I keep two spare computers (one ISA, one PCI) all set up with galil cards and IO cards for my CNC controls.
Needed a spare last fall and the battery was dead. Got it going after waiting a week for a new battery. (It would run but PITA to reconfigure CMOS every time)
I just bought a galil card off flea bay. It arrived today. Darn, the battery is dead on this box too. I'll be able to test the galil card, but I don't like the boxes not being totally ready to go.
Do you have to keep computers plugged in and then booted regualrly to keep the battery up? Or have I just had a bad run of luck?
Depends on how old the beasts are. Some of the early 386-486 class boards had rechargeable nicad CMOS batteries, they'd go flat and leak in storage, which would ruin the board. If kept powered up, they had voltage on them and were fine. Current crop of boards uses lithium cells, really doesn't make any difference whether they're powered up or not, the cell life is basically shelf-life if powered or not. The circuitry they're attached to just doesn't draw much more than the cell's internal discharge would normally. I usually have some CR2032s sitting around for other things, so a flat CMOS cell isn't a disaster, just an irritation. And you can still boot a box with a flat cell, just have to manually set the time like with every other computer in the pre-RTC era. Make a setup list and tape it to the lid for the other parameters. And eventually the electrolytics on all the old boards go bad and nothing works anymore.
OK, makes sense. I don't want to leave them plugged in all the time. Lightening lets the smoke out every now and then. How about a cheap timer? and how many minutes a day need to be on?
Ok, I'll just get some spare batteries to have on hand. Sounds like way less trouble than keeping them plugged in.
My boxes with ISA slots look to not be replaceable. So I'm buying Galil cards with PCI and figuring on scrapping the ISA ones (nobody is buying these on eBay)
Now, I'm starting to wonder how long PCI will be around. My new office box doesn't have a PCI slot.
No, the CMOS backup batteries may discharge slightly faster if the system is not powered on, but generally it is just a matter of time. The Lithium batteries have about a 10 year life, depending on type and size, and they just quit after that time has expired.
It won't make any difference to have the system plugged in, there is no battery charging circuit on most of these systems.
PCI-X, the short little slot is the next hop I think. PCI should still be around for a fair while since a lot of server class PC platforms use PCI fibre channel adapters and whatnot. You may need to start looking to used server boxes rather than consumer PCs.
Things are normally set up so the power comes from the power supply when the unit is plugged in and powered up. (The voltage supplied is a bit higher than that from the coin cell, and with a diode added to each line, whichever has the higher voltage will power the COMOS RAM and (perhaps) clock chip).
At least you don't have the problem with older Sun computers, where the coin cell was potted into the same chip body as the clock/NVRAM chip. Under normal conditions -- these computers are usually kept powered up full time -- no problem. The longer the computer is unpowered, the sooner the cell will die -- and on those, the HOSTID and the MAC address (raw 6-octet ethernet address) are kept in there along with the optional settings. When that dies, you either need to get a new one from Sun (based on the barcode on the chip) or to have saved the information and downloaded the information off the net on how to program this into a new NVRAM/Clock chip. (The HOSTID does not matter most of the time, but if you have license keyed software, it just won't work with the wrong HOSTID.
If you want to keep the system in storage, with no power on it
-- write down *all* the important settings, then pull the coin cells and store them in an insulating envelope in the computer.
With the exception of the Toshiba NVRAM/Clock chips which Sun once used. Once they are booted in a system, a flag is set which turns on the clock, and that keeps running (and drawing a bit more than the shelf life current). Solbourne (once an alternate maker of Sun compatible systems) used a bipolar ROM chip for the hostid, and put a coin cell holder on the system board to run the clock chip and save the more transient variables. I considered this a *much* nicer way than Sun's way (otherwise I really liked the construction of the Sun machines). These days, the y have a SEEPROM chip, which does not need power to retain data, and a separate clock chip with a holder and coin cell for the TOD information. (TOD (Time Of Day) -- not RTC (Real Time Clock), which really means a constant clock tick to tell the system when to switch tasks.)
Most -- but not all. The Sun Fire 280R (and other server systems) has power full time to a board called the RSC (Remote System Control) normally used to control a system living in a machine room without having to go into that room unless you need to replace some hardware. They make a big point that you are supposed to unplug *both* power cords (separate hot-swappable power supplies) before pulling any cards out of the system.
You can even install a new OS via the RSC, and booting via the net. (But this is not PC hardware which I'm talking about, so it does not apply to you. :-)
IDE / PATA hard drives are also becoming rare. I bought a larger replacement for this laptop and kept the old one as a backup, and I label and static bag all old memory when upgrading.
I _think_ its more a matter of if the mains power is not applied the backup battery runs the clock and CMOS back up etc. Most of them are not rechargeable. So whenever the power is off the battery is holding it all up. I've seen more than once where guys have hacked AA alkaline batteries in a holder to run the clock/cmos. I imagine that they would last the shelf life of the batteries.
My first homebrew computer had CMOS static RAM with NiCad backup power, so it booted instantly. Instead of blocking diodes I used a saturated PNP power transistor in series with main power, with its base current charging the batteries. The forward drop is only about
50mV. When the power is off the base-emitter junction is reverse biased by 2.4V, comfortably less than its ~ 6V rating.
If the batteries died I had to toggle in the bootstrap loader, then hook up the Teletype to read first the boot program, then the operating system.
thanks for the link. The motherboard would be a better deal than tossing the ISA slot galil board and replacing it/ (I just paid $500 for a PCI galil card, 3K new)
My first homebrew computer had CMOS static RAM with NiCad backup power, so it booted instantly. Instead of blocking diodes I used a saturated PNP power transistor in series with main power, with its base current charging the batteries. The forward drop is only about
50mV. When the power is off the base-emitter junction is reverse biased by 2.4V, comfortably less than its ~ 6V rating.
If the batteries died I had to toggle in the bootstrap loader, then hook up the Teletype to read first the boot program, then the operating system.
I took a lot of flak from wife and friends for building rather than buying even after I explained that it was very good training for my electronics technician job. I gave it up after the PC-AT came out because I was falling behind too fast. Anyway they let me design a hard disk controller and several test and calibration stations and made me a Test Engineer at work because of the homebrew.
It isn't about battery charging; when the gizmo is plugged in, there's a STANDBY power source (always-on) that might (or might not) take the load off the battery.
Laptops frequently have rechargeable CMOS batteries, perhaps just because it's hard to open 'em for replacement.
PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.