0
\$\begingroup\$

I am building a 555 timer circuit with a frequency of 15, 30 and 60 Hertz. I will most likely use a 1µF capacitor but I was wondering what kind of capacitor would be best for keeping a stable value even when the temperature changes (goes down most likely)?

I will be putting potentiometers so that if the values do change they can be calibrated again, but that will only be done before we deploy the systems. Once deployed, we cannot calibrate again until retrieved days later).

Basically, I am asking what kind (ceramic, electrolytic, etc.) of capacitor will keep its capacitance constant at a value of 1µF and with varying temperatures.

\$\endgroup\$
  • 3
    \$\begingroup\$ Are you sure a 555 timer is the best option? Why not some kind of crystal-based clock? \$\endgroup\$ – Justin Jan 9 at 21:17
  • 1
    \$\begingroup\$ Not a very good idea to design without tolerances and environmental specs (!?!) Even some MHz Xtal and a couple binary counters would be better, than 3 pots \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Jan 9 at 22:06
  • \$\begingroup\$ considering the answers i'm getting, I am definitely going to change my design to a crystal oscillator and divider. \$\endgroup\$ – Francois landry Jan 10 at 15:19
6
\$\begingroup\$

As DKNGuyen said, if you want a stable capacitor, use C0G/NP0 ceramic.

However, if your actual goal is to have a stable frequency rather than discuss capacitors, then your original idea of using a 555 timer is not the way to go, as this chip will have worse drift than the C0G capacitor.

A much better option would be to use a quartz oscillator, for example a 2.4576 MHz oscillator which will cost about €1 in 50ppm/°C stability, then divide by 4096 using a 74HC4040 ripple counter. You could use a 74HC4060 too, with a crystal instead of an oscillator. Also the quartz oscillator is pre-calibrated and you don't need to adjust it.

EDIT I misplaced a decimal point...

I mean, a 1.2288 MHz oscillator followed by 74HC4040 to divide by 4096, resulting in 300Hz. Then a 74HC390 or similar which can divide by 5 then 2 then 2 which gives 60Hz, 30Hz, 15Hz.

Or a 192kHz oscillator then the same two 74HC chips, but this oscillator is only available on digikey in tiny MEMS flip chip so maybe not the best option. Anyway you get the idea, pick a frequency and a convenient division ratio with 74HC chips...

The BOM cost for both solutions will be less than a precision capacitor and a potentiometer, and that doesn't count the salary of the intern who gets to tweak the frequency just right...

If you need better stability than what a cheap XO will provide, you can use a TCXO. That will cost a bit more (a few €) and it will be available in less convenient frequencies. So you can use, for example, a 12.288MHz TCXO and divide by 60000, you'll get a few ppm stability over temperature.

Note a cheap microcontroller makes a nice programmable divider if you need one.

If you already have a microcontroller in your project, why not use that?

|improve this answer|||||
\$\endgroup\$
  • 2
    \$\begingroup\$ why is it that the simple solutions are the hardest to see. i will definitely be going with a crystal or quartz oscillator for my circuit. \$\endgroup\$ – Francois landry Jan 10 at 15:23
  • 1
    \$\begingroup\$ Also, I do not have a microcontroller in that system. \$\endgroup\$ – Francois landry Jan 10 at 15:41
  • \$\begingroup\$ Happy to help! Thanks for the accept. \$\endgroup\$ – peufeu Jan 10 at 17:31
  • \$\begingroup\$ You mention that a 2.4576MHz oscilator would work but I don't seem to understand how \$\endgroup\$ – Francois landry Jan 10 at 19:29
  • \$\begingroup\$ If you have a microcontroller or other programmable divider you can use any division ratio you want (for example, take 12 MHz clock and divide by 200.000 to get 60 Hz. But if you want to keep it simple and use a hardware ripple counter like 4040 or 4060, these can only divide by 2^N... so 4096 is 2^12 and 4096*60 Hz = 2.4576 MHz. Feed 2.4576 MHz into 74HC4040 clock input, get 60Hz on output 11. \$\endgroup\$ – peufeu Jan 10 at 19:47
4
\$\begingroup\$

Regardless of the caps you use, a 555 is a 0.1-1%-ish stability device over a modest temperature range.

Typical (not maximum) drift of the bipolar NE555 is specified at +/-150ppm/°C, so to get 0.1% typical drift with perfect external components you only need to change the die temperature by 7°C (either from ambient changes or from the die heating).

You can slap 10 C0G 0.1uF 1206 caps in parallel and get 1uF with +/-30ppm/°C tempco and use some nice expensive 10ppm or better resistors but you're just putting lipstick on a pig.

|improve this answer|||||
\$\endgroup\$
2
\$\begingroup\$

Neither. What you want (or rather, what is practical and most cost effective) is a C0G/NP0 ceramic capacitor. Not any ceramic capacitor, but a C0G/NP0 dielectric. Good luck finding that in a size as large 1uF. Even 100nF is quite rare, and expensive.

Other stable alternatives such as mica, glass, or poly max out at even lower capacitances and are much more expensive.

Also, temperature isn't the only thing that affects capacitance. The DC-bias the capacitor experiences does too.

|improve this answer|||||
\$\endgroup\$
  • \$\begingroup\$ Agreed that C0G/NP0 is the way to go. Typically with 555 timers, designs are set with both an R and a C. You should be able to choose a readily available, inexpensive C0G/NP0, but will probably be much less than the 1uF value you mention. Chose that value first, then select the resistor value. As DKNguyen mentions, DC bias on the capacitor will also affect its value, so only operate the capacitor at 50% its max voltage rating. \$\endgroup\$ – BEE Jan 9 at 21:10
  • \$\begingroup\$ A quick search on digikey brought up many options link. Are these the ones you are talking about? \$\endgroup\$ – Francois landry Jan 9 at 21:17
  • \$\begingroup\$ Yes. 1000pF (1nF) is a very common value for these. Must say "C0G/NP0" somewhere. \$\endgroup\$ – rdtsc Jan 9 at 21:22
  • \$\begingroup\$ so only operate the capacitor at 50% its max voltage rating. in general for ceramic capacitors. You might do the same with other capacitors for voltage derating purposes (like tantalums). But C0G/NP0 should be relatively unaffected by DC bias. Most ceramic dielectrics are heavily affected by both DC-bias and temperature. My understanding is that electrolytics aren't as affected by temperature but are still affected by temperature \$\endgroup\$ – DKNguyen Jan 9 at 21:24
  • 1
    \$\begingroup\$ "temperature isn't the only thing that affects capacitance" and capacitance change isn't the only mechanism by which temperature affects the frequency of a 555 multivibrator. \$\endgroup\$ – TimWescott Jan 9 at 21:26
0
\$\begingroup\$

use a 30.72kHz crystal and a CD4060, digikey is sold out and says it's obsolete, but some other places still have this crystal

Q8 will give 60Hz, Q9 30Hz, and Q10 15Hz

|improve this answer|||||
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.