Date: Mon, 08 Sep 1997 15:54:56 +1000 From: Mike Smith <mike@smith.net.au> To: "Jamil J. Weatherbee" <jamil@counterintelligence.ml.org> Cc: hackers@FreeBSD.ORG Subject: Re: Speaking of device drivers. Message-ID: <199709080554.PAA01633@word.smith.net.au> In-Reply-To: Your message of "Sun, 07 Sep 1997 23:13:24 MST." <Pine.BSF.3.96.970907225525.1391B-100000@counterintelligence.ml.org>
next in thread | previous in thread | raw e-mail | index | archive | help
> > > > Sounds like a lot of overkill, IMHO. This isn't the sort of thing you > > want in an industrial environment. > > I didn't really have a word in this respect, I am just supposed to deliver > a working software--- people put hardware together demonstrate that it > functions correctly and then expect it all linked. Yuck. I hope you are billing the crap out of these people. Unless there's something really compelling about having multiple machines, it's basically a losing architecture. > > >From my POV I would be using one or more RS-485 links and either custom > > slaves at suitable points or off-the-shelf 485 slaves like the > > Advantech ADAM modules. The only time this breaks down is when you > > have precise timing requirements between multiple slaves, and often the > > easiest way to go then is to have a separate transmit-only time-sync > > bus. > > I actually have looked into this I've done some significant programming > on a 68hc11 board with RS485 ports We use these a *lot*. > also I am working on an unrelated > solutions to another problem on the 68hc11KA4 a truly versatile processor We mostly use the HC811E2/711E9/711E12; the '2 is an all-EEPROM part and thus great for field upgrades, the E9 is easy to get, and the E12 has loads of ROM space. > very hard to get for your run of the mill embedded solutions), in fact if > it was up to me I would of found a multiport rs485 board (i think > Industrial computer source sells some of these) and done the whole thing > with a network of 68hc11's (actually you can implement rs485 stuff on a > bus which is really neat -- and next time I design this kind of stuff I'll > keep that In mind). Heh. I have a driver for stock '485 boards (we use the Advantech PCL-740) which does the 9-bit address-prefix mode, which I keep meaning to commit. As you might guess from this, we use '485 pretty heavily. > One of the major problems with the networked embedded > design was analog ports, I have 48 12 bit analog ports and I could not > find any embedded hardware capable of sampling 256 ports at that > resolution (also 10,000 samples/second) , so I am using a analog io card > in a remotely connected > machine, plus the processing overhead for the kind of operations being > done are somewhat significant. Is that 10Ksamples total, or 10K per channel? For that sort of situation, I'd go ahead with the card-in-PC approach, but that's still just one PC. > sorry about the run on > sentences I have written that way since I was 8 years old in really long > sentences without periods or punctuation I guess just wanted to be Cognitive dissonance is all well and good, but remember that your goal here is to communicate, not to piss other people off. > In other words I needed a little more power than I could of gotten out of > some embedded hc11 system for what I am doing Not just one embedded system, but lots of them. Parallel processing 8) Speaking of which, does anyone remember the old Byte daisy-chain Mandelbrot processor? mike
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199709080554.PAA01633>