We are using Mettler Toledo Probe, we already replace the electrode with the new one (to try solve this problem), and we have not yet used pH controlling or Cascade so, should the PID not be involved in this problem?
Measurement and control is different ball game altogether. Is your medium very viscous? Certain pH probes may not be suitable for viscous medium. From a mixing point of view, if your medium is viscous then the long delay time needed to stabilize the reading is reasonable. You should make sure that the measurements are what you want with a suitably fast response time before putting in your control system.
There may be two reasons 1. viscosity of fermentation broth and mixing conditions 2. Sensitivity and control system of probe. Please check your probe with different viscosity solutions if it is working properly then the problem is in mixing method or low rpm (if you are using agitator).
I also faced the same problem but increasing rpm resolved the problem.
I don't know the details of your fermentor system but it seems that you are facing some 'stability' criterion that is being currently unable to cope with the noise and drift of the signal being processed. If that is the case there are two possible approaches: either you act to reduce noise and/or drift, e.g. by selecting a pH electrode less sensitive to noise, by relocating the electrode, or by improving mixing; or you can change the previously mentioned stability criterion so that it becomes less stringent, thus accepting less stable readings.
Improving PID controller’s parameters may be helpful while following the first mentioned approach. However, this should not be a significant contribution, as long as the controlled variable (possibly the temperature) remains reasonably stable.
In case you have on-line data being processed by a computer, you may also have some data acquisition trouble, what may be solved by properly selecting the relevant data acquisition parameters, as the sampling frequency or the incoming data buffer size.