29-10-2015, 01:44 PM
Well this was an interesting one, a ZX Spectrum 4S board, in which nothing was working whatsoever. Some research found the 4S motherboard was an oddity, but the internals closely matched the 4a revision so a circuit diagram was printed off the house server archive and off to work we went.
Initial testing found no +5 volt supply rail and some metering revealed the 7805 regulator was shot. I keep a good back supply of the more usable items in stock and a new reg was a five minute search in the storage box. Power back on and the +5 rail was back up but still the spec did not want to boot.
A nasty issue with the lower 16k memory chips, type 4116, was their tendency to internally short out. There are some good quick fixes for this, Grant Searle over at ZiggNet does a complete replacement module that fits into the space vacated by the removal of the duff chips. I'm somewhat traditional and replacement 4116s were sourced for a £10 in postage from a Euro supplier, quite a good deal considering their rarity. What happens is one or more 4116 internally shorts and blows the entire voltage convertor stage, and in this case the 7805 with it. All of the 4116s were removed and 16 pin turned sockets fitted ready for memory replacement.
With the old memory out and no new chips fitted, TR4 was removed, tested and found wanting, a replacement ZTX651 was fitted. TR5 was tested and found to be internally short circuited. TR5 when shorted turns TR4 hard on and both quite happily cook with copious amounts of smoke. The bobbin transformer was removed, tested ok, and refitted. Power up and zilch, but TR4 started getting very warm with +19v on the +12v rail so a quick power off and a sit down with a meter and a diag to analyse what was wrong.
In the end it turned out to be quite simple, over the course of revisions the pad positions of TR4 and 5 have changed. The Mk1 and 2 motherboards used the standard CBE arrangement but with the use on later boards of 2SC series transistors the pad positions were changed to BCE on one, then both transistors. Mental note to self, just because your used to an arrangement, don't get complacent. Both transistors were re-soldered correctly into place, power up revealed the correct voltages on the +12 rail but nothing on the -5v rail. TR4 is part of a saw tooth oscillator with the bottom third of the wave dipping below the virtual 0v rail, this - dip provides the drive for diodes D10 and D11, both in this case being 1N4148 types backed up by a 5.1v zener. Both 1Ns were found to be blown and replacement of both restored the -5 volt rail to full working order.
We now had the correct +5v, +12v and -5v supplies ready for fitment of the eight 4116 memory chips. Power up revealed one other issue in that the ULA chip was unhappy and would require replacement once one was sourced but other than that a satisfying repair. This repair was a tough one and the first time I have had to delve into the intricacies of the Spectrum internal power supply.
Initial testing found no +5 volt supply rail and some metering revealed the 7805 regulator was shot. I keep a good back supply of the more usable items in stock and a new reg was a five minute search in the storage box. Power back on and the +5 rail was back up but still the spec did not want to boot.
A nasty issue with the lower 16k memory chips, type 4116, was their tendency to internally short out. There are some good quick fixes for this, Grant Searle over at ZiggNet does a complete replacement module that fits into the space vacated by the removal of the duff chips. I'm somewhat traditional and replacement 4116s were sourced for a £10 in postage from a Euro supplier, quite a good deal considering their rarity. What happens is one or more 4116 internally shorts and blows the entire voltage convertor stage, and in this case the 7805 with it. All of the 4116s were removed and 16 pin turned sockets fitted ready for memory replacement.
With the old memory out and no new chips fitted, TR4 was removed, tested and found wanting, a replacement ZTX651 was fitted. TR5 was tested and found to be internally short circuited. TR5 when shorted turns TR4 hard on and both quite happily cook with copious amounts of smoke. The bobbin transformer was removed, tested ok, and refitted. Power up and zilch, but TR4 started getting very warm with +19v on the +12v rail so a quick power off and a sit down with a meter and a diag to analyse what was wrong.
In the end it turned out to be quite simple, over the course of revisions the pad positions of TR4 and 5 have changed. The Mk1 and 2 motherboards used the standard CBE arrangement but with the use on later boards of 2SC series transistors the pad positions were changed to BCE on one, then both transistors. Mental note to self, just because your used to an arrangement, don't get complacent. Both transistors were re-soldered correctly into place, power up revealed the correct voltages on the +12 rail but nothing on the -5v rail. TR4 is part of a saw tooth oscillator with the bottom third of the wave dipping below the virtual 0v rail, this - dip provides the drive for diodes D10 and D11, both in this case being 1N4148 types backed up by a 5.1v zener. Both 1Ns were found to be blown and replacement of both restored the -5 volt rail to full working order.
We now had the correct +5v, +12v and -5v supplies ready for fitment of the eight 4116 memory chips. Power up revealed one other issue in that the ULA chip was unhappy and would require replacement once one was sourced but other than that a satisfying repair. This repair was a tough one and the first time I have had to delve into the intricacies of the Spectrum internal power supply.
