2
\$\begingroup\$

I'm building a LED dimmer using a 555 timer (yeah, lame I know). I'm using it to produce photographic flat frames which require perfect illuminated fields even with shutter speeds as low as 1/1000 of a second. I wanted to build a variable light dimmer which operates on a PWM frequency higher than 10kHz. I've built everything on a breadboard and it was working nicely. Now I've soldered everything to a prototyping PCB and things get weird. Not sure if something got lost in translation but I remember running this circuit for almost 1 hour, testing currents and voltage drops.

schematic

simulate this circuit – Schematic created using CircuitLab

Simulation with circuitjs: Run Simulation

enter image description here

enter image description here

The Setup

I'm driving a LED system which is kind of a black box for me. All I know officially is that it's a 12V system. I've measured the load and there's 350mA of current flowing through my power supply when I connect the LED system with 12V. This indicates a total resistance of ~ 34.3 Ohm.

Now I'm driving the LED with an NPN darlington BJT with an hFE of ~ 1000. It's a TIP212 transistor. Maybe not the perfect choice? That's what I have, along with the PNP TIP version and some low current BJTs. The 555 timer drops my voltage down to 9.86V on the output and as far as I understood, to keep forward bias, Vc cannot be greater than Vb - 0.7(?!). Since this is a darlington setup, I guess that's why I only measure 8.66V at Vc, although connected straight to the 12V supply.

I decided, due to the sufficient quality I get with 8.66V (I need a really dim light), I'll just go with that. I naively estimated the load of the LED to be around 252mA. As per the theory, I should, therefore, have 0.25mA at Ib, right? That would mean a resistor Rb of ~40kOhm right? I tried that, but the current Ib was limited and I couldn't figure out why. I started trial and error exercises and landed at 10kOhm. My testing method during trial and error was rather scientific, but I just gauged the light intensity and compared it to direct 8.66V brightness.

Problem and Symptoms:

After I soldered everything to a prototyping PCB (Schema is attached), my system started to show symptoms. There was a flickering that started to get worse and then the dimming function was gone. The LED was the same brightness even at the lowest variable resistance. I measured the output of the 555 timer using my oscilloscope and there was just straight current without any modulation. If I disconnect the circuit for some time and reconnect it, the PWM is showing for about 3 seconds, before it flattens out again. Also, the flattening is immediate.

I know a lot of you guys probably already see 10 things I'm doing wrong, and I'm so much looking forward to hearing what it is :-)

Happy new year!!!!

Update 01.01.2019

Meanwhile, I've done some more testing and thanks to the Q&A on this question I've been running some simulations with circuitjs. I've also tested to switch from the darlington NPN BJT to a power MOSFET (HEXFET from IR). Running some tests in simulation made me realize a problem which can occur if the variable resistance (potentiometer) is turned all the way down to have only minimal resistance. The circuitjs simulator must have a potentiometer with very low resistance min/max values. I've noticed that in combination with the 10nF capacitor, there was a negative voltage spike on the trigger and threshold path in the circuit everytime the cap got discharged. To prevent that, I've introduced a 150 Ohm resistor in series with the potentiometer. I've resoldered my PCB and the problem seems to be gone. I will also apply this change to the previous PCB board wich is the one I've posted this question and verify if the negative voltage spike is causing the issue there too. Does anyone know is what I observed could be the root of my problem and the symptoms I'm seeing?

\$\endgroup\$
18
  • 3
    \$\begingroup\$ Atfer a few initial glances, I would like to say that the schematic layout can be improved (so it can be easily read by everyone, including you): power supply should be on the left. There are also too many wires going to ground from below: they should go to ground from above. This might be a possible clue pointing to a possible mistaken translation to the pcb. Hope this helps :) \$\endgroup\$
    – Daniel Tork
    Commented Dec 31, 2018 at 19:55
  • 2
    \$\begingroup\$ With no current limit, I think it's your LED that stops working, permanently. \$\endgroup\$
    – Unimportant
    Commented Dec 31, 2018 at 20:05
  • 2
    \$\begingroup\$ What are your actual capacitor values? Those ones don't make any sense. 10pF and 100pF? They should be around 100nF each. Are you using an NE555 or one of the many CMOS versions?There should be supply bypass capacitor such as 100uF electrolytic in parallel with 100nF ceramic. \$\endgroup\$
    – Spehro 'speff' Pefhany
    Commented Dec 31, 2018 at 20:27
  • 1
    \$\begingroup\$ gkunz, no worries. I have redrawn the first schematic to make it more readable. It will take a while till it becomes visible, but once it does (it should), I hope you will take a look at it, for reference. \$\endgroup\$
    – Daniel Tork
    Commented Dec 31, 2018 at 20:27
  • 1
    \$\begingroup\$ The trick is to click the edit link under your question and then click the "edit this schematic" link under the schematic. No CircuitLab account is required when you launch it from this site. If you roll back the edit you can fix it. \$\endgroup\$
    – Transistor
    Commented Jan 1, 2019 at 9:29

1 Answer 1

Reset to default
1
\$\begingroup\$

Yes you are certainly over driving the LED without a current limiting resistor.

Define the LED max Current and choose a series R , Imax=Vdrop/R

How bright must this LED be to work well in 1ms? I suggest you need several watts.

e.g. 3 large SMD white LEDs in series capable of 350mA pulses for 9.x V then drive direct from 555 output is possible with 10 Ohms in series.

enter image description here

Simulation

If you buy this 8.4V LED from Digikey, also get some Resistors in the range of 2 to 10 Ohms enter image description here
then consider 555 Vout has 2.5V drop @ 200mA so 12V ecomes 9.5V max compute (Vout - Vf(LED))/If= R e.g. (9.5-8.4) /220mA = 5 ohms, 400 mA maybe 2 Ohms, never 0. ( there are better designs)

\$\endgroup\$
7
  • \$\begingroup\$ Thanks for your quick answer! The thing is, that this LED system (I believe its a LED strip with some light diffusers) itself must have a current limiter. I'm driving the 12V input of this black box system (I don't really know what's inside). But given the fact, that this system is made of operation using direct 12V DC, doesn't this mean that it already limits current correctly by itself? \$\endgroup\$
    – gkunz
    Commented Dec 31, 2018 at 20:13
  • \$\begingroup\$ Also the maximum load on the 555 timer output should never exceed 200 mA, which is why I'm using a transistor. \$\endgroup\$
    – gkunz
    Commented Dec 31, 2018 at 20:18
  • 2
    \$\begingroup\$ Provide datasheet links . Did anything get hot? Use a FET for a 12V switch not a darlington or at least the collector out. \$\endgroup\$
    – D.A.S.
    Commented Dec 31, 2018 at 20:20
  • \$\begingroup\$ The 200mA is steady but for pulses more is ok, imagine a 2 Ohm Rce in series with driver for voltage drop. Did you try my simulation>/ LEDS are like Zeners with different Vf and series R equivalent circuits \$\endgroup\$
    – D.A.S.
    Commented Dec 31, 2018 at 21:08
  • \$\begingroup\$ I think that the base resistor prevents too much current from going through the LED. Also, OP mentioned in comments that the LED can work with up to 15V. I would always add that resistor connected to collector and put the LED on that side, too. I agree that there are better designs. \$\endgroup\$
    – Daniel Tork
    Commented Jan 1, 2019 at 19:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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