Now beleive it or not under this pile of probes is the heart of the prozboard, the Polyphonic Phase Accumulator section. The concept of the phase accumulator oscillator has been mentioned else where in these pages, so I wont go into too much detail. The PPG has 16 of them and uses a pile of SRAM chips (74s189) to do this, these chips are VERY hard to find, but I now have a small supply of them should one die. finding the fault can take ages.

In this case the problem was that playing a scale would play the wrong notes, ie Pressing a C would giev a note, but not a C, pressing a D would give anything but a D. the clues which pointed me to the proz board were as follows,

A) It happened on EVERY voice, so it was not likely to be a voice card (remember there are two in here)
B) The pitch was consistant, despite the fact that it was the wrong note it was constant, pressing D 8 times got 8 notes of the same pitch.

To prove this further I tried the board in Johns 2.3 and sure enough it was deffinatly the PROZ board.

The 74S189
before I go on I need to explain a little about the 74S189 chip. It is described as a 16by 4 scratch pad SRAM. that is to say it holds 16 four bit numbers. The other interesting points are that it has seperate Inputs and Outputs, most SRAMs only have one common Data port. and to further add to the confusion the outputs are inverted, ie if the value in the RAM is '1001' the value that shows on the Ouput is '0110'.

With that explained here is a screen capture from the Logic Analyzer of the faulty board showing the fault.


Before starting I traced the 74S189s and made a note on a peice of paper and named them A,B,C,D,etc,etc. I did the same on the logic analyzer. With the logic anaylzer I monitored the input pins, output pins, Write and also the address pins.
I did this for BOTH parts of the osccilator section SRAM, the Control register (this sets the pitch) and the Accumulator (this holds the result of the addtition) It was in the Accumulator (ACC) section that I found the problem. You can see that for a normal operation (shown in blue) when a value is present at its input and the write pin went high again, that the outputs changed to follow the inputs (remember they are inverted!). All were fine except one (shown in red) for some reason only bit 2 was correct. I checked this with all address values (ie all 16 oscillators) and sure enough it was consistant, BIT 2 was correct BITS 0,1 and 3 were always wrong.

 
Another example of a proz board problem and repair, one of the 74LS379s had died and a replacement couldn't be found.