Signal Generator Needs Improvement

Thread Starter


Joined Oct 17, 2011
I have this circuit which I would like to use as a signal generator. I could use an IC, but then I would not have this much fun building it. Also, since I will be using this in some solar applications wide input range is essential, all the way up to the limit of the discretes. Finally, when it blows up I don't have to replace an expensive IC, just swap some resistors or transistors.

I have some issues with it, and if you are kind enough you could tell me how to improve it.

- At low load-resistance the wave-form is rounding a little too much. I do not believe this is big of an issue though.
- The 5.1 Volts peak is drifting somewhat between 7.6 minimum and the 28 maximum Vin. Not really an issue.
- At high load-resistance the output floats above 1 Volt. I need this to be as close to 0 as possible at all times - but at least under 200 mV. This I would like to fix.
- I cannot figure out what is the best way to modify the duty cycle. It would be nice to be able change by 10% increments.
- Please let me know If you think it would be safe to farther reduce current draw. I never actually built this circuit yet, but I read 4148 diodes need at least 1mA to produce the 600mV or so drop.

Anyhow, thank you for reading.



Joined Dec 20, 2007
You have a current limiter and a voltage regulator in the power supply. The voltage regulator is a shunt type that varies its current to ground so it fights the current limiter.
Use a series voltage regulator with a stable voltage reference.

The current limiter has too many parts, it needs only a PNP transistor, two diodes and one resistor.
Your voltage regulator does not have a stable voltage reference so its voltage varies when the load current changes, when the input voltage varies and when the temperature varies.

The voltage gain of your output differential amplifier is too low. The output will have a better square-wave if the gain of it is much more.


Joined Jul 17, 2007
Many folks don't have LTSpice installed, so they won't be able to comment on your schematic. However, I've taken the liberty of removing some "white space" from your schematic so that it will be easier to read (text will be larger) when running a simulation, both for you and other people who might want to comment.

Have a look at the attached.



Joined Dec 26, 2010
1N4148 diodes have logarithmic V/I characteristics typical for silicon diodes, extending down to at least tens of microamps at room temperature. The voltage will decline at maybe 100mV per decade of current reduction, but you will not expect it to fall off a cliff edge anywhere near 500μA.

A dummy load of some tens of kohms should keep the circuit working independently of any external load.

As presently designed the output transistor can supply only a limited current before coming out of saturation. This might be improved somewhat, but note that presently less than 1mA is available from the power rail. A buffer stage could provide the grunt to move 50 ohms if needs be, but not if it is powered from a 1mA trickle.

The small timing capacitor and huge 10Mohm timing resistor seem a bit optimistic in a BJT application. Could this lead to undue sensitivity to current gain variations?

Thread Starter


Joined Oct 17, 2011
Thank you all for your time and your comments.

Some of it I already suspected, and wanted to see what someone, who actually knows this, have to say.

Some of the comments I do not really understand, but I will read up on them. I have only been tinkering with circuits and simulators a little while.

Actually, I got more answers than I hoped for. The dummy load idea solves my worries about the solar array voltage swinging too high when the load eases or disconnects. Of course! Things will not burn up after all.

sgtWookie, thanks for you troubles. I have a few more circuits I would like to know more about and I will post again soon - both asc and png. Cheers.