From: Rick Bensene
Email: oldcalcs@bensene.com
Hi, I have done some reverse engineering on this machine, though I don't have schematics. I did, though, reverse engineer a schematic for the row/column driver circuits for the core memory. It is hand written, but I could scan it and EMail to you if you wish. When I received the machine that I have, it was malfunctioning, but the machine is of a pretty straightforward design, and I was able to track down the faults and repair them. In general, the most common failure with these old core memory-based calculators is problems in the core memory system itself. Since the core memory is the heart of the serial architectures of these machines, if there is any kind of problem with the core memory, the machine won't function properly. Core memory systems are pretty straightforward...there are the core row and column drivers, the sense amplifiers, and the inhibit drivers. The common failures are in the drivers (row/column/inhibit). It's a pretty simple matter to figure out which boards are the driver boards, and them to simply test out the transistors in each driver in-circuit using a good VOM, or an in-circuit transistor tested. Once bad transistors are identified, you can generally find suitable current-day replacements, and most of the time, that'll fix the problem. Most of the calculators of this era that I've come across that are malfunctioning can be fixed by tracking down the defective core drivers, and fixing them. Less frequently there are problems with the sense amplifiers...a noisy transistor, or power-supply ripple that causes problems with the sense amplifiers. The key thing to check first is for ripple on all of the power supply lines. This can only be done with an oscilloscope. Frequently on these old machines, the power supply filtering capacitors have malformed due to age, which reduces their ability to filter out AC. The resulting ripple on the DC supply rails will freak out the sensitive amplifiers used to catch the sense signal from the core, not to mention cause chaos with flip flops. Unfortunately, I don't have the information on the supply voltages for the IME86 with me. Generally transistor machine run at a fairly symmetrical bipolar supply in the range of +/- 10 to 15V DC. Nixie tubes run at between +150 and 220V DC (watch out...they'll bite you pretty good!). Core memory drive systems utilize constant current sources (the voltage isn't as important, but generally ranges from around 6V to 18V depending on the characteristics of the core) that are either independently generated, or are derived from the main logic rails. Backtracing the power rails for the row/column/inhibit drivers can allow you to identify where the current source is, and you can check to make sure that the current source is delivering clean, ripple-free power to the driver transistors. As I recall, in the IME 86, the core array itself is hard wired into the chassis. Located in slots adjacent to the (fixed) core card are a number of cards that contain the drivers and sense amplifiers. The syptom that I noted on my machine when there were bad drivers was that some digits wouldn't 'latch in' when a key was pressed. The digit would very briefly flicker into the display, then the digit would revert back to zero, though some other digits would accept entries. The way to test this is to use the .> and .< keys to move the decimal point around, and then enter a digit. Figuring out which digits take hold of an entry can give clues as to which core drivers are bad. Faulty inhibit drivers can have more far-reaching effects, as they'll manifest themselves as a whole 'bit' being dead, e.g. if you enter a 5, you might get a 4 if the 2^0 inhibit driver is bad. Please feel free to write me with details of how your machine is behaving and I might be able to offer more than just generalities as to how to repair. Best of luck, Rick Bensene The Old Calcualtor Web Museum http://www.geocities.com/oldcalculators
[arch3/_private/calc1_aftr.htm]