I see that LED light is replacing nowadays every old style light. I open one LED bulb and found that had a little SMPS inside it. Those convert AC line voltage to DC to supply the LED.

Wondering we can design to connect LED's on directly on AC line . It will improve efficiency

As per attached circuit. If we can connect LED on Direct AC line

  1. Will it work?
  2. Will it be safe and reliable?
  3. Whether it needed any limiting resistor, since all AC line voltage will be drop on LED . (Vline = 230 V AC, Vp = 230 * 1.414 (+15% tol) = 360 V approx. Total no led is 120 (changed 76 to 120), each led has 3V forward drop).
  4. How much line fluctuation will effect on LED?

enter image description here

  • \$\begingroup\$ Imagine what happens when one LED fails open in the string. \$\endgroup\$
    – stark
    Aug 15, 2016 at 16:44
  • \$\begingroup\$ No. At the AC peak voltage you have 230*sqrt(2) / 76 V = 4.3V per LED, at which they will draw excessive current and die. The other 76 LEDs will do the same half a cycle later. You will need some form of current limiting; the most efficient form will be a small SMPS. \$\endgroup\$ Aug 15, 2016 at 16:49
  • \$\begingroup\$ These exist (with some form of current limit) but they have a very bad power factor since they only conduct at the peak of the waveform. There are parts to take care of several of these issues: ti.com/lit/ds/symlink/tps92410.pdf \$\endgroup\$
    – Daniel
    Aug 15, 2016 at 19:49

4 Answers 4


230V is RMS value. Peak value is 325V. So you would need 109 LEDs, not only 76.

And 230V is not precise. It can be ±15%, so peak voltage can be between 290V and 360V.

It is not very good to rely on mains voltage.


230V AC means ~325V at peak. So 325V/76 = 4.27V. At that voltage and with virtually unlimited current LED would not have a chance to survive. :)

  1. No, look explanation above.
  2. No.
  3. Yes, at least some current limiting resistor is needed.
  4. They will change light intensity depending of the voltage. Keep in mind they are current driven devices, so very low light if voltage is below at least 2.5V.

If an ideal diode is given 17 mV more voltage, it draws twice as much current. Only very non-ideal diodes would survive in the AC connection shown, because an AC line may (even if there's local protection) have occasional peaks, nearly 1V across each diode. Even a line-voltage allowed variance, of 10%, would make a normally biased (3V) LED see 3.3V, and that causes a current excursion of a factor of 2**18 (about 250 thousand times) the 'normal'.

What one CAN do, is put a series inductor, similar to a fluorescent lamp ballast, into the circuit. Then, instead of 10% voltage deviation causing 25 000 000% current increase, it causes 20% current increase.

  • \$\begingroup\$ Your answer would benefit from some real numbers. Take some LED that suits you and use the data sheet values for example. \$\endgroup\$
    – Rev
    Aug 15, 2016 at 19:48
  • \$\begingroup\$ It's not possible to extrapolate from data sheet values; they represent only test results (and destructive currents are not in the test regime, of course). For extrapolation, I've used Shockley's diode equation, I = I_s * exp (Q_e * V / (k * T)) which ignores series resistance. Data sheets DO suggest resistance, about 12 ohms for Cree CLA1A-WKW, but that is not guaranteed. \$\endgroup\$
    – Whit3rd
    Aug 16, 2016 at 7:41

Will it work?

It will work for some time, perhaps.

enter image description here

Figure 1. An graph of LED current versus voltage. Note the huge variation in current for a small variation in voltage.

Will it be safe and reliable?

No. The current will vary with voltage. A voltage spike or rise will cause a disproportionately large change in current - possibly to the point of destruction. Also a reduction in voltage will result in a large dimming of voltage.

Whether it needed any limiting resistor, since all AC line voltage will be drop on LED . (Vline = 230 V AC, Vp = 230 * 1.414 (+15% tol) = 360 V approx. Total no led is 120 (changed 76 to 120), each led has 3V forward drop).

Yes, current limiting is required.

How much line fluctuation will effect on LED?

See above.

The big problem is that the LEDs will only blink briefly at peak voltage. You can see from the graph that if we were working at 2.75 V per LED at peak mains voltage that we would get 30 mA. When the voltage drops by 20% the LED would effectively be turned off. There would be severe 100 or 120 Hz flicker.

The SMPS supplies current limited DC which solves all these problems.


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.