Talk:Embedded system/Archive 1

From Wikipedia, the free encyclopedia
Jump to: navigation, search

Clean up

I have done a major trimming of the more lengthy and opinionated parts of this article. It still seems too long, but at least it no longer shows the "too long" warning from Wikipedia itself. Some parts might be better split into other articles, such as the discussion on reliability.

I expect I added lots of mistakes too, please feel free to correct :)

Aaron Lawrence 23:47, 9 July 2006 (UTC)

The 1st embedded system

Regarding the first emdedded system: wasn't it actually the Whirlwind? -- Jörgen Nixdorf

Hmmm. I suppose you're thinking of SAGE. It all depends on the definition, of course, but machine room located "big iron" computers making up part of large networked systems tend to be considered as computers as such, being visibly important parts of the greater system, as opposed to "unseen parts" of an embedded system operated via a restricted UI (and display, sometimes) instead of computer terminals. So I would still say the AGC qualifies as the first real embedded system. --Wernher 19:29, 2 Apr 2004 (UTC)
Why was SAGE taken out?
-- Yomomma —The preceding unsigned comment was added by (talkcontribs) .

Anti-RTOS bias?

There seems to be a bit of anti-RTOS bias here. Add little pro-RTOS text to bring the balance back to NPOV. -- DavidCary

There do tend to be two camps regarding the use of an RTOS in embedded systems. However, it really comes down to the overall complexity of the system being developed. Using an RTOS in more simplistic systems (one with very few tasks to perform and possibly with a lower powered processor or very limited memory) would probably just make the entire system slower and more complicated than it needs to be. However, in more complex systems (with higher powered processors such as an MPC82xx), an RTOS actually makes development simpler. I've developed extensively in both environments and am quick to take either point of view, depending on design goals of the system. -- JimWilliams57

I added a bit to explain the rising popularity of embedded linux. Amnonc 22:47, 16 September 2005 (UTC)

PDA's are NOT Embedded Systems

PDA's (PocketPC) aren't really considered Embedded Systems anymore. They have become as generic as Personal Computer with a very large user space. -Abe

Not sure about that.IMHO, PDAs are included on embedded systems, as they are based on ARM, M68K, MIPS, H8, PPC, ... processors. They need cross-compilation support and they are specific use circuitery. For me they are Embedded systems. -Hector

One of the things that would indicate not being an embedded system, the ability to develop and compile software on a device, is present (at least with third-party applications) on the Palm platform; I'm not sure about others.

Many programing tools such as Python and the GCC C compiler have also been ported to Windows CE.--Tfgbd 23:55, 18 April 2006 (UTC)

PDAs are embedded systems if they have memory, peripherics, etc. in the same encapsulation. -- alex
PDAs can be embedded systems if they are used for specific applications, just like PCs can be embedded computers. In my opinion a standard PDA is not an embedded system, but a general purpose computer. An embedded system is a computer (or computer platform) designed and used for a very specific purpose, and usually is part of larger system. -- Johan (
Johan is definitely corect in saying that PDAs can be an embedded system. In a system that I support, we considered using a device (which the vendor was selling as a PDA platform) as the principle UI and controller of a larger embedded system. The "PDA" just happened to have the right combination of display, touch-sensitive input, computation, and communication ability needed for our application (which is definitely an embedded system).
On the other hand, I think one could debate whether a PDA-as-PDA doing ordinary PDA things is, itself, an embedded system; points could be made both for and against the claim.
Atlant 14:50, 16 May 2006 (UTC)
The characteristic distinction the above learned discussion is apparently missing is that embedded systems are 'Turn Key'—whether there permutations by adding different software are moot, whether their processor and other computer hardware may or may not be general purpose is moot. 'To their operators', they have fixed application tasking. Period, no appeal to higher court, alternative operating system, or adding programmability. Those of us who make a living creating such systems care not a whit whether some academic defininition by some inexperienced professor sans real world experience attempts to makes an artificial distinction on the basis of some other set of attributes. The thing boots, does it's job, and only that job. Voila--it's an embedded system processor in an embedded system doing an embedded system application[1]. Such does not mean I, having the proverbial keys to the kingdom can't go in and change or update it's tasking and such, including even substituting a PDA for a IBM PC—assuming it's good to go for that particular task. e.g. Fix an XBOX with a HDD, add a minor patch, and it makes a dandy firewall—an embedded system task. // FrankB 17:57, 22 May 2006 (UTC)
I'm quite sure that your iconic "inexperienced professor" would take issue with your creating, as you said, an artificial definition based merely upon your personal experiences. Yes I believe that artificial academic definitions are often mundane and overly pendantic, but by the same measure, industry definitions can be equally convoluted. Academia exists for arguing about definitions, anyway; let the standards organizations form committees to decide what's what while you fine fellows spend your valuable time making constructive edits to this article.
I'd say that anybody here who tries to put so fine a point on what an "embedded system" definitively is or is not is fooling only themselves. The discussion on how to define the term could easily be as exhaustive and fruitless as the ones we've seen before on how to define "computer". I think the best strategy for this sort of thing is to acknowledge that definitions vary and continue writing the article to address the classes of devices that most definitions agree upon as being embedded. It's really not worth devoting more than a sentence to PDAs to the tune of, "some classify PDAs as embedded systems while others do not". (personally I could go either way; I'd say late PDAs are less "embedded" than early-90s examples) In other words, why devote paragraphs trying to place PDAs into a strict classification system when you could just acknowledge the rather broad meaning of the term "embedded" in this context? -- uberpenguin @ 2006-05-22 22:47Z
Side note: computer-related terms are often difficult to strictly define because their meanings can change rapidly (see the discussion on Talk:Computer from awhile back). The FOLDOC entry that was cited was written in 1995; a time during which you'd rarely see a superscalar "general-purpose" microprocessor placed in an embedded application. Now that sort of thing is commonplace, which only expands the class of microcomputers that may (or may not) be considered embedded. -- uberpenguin @ 2006-05-22 22:51Z
Today, almost all PDA has operating system and add-remove application capability. We can also connect them to other larger system via it's I/O. I think Johan is right to say that PDA is not embedded-system in nature. To become part of embedded system, PDA should be set to specific function. for example as brain of video game (a game board with attached naked PDA is exists in market years ago) -- Kerdip 11:46, 27 August 2006 (UTC)

