Issue 3 upper RAM troubleshooting

[TakuikaNinja]

I've decided to move the topic of troubleshooting my issue 3 speccy over here instead of continuing in a music thread. To summarise the problem, pins on IC26 (TI 74LS157) seems to have been bent and broken during the process of an upgrade to a 48K system by a previous owner. The unit was not reporting any memory errors (very strange), and the issue only manifested itself when actually trying to access certain parts of the upper RAM.
I inserted the replacement IC today and now the system reports the upper RAM as not being detected at all. The upper RAM ICs are all OKI chips but the RAM selection links are soldered as "TI" and "3" (from the factory, judging by how dirty they are). I'm guessing I would need to switch it to "OKI" and either "L" or "H"? How would I know which one to solder?
(It could also be the fact the I managed to slightly damage the socket for IC26 in the process of removing the IC... I don't have any soldering experience and was only planning on soldering a new voltage regulator.)

[1024MAK]

That explains why the chip fitted in position IC26 had been modified. OKI 32K DRAM needs a different DRAM address (row/column) compared to TI 32K DRAM.

This is what the OKI / TI DRAM links select.

The H / L links select either the upper (high) or lower half of the DRAM (32K DRAM are 64K chips where only half has been certified as free from defects by the semiconductor manufacturer).

The answer is in the service manual. Page 11 and page 23. And pages 37 onwards. These page numbers are for the version from Spectrum For Everyone.

Mark

[TakuikaNinja]

That makes much more sense now. Pin 10 was bent so it could be connected to A14 with the existing "TI" configuration. Would pin 11 being broken imply that it was supposed to select the lower half of the DRAM ("L" configuration)? Or was that modification entirely incorrect on part of the previous owner?

[1024MAK]

Officially, for 74LS series logic chips, it's strongly recommended to tie unused inputs to the +5V supply rail via a 1kΩ resistor.
However, most 74LS chip inputs will float to a high logic level if left unconnected (due to the design of the internal circuitry). Floating high here is equivalent to fitting link H.

Given that setting the links correctly is easier than messing about modifying a 74LS157, I think it is more likely that the owner bought an upgrade kit designed for an issue two board, Issue two boards don't have all the same links, they can officially only use TI DRAM. Hence upgrade kits using OKI DRAM were sold with 74LS157 modified by the supplier or with a daughter board (which rerouted the relevant signals). Then the owner just fitted the parts into the existing sockets on their ZX Spectrum board.

Presumably OKI DRAM being cheaper than TI DRAM at the time.

Mark

[RST#08]

Hence upgrade kits using OKI DRAM were sold with 74LS157 modified by the supplier or with a daughter board (which rerouted the relevant signals). Then the owner just fitted the parts into the existing sockets on their ZX Spectrum board.
Mark

Yes, this is what I recall. The Fox upgrade used a daughterboard, but the ELR SP48 kit, which is the one I bought at the time, used a modified LS157 with a pin removed, and another bent, inserted into a DIL socket, which itself was plugged into the motherboard. I should take a photo of this next time I open my Issue 2 machine (mine's a more common version with transistor soldered onto the ULA).

[1024MAK]

The extra transistor (TR6 for issue 3 and later boards) for issue two boards was normally placed on top of the processor (where it is known as the ‘spider’), not the ULA , or on the PCB between capacitor C47 and IC19. Although it is occasionally found in other locations.

Although it's correcting a design limitation of the ULA.

Mark

[RST#08]

The extra transistor (TR6 for issue 3 and later boards) for issue two boards was normally placed on top of the processor
Mark

Of course, yes ... my mistake ... it's been a while since I opened it up

[1024MAK]

No worries

Mark