's range is very limited and as far as I can tell the only legacy devices it connects to are old laptops that cost $1000's when new and the odd printer, the IrDA spec is actually incompatible with the vast majority of Infra-red devices.
I was not aware of 10BaseT/100BaseT connector fragility. That can certainly be a problem Maybe the answer is assuring that schools/clubs have wireless coverage. That is inexpensive even without volume purchase
Palms, IPAQs, cell phones etc. have infrared capability.One would hope the laptop would seed an add-on after-market in toys and games - say wired an wireless controlled/sensed "Fun Things" (see below). If there was infrared support some manufacturers may chose to use it\ rather than WiFi and presumably the costs for infrared are lower in such "peripherals". Also, there is a message in having options and choices, such as BOTH WiFi and infrared for wireless. Options and Choices are not even taken for granted in a country like the USA and even less so in Europe. In poorer countries there is lesser empowerment and encouraging children to think in terms of choices is important. L Pfeffer March 18, 2006


It's very difficult to damage a USB socket. Before now I've managed to break the connector off a USB-Flash drive in the socket, the socket was undamaged! Even if you manage to break the tongue in the socket it can still work.
The problem with power on a school bus is that it's have to come from the diesel powering the bus, diesel costs money and is an inefficient power source even here. Much better to provide power in the classroom where it can be cabled in from whatever source is cheapest/easiest at the time. 62.252.0.11 08:14, 18 March 2006 (EST)
Probably there are cases where busses may not have to burn much if any excess diesel to generate the kind of power needed for charging some laptops. Would be useful to see a back of the envelope engineering analysis of the possible "spare power" opportunity. Ideally the unit's charging section would be "smart" and not put drain on the charging system once the battery is charged. Also, the power generator add on to a bus would be sensor based and have an option to couple to the bus mechanism only if the bus is going downhill or at least on level ground - and de-couple while going uphill. Etc. There must be an optimization opportunity here (and room to think of the bus too as a "system" - and find a way to add a computer to it for this and perhaps other tasks which are cost or otherwise justified. (As with many choices there is also a BOTH option - support both in school and on busses ...) L Pfeffer March 18, 2006

Hopefully it will be easy to add memory for program execution - maybe the way FLASH memory is added to cameras. Similarly, it should be possible to add FLASH capacity via FLASH memory cards from consumer electronics appliances (e.g. one of the standards from digital cameras). NOTE: neither of these should be via the USB port, but via recessed slots in back or bottom of case. They should be non-intruisive augmentions - much like FLASH cards in digital cameras.

Designing the the system to be an extensible framework is a key architectural choice.

Two exciting extensions or a separately packaged/priced augmentation unit ought to be considered.

a) a laser/etc. based MEMS chip+optics based projector which could give the system a large screen on a wall etc.

b) some linkage to electronic paper. In fact this endeavor could accelerate commoditizing ePaper technology

Seems that even if a few add-on hardware features would raise price target to say $120 this could be well worth it.

L Pfeffer March 17, 2006

It should be assumed that in some cases the system may run off a car battery. Would be interesting to know how long it could operate off that source. Would be useful to have as part of the product line at least three low-cost kit based accessories for power based on solar, hydraulic, and wind based generators + battery. There must be such kits already developed for use in poorer countries.

I actually live in the third world and let me tell you there are no school buses. Children either walk to school or take a regular bus. I think it's very important to make these cases waterproof and shock resistant. Gigi (From Venezuela)


Any chance for VGA connector?

Omission of VGA connector is what kills this device for me and my kids ...

Again, USB comes to the rescue. There are USB-VGA adapters and some people have created Linux device drivers for these. If memory serves correctly the Linux drivers supported the USB-VGA adapters with SiS chipset. Admittedly, this is not a common solution, but it does exist so the principle has been proven to be possible. Again, if memory serves me, the Linux driver was developed by someone who wanted video on an ARM-based NSLU2 which has two USB ports.

Ability of attaching the laptop to external projector/monitor/TV is very important for me ...

Projectors mostly work off a VGA port so the USB-VGA device mentioned above should solve that problem. Also, something involving a base-station PC and X-Windows over wireless could be worked out by any Linux sysadmin.

Kids can then easily share their results/works with classroom ...

[With the OLPC's being mesh-networked, can't sharing results/works be done via a VNC-style solution or some shared whiteboard software? Richard Lloyd, UK]

Also since 17" monitors are down at $120 - for $220 +keyboard/mouse one could have a nice terminal for more comfortable work ...

[Spending more than the OLPC on adding a monitor seems to defeat the point of the project somewhat. If you can afford a monitor, why not buy a second whitebox PC as well and use the OLPC to ssh into or run X programs from back to the second PC? It's how I use my Sharp Zaurus PDA when I'm back at home... Richard Lloyd, UK]

There are USB video card devices based on the SIS chipset and drivers are available for Linux. There is no reason why this could not be used on the OLPC.

Server and Repeater Units

In areas where no electricity is the Internet will probably not be available. The mesh needs a boost to reach other networks, and something to fall back on when the larger network in unavailable. A server would be nice to have for village email, messaging, webspace and fileserver. The server needs no display and no keyboard, but much more storage.

A USB WIFI device with a long USB cable that can be strung up outdoors would be useful (long USB may need a active repeater). A directional antenna could connect locations separated by long distances. A good installation with line-of-sight can reach 20+ miles with quality normal 100 mw radios (200 miles has been done with 6000 mw radios). The Linux kernel has bridging that can be loaded as a module. The internal WiFi device could supply network access to the local users.

In effect, the server would be a bridge to a distant mesh node that also cached content and ingoing/outgoing messages. The server can reach far networks because it has the advantage of height (hilltop, not valley) and the directional USB WiFi (possibly with a higher-powered radio).

--Hckhckhck 08:52, 24 May 2006 (EDT)

Any user connected to the internet in any way could share the access to the whole mesh.
Just imagine a village with a satellite connection, a teacher connecting through his mobile phone, a few landlines serving internet over modem at the school, the hospital and the city hall. As we're not emuling any huge software package and a big part of the contents aren't multimedia, THIS is BROADBAND!!!--S112 05:04 16 August 2006 (EST)

I wonder if one of the laptops could be the screen for a server, the idea is simple, just a dumb pci graphics card that instead of providing a VGA output provides a USB connector than can be connected to the olpc. The olpc would see it as a VNC (or similar) server the server would see it as a VGA card (unacclerated!). The only problem I see is that it intrudes on the KVM market.

BTW: Wi-Fi repeater: http://www.meshcube.org/english/specs.html BTW: Headless server: http://www.realweasel.com/intro.html

62.252.0.11 05:57, 22 October 2006 (EDT)

Reinventing the OLPC