Diagnostic LEDs

Someone added:

A common scheme is to have the electronics turn off the LED(s) at reset, whereupon the software turns it on at the first opportunity, to prove that the hardware and start-up software have performed their job so far.

Actually, the smart play is to rig the hardware so that power-on reset illuminates all the LEDs and the software then starts changing things from that pattern. This serves as a lamp check and also tells you that +5 (or whatever) is present in the system. If the LEDs are all extinguished by power-on-reset but the software is stone cold dead, you can't tell whether it's missing +5, dead LEDs, or a software failure.

Atlant 15:18, 8 May 2005 (UTC)


The style of this article sucks like a vacuum-cleaner. I wish I had the time to do something about it, but it will have to wait. In the meantime, I encourage everyone to not only add info, but also to improve the style and flow. Bye, Shinobu 20:20, 26 May 2005 (UTC)

So fix it. If others had the time or desire to fix the article, it would have already been done. -- uberpenguin 18:15, 11 March 2006 (UTC)

A copy or ...?

Hmmm. Shinobu 06:01, 8 August 2005 (UTC)

Wiki content is licensed for re-use. Lots of sites mirror some or all of the article space.
Atlant 12:56, 8 August 2005 (UTC)

I know that dude... When I last looked at that link it (probably) contained a very old version of this article to the point that I couldn't tell wether we had copied the content and reworked it or something else. Since it didn't contain proper attribution (as required by GFDL) at the time I didn't know it was a mirror. These problems seem to have been fixed, so all's well. Shinobu 15:59, 8 August 2005 (UTC)

I suggest someone can use the wikibook on embedded systems to revamp the article. A few issues and basics are better dealt there.

Removed reference to using MS-DOS as an embedded OS

I have removed the reference to MS-DOS as an embedded OS because:

  • It is not used in any embedded products I am aware of.
  • It only runs on x86 processors and can not be ported
  • MS-DOS's main functionality is basically a file system, so is of little use to diskless embedded systems.

Amnonc 05:49, 17 September 2005 (UTC)

I respectfully disagree. Some low volume embedded systems use DOS on a PC104 platform with a flash disk. DOS lets people use cheap compilers (e.g. that wonderful port of GCC to DOS (DJGCC), Turbo C 1.0; etc.). The debugger is a standard remote debugger via a standard PC serial port, except you can debug it on another PC. There's an open source version of DOS now (FreeDOS). There's direct access to hardware. No moving parts, easy software installation, files, maybe even code in GW-BASIC and a screen interface. You copy the software from a PC, stick the flash into the computer, and it just works (or not... but the download is painless). User:Ray Van De Walker
There probably should be some distinction in which DOS we are speaking of... I can think of about 8 OSes labeled "DOS" off the top of my head; running on everything from IBM midranges and mainframes to the Z80 and 8080. The article currently says DOS, it should probably say something more like "IBM PC-DOS variants," since that's what most later OSes labeled "DOS" are/were. -- uberpenguin 04:28, 15 October 2005 (UTC)

Change title to "Embedded computer"

"System" is a very general term that is applicable to many fields. This article definitely talks about embedded computer systems, and I think the title should be changed to reflect that. Thoughts? -- uberpenguin 04:16, 15 October 2005 (UTC)

Yes, "system" is a very general term. However it appears that Embedded Systems Design magazine (formerly called Embedded Systems Programming magazine) only talks about systems that include some sort of CPU. Would you recommend that magazine also change its title to Embedded Computer Systems Design magazine?

