74xx393 strange behavior

[OlPhart]

Howdy, I've used this dual 4 bit binary counter in a Lot of designs. This one has me pulling hair: tie the B side clock to bit3 of A for an 8 bit counter. Feed A side clock with a rotary encoder. Take all 8 outputs into a CD4021 parallel->serial shift register. Read the encoder count by a pulse of the shift/load line then 8 clock pulses into the micro. Subtract the previous count from the current count to get # counts this period.

Simple, can't miss an encoder pulse (if period vs encoder speed doesn't overrun 256), perhaps elegant. Here's the rub: about half work as expected, the other half have counter problems. Most failures are the 1st F/F having same waveform as the input: not dividing by 2. Some exhibit this on the 1st and 2nd F/Fs. A couple have the 2nd F/F adding transitions: divide by 2 then only by 1... erratically. Some boards you can make behave by putting a locally ungrounded (!) scope probe on the problem line, others ignore any such "tapping".

I started out using the HC family, but tried changing to the LS... no avail. I have both S/R & counter adequately bypassed, all boards are built the same, with minor layout differences. No apparent correlation of layout or part family to behavior. I'm missing something, but what? TNX <<<)))

 

[ericgibbs]

hi Ol.
Do you have a circuit diagram you could post, it will help us see what the problem might be.
Eric

 

[Ian Rogers]

Trouble is The second 4 bit counter needs its clock high to start with, bit 0 (bit 4 in this case), so you need a soft start on the clk/q3 connection.. Other wise it starts counting at 0x10....

I put the soft start on both MR pins.... That seems to have corrected it..

 

[OlPhart]

Howdy and thanks for the responses. I knew what I posted wasn't sufficiently described, and didn't take time to sketch a schematic.

However, a reapproach pointed out the answer: replacing the encoder with a function generators' square wave works fine. I'm not conditioning the encoder signal right. Now that I know the cause, solving it is fairly straightforward. Perhaps just posting was enough to sweep the cobwebs from my thinking... Thanks again <<<)))