design an LED driver with 12VDC output and 340 mA output

[yassin.kraouch]

Hi i want to design 12 VDC 340 ma driver, to driver LED from main AC voltage,
have you please a reference design or link to help me do that ???
thanks

 

[pradeep.g.belchada]

use powerintegaration ic

http://www.pi.com
or
use uc3844 and moosfet flyback smps

 

[steveelliott]

I think pradeep's suggestions are good. The proper links are:
**broken link removed** or
**broken link removed**

 

[yassin.kraouch]

thank you for the link, the first one is not safe because it is not isolated, i will see the second one and i will come back to you

 

[yassin.kraouch]

i just found another cheaper one, **broken link removed**
it is PFM ( pulse frequency modulation)
and few external component

 

[steveelliott]

Good call, yassin. The BCD part looks pretty nice. The PFM mode is helpful because the pulses are all the same size - it makes your bench testing easier (the EMI testing is a bit more lengthy, but not bad). With PFM the pulse (i.e. peak current) doesn't change depending on load so it's easy to make sure your voltage overshoot and snubbing are right.

 

[yassin.kraouch]

Good call, yassin. The BCD part looks pretty nice. The PFM mode is helpful because the pulses are all the same size - it makes your bench testing easier (the EMI testing is a bit more lengthy, but not bad). With PFM the pulse (i.e. peak current) doesn't change depending on load so it's easy to make sure your voltage overshoot and snubbing are right.

thanks for the explanation, can you please give me the difference between PFM and PWM ?

 

[steveelliott]

Sure, glad to. In PWM, the frequency is fixed and the pulse width varies from 0% to 100% (sometimes the limits are not 0 and not 100 but pretend they are for now). Assume we have a 20kHz PWM frequency. The period of a 20 kHz waveform is the reciprocal of 20kHz = 50us. The pulse width varies from 0 us to 50 us. If the pulse width is 5us, that means the pulse width is on 5us out of 50us, or 10%. If the input voltage is say 75V, then the *average* output voltage is the 10% times the input voltage of 75V = 7.5V. If your circuit needed 15V average on the output, then it would adjust the width of the pulses to 10us or 20% because 20% of 75V = 15V. If the input voltage dropped to 60V, your circuit would automatically change the pulse width to 12.5us because 12.5us is 25% of 50us and 25% of 60V is 15V.

In PFM, we make all pulses 50us wide. (the maximum frequency would be the reciprocal of the period of the single pulse: 1/(50us) = 20kHz). We if we want 10% duty cycle, we put out the 50us pulse and then wait 450us and the fire another pulse (frequency = 1/(50us + 450us) = 2kHz). The average value is still 10% (50us pulse and 450us no pulse). if we need a 16% duty cycle, the wait time would be 263us (50/(50+263) = 20%) and the frequency would be 3.2kHz. The advantage of PFM is that each (50us) pulse is identical - nothing changes between pulses; in PWM the size and energy of the pulse changes so you need to test it very carefully to make sure that the possible pulse widths all look right on an oscilloscope. The second main advantage of PFM is that (almost always) the energy in each pulse is completely transferred to the load.

In contrast, in what is called Continuous Current Modulation, the storage element (the inductor or transformer) doesn't transfer all of its energy every cycle (current stays flowing in the inductor "continuously" - hence the name). When the power devices turn on, they not only have to switch the voltage, they also need to switch the inductor current too. Switching current and voltage means power is dissipated in the semiconductors and you have to check that very carefully on the oscilloscope. The advantage of CCM is that the inductors are smaller and dissipate less power than "Discontinuous Current Mode" operation.

Some circuit topologies, Flyback in particular, are natural candidates for PFM because they MUST transfer all the energy in each pulse to the output or very bad things will happen. The penalty for PFM is that you need to make sure the rest of your circuit does not respond to ANY frequency from 1kHz (5% duty cycle) to 20kHz(100% duty cycle) since your PFM circuit can produce any frequency in that range; with PWM you only need to check the rest of your circuit at the PWM frequency. (Conservation of Misery applies).

Hope this helps.

