Is there any circuit to make incandescent bulbs (60-100W) light up more gradually? I've seen many different devices but none of them would work for me. I do not want to just stop the high current spike while the filament is warming up. I want the bulb to have 1.5-2 seconds delay before glowing fully bright. My only find is this Russian video of a light bulb soft start(Sadly I can't just buy it from the Russian store) Most other designs just focus on the inrush current limitation. I want is because I just like the effect of a bulb gradually increasing its brightness. Can this be done without fancy micro controllers or specialized IC's? Could you send any schematics that would work? (The mains voltage in my country is 230V AC)

  • \$\begingroup\$ There are commercially available dimmers with pushbutton on-off which does this for you automatically even if the setting is 100 %. NTC would do the job too. \$\endgroup\$ – winny Sep 1 '17 at 13:03

You can run LEBs (light emitting bulbs) with PWM. In fact, soft start actually increases their lifetime.

It is possible to chop AC, but generally easier to rectify the AC then chop the resulting DC. Four diodes and a capacitor are good enough to make DC to run the bulb from. You also need to make a small low voltage supply to run the processor from.

The switch itself is probably most easily realized by a NPN transistor rated for the voltage. It can be controlled directly from the low voltage output of the microcontroller PWM.

Note that the rectified DC will have a higher RMS voltage than the AC. You compensate for that by not going to 100% duty cycle for full on. You could even measure the DC and adjust the duty cycle on the fly to make the light intensity relatively insensitive to voltage fluctuations of the AC. You could even make a "universal" powered light. That's one that can run from 90 to 250 V, 50 to 60 Hz.

  • \$\begingroup\$ I was thinking about chopping AC (i.e. leading edge dimmer), but your AC->DC preconversion is a better idea. As the OP did specify he does not want to use MCUs, you can generate that slow turn on using a good old 555 timer in astable configuration. The slow-turn on can be achieved for instance slowing changing the voltage at the control pin, with an R-C. \$\endgroup\$ – next-hack Sep 1 '17 at 11:52
  • \$\begingroup\$ "Note that the rectified DC will have a higher RMS voltage than the AC." You can avoid this by not filtering the rectified AC. This would eliminate the need for PFC as well. \$\endgroup\$ – Dave Tweed Sep 1 '17 at 11:57
  • \$\begingroup\$ This solutions seems quite easy and convenient but I would really want to avoid using a micro controller. I want to make it compact and cheap. Adding a small dc supply for the PWM controller would be another problem. Also if I end up using a Pulse Width Modulated DC I will switch from Incandescent Bulbs to LED strips/lights. Sadly most of the LED bulbs can't be dimmed and have no soft start functionality built in. And no LED will replace the warm glow of a regular light bulb. EDIT - The 555 timer idea is nice but I would still like to avoid making a low voltage supply. \$\endgroup\$ – Velociraptor Sep 1 '17 at 11:57
  • 3
    \$\begingroup\$ @Velociraptor I thinkg the suggestion of DaveTweed is also nice. You can still use a capacitor dropper + zener (on the AC side!) to create a very crude low voltage supply for the 555. Then use an optocoupler to drive the BJT on the rectified - but not filtered - mains voltage side. \$\endgroup\$ – next-hack Sep 1 '17 at 12:36
  • 2
    \$\begingroup\$ @Velociraptor The way to make it compact and cheap is to use a microcontroller. Any other solution will be larger, more expensive, require more parts, and be less flexible. \$\endgroup\$ – pipe Sep 1 '17 at 21:39

The common inrush limiters are just NTC (negative temperature coefficient) thermisters that are placed in series with the bulb. You just need one that has a longer thermal time constant. Pick one that has suitable voltage and current ratings, and also make sure that its steady-state power dissipation is at an acceptable level. You'll probably need a different one for each type of bulb you want to control.

  • \$\begingroup\$ Thanks for the suggestion. I know that an NTC can be used but finding the correct one (with the correct values) is very difficult (there aren't any stores which sell them near me) and I want to avoid having a hot thermistor in my lamp. \$\endgroup\$ – Velociraptor Sep 1 '17 at 12:10
  • \$\begingroup\$ I remember when you used to be able to buy a thin disk containing an NTC device that you put into the light socket, and then you screwed the bulb down on top of it. \$\endgroup\$ – Solomon Slow Sep 1 '17 at 17:26

You were right, all the schematics on google were either quite outdated or not really compatible.

enter image description here

This timeless classic has two issues: unlike a comparator, the diac will discharge the capacitor on every triggering, and the voltage on the diac and gate is AC, so we can't make a ramp by charging a capacitor, because this requires DC.

