It is a composite chip. Basically a static ram with an oscillator and a lithium battery piggy-backed on top.
The PC won't work at all without the static ram and oscillator so unless you can remove the chip and cut or grind off the top-hat holding the battery and somehow glue a replacement,connecting to the internal chip connections, you are stuffed.
The PC will work without it - all the oscillator is doing is keeping the RTC running. The Static RAM is for holding the CMOS settings. (This is nothing to do with the CPUs own oscillator, or main memory.
The difficulty is having the PC retain its HDD settings between power cycles without it. It will be happy to live without a RTC (as did all early PCs anyway, they simply included a call to the time and date programs in autoexec.bat for the user to set the time at startup).
It's just something I've read. Maybe there is something in DOS that relies upon part of those PCs that requires a clock tick that is provided by the RTC chip (it has a programmable square-wave output pin). Presumably the software on the manufacturer's boot disk doesn't use DOS and doesn't rely on anything needing this tick.
SteveW
The 'static ram' in this case being the 70 odd spare registers in the RTC itself (cmos ram because the RTC, as per every quartz controlled electronic timepiece which took advantage of the extremely low power consumption of cmos, was cmos to start with). The Dallas RTC was simply a slightly modified RTC chip with a couple of tiny lithium coin cells strapped onto its top, along with a 32768Hz quartz crystal embedded into an extra layer of potting compound laid on top.
With care, it's quite possible to dig away the potting compound to dig out the two tiny lithium cells to make way for a new set. Just look it up on t'internet to find out where to start digging away at the top of the package - hint: aim for the lithium cells, they're expendable and you can damage them all you like when extracting them.
I'm not sure how the POST would handle the absence of the IOByte switch and all the other missing setup 'switches' represented by the RTC's registers. It may or may not boot up (or, at best, only boot up from a floppy disk plugged into the A connector on the end of the 34 wire ribbon cable plugged into the MoBo's 34 pin FDD header, assume the local keyboard as the STDIN device and a VGA adapter as the STDOUT device). It's not an experiment I recall running with any of my 286/386/486 AT clone systems. :-)
Luckily, even if that were true (I can't say for certain), it's not at all difficult to perform the minor surgery required to do a 'battery' transplant operation on the Dallas chip.
What was special about the Dallas chip was that it had been designed to use two separate lithium cells for redundancy against the risk of a 'faulty from new' cell being used. The cells weren't just wired in parallel to create a larger capacity 'cell' out of two cells small enough to be potted onto the top of the RTC chip as you might have been tempted to conclude. The chip monitored and used them independently of each other to maximise usable 'battery life'.
The other way round actually. By default MSDOS would *always* prompt the user to set the date and time if there *wasn't* an autoexec.bat file present in the root of the boot drive. Whenever there was an autoexec.bat file present, MSDOS would stop bothering the user to enter the date and time.
On even more ancient PC and XT hardware sans any adapters sporting an RTC chip, if such manual entry of date and time was required of the user, the DATE and TIME commands could be added to the autoexec.bat script. However, on upgraded PCs and XTs, there could be as many as three RTCs to choose from[1] (multi IO adapter, Graphics adapter, serial port adapter, parallel port adapter, memory expansion adapter and so on).
Each adapters' RTC were supplied with a driver/setup program which could be invoked from the autoexec.bat file with regard to supplying the date and time inputs to the dos date and time commands from the RTC's own date and time registers.
As for brand name computers like Elonex and Compaq in particular, compared to a home built AT class 80286 or 80386 machine made up with a clone MoBo, they were so fekin' retro with that cmos setup boot floppy nonsense, it was almost unbelievable.
Compaq in particular persisted with this antiquated system for well over a decade after the clone Mobo manufacturers had included the cmos setup program as a part of the main bios which could be invoked by simply pressing the DEL key during the POST. Proprietary practices at their worst. Why anyone would put up with such nonsense when they could build a better system quite easily with industry standard parts for less money could have only been down to customer ignorance.
[1] I landed up in just this very situation with my second home built PC (XT in this case) as a result of having RTCs on three of the 4 or 5 installed adapter cards. It was, as had been true of my very first IBM PC, a system built from 2nd hand parts some time around 1983/84 as best as I recall.
AFAIAA DOS does not require a RTC chip at all - (early PCs did not even have a RTC, which was not introduced until the PC-AT)
Yup, the system would be missing lots of details about configuration like base addresses of parallel and serial ports and details of the display adaptor etc. As long as the installed hardware is fairly stock, you should be able to get a boot from a floppy.
Unless you included them in the autoexec.bat (as most default startups did on machines without a RTC) - and you typically wanted the autoexec etc to set a code page and keymap etc. (at least for us this side of the pond)
IBM took similar nonsense a stage further on the MCA PS/2 machines with their hardware parameter floppies to configure expansion cards. Basically without the disk the add in cards could be impossible to configure.