 

[ahmed-agt]

Sure, glad to. In PWM, the frequency is fixed and the pulse width varies from 0% to 100% (sometimes the limits are not 0 and not 100 but pretend they are for now). Assume we have a 20kHz PWM frequency. The period of a 20 kHz waveform is the reciprocal of 20kHz = 50us. The pulse width varies from 0 us to 50 us. If the pulse width is 5us, that means the pulse width is on 5us out of 50us, or 10%. If the input voltage is say 75V, then the *average* output voltage is the 10% times the input voltage of 75V = 7.5V. If your circuit needed 15V average on the output, then it would adjust the width of the pulses to 10us or 20% because 20% of 75V = 15V. If the input voltage dropped to 60V, your circuit would automatically change the pulse width to 12.5us because 12.5us is 25% of 50us and 25% of 60V is 15V.

In PFM, we make all pulses 50us wide. (the maximum frequency would be the reciprocal of the period of the single pulse: 1/(50us) = 20kHz). We if we want 10% duty cycle, we put out the 50us pulse and then wait 450us and the fire another pulse (frequency = 1/(50us + 450us) = 2kHz). The average value is still 10% (50us pulse and 450us no pulse). if we need a 16% duty cycle, the wait time would be 263us (50/(50+263) = 20%) and the frequency would be 3.2kHz. The advantage of PFM is that each (50us) pulse is identical - nothing changes between pulses; in PWM the size and energy of the pulse changes so you need to test it very carefully to make sure that the possible pulse widths all look right on an oscilloscope. The second main advantage of PFM is that (almost always) the energy in each pulse is completely transferred to the load.

In contrast, in what is called Continuous Current Modulation, the storage element (the inductor or transformer) doesn't transfer all of its energy every cycle (current stays flowing in the inductor "continuously" - hence the name). When the power devices turn on, they not only have to switch the voltage, they also need to switch the inductor current too. Switching current and voltage means power is dissipated in the semiconductors and you have to check that very carefully on the oscilloscope. The advantage of CCM is that the inductors are smaller and dissipate less power than "Discontinuous Current Mode" operation.

Some circuit topologies, Flyback in particular, are natural candidates for PFM because they MUST transfer all the energy in each pulse to the output or very bad things will happen. The penalty for PFM is that you need to make sure the rest of your circuit does not respond to ANY frequency from 1kHz (5% duty cycle) to 20kHz(100% duty cycle) since your PFM circuit can produce any frequency in that range; with PWM you only need to check the rest of your circuit at the PWM frequency. (Conservation of Misery applies).

Hope this helps.

Good explanation! please can I have some reference on PFM and PWM.

 

[yassin.kraouch]

Sure, glad to. In PWM, the frequency is fixed and the pulse width varies from 0% to 100% (sometimes the limits are not 0 and not 100 but pretend they are for now). Assume we have a 20kHz PWM frequency. The period of a 20 kHz waveform is the reciprocal of 20kHz = 50us. The pulse width varies from 0 us to 50 us. If the pulse width is 5us, that means the pulse width is on 5us out of 50us, or 10%. If the input voltage is say 75V, then the *average* output voltage is the 10% times the input voltage of 75V = 7.5V. If your circuit needed 15V average on the output, then it would adjust the width of the pulses to 10us or 20% because 20% of 75V = 15V. If the input voltage dropped to 60V, your circuit would automatically change the pulse width to 12.5us because 12.5us is 25% of 50us and 25% of 60V is 15V.

In PFM, we make all pulses 50us wide. (the maximum frequency would be the reciprocal of the period of the single pulse: 1/(50us) = 20kHz). We if we want 10% duty cycle, we put out the 50us pulse and then wait 450us and the fire another pulse (frequency = 1/(50us + 450us) = 2kHz). The average value is still 10% (50us pulse and 450us no pulse). if we need a 16% duty cycle, the wait time would be 263us (50/(50+263) = 20%) and the frequency would be 3.2kHz. The advantage of PFM is that each (50us) pulse is identical - nothing changes between pulses; in PWM the size and energy of the pulse changes so you need to test it very carefully to make sure that the possible pulse widths all look right on an oscilloscope. The second main advantage of PFM is that (almost always) the energy in each pulse is completely transferred to the load.

