Date: Sun, 4 Feb 96 13:25 WET From: uhclem@nemesis.lonestar.org (Frank Durda IV) To: hackers@freebsd.org Cc: uhclem@nemesis.lonestar.org Subject: Re: Watchdog timer Message-ID: <m0tjA3e-000CN8C@nemesis.lonestar.org>
next in thread | raw e-mail | index | archive | help
[14]Terry Lambert writes... [14]** ISA MUST DIE. THIS IS NON-NEGOTIABLE. ** I have been reading the "build a watchdog timer board" thread with detached interest, and I thought this statement was extremely funny, and made me decide to respond. By going to PCI, you will ADD for a small-volume board (<1K units) roughly $40 per board JUST for the PCI, both for fabrication, electronics and mechanical packaging. Those PCI chipsets aren't cheap in low volumes, and the packges they come in eliminate do-it-yourself manufacturing. Also, if you want to go commercial and get a FCC Part 15 approval (technically you are supposed to do this for any volume of digital boards if they are sold as something other than educational, experimental or "kits"), a PCI board will cost even more to get certified since not all certification shops handle PCI and so there is less competition. If you go back to the original task of this "device", which was to somehow monitor the lifesigns of a PC and if there are none for a certain amount of time, RESET the machine, and perhaps if that doesn't work, cycle power on the machine too, a board (even an external box) could be built in small volumes (100 to 500 pieces) to do this job and sold for a reasonable return for around $30US. ($50 for power cycling) If you wanted a board/box that could be optionally stuffed to monitor up to three other computers and reset them independently, then you are probably talking about a $60US box. ($100 for power cycling) As an example for a simple system: in the external model, you would simply provide a chassis spine (those metal things on the ends of cards) with two plugs. Inside, you would connect one cable to the RESET connector on the MLB, and the other to either the SCSI activity light or the IDE activity light connector on the MLB or SCSI/IDE card. Externally, the cables (shielded) would be run to the monitoring box, where a couple of properly tuned NE555 timers would do an excellent job of monitoring dead time by looking for disk access. To ensure delayed writes idle times, and other stuff don't nail us, the system would be required to run a daemon that wakes up every so often and reads one block from a raw device on the SCSI/IDE being monitored. Such a simple external device would probably be exempt from Part 15 testing because it has no oscillator. Power would come from a wall adapter. If one timer expires and generates a RESET and things go dead again, a second and much longer timer (five to ten minutes) could cycle a relay and kill power to the computer chassis for ten seconds or so, and then let things reboot. In an internal board that did the same thing (minus the hard power cycle which must be done externally for UL/CSA approval), you would save on cable costs external box costs, and you would probably just stick the board somewhere inside, ie, not in a PC slot since it doesn't need access to the bus. Power would come via one of the drive connectors. Now, if you start getting fancy and want a ISA card that you can program how long "dead" is with something fancier than a potentiometer and have it accurate down to the microsecond, then your costs start going up and you are obliged to obtain FCC certification since you now have a clock present in your design. And if you don't want it stopped and seized at foreign customs, don't forget the international approvals, CSA, VDE, CISPR, etc. I have designed, prototyped and built several different ISA and non-ISA boards that became retail products for various things over the years and so I am fully aware of the issues involved. It's messy, expensive and slow. My point is that you should really decide exactly what is needed and what is nice and try to control the leaping elegance demonstrated in this thread. I can't believe anyone would want to waste a precious PCI slot on something like this, but then I guess it depends on how much other stuff you are willing to add (and pay for) on the board. You also limit your potential buying audience for what is already a low-volume device, since a lot of new machines only come with one or two PCI slots, and it seems we can demonstrate that a lot of FreeBSD users (and potential buyers of this gadget) don't have PCI slots at all or they are full. I'd be happy to create a design (analog or digital) and even prototype one of these "whatever it becomes" boards, and even know someone who does PCB layout who has done contract work for me in the past, but all of these "add-ons" are really going to drive up the cost, increase the complexity, and delay delivery of the gadget. If it is simple and affordable enough, I would even handle having the boards made, stuffed, and retailed since I have channels for doing that. But I have to see a design that has plenty of interest first, since past experience has shown me that only 10% of the people who say they really want something actually buy it. Frank Durda IV <uhclem@nemesis.lonestar.org>|"The Knights who say "LETNi" or uhclem%nemesis@rwsystr.nkn.net | demand... A SEGMENT REGISTER!!!" ^------(this is the fastest route)|"A what?" or ...letni!rwsys!nemesis!uhclem |"LETNi! LETNi! LETNi!" - 1983
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?m0tjA3e-000CN8C>