How to choose a firmware partner

[...CREDIT story...]

Coincidentally, the paper I was writing was sort of a farewell letter, describing lessons learned and possible future directions the division I was working for could take to improve products and the development process. I was leaving for a new job in another division of the same company, where I would be using EMACS under VAX/VMS.

As did ours. The system disk was in the first drive, and the data disk in the second. It's been far too long for me to remember the actual command and file syntax, but to use an MS-DOS analog, I invoked the "save file" function, was prompted for a filename, and I entered "a:memo.txt" or perhaps just "memo.txt" when I meant "b:memo.txt". IIRC, ISIS used numbers rather than letters for the drive specifier, but you get the idea...

Regards,

-=Dave

--
Change is inevitable, progress is not.

Roman soldiers sounded off in turn while pulling guard duty. If the voice from the east wall didn't happen, the entire regiment would be woken up to investigate. That system was a watchdog timer.

--
Guy Macon, Electronics Engineer & Project Manager for hire. 
Remember Doc Brown from the _Back to the Future_ movies? Do you 
have an "impossible" engineering project that only someone like 
Doc Brown can solve?  My resume is at http://www.guymacon.com/

I have never even considered of using a watchdog timer, but might reconsider, if I had to design something for high radiation (space) environment with SEUs etc.

In any safety critical systems, I would not even consider to use any processor based (or even sequential logic based) systems, but instead rely on mechanical, pneumatic or hydraulic systems.

I have not been working with hardware recently, but by impression was that the CE requirements required that the device should withstand a few kV charged into a few nF capacitor and then discharged at the metalwork _and_ the input and output pins.

In addition to pin protection, this also requires that you pay attention to ground plane design.

Paul

Yes.

-Hershel

Sorry, I'm tired. I'll let someone else point out why you're not entirely wrong, and why watchdogs are indispensable anyway.

Steve (who has been working with hardware recently)

A full megabyte....in the 60's?....that's pretty big. I programmed on WWMCCS GE/Honeywell mainframes in the early 80's that didn't have a full megabyte of magnetic core memory.

I would be interested (and surprised) to know which computer had a megabyte of core in the 60's. see

to help "refresh your memory" (bad pun). I worked on Elliot 920B's (not on the list) perhaps because it was military??

Phil

TROLL ALERT

Stop, please! I'm getting all misty eyed.

My first job out of university was with a small custom microelectronics company developing embedded systems for the oil and gas industry. We used the MDS but we only had one so we had to write the code on paper and our secretary typed it in for us. She became so used to this that she did the first level syntax check.

We were also using PL/M as a programming language, 1702 EPROMs or battery-backed memory we used to load from the MDS and carried to the lab.

I loved it when we could afford to get a PDP-11 with VT100 terminals. The editor was one of the best I've ever used.

Sniff, sniff...

Oh, and we were using watchdog timers in the mid-70s. Can't imagine a system without one or something similar, unless you're talking sync-matched processors like the old Northern Telecom DMS-100s I used to program for.

...Tom

"Phil" wrote

on

full

I would like to know as well. I worked on Honeywell 6000's and when I say early 80's, I mean July of 1980 and on. IIRC (and I think I do ;-) they had about 128K words of core, and each word was 36 bits. They really didn't use "bytes" allot, as the native character set was BCD (6 bits/char). They also had ASCII support (sort of ;-) where there were 4 "must be zero" bits per word. Bits were numbered from left to right, IOW Bit 0 was the MSB. The first 64K words was the "hard core monitor" or HCM and was the nucleus of the OS. The machine was an evolution of the GE 635 which was a bit simpler, and before my time. I could go on, but I'll stop before I bore you to death. I really miss that architecture from a low level point of view, it was a real beauty in it's own right. :-( For real-time response and transaction processing they always kicked IBM's butt. Shows the power of a good marketing program.

If you write 100% correct code on a 100% correct platform, you don't need it. But the more complex an application gets the more likely it is that there might be a race-condition not foreseen. (Or do _you_ test your system with __all__ possible input values ? I bet no !)