Some people got creative:

enter image description here

Here we have a rectifier, charging a capacitor with a ramp, then a P.U.T no less, acting as oscillator and driving the triac gate through a pulse transformer. Two components which should be difficult to source...

enter image description here

One solution would be to put the whole thing inside a diode bridge, so we're able to work on DC... although this one isn't a soft start, but you get the idea. However, I'd like the switching noise to stop completely after the soft-start is finished.

Of course, everyone would use a microcontroller, these days.

enter image description here

This one has a transformer because it's for low voltage bulbs. But it uses two 555s to create a PWM, also a voltage regulator... so high tech!

So, I propose this:

enter image description here

First, the AC mains is rectified. No need for a smoothing cap, we're driving a lamp.

Next comes one of those lovely "kill yourself" power supplies, basically R1/D2 are our dropper resistor and 12V zener, D5 rectifies, D3 mostly decorates (forgot to remove it), and C4 jump-starts the charging of C2, which would otherwise take a while, as I used as little current as I could get away with.

Next the supply is split in half (R4/R10) and filtered by C3 to make a slow ramp, on node "REF".

The rectified mains is used as PWM sawtooth, after division via R5/R6, and compared with the reference.

The comparator drives a MOSFET.

Parts selection advice after dinner (unless I made a mistake, this was kind of a rush job).

So, shopping advice for 230V:

  • FET - 5-600V to be safe, like these.
  • R9 is the lightbulb
  • R6 R1 R8 use resistors rated for 400V or two resistors in series. R1 should be 0.5W... maybe 1W...
  • Diode bridge: pick from junk parts box, 5-600V, a few amps...

Alternate Solution:

In the first schematic above (triac dimmer) replace pot with LDR (need to check voltage rating, though), light it with LED powered by ramping current. Might be simpler... you'll need a capacitor dropper for the LED though.

  • \$\begingroup\$ There are many schematics but I don't know which is good. Most of them are designed to run a high-powered motor not a small light bulb. Some of them are overly complicated. If someone could show me the correct one I would be very thankful. \$\endgroup\$ – Velociraptor Sep 1 '17 at 15:52
  • \$\begingroup\$ Here's one ;) check the edit \$\endgroup\$ – peufeu Sep 1 '17 at 17:49

I use this dimmer, it has to be connected in series with the lightbulb and it gradually increases the brightness to full over ~5 seconds. If, while the brightness is still increasing, you quickly switch the lighbulb off then back on, the brightness stays at that level.

I bought those devices specifically to increase the lifetime of the bulbs by having them slowly warm up to full brightness. It seems that they help as I am now very rarely replacing bulbs.

(I am sorry if recommending a commercial product is against the rules. I have no affiliation with the company that makes those devices, I just happen to use a couple of them and am happy with their function)

  • \$\begingroup\$ That is what I've been looking for. If only there was a schematic for it do do it by myself. \$\endgroup\$ – Velociraptor Sep 1 '17 at 13:18

Treat it as a constant-current device

Why not? Every other illuminator - LED and discharge (neon, fluorescent, LPS, HPS, MH, MV) - all need to be driven in constant-current mode, because they are non-linear. Discharge lighting particularly, once the arc is struck, resembles a dead short. LEDs are more linear but still pretty non-linear.

Incandescent lights are linear - once lit. Before they are lit, their resistance is very low, sort of the inverse of the discharge light. This low resistance causes inrush current, and it quickly gets hot enough for its resistance to increase and it becomes linear at that point.

Why do we attach them directly to the power line? Because we can get away with it if we're willing to endure the in-rush current. The benefit is we can get away without a ballast/driver.

But what if you used a constant-current driver anyway? On startup, the low resistance at spec current would result in low voltage, so low wattage, and slow warmup. Resistance would increase as it warms up, causing voltage @ spec current to increase and thus total power to increase. It should "spin up" slowly. Whether that's the 2 seconds you want would need to be a matter of experimentation.

Need to be matched to bulb?

Suppose your constant-current driver is calibrated for an 100W bulb (0.8333 amps at 120V). And you put a 60W bulb in there, which wants 0.5 amps. The constant-current supply will push 0.8333 amps initially, which will cause warmup somewhat faster. Once the bulb is at operating temperature, the driver would have a problem: In order to push 0.8333 amps through the bulb, it would have to increase voltage above its input voltage of 120. If the driver is operating in buck mode, it would not be able to do that. It would simply deliver near 120V to the bulb, and the bulb can handle that, drawing 0.5A. So driving a smaller wattage bulb would still work.

If you put a 150W bulb (wants 1.25 amps), the driver would limit at 0.8333 amps. The 150W bulb would start slower, and slowly get up to 0.8333 amps, which would occur around 80 volts. And that's as bright as the 150W bulb would get: 66 watts. It would fail, but fail soft.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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