Although I did lots of development on PCs from the late 80s onward, and used to fix and upgrade other people's - I think I managed to avoid actually buying my own til the early 90s (was happy with my Amiga 2000!).
IIRC The first PC I bought was a 33MHz 486DX, 4MB ram, Trident 9600 VGA and a 200MB HDD (and that disk was s special order since it was so obscenely large!), so that should date it...
====snip====
I'd guess at late 91, early 92 since I upgraded to a 486SX with a decently sized 200MB HDD around the middle of 93. Thinking about it now, that was only ten times the size of my very first (ST225 ?) HDD around
85/86 or so (as per usual, a 2nd hand drive), after an interval of 7 or 8 years which seems a long time for drive capacity to increase by just one order of magnitude.It's all a far cry from the 6TB and larger drives we have today. I'm seriously considering replacing one of the two 4TB drives running alongside of the 6TB drive in my NAS4Free box to an 8 or 10 TB 'monster' some time in the next 6 to 12 months. I'm down to less than 750GB of free space in the NAS right now and in danger of running out unless I start culling or archiving off the less interesting TV programmes I've recorded over the past 13 years.
I've also got a 3TB drive still physically mounted in the NAS box which I retired several months ago and disconnected to reduce energy consumption. I was expecting to save 7 watts or more but only saw a 3 or
4 watt reduction so I'm seriously thinking of just adding the new drive to bring the drive count back up to four and hang the 3 or 4 watt increase. It's so much simpler to just throw extra capacity at the problem of dwindling free disk space than mess about with culling/ archiving and/or moving the data off the hotter of the two 4TB drives onto its larger capacity replacement. :-)That 200MB HDD isn't even big enough to hold the NAS4Free image which now needs a 4GB SDHC card to allow the image updating process to function. There was a time only some 7 or 8 years ago when it only needed a 64MB SD card to hold the very early FreeNAS boot images. The flash memory size requirement just kept creeping up, first to 128MB, then 256,
512, 1GB and then 2GB and now 4GB which only has a 1905MB partition on it.Oh, how things have changed beyond all recognition in just the past 35 years of IT development. I might eventually get back to fixing up my Transam Tuscan S100 bus computer for the sanity of 8bit retro computing when it was still possible to understand how all the hardware and software actually functioned. :-)
Amen to that!
Indeed. Its npt te TRTC clock that is teh issue: Its the battery backed memory to hold not Dos but BIOS defaults.
Without *those* the system may well not know what to boot from.
Removing the chip entirely will trash that. It may nb[ot boot at all, but a flat battery simply means go into the Bios and set it up on power on before it will boot.
EISA bus ? Had a clever backward physical compatibility with ISA, but doubled the number of connectors.
Which is what I said in the message before...
If you have a 360K FDD in the drive A position then there is a resonable chance it will boot from that even without the CMOS data. Anything non standard will pose more of a problem.
Not all machines of that era can "go into the BIOS" - there was not always a built in bios configuration screen. The basisc configuration was done by jumpers. With some of the more esoteric settings you needed a system board specific utility that was run after boot.
I'd forgotten that. The Apricot Qi (an almost clone of the PS/2) was similar. IIRC you needed the correct floppy that came with each card?
What a ballsache.
No, MCA
Ah, of course - IBMs (doomed) attempt to wrest back control of the PC market (they never believed in) by locking down the MCA. I think I can be forgiven for it slipping my mind.
I really liked EISA - it seemed to solve a lot of problems. But I guess the complexity was too much.
Maybe you could empty your 'Sent Items' folder :-P
ISTR you could combine multiple .ADF (adapter description files) onto a single floppy as cards used an ID (I suppose the fore-runner of PCI and USB VID/DID numbers).
You may well be right. It's a long time ago now, and I only worked an a small handful of MCA machines, thankfully. I do recall an MCA-based Apricot server that had a built-in UPS- the size of a small filing cabinet.
*googles*The VX FT.
It also had to compete with quick and dirty solutions like VESA bus that were "good enough" for most video card applications, which was where the real bottleneck was...
A lot of the RM Nimbus machines are not actually PC compatible - they run MSDOS and Windows but it has to be the customised RM versions.
Mike
MCA was overpriced crap.
I ran a CAD/CAM (AutoCAD/Pathtrace) suite of 2 386 25Mhz clones, 5 Tandon 486SLII (486 33Mhz) and 1 IBM model 70 20 MHz.
Network cards cost £25 for the ISA bus. £125 for the MCA IBM. It was the only one that dropped packets. I had set the ISA bus to 12MHz on the PC clones and 16 MHz on the Tandons.
HomeOwnersHub 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.