I'm trying to make a simple astable NE555. The output frequency increasing alone. For example it start from for example 4.2 kHz and immediately change to 4.3 kHz, 4.6 kHz, 5kHz, then play around 4.5 kHz up and down.
I tried different NE555 IC like TLC555 and a different 9V battery, same problem.
I am using breadboard.
C1 is a ceramic capacitor. I measured the frequency using a PIC microcontroller based frequency counter.
What you have discovered is one of the major application limits of what are known as Class II ceramic capacitors -- they can't be used for timing. This is because the dielectrics used vary in dielectric constant, and thus the capacitor varies in capacitance, as the voltage across the capacitor changes -- this effect is known as voltage coefficient, or just voltco for short, and is the main reason for voltage derating of these capacitors in their application domain as well.
Class II ceramics are easy to identify from their temperature coefficient rating -- all X?? (such as X7R and X5R) ceramics are Class II, while C0G (NP0) ceramics are Class I, i.e. temperature and voltage stable. However, C0G capacitors are not available for values greater than 10nF as a practical matter, so for larger values, film caps (polyester/Mylar, polypropylene) are used instead when capacitance stability with voltage is a concern.
First, you must have a bypass capacitor across the 555 power pins, especially for the non-CMOS type. 100uF/10V is good. The 555 derives its switching thresholds with a resistive divider from the power supply and any noise (internally or externally caused) during the cycle will change the timing. Use a 'clean' power supply with low impedance, such as from a dedicated regulator. It has a low voltage coefficient with static voltage changes, but not if it's bouncing around during the cycle. There is also the voltage coefficient issue which @ThreePhase points out, which will mostly lead to non-ideal waveforms but will change the timing if the supply voltage changes.
Secondly, the X7R type caps have a fairly large temperature coefficient so small changes in temperature can lead to timing changes (graph from here).
As you can see, a 1°C change can lead to almost 0.1% change in frequency.
This might be way too late for you.
There is a capacitor missing in your circuit - there should be, at least, 100nF over from #1 to #4/#8. The 555 has a quirk in which it consumes lots of mA when switching its output. The internal resistance of the 9V battery can't cope with that. Better is putting a 10uF in parallel with 100nF, near the 555.
