Editing Sinclair ZX Spectrum+ Troubleshooting Guide (section)

== Finding a faulty RAM chip ==

If the machine boots but a RAM fault is suspected, the Spectrum can name the failed chip itself.<ref name="sm48" /><ref name="ian" />

Type:

 PRINT PEEK 23732 + PEEK 23733 * 256

A healthy machine returns '''65535''' (48K) or '''32767''' (16K). Any lower value ''n'' means the first faulty location is '''n + 1'''. A result below 32767 puts the fault in the lower 16 K (IC6–IC13); above 32767, in the upper 32 K (IC15–IC22).

Then, using that address:

 POKE n+1,85 : PRINT PEEK n+1
 POKE n+1,170 : PRINT PEEK n+1

85 and 170 are 01010101 and 10101010, so between them they exercise every bit. If either value reads back wrong, look it up:

{| class="wikitable styled-table" style="width:90%; text-align:center;"
|+'''RAM fault location table'''
! Reads back after POKE 85 !! Reads back after POKE 170 !! Error size !! Bit !! Lower RAM (&lt;32767) !! Upper RAM (&gt;32767)
|-
| 84 || 171 || 1 || 0 || IC6 || IC15
|-
| 87 || 168 || 2 || 1 || IC7 || IC16
|-
| 81 || 174 || 4 || 2 || IC8 || IC17
|-
| 93 || 162 || 8 || 3 || IC9 || IC18
|-
| 69 || 186 || 16 || 4 || IC10 || IC19
|-
| 117 || 138 || 32 || 5 || IC11 || IC20
|-
| 21 || 234 || 64 || 6 || IC12 || IC21
|-
| 213 || 42 || 128 || 7 || IC13 || IC22
|}

Values from the Service Manual.<ref name="sm48" /> If more than one location is faulty, the first must be repaired before the test can proceed past it.

If the value read back is not in the table at all, suspect bent or straddling legs on socketed upper RAM, or faults in '''IC23 (74LS32)''', '''IC24 (74LS00)''', '''IC25''' or '''IC26 (74LS157)'''.<ref name="ian" />

'''Modern alternative.''' Purpose-built diagnostic ROMs make this much faster: Brendan Alford's '''ZX Diagnostics''' test ROM will run from the ROM socket and identify bad bits even when the display RAM itself is faulty, reporting the failing data lines and IC numbers.<ref name="tyne" />

=== Checking for shorted or leaky RAM ===

* After the supply rails are proven, feel each lower RAM chip. '''None should be even warm''' — any that is should be removed.<ref name="ian" />
* The manual's method is to switch on and check whether any RAM gets hot; a very hot chip is a sure sign it is short circuit. For open-circuit failures, bypass each RAM in turn with a known-good device — and note that after switching off you should '''wait about ten seconds before removing the test chip''', or it will damage itself.<ref name="sm48" />