Re: weird reboots of embedded pc
From: john jardine (john_at_jjdesigns.fsnet.co.uk)
Date: 12/12/04
- Next message: radio913_at_aol.com: "Re: BJT Can be Considered a CCCS in some situations"
- Previous message: Joerg: "Re: OT: Hyperlinks in MS-Word stuck"
- In reply to: Guy Macon: "Re: weird reboots of embedded pc"
- Next in thread: John Woodgate: "Re: weird reboots of embedded pc"
- Messages sorted by: [ date ] [ thread ]
Date: Sun, 12 Dec 2004 00:13:17 -0000
"Guy Macon" <_see.web.page_@_www.guymacon.com_> wrote in message
news:10rmadrilepen91@corp.supernews.com...
> john jardine wrote:
>
> >I'm not sure what "embedded" refers to nowadays but occasionally
> >a query will come in for some kind of industrial machine controller
> >that is easily and cheaply satisfied by use of a standard PC running
> >just DOS.
>
> While I agree 100% about DOS, I strongly disagree with your conclusion
> that a standard PC is OK for an embedded system. It simply will not
> meet any reasonable requirements for reliability or durability.
Ah! but what is a 'reasonable requirement'?. If I was knocking up mil' spec'
gear, then a PC could be a nightmare. Most applications dont' require shock,
dust, temp' and vibration. Even then, an external enclosure can solve most
environmental problems. Don't know about you but for home and work, I've
been buying and using PC's for 15 years and not one has had a hardware
failure. That's more than can be said for the fancy electronic test
equipment I use.
> >A recent request was for a turbine blade checking unit. Essentially a
> >bed-o-nails type setup using 100's of LVDTs. A PC is ideal as the LPT
port
> >can interface to the real world black box stuff. Run a DOS PowerBasic
> >control prog from power up. Calculate and display numerous visual items
and
> >store and recall data for setups, auditing ect.
> >If anything fails, the idea is that the PC is a cheap item and can be
> >easily replaced using a spare from the office down the corridor.
>
> In the system described, the "real world black box stuff" is the
> embedded system. The PC is filling the role of a desktop PC, which
> is what it is good at.
Yes, Rich also clarified this. I'd mentally deleted the word after noticing
sales people pick up on it a few years ago.
>
> >One similar system I did 5 years ago using a standard PC has been
> >running 24/7/365.
>
> Either you got lucky, or your "standard PC" was by a company such
> as Compaq which was making PCs to the high quality levels needed
> for embedded work back then (today's Compaq / HP hybrid is another
> matter). Try that with a modern asian motherboard and you are
> unlikely to see it survive to its 5th birthday.
>
> >I've no worries with hardware reliability of the commmercial PCs.
>
> You should have worries.
I dropped lucky!. That one was a chinese clone bought from the shop down the
road. But ... I enjoy a walk on the wild side, and I've pushed my luck a
few times since :-).
Years ago, a previous employer was planning to buy a PC. The company had
just been bought out and word came down from some imperial, corporate I.T.
section that we must buy a quality 'named' PC. Shelled a barrowload of money
out for a Compaq. Yes, for the money we got good quality casings,
components and circuit cards. We also got a non standard PC, that locked us
in to Compaq for the next 2 years.
Compaq had made enormous effort to enginner the PC to be just capable of
running mainstream PC progs' yet not allow any part of it to be upgraded
without having to buy expensive and inferior Compaq parts. We ditched it a
the first opportunity. Lesson learnt for the future.
>
> >This path has recently been nipped off, as PCs no longer have
> >LPTs and ISA sockets. I must look to the 'industrial PCs' on
> >offer. Eg PC104 and upwards
>
> That is the wise choice. Have you used a scope to look at the
> noise and timing margins on the motherboards that are shipping
> nowdays?
Yes I have. The logic is so noisy the signals are invisible.
I'm in a minority here but I prefer to use equipment that has withstood the
major tests of time and mass production.
Doesn't matter how cheap it appears, or how few bits it uses, or how noisy
the electrics are. I know it'll work and carry on working.
Like my HP printer is absolute crap quality but just runs and runs. Yet
those beautifully engineered IBM golfballs and line printers used to break
down weekly.
>
> >and what their software setups and special requirements are.
>
> If your PC104 has any special requirements, then something
> is wrong. It should act just like any other PC.
>
> >Problem is, they are soooo damned expensive
>
> $300 will get you a fine 386SX PC104 stack. That's three
> hours of my time at my standard consulting rate.
Last time I looked (UK) they were over £300. Hopefully the prices are
dropping.
I envy being able to charge an hourly rate thing :-). Round here the buggers
won't stand for it. They demand a fixed price quote for all jobs, start to
finish. Tricky but can have its upsides.
> >and sooooo specialist and not particularly standard.
>
> If your PC104 isn't standard, then something is wrong.
> It should act just like any other PC.
I agree yes it should. I'm jaundiced though by the knowledge that given even
a remote chance, any specialist supplier will (dutifully!) try and lock a
user in. Last thing they want is to compete only on price. The earlier
Compaq was advertised and sold as standard, but wasn't.
>
> >The customer would be locked into a particular
> >industrial PC supplier from day one.
>
> If your PC104 locks you in, then something is wrong.
> You should be able to replace it with any PC104 board.
>
> >A bit like the risks involved in basing a companys product
> >range around one specialist IC.
>
> I have no idea what you are getting at here. PC104s use standard
> parts such as Intel 386SX or 386EX - parts that Intel will be
> cranking out long after they stop making the latest Pentium X.
>
It's this non-standard-standard thing again. I've no choice now but what I
didn't want, is to buy some companies industrial PC type gear and then
discover much later down the line that the makers by clever engineering have
subtlely managed to pull a fast one.
I never ever, want to be in the disasterous position that posters here some
times indicate. eg "Having customer returns problems when using an Acme
Mk4/b2 message translator unit being driven by a Putzin model 1234 modem,
fed from a Yanking EZ USB driver. The USB driver company went bust and am
having to use so and so's later version but it don't seem to work the same".
regards
john
- Next message: radio913_at_aol.com: "Re: BJT Can be Considered a CCCS in some situations"
- Previous message: Joerg: "Re: OT: Hyperlinks in MS-Word stuck"
- In reply to: Guy Macon: "Re: weird reboots of embedded pc"
- Next in thread: John Woodgate: "Re: weird reboots of embedded pc"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|