Hi everybody
I am an unexperienced hobbyist in electronics. Recently I have build a sunrise lamp, based on triac switching. The PWM scwitching has been accomplished by a linear 30 min voltage ramp and a 100 Hz sawtooth voltage, which are combined to a variable duty cycle rectangular voltage by a comparator. The system works fine, but the increase in brightnes is not linear with the voltage ramp: at first the brighthess increases slowly, it is most rapidly increased at about 15 minutes, and the brightening slows again at the end of the (linear) voltage ramp.

I am not sure, but I think the reason for this unlinear increase is the sine waveform of AC, which seems to cause the increase in voltage to follow an "S-shape" allthough the increase of the duty cycle of the triggering singnal is linear.

How to solve this unlinearity? Is it possible to construct an "inverted S" voltage ramp to compensate the unlinearity. Bellow is a crude modell as a character graphics.

Voltage x
I x
I x
I x
I x
I x x x
I x
I x
I x
I x
I x
__________________________________> Time

Mikko

 

Can you use a microcontroller with PWM and a lookup table.

__________________
Bill
Smart Kits build Smart People
http://www.blueroomelectronics.com

 

Unless you're using a true RMS meter you can't measure the voltage properly.

The human eye has a logarithmic response but incandescent lamps increase their brightness exponentially with the increase in voltage so it should cancel that out.

You could use an RC circuit to increase the signal voltage going into the dimmer but a 30 miniute delay would require a huge capacitor, resistor and high impedance buffer.

__________________
I also post at the following sites:
http://www.stop-microsoft.org http://www.heated-debates.com
Screen name: Aloone_Jonez
And http://www.silicontronics.com, same screen name as here.

 

I agree with Bill that a microcontroller is the way to go. As noted, the apparent brightness change is affected by both the nonlinearity of your eye and the nonlinearity of lamp brightness verus ramp position on the sinewave. You will thus likely need to determine the best ramp shape by trial and error, and a µC makes it easy way to change the ramp slope and generate any arbitrary shape.

__________________
Carl

 

Thanks of your replies.

Unfortunately I am too unexperienced to playt with microcontrollers and I am looking for an analog solution. One could be summation of two voltage curves, one convex and the other concave

The first convex part of the "inverted S" curve seems to be easy: it is just the charging curve of a capasitor via a resistor at constant voltage. The second , concave part is the problem: the voltage should be rising exponentially. I suppose that here some kind of amplifier with feedback is needed. But can anybody give mee advise how to build such an amplifier with controllable feedback? I should be able to control the exponential voltage rise within 30 minutes time.

An other possibility might be a potentiometer with its wiper connected to a fixed voltage divider constructed bu two resistors from the wiper to the ends of the potentiometer. The resistors make the voltage change vs viper position curve "inverted S" shape. The potentiometer could be turned by a constant rate, or it could be a digital potentiometer with a step controller

See http://sound.westhost.com/project62a.htm

Mikko

 

This kind of curve is inherent with SCRs and TRIACs and maybe has to do with device physics. A linearizing look up table with digital control may fix it, but it's a lot of complexity for a small improvement. There doesn't seem to be a consumer demand for this kind of fix.

 

Hi
I agree that the nonlinearity of my sunrise lamp is not very disturbing. Maybe I will let it be as it is.... Maybe we hobby electronics builders feel that we never get our projects finished and make changes untill we have a handful of spoiled components in our hands.

MikkoJ

 

On second thought, if you work this out graphically using the conduction angle on a sine wave, I think you will get this waveform even with a perfect triac.

 

Hi everybody

I think I have resolved the problem of creating an ”inverted S” voltage curve in order to correct the inherent nonlinearity of my sunrise lamp’s triac power regulator. After posting my first question on this forum I already gave up, but it did not stop my thinking. While lying on my sofa I realized that one way to solve the problem is combination of two exponential voltage curves, one convex and another concave. But how to make such voltage curves?

I have been reading the basic tutorials of electronics in the net, including those dealing with bipolar transistors. I remembered that when the base voltage is very low the transistor’s response to base current is nonlinear. I also realized that the collector (or emitter) current curve as a function of base curve of a PNP transistor is convex and that of a NPN transistor concave. I decided to make a circuit of one PNP transistor (BC557) and one NPN transistor(BC547) with their emitters coupled together and a voltage divider (R4-R5) giving the emitter voltage as half of the total voltage of the circuit. My first experiment was that I joined the base resistors (R2-R3) to the wiper of a potentiometer (R1) (see the attached circuit image, with points A and B directly connected). Using this potentiometer I increased the in voltage by 1 V increments (point A) and measured the voltage at the emitters of the transistors (point B). I was quite excited when I found that my idea was correct!

I, however, had a doubt that maybe the system is not working similarly with the voltage ramp generated by an op amp in my sunrise lamp. That is why there is an op amp (UM741) voltage follower between the potentiometer and the base resistors in the attached circuit. This final circuit is in the attached image, together with the input-output plot obtained with the given component values.

The curve does not exactly follow the optimal voltage course to correct the triac distortion, but if the inversion point of the ramp is adjusted to a voltage resulting in an on-time of ¼ cycle of AC, the increase in the lamp power should be more linear with time than without the circuit. The adjustment of the inversion point is easy: you just adjust the emitter voltage by the bridge R4-R5.

Around the inversion point the voltage rise almost stops. I think the reason is the 0.5 V threshold voltage of a silicon transistor. A total of 1 V change in input voltage (0.5 V for the PNP + 0.5 V for the NPN) is needed before the output begins to increase. This plateau could be shrunk to about 0.6 V by using germanium transistors. Maybe there are other ways to affect it, but I am too inexperienced to image how.

The input voltage in the attached circuit starts at about 2V. I think the reason for this is the UA741 op amp, which cannot reach the 0-voltage rail. It is not very important in a sunrise lamp. The triac must be about 10-20 % conducting before any light is produced by a 60 W incandescent lamp.

mikkoj

Attached Images

Inverted-s-mj1.jpg (79.1 KB, 3 views)

 

Diodes and resistors will give piece-wise linear approximations of many curves, with more diodes = better approximation.