What other sorts of "embedded system" are you thinking of that does not include some sort of CPU? --DavidCary 07:08, 1 November 2005 (UTC)

"Office-style" operating system?

What, exactly, is meant by an "office-style" operating system, or "office work"? The section seems to talk mostly about Linux, which is by no means restricted to office or business use, and the author seems to imply that screen+keyboard == office work.

On the other hand - if this means I can justifiably say that I'm performing office work when I have UT2004 up onscreen... Cammy 17:09, 27 October 2005 (UTC)

This might be talking about the use of WinXP etc. for ATMs and the like.

Actually this confused me too, I am of the opinion that this entire section regarding the architecture should be changed. There is no discription of a RTOS architecture (except a throw-away line and link) and I have never heard of an "office style embedded operating system" before. If the author's intention was to refer to OSes like Win CE or Embedded Linux, perhaps that should be more clear. I am not really sure if anyone has ever referred to those operating systems as "office-style" which tends to conjure up images of Office 2003 for I'm sure many people. -thomaslw 4:39 PST 2/17/06

i think that this part of the artical does not conform to the npov policy Someguyonearth 19:13, 13 April 2006 (UTC)

It mentions not wanting to use an office-style operating system for things like telephones, but versions of Linux marketed as telephone hardened systems. -Rick


Can we have a cleanup tag for this article? There's a lot which reads like a tutorial, course notes, a collection of design ideas, or as the work of one person, perhaps suitable for a wikibook rather than the encyclopedia entry. Plus lots and lots of really strange phraseology, e.g.:

  • "There are well-known tricks to construct [safe, timely, reliable] systems: e.g. Hire a real system programmer."
  • "Hardware events fail about once in a trillion times" without explanation of what that means
  • "Developers should insist on debugging which shows the high-level language, with breakpoints and single-stepping, because these features are widely available."
  • "If a design needs a screen, many designers use plain text. It's preferred because users have been reading signs for years." - it seems mobile telephone designers didn't get that memo...

Ojw 15:40, 11 March 2006 (UTC)

I'm not sure why you ask. I have added the tag for you. This article is in pretty bad shape, starting from the lead section (typos such as "electo-mechanical", repetitions etc.) --Gennaro Prota 17:10, 11 March 2006 (UTC)
"In the opinion of at least one author of this article" is a terrible thing to say in an encyclopaedia, as well! I don't really have the knowledge to fix this though. -- Mithent 12:31, 26 April 2006 (UTC)

Embedded system definition

The notions of embedded system and embedded system design have today several meanings, specific to various communities.

  • In the largest sense, an embedded systems, or embedded computing system is a computerized control system that is defined by its control function (by opposition to general-purpose computing systems). If you take a look onto embedded systems publications, you will see that nuclear plants, railway systems, or cell phones are often considered as meaningful examples.
  • Few people use today the stricter sense mentioned in the article.

Example: [2] Dpotop 11:33, 7 April 2006 (UTC)

This is Wikipedia, so you know what to do: Be bold and update the article!
Atlant 11:36, 7 April 2006 (UTC)
Just want to endorse the above point opening this section. I've made quite a few house payments designing hardware special purpose to a task but utilizing a stripped down IBM PC as a controller. Usually, on the factory floor, they have neither keyboard nor disk access, and there have been the occasional black box variant sans display monitor, utilizing the dedicated control hardware for that. Ditto applications like airport security scanning equipment. Utilizing a special purpose CPU board is frequently less cost effective than the cheap mass produced PC. FrankB 17:37, 22 May 2006 (UTC)


After scanning this article, I see no basis for the above 'In your Face' template to remain. It seems like the above talk indicates it was dealt with long ago, and no-one had the gumption to remove it. If I missed something—I'm sorry— by all means, put it back, but do so with a detailed renewed assesment of why you are doing so. The Anti-RTOS bias last September, seems a bit weak as a reason for such a damning tag. Best regards, FrankB 17:37, 22 May 2006 (UTC)

Sentence under Characteristics does not make sense=

The sentence I noticed reads : "Engineers minimizeapproach is used to select hardware that is just “good enough” to implement the necessary functions." I'm changing this to read : "Engineers typically select hardware that is just “good enough” to implement the necessary functions."