In contrast, in what is called Continuous Current Modulation, the storage element (the inductor or transformer) doesn't transfer all of its energy every cycle (current stays flowing in the inductor "continuously" - hence the name). When the power devices turn on, they not only have to switch the voltage, they also need to switch the inductor current too. Switching current and voltage means power is dissipated in the semiconductors and you have to check that very carefully on the oscilloscope. The advantage of CCM is that the inductors are smaller and dissipate less power than "Discontinuous Current Mode" operation.

Some circuit topologies, Flyback in particular, are natural candidates for PFM because they MUST transfer all the energy in each pulse to the output or very bad things will happen. The penalty for PFM is that you need to make sure the rest of your circuit does not respond to ANY frequency from 1kHz (5% duty cycle) to 20kHz(100% duty cycle) since your PFM circuit can produce any frequency in that range; with PWM you only need to check the rest of your circuit at the PWM frequency. (Conservation of Misery applies).

Hope this helps.

Very good explanation, i understand the difference now,
but in PFM mode how we avoid that our circuit not respond to any frequency ?

 

[steveelliott]

You just need to check. Most of the time any EMI problems (in any circuit) that occur are caused by the high frequency components of the switching, not the low frequency. In our 20kHz example, the pulse is a 10kHz pulse (the maximum AC energy content is at 50us on/ 50us off) occurring at a rate of 1kHz to 20 kHz. The high frequency components are caused by the leading edge (usually not a problem since the current is 0 at that point) and the trailing edge (the worst problem because the current is at maximum when the power devices are switched off). You'll see 10kHz noise in the output of your op-amps or in the results of your a/d conversions et cetera. This test is done when you are at 50% power (10kHz pulse rate) so you have the maximum 10kHz components and the high frequency components are riding on top of the highest amplitude "low frequency" (10kHz) ripple.

The 1-20kHz stuff affects the input DC power supply because you are pulling power from the input at a 1-20kHz rate. That's where you need to put your oscilloscope probes - to see if any part of the DC input (especially the input LC components or PFC circuit) is sensitive to the varying frequency.

Usually we test both for "steady-state" instability by varying the frequency slowly by changing the output load slowly and for "transient" instability by sudden load changes. The tricky part about transient instability is that you have to change the load from "all" combinations: step-up load changes: 10% to 80%, 11% to 80%, 12% to 80%,.. 10% to 79%,.. 10% to 81%... 10% to 50% and also the same step-down load changes: 80% to 10%, et cetera. Mostly you look at spots where didn't see a perfectly smooth DC link during your slow "steady-state" testing.

 

[yassin.kraouch]

you are talking about DC input, right ? the problem is that i want to use PFM with AC input (the main)
is it the same coonsideration ?

 

[steveelliott]

Exactly. Your AC mains voltage is converted to DC in the first rectifier stage. So yes you need to look at the both the mains voltage and the rectified DC after the bridge rectifier.

 

[Ice-Tea]

i just found another cheaper one, **broken link removed**
it is PFM ( pulse frequency modulation)
and few external component

Be aware this is not an equivalent to the mentioned PI device.

- The PI device is PF corrected. The result is that you need a far smaller primary buffer cap (1 uF or so).
- The PI device uses 1 less winding on your trafo. In addition, there's that additional rectifier, cap and zener to make Vcc. The PI device has this internally.
- The PI device has an internal FET. Not only is this cost and space saving, it means the FET is matched perfectly to your driver.
- The PI part is overall better protected (thermal and overcurrent shutdown)
- The PI part runs at 132kHz, the BCD part probably less than half of that. This has direct effect on your magnetics.
- PI has grade A++ design support and excellent design tools.

So, make sure you have the full picture before you make your choice

 

[yassin.kraouch]

Good and thanks for the explanation, have you an idea from where i can get the price of this chip ?

 

[Ice-Tea]

Digi-Key gives an indication. In my experience, distributors still have some margin to beat digi for PI if they want

 

[yassin.kraouch]

i have another question, from where did you get the value of 132 khz ?

 

[Ice-Tea]

The datasheet.

 

[yassin.kraouch]

in the datasheet they specify just a frequency of 66 khz, right ?

 

[Ice-Tea]

Lytswitch-4 datasheet p13