I've been experimenting with oscillators, and am particularly fond of a very low-current MOSFET design that uses ceramic capacitors and oscillates at about 1 to 10 Hz. However, I am having trouble adjusting its frequency and duty cycle.

The resistors involved are the largest I've got, so to slow it down a little I decided to use larger capacitors, like a 1μF electrolytic instead of 0.01μF ceramic. Instead of slowing it down by a factor of 100, it just had some weird smoothing effect on the square wave, depending on which side was connected to the load. In fact, 100μF and 1000μF had the same frequency, just more smoothing.

I figured I just had the capacitor in backwards, and flipped it around, but this often made no difference, and it never made the frequency "correct". Also, even with 100μF on one side and 0.01μF on the other, the duty cycle was as close to 50% as I could measure.

I used a multi-meter to check which way the capacitor should really go, and was shocked to find it was always backwards. Eventually I discovered that the voltage was swinging from -130mV to +130mV, and if the capacitors weren't balanced or the phase of the moon was wrong, it would sort of get stuck on one side, and the other capacitor would handle all the oscillating.

Through many hours of simulating oscillators in spice and the falstad.com simulator, I have found that some of my oscillators maintain 0 to +V across the capacitors over the entire wave, but some configurations are more like -0.5V to 3V across the capacitors. I have no idea what controls this.

How does one design an astable multi-vibrator to have a specified voltage range across the capacitors? (Ideally, given a fixed frequency and duty cycle, and maintaining low current.)

If it matters, I substituted ON 2N7000's for the depletion mosfets, and this works with ceramic capacitors, just not polarized capacitors.


2 Answers 2


Have you considered bipolar electrolytics? Normal electrolytics only have one plate doped (if that's the right word), so they can only withstand a unipolar signal (in reverse they act as a short, if I recall correctly, which is what causes them to bulge and eventually vent/explode.) Bipolar electrolytics usually internally have two capacitors connected in inverse series, or, in some cases, both plates equally doped. You can create the same effect in a pinch by wiring two electrolytics in inverse series, like this.

  |  / +  + \  |
--| |--------| |--
  |  \      /  |

You can also get axial ceramics in values up to around 1µF, and bigger in surface mount sizes (I use 22µF 6.3V 0805 capacitors.) For example here is a 1µF cap rated for 50V.

  • \$\begingroup\$ In reverse, the dielectric breaks down, and they eventually become a short. Initially, though, they'll behave just like a normal capacitor. \$\endgroup\$ Nov 12, 2010 at 16:53
  • \$\begingroup\$ @reemrevnivek , have you seen the people who claim that such back-to-back electrolytics cause "no problems"? See electronics.stackexchange.com/questions/3062/… \$\endgroup\$
    – davidcary
    Nov 14, 2010 at 1:42

That is a high-impedance circuit. The capacitor is in series with a 22 megohm resistor. Any capacitor leakage at all is going to cause problems.

Electrolytic caps are known for their high leakage. They are not a good choice for this circuit. You will be better off to use a series of flip flops (also knows as a divider) to get a lower frequency output from a higher frequency oscillator.

Here is a datasheet for a low-leakage electrolytic: http://products.nichicon.co.jp/en/pdf/XJA043/e-kl.pdf Notice it has a leakage spec of 0.2uA, and this is only after settling for a minute. 0.2uA through the 22 megohm resistor will drop 4.4 volts. Since the circuit is only powered by 1.5V, the capacitor will simply not charge through 22 megohms.

  • \$\begingroup\$ Can you say a little more about the series of flip flops / divider? Since some people say "flip flop" for any kind of multi-vibrator, I am having hard time sorting the results. I know about clock dividers and prescalers from the "software" side, as in, this little chunk of code turns one on, but I don't actually know how to build one. \$\endgroup\$ Nov 12, 2010 at 18:07
  • \$\begingroup\$ Also, probably I should just get the larger ceramic capacitors to fix the leakage? 0.1μF to 1μF capacitors seem like they would be fine, especially if I get a few more megaohm range resistors. \$\endgroup\$ Nov 12, 2010 at 18:09
  • \$\begingroup\$ Ceramics should be fine, but check the datasheet. \$\endgroup\$
    – markrages
    Nov 12, 2010 at 18:12
  • 1
    \$\begingroup\$ Consider the diagram on the humble 4020 divider: st.com/stonline/books/pdf/docs/1961.pdf The clock goes in, and a frequency of clock/2, clock/4, etc comes out. The lowest frequency is clock/32768. So if you have a 32.768 kHz input, that last pin on the divider chain will be a 1 Hz square wave. (Not coincidentally, if you take the back off your watch you will find a crystal cut to oscillate at 32.768 kHz.) \$\endgroup\$
    – markrages
    Nov 12, 2010 at 18:21
  • \$\begingroup\$ Oh, a counter! Marvelous. It's fun to see that the impossible-to-imagine device is something I've looked at a hundred times. How to divide by 2? Wait for every other one. How to divide by 4, count 4 between each tick. Amazing. I've also wondered about that strange frequency, now it makes perfect sense. I wish I could +2. \$\endgroup\$ Nov 12, 2010 at 18:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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