Well its 35 years ago but I'll have a go. The store was made up of planes. Each was (I think) 64x64 cores. Wires ran Horizontally and Vertically through each core plus a sense wire that snaked through all the cores diagonally from one corner of the plane to the other. The cores could be magnetized in one direction or the other. It was driven by current pulses of differing magnitude depending on reading or writing. Reads came out on the sense wire as serial pulses of differing polarity to indicate 0 or 1 and you had to write the answer back each time. We had a test bench with power supplies and test boxes to write then read the store. The setting up consisted of setting the shape of the pulse amplifers output to a drawing we had. If it ran the read/write pattern ok for an hour it was a good one. If you had good electronics (ie a working board set and pulses present at the core box input)and no output you checked the date of manufacture of the core box. If it was still in warranty you went to he stores with it and said "Another one to go back to Mullard" and got a new one. If it was out of warranty you threw it out of the last window on the left (That's where the skip was) as you left the building to go and get a new one from the stores. -----Original Message----- From: cctech-bounces at classiccmp.org [mailto:cctech-bounces at classiccmp.org] On Behalf Of Roger Holmes Sent: 16 May 2007 23:28 To: cctalk at classiccmp.org Subject: Re: Harwell (was RE: A local computer history group for my area. ..) Maybe it had gone by then, but there are photos on the web of the building when it housed a Ferranti Atlas. Specifically: http://www.chilton-computing.org.uk/gallery/harwell/ slide8.htm or the view the whole collection of photos at http://www.chilton-computing.org.uk/gallery/home.htm What was involved in refurbishment? My 1301 has five core stores, each of 400(decimal) words x 50 bits (48 data + 2 parity). Three of the stores work perfectly after they have warmed up, but two of the stores are totally dead. We've tried everything, starting with the obvious - looking for a fault in the selection logic, but after replacing the same boards about five times with no effect (especially when the 'scope also says they're working properly), we had to started looking for a fault in each of the stores. But everything there looks OK too, at least measuring voltages they look OK, I don't possess a current probe and I'm not sure if I had one I'd be able to use it.