Fading LED a different way
A different approach...
I came up with a different approach to let a LED fade in and out.
When two separate squarewaves, with a slight frequency difference of about 0.5 Hz, are combined, the output squarewave pulsewith will vary between 50% and 100% ("due to phase shifting"). But the LOW state varies between 0% and 50%.
So I used two CD4093 NAND gates, with Schmitt-trigger action, to build two 200 Hz oscillators, one with with a fixed resistor, and the other with a 20 turn potmeter, to be able to adjust a 0.5 Hz frequency difference between them. This results in a linear fading LED, driven by a FET, as can be seen on the video.
The reason why I used two separate CD4093's is the fact that I found out that the schmitt-trigger points are influenced internally by the output sink-current, resulting in 'locking' the two oscillators when the frequency is nearly the same!!!
I tried to minimize this phenomena by separation, but the effect remains.
After decreasing the frequency from 2 KHz to 200Hz, the effect was gone.
This circuit is a bit temperature sensitive, but the principle works well :-)
When two separate squarewaves, with a slight frequency difference of about 0.5 Hz, are combined, the output squarewave pulsewith will vary between 50% and 100% ("due to phase shifting"). But the LOW state varies between 0% and 50%.
So I used two CD4093 NAND gates, with Schmitt-trigger action, to build two 200 Hz oscillators, one with with a fixed resistor, and the other with a 20 turn potmeter, to be able to adjust a 0.5 Hz frequency difference between them. This results in a linear fading LED, driven by a FET, as can be seen on the video.
The reason why I used two separate CD4093's is the fact that I found out that the schmitt-trigger points are influenced internally by the output sink-current, resulting in 'locking' the two oscillators when the frequency is nearly the same!!!
I tried to minimize this phenomena by separation, but the effect remains.
After decreasing the frequency from 2 KHz to 200Hz, the effect was gone.
This circuit is a bit temperature sensitive, but the principle works well :-)
Discussion (3 comments)
Roel Arits 7 years ago
Good idea for experimenting !
Roel Arits 7 years ago
You could use multiple VCO's (that can be found in a HEF4046 or CD4046 PLL chip).
You can use a voltage divider to generate input voltages for the VCO's that are very close together, so the output frequencies of the VCO's are also very close together.
F.e. when using a voltage divider with 3 resistors : 1K, 10E, !K, The voltage at the top and at the bottom of the 10E are than 0.05V apart when using a 10V voltage.
Because working with voltages, it is also easy to modulate the frequencies slightly, so the "beat-frequency" will also swing slightly.
Maybe that also gives a nice extra effect.
Roel Arits 7 years ago
It is an interesting circuit to tame ...
Arnoldus 7 years ago
Arnoldus 7 years ago
I think you are wright, that can be the only reason why the separate oscillators 'see' each other. After using two separate 4093's, I used an IC socket with a build-in decoupling capacitor, but only for the second 4093.... :-(
Roel Arits 7 years ago
If i may give you a tip :
I guess that the oscillators have the tendency to lock to eachother (synchronise) via the power supply connections were all the switching spikes of the oscillators are present. So i guess you need to provide individual 100nF decoupling capacitors directly over the power supply pins of each oscillator to improve the decoupling of the oscillators.
Arnoldus 7 years ago
but not tried sofar. I think I am going to try soon :-)
David Ashton 7 years ago
Arnoldus 7 years ago
I have thought of using the mains frequency as base signal, but it seems to vary slightly during the day, but the total cycles/day are always kept at 4320000, (some older clocks use the mains frequency as time reference).
I like analog circuits a very VERY long time..... ( >40 years )
ClemensValens 7 years ago
Arnoldus 7 years ago