befuddled

Dear experts,

The circuit on the left on the attached figure is a low voltage light flasher as I transcribed it from DAVE JOHNSON’s UNDER VOLTAGE LED FLASHER ( **broken link removed** and https://www.electro-tech-online.com/custompdfs/2009/01/undervol.pdf ). C2 was changed from 0.68 to 1. uF as shown. The circuit I copied used PN2907 and PN2222 transistors rather than the BC557 and BC547 as shown, but both sets are “General Purpose.” At 3.3V the circuit appeared to work properly, although the duration of the flash was on average about ten-fold greater than expected and had a huge variance ranging from 6 ms to 71 ms. But, it flashed and I was happy. I then changed from my power supply to a pair of somewhat spent NiMH batteries with a voltage of 2.6V. While this is still above the designed 2.4V low detection limit it is significantly lower than the 3.3V of the power supply. The LED simply lit up and stayed lit. Occasionally, when I plugged the battery in it would flicker once or twice before lighting the LED permanently. I tried to simulate the circuit using the java applet from Circuit Simulator Applet, and discovered quite by accident that the circuit would work when transistor Q1 was oriented properly and when it was oriented improperly (circuit on the right) down to a voltage of 2.2V. On a lark, I tried reversing the Q1 transistor and much to my surprise, the circuit sprang back to life. The new circuit works from 2.6 to 5.0V (I have not tested below this range, it does not work at 9V) with a flash ON time of 3 to 4 ms (still not 2 ms, but the capacitor is different).

So, what I want to know is did I make a mistake in the original transcription of the circuit? If not, why does this circuit work in either transistor orientation at 3.3V, whereas the “forward” circuit fails at 2.6, while the backwards circuit keeps flashing? I clearly do not have any real insights as to how this circuit works, if someone could take me through it, it would be appreciated. Alternatively point me towards a reference that will help me understand this better. Ultimately, I want to make a flasher to warn when a pair of NiMH batteries is running low (~2.2V).

DavidBear

The capacitor must be non-polarized. Try it.

Audioguru,

It worked! The store bought metal film capacitor is huge, but it worked. Thanks! Now, would it be asking too much to explain where the reverse charge is coming from? Is there a good text or online tutorial that would help me see the light?

DavidBear

I have some 1uf/5%/63V rectangular film capacitors that are small with a 0.2" leads spacing.
Here are two voltage measurements showing that the capacitor voltage reverses.

One last question, if I may? How do I calculate the times t(ON) and t(OFF), and which components are contributing to t(ON). I think that t(OFF) may be 0.63x4.7(M)x.68(u), where 0.63 is derived empirically (what should it be?). I have never achieved the 2 ms ON pulse in spite of fiddling with the 680n resistor and the 4.7K resistor. I do note that if the 47n resistor is reduced, t(ON) also is reduced, but this does not seem to be a rational way to control the timing.

Here's you a real simple flasher. Changing the size of C! will change the flash rate.

Space Varmint,

I appreciate the reply, but at this point I am not trying to get an LED to flash, I want to understand the design and function of this very low current (0.6 uA when the LED is off), low voltage (gives up the ghost at ~1.9V) oscillating circuit. It is my hope that if I can better understand this seemingly simple circuit that I will be able to think more clearly about more complex (and interesting) circuits in the future.

When I measure t(ON) at 2.8V, I get 5-6 ms rather than the stated 2 ms in the original circuit. I would like to reduce the pulse down to that level, and I have not been able to do so by arbitrarily fiddling with the resistors in the circuit. On the other hand, it is simple to manipulate t(OFF) by adjusting the size of the 4.7M resistor, but I don’t know the theoretical basis for t(OFF). I know it is t(OFF) = nRC, but I don’t know what n is supposed to be and why.

If you can help me learn this either by directly answering or pointing me to some reference that would help me understand it, I would be appreciative.

Thanks again,

DavidBear

I too would like to understand how this circuit works.
I don't see why q2 turning on hard raises the voltage at the base of q1 turning it off.

The base of q1 changing from 2.4v to 2.3v (as AG posted) does not help me to understand what is happening to turn q2 on and off.

The easiest way to see how it oscillates is to remove R2 and C2 from the circuit. Then you'll see that it is now simply a non-inverting amplifier with a quiescent current of apx 5ma assuming a beta of 100 for both transistors. Add R2 and C2, and you have positive feedback which causes the circuit to oscillate.

kchriste,
Thanks for the reply, that was helpful. I'm not sure how you calculate 5 mA quiescent current. I also don't see how t(ON) is determined. Perhaps the two are related?
DavidBear

First I had to assume that the beta of both transistors is 100 and Vf of the LED was 2V. Beta varies with temperature, which week the transistor was made, etc, so it can be hard to pin down even if you look at the datasheet.
Next, we know that transistors have a typical Vbe drop of 0.7V so this puts 3V- 0.7V = 2.3V across R1, the 4.7MΩ resistor. This means that we have 489nA of base current on Q1. If the beta of Q1 is 100, then 48.9µA will flow in the collector as long as enough voltage is across Vce of Q1. Since the 4.7KΩ (R3) will only drop 0.23V and the Vbe of Q2 is 0.7V we still have 2.07V left across Q1 so 48.9µA will flow without saturating Q1. Since there is now 48.9µA going into the base of Q2, which has an assumed beta of 100, 4.89ma of current will flow in the collector. Since the 47Ω (R5) will only drop 0.23V and the Vdrop of D1 is assumed to be 2V we still have 0.77V left across Q1 so 4.89mA will flow without saturating Q2.
The ON time is most likely a product of transistor beta and the value of R2.

kchriste,

I can not begin to tell you how helpful that detailed explanation of your circuit analysis is to my understanding. Based on your analysis, I will focus on R2 and try to derive an empirical relationship between R2 and t(ON).

Thanks,

DavidBear

kchriste, you make it sound easy, thank you
I guess analysis takes practice. Start simple and build on it.