First computer with real-time clock?

I don't rememebr a clock card for the '11 -- there were line-time and programmable interrupt cards that gave a periodic 'heatbeat', and that's what DEC normally regarded as a real time clock... Looking back through my manuals, I've found a couple of somewhat interesting ones : There was one of the P851 (1978-ish) that was a battery-backed digital clock card (which either counted up to 100 days or 10000 hours in 1s increments). You could even hang an LED display off it if you wanted a human-readable clock in your machien room. I have schematics, but alas not the card (I do have a P851...) There was a real-time clock model (98035) for the HP9825 'calculator' (really a computer!). It actually contained 2 clocks. When the machine was powered up, it used an internal microcontroller clocked at a known rate. When the machine was powered down, there was a battery-backed chip (looks to be a repackaged chip from a digital watch) that kept the time, and which was read by the microocontroller at power-up (the interface to this chip was strange -- a couple of lines corresponding the the 'buttons' on an LED watch, 3 digit stobes, and the 8 lines for a 7 segment display (7 segments + annunciators for things like the leading '1' on a 12 hour/US date display, that's how they got away with only 3 digit lines). Almost certainly not the first RTCs on computers, though. If you go back to data loggers, the Solartron DTU (1960's vintage) has a clock card. This was a rack of mostly DTL logic that connected to a DVM or DFM and to an output device like a Facit 4070 punch or an ASR33 teletype. Even so often (the rate could be set using a switch on the clock card, or it could respond to an external contact-closure input), it would scan up to 20 channels (using relays in the DTU), connect each one to the input of the DVM, and record the result on the output device, along with the time. Thsi unit was sort-of a forerunner of lab minicomputers, in that it let you take results automatically and feed them (via paper tape or punched cards) into a mini or mainframe computer I also have a much later Fluke logger, controller by a pair of 4040s. It has a real time clock built in (periodic interrupt to the main processor IIRC) which goes up to 1000 days in second increments. Several of the early RTCs keep time in hours/minutes/seconds as you'd expect, but ject count days up from 0-99 or 0-999 or whatever. This gets round the stupid features of the calendar (and incidentally avoided the Y2K bug :-)) -tony