0
\$\begingroup\$

I'm trying to get a 19.2 KHz frequency out of the MIC1557 clock generator.

I managed to get close enough (enough for my application) to this frequency with off the shelf components. The shematic is basically the same as in the first page of the datasheet under "Typical Application", the schematic on the right "MIC1557" ( not the 1555).

Resistance is 3.4K (composed of 3 resistors in series) (instead of 1k in the schematic) and capacitance of 0.01uF (one capacitor).

I also tried the same combination with 34K (also 3 resistors in series) and capacitance of 0.001uF. All resistors and the capacitor are SMD.

In both cases, I get a frequency of approximately 19.2 KHz (+- 0.1 KHz).

The problem is that this frequency is increasing constantly and regularly. After half an hour it's about 0.5 KHz higher.

It happens also when the oscilloscope probe is disconnected. When I reconnect it the frequency incrementation is the same. And nothing seems to stop this. Please help.

(I know using a crystal oscillator is a better solution. But this one, with the MIC1557 is much cheaper and doesn't generate any high frequency.)

\$\endgroup\$
4
  • \$\begingroup\$ What type of capacitor are you using? \$\endgroup\$ Mar 30 '20 at 23:32
  • 1
    \$\begingroup\$ As Kevin is implying, it is often the poor temperature coefficients of the passive parts that cause this kind of drift. Some ceramic capacitors are really bad. Try blowing on the parts, or use a heat gun ( hair dryer?) to see if you can change the frequency. \$\endgroup\$
    – user69795
    Mar 30 '20 at 23:38
  • \$\begingroup\$ @Kevin White These caps are MLCC 0804, from Wurth El. See links on the edited question. The accuracy is much better than the tolerance indicated on the seller website (10%) because the RC ratio is very close to what I calculated theorically. I didn't chose these components specifically for this application, but for general purpose.Resistors are cheap one, SMD 0603 from Vishay or Yageo with a tolerance of max 5% but also with a decent accuracy when mesured. They were soldered manully. (that's why I chosed the 0804 size). \$\endgroup\$
    – Fredled
    Mar 31 '20 at 0:03
  • \$\begingroup\$ @user69795 Ok I'll test that. What type of cap would you suggest. I heard of Niobium Oxyde cap. Are they worth it? \$\endgroup\$
    – Fredled
    Mar 31 '20 at 0:06
2
\$\begingroup\$

From the datasheet the capacitor changes it's value significantly with temperature.

That is probably the cause of the frequency instability.

As @user69795 states intentionally heating or cooling the capacitor can help diagnose the problem area.

https://redexpert.we-online.com/redexpert/#/module/71013/selecteditems/885012207122/productdata/=885012207122

\$\endgroup\$
2
  • \$\begingroup\$ 2.5% seems a lot though, considering its probably increasing temp by like <10C. This graph suggests that you might expect a change of 0.5% \$\endgroup\$
    – BeB00
    Mar 31 '20 at 2:04
  • \$\begingroup\$ Heating the capacitor to +- 60°C causes a frequency increase of 3.6% almost immediately. Without heating, there is a slow but gradual increase reaching 2.6% after half an hour. Then it keeps increasing but very slowly. I'm very surprised that self heating caused by the operation is having such a strong effect. I rather expected changes with ambient temperatures. \$\endgroup\$
    – Fredled
    Mar 31 '20 at 10:25
0
\$\begingroup\$

Both your caps are X7R, very drifty with temperature. Use C0G if you want the osc to be more stable.

\$\endgroup\$
1
  • \$\begingroup\$ This was an experiment. If I use this technology for a finished product, I plan to use C0G capacitor. \$\endgroup\$
    – Fredled
    Mar 31 '20 at 10:28

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.