And I'd rather see nuclear-powerplant computer reboot than staying in an endless loop :-)

--
42Bastian
Do not email to bastian42@yahoo.com, it's a spam-only account :-)
Use @epost.de instead !

And it is even not an invention of electronic-age.

- railway-signals used to have something alike

- dead-man button in trains

--
42Bastian
Do not email to bastian42@yahoo.com, it's a spam-only account :-)
Use @epost.de instead !

Fine. That is why I have to power-cycle my settop box from time to time. If they only had build in a WDT, I would not have to climb behind my TV (of course the settop box has no switch, also a new _invention_).

--
42Bastian
Do not email to bastian42@yahoo.com, it's a spam-only account :-)
Use @epost.de instead !

Can't agree here. GNU ok. PC ok. But for embedded debugging I really appreciate a full blown emulator (of course these are dieing).

--
42Bastian
Do not email to bastian42@yahoo.com, it's a spam-only account :-)
Use @epost.de instead !

Damn, I got trapped :-)

--
42Bastian
Do not email to bastian42@yahoo.com, it's a spam-only account :-)
Use @epost.de instead !

In the very late '60s I worked on a UNIVAC 1110 which had that much memory but I can't say now if it was core or semi.

Norm

So this may seem (a troll), to some people, but won't anyone even consider this for a moment:

1) A one byte loop is lockup-proof.

2) So for a standalone, MCU, lockup proof code is only a matter of scale.

3) Evidently the MCU hardware itself cannot become inherently "stuck" because that would make the clearWDT instructions inaccessible and a mockery of the whole watchdog scheme.

I am guessing that the above posters are talking of complicated, multiproccessor, asynchronous systems; which is a little narrow minded.

A better response could be: "The OP probably means 1/2k of code running in a 12C505, heh heh, he'll grow up eventually!"

Cheers Robin

In the 60's, it must have been core.

In the mid-70's, the PDP-11/70 physical address space was 4 MB, but with original DEC core modules, two full size cabinets would be needed to hold all those core modules. However, connecting all these would have violated the memory bus length limitation. Thus, with core memory, the maximum memory capacity would have been IIRC 1 or 2 MB.

In 1977 or 1978 Intel sold semiconductor memory modules built around

8 kbit x 1 DRAMs. With these modules 4 MB would fit into one cabinet and the memory bus length was within specification.

In those days the largest DRAMs Intel made was 16 kbit x 1, so apparently they used those 16 kbit DRAMs with a defect in one half of the memory plane and used as 8 kbit devices in their own modules.

In those days DRAMs suffered from soft errors, apparently due to alpha particles emitted by the plastic package, thus the modules had ECC instead of parity, which was the norm for core memories in those days. Operators manually logged the error status registers on the module before resetting the counters. Frequent failures in a particular chip required replacing it sooner or later.

While computer systems with 1 MB of memory could have been done in

1975/76 using 4 kbit x 1 (requiring 2000-2500 DRAMs), but I doubt that making 1 MB with 1kbit x 1 SRAMs (requiring 8000-10000 chips) would have been very practical due to the high power consumption.

Paul

1n 1965 The Sperry Rand UNIVAC 1108 II (aka 1108A) had two memory cabinets with 262,000 64 bit words each, which adds up to nearly 2MB.

Are you sure it wasn't an 1108 or 1106? The 1110 was introduced on November 10, 1970. (Many web pages say 1972, but they are wrong.)

The 1110 had a full megaword of memory, which is over four megabytes (36 bit words). It also had plated wire memory in addition to the core. In 1976 the 1110/40 came out, which was an 1110 with a megaword of (faster) RAM chips instead of core.

--
Guy Macon, Electronics Engineer & Project Manager for hire. 
Remember Doc Brown from the _Back to the Future_ movies? Do you 
have an "impossible" engineering project that only someone like 
Doc Brown can solve?  My resume is at http://www.guymacon.com/