adaptive Constant OFF time LM3445

lm3445

can anyone explain the function of R4 , Q3 and C11 for constant turn off time circuit of this LED Driver LM3445 from NS.

**broken link removed** - page 13 figure 11. The functions of these 3 devices were not discussed in the data sheet.

off time

Q3 and R4 are used to generate a "constant" current (along with the LED's) to drive into the capacitor. while the mosfet is turned off, the current is charging the capacitor. When the cap has a voltage of about 1.25 volts, the mosfet is turned on.

The time to charge the capacitor to a specified voltage with a constant current, will take a constant time. Hence the constant time off. (C V) / I = time.

Oh, it looks like the chip will discharge the cap through about 33 ohms when the mosfet is on, and then start the charging again whent he mosfet is turned off.

constant off-time

Hi stefannm

Thank you. its very informative.
Can you please elaborate more? What about the PNP?
Especially the 33 ohms you are talking about? Is it necessary for the Constant Off Time?

_andrew

because I am doing my own LED Driver, and I want to do the Constant off time mechanism for the controlling the current. Please help.

lm3445.pdf

Assuming continuous conduction mode, the inductor is going to keep some amount of current in the LED's (DC wise, you can think of the average current here). This current is going to create a voltage drop across the LED's. This voltage minus the Vbe of Q3 should be divided by R4 to set the current charging the cap (approximately).

Then if you look at the simplified block diagram (Figure 1), You can see that the cap voltage will be fed into a comparator. Notice he 33 ohm mosfet pulling that path to ground. If you figure out the logic, you will see that this switch is off while the cap voltage is below the reference 1.276 volts. Then the switch is turned on when the cap voltage meets that voltage.

The 33 ohm switch is important, but the actual impedance would be negotiable. Smaller is better. You want to keep the discharge time as fast as possible so this circuit does not affect the on time.

The off time is timed by the cap charging, hence the name constant off time.

lm3445 design

stefannm,

Can you check page 22, why is that in the computations, he used different values for the Fsw (number3 and 12).
The only Fsw I know is the pulse freq for the Gate of the PowerMOS. Is there any other Fsw in the circuit?

lm3445 mode

It looks like the change in frequency comes from picking a common capacitor value in step 10. If you use equation 9 to calculate the new Toff with 120 pF cap. Then f = 1/(Toff + Ton) ~ 350K.

constant off time

stefannm,

i tried to do the same design but my result is not accurate, i think. Does the controller do something about the constant off time? because I can see an "OR" before it is applied to the S of SRfipflop.

I tried to connect the output of the comparator directly to S. And I also have the MOS 33ohms as what you have mentioned before.

Please help.

------------------------

actually, before I am just using an Oscillator for S. And I have a very constant current. but now I am trying to change the oscillator to an constant OFF time control but the output current is not stable anymore. I dont know what is my mistake here.

Andrew

It looks like the controller initiates the start (the part that goes into the or gate) and it resets the flop (Likely when the pull down mosfet (33 ohm) gets turned on, with the appropriate delay of course)

Hope this helps.

hi stefannm,

you mentioned before that "while the mosfet is turned off, the current is charging the capacitor. When the cap has a voltage of about 1.25 volts, the mosfet is turned on. " .... are you extremely sure about this?

The capacitor will only charge if the mosfet is turned off? and will turn on if the capacitor voltage increases morethan 1.25?

****************************

can you also check if my understanding is accurate. I believe that the capacitor is continuously charging, but if the Gate of MOS is ON then the 33ohms resistor will just absorb the current/voltage? and when the powerMOS is turned off, the 33ohm will be inactive and there will be voltage input in the COMPARATOR? Is my understanding correct?

Please Guide me

--andrew

By mosfet, I was referring to the internal 33 ohm mosfet, not the PowerMOS of the buck converter. But, they essentially share a gate signal so the point is moot I guess.

Anyways, the current