I am using a CD4040 to divide a oscillator frequency and reading its period using PCA. The problem is I was using 3 oscillators, 120kHz, 70kHz and 30kHz and dividing it by 32. The problem is, when I shake my board, the measuring get really screwed. I tried my oscillator, I can't see any change on it using the oscilloscope, but the Capture value shows that it changed. What could the problem be? I'm starting to think that I could have harmonics between CD4040 and 8051 input.
I can't send my setup pics unfortunately but about the mechanical problem, if I hold my board and try to hit it softly at random places, the only place my measurement becomes different than it should be, is when it hits near the 8051. I can hit my LC tank, I can hit my 1:1 transformer, I can hit my transistors and etc. I tried to resolder all my SMD components again and even got a new 8051, but the problem continues. Later today I'll change my CD4040 exit to a lower divider so my frequency gets a bit higher, because I'm afraid my lower frequency can be causing this sensitivity.
About the PCB, it's a real PCB, we designed it, made it at a PCB house and soldered the components.
My major problem is: My uC can detect the period variation, but my oscilloscope can not. I'm using a Tektronix TDS-1012.
It's a "Differential Amplifier" configuration, but one of my NPN collectors has the LC Tank. My L is a 1:1 transformer inside one of these and my Inductor is a loop for metal detection. So for values I don't know them very well, because I didn't measure the loop inductance nor the transformer's mutual inductance. The Tank's capacitors are polyester, mostly bought from Digikey.