I design a digitalization electronic system for an ultrasonic device.
Problem:
The FIFO SN74V245 goes first well after power-up. After a few (ca. 100) write and read cycles it begins to manipulate the output data. First its like an alias effect, then the second half of the data remains nearly constant. When a reset is performed it works well first but the problem repeats. The longer the power supply is present the shorter time it takes before the problem appears. The flags do sometimes not react at all, even though these are double buffered. They are then both on a low stage. My question is, if this is a known problem and if you could help me finding the source.
Steps Needed to Recreate Problem:
The FIFO is placed on a four layer board with three ground planes, nevertheless I think it could be a ground problem. There are no decoupling capacitors near by the device. There is probably no timing problem in the read or write cycles. It is used in the standard mode with double buffered asynchronous flags. It is not used in a depth expansion. The read clock is directly connected to an oscillator, whose output is decoupled to ground. The input signals come from an analog-digital-converter, these are ok. The frequency of the write clock is 40MHz, the one of the read clock 12MHz. The write and read progress do not happen at the same time but sequentially.
