# What is the typical frequency drift of an astable 555 circuit over its lifetime?

I am involved in qualifying a sine wave generator that is using a first order LPF and a 555 timer in astable configuration to generate a sine wave(-ish) output. I would have designed it differently, but we already made the board, so I have to work with the design the best I can. The designer has placed the corner frequency of the first order LPF at 100kHz and the oscillator is running at 230kHz. A variable gain stage (not an op-amp, but similar, with 7M ohm input resistance) calibrated with a variable resistor is being used to set the final gain, so output amplitude from the filter doesn't matter except that it must remain constant. The output frequency isn't critical, but the peak output voltage is.

Because the oscillator is running in a region of the filter where gain changes -20dB/decade, the frequency has an impact on the gain. I was hoping someone could tell me how much 555 timers change in frequency from use to use to help me assess how much trouble this design is in. All components are surface mount. I know capacitors and resistors change with temperature; the device should always be within a 15 degrees Fahrenheit range when in use. Also, I am unfamiliar with how much surface mount capacitor capacitance changes over long periods of time.

• Man, that is just wrong Oct 18, 2013 at 22:41
• As always, you need to have some idea of how much amplitude error you can tolerate, since no circuit is perfect. Oct 18, 2013 at 22:43
• Do you mean the lifetime of the Si or lifetime of the application? Oct 19, 2013 at 2:19
• @Chetan: Life time of the application.
– Bob
Oct 19, 2013 at 2:25
• @Dave Tweed's first comment, I know, right? As for your second comment, I'm still working on trying to get a spec. Secretly, I'm hoping the variation will be enough that I can just say "we need a different solution." I wouldn't have considered this design a solution. Ever.
– Bob
Oct 19, 2013 at 2:26

555 has temperature sensitivity of ~50ppm/C. Delta of 15°F is about 8°C. This gives you 400ppm.

So your 230kHz clock could be off by 92 Hertz over temperature due to 555 tempco.

If you use capacitors with NP0 dielectric you most likely get very decent temperature and age stability, but this limits you to capacitances <500pF. Vanilla resistors though could have tempco of -500ppm/C, which will shift your frequency by ~920 Hertz.

Aging of the chip itself likely to give much weaker effect.

So if 1kHz frequency error is ok for you - you are good to go (with NP0 capacitor!).

PS. Check tempco of your LPF too ;-)

• I wasn't thinking the aging silicon would do much, but rather the aging caps. I know big electrolytic caps can dry out, but I don't know anything about how surface mount caps hold up. Excellent answer, thank you!
– Bob
Oct 19, 2013 at 2:34

It's generally tough to design a good sine-wave generator without using some sort of amplitude feedback. I would not rely upon any circuit such as you describe to output a consistent amplitude. Even if the frequency was stable, there's no reason to expect that everything else about the filter will be as well. Add in the uncertainties in the frequency, and you're begging for trouble.

Why are you using the LPF with a corner frequency so far below the frequency of interest? Is your goal to ensure adequate stifling of the third harmonic? There are some other easy ways of generating waves whose third-harmonic content is zero. For example, if you generate two square waves 60 degrees apart and add them together, the sum will have harmonic content just like a square wave except with all third-harmonic content eliminated.

• that thing about adding the square waves is kind of amazing.I'm going to have to look into that. I think the designer's intent was to push the corner frequency far enough away from the driven freq that he would get a sine wave. I agree that this is no way to solve this problem. I'm going to check out amplitude feedback, because I'm not familiar.If you could edit a good link into your question,it would be appreciated.Good point about the filter changing as well.
– Bob
Oct 19, 2013 at 2:32
• @Bob: You do understand, don't you, that once you're past the corner frequency, the relative attenuation between the fundamental and the harmonics doesn't change any further? 6 dB/octave is 6 dB/octave, no matter where you are on the slope. I'm guessing that you get it, but the original designer certainly didn't. Oct 19, 2013 at 3:43
• @DaveTweed I did think of that, but it feels really good to hear someone else say it. I can't explain why the designer did what he did... I am speculating when I write about "the designer's intent."
– Bob
Oct 19, 2013 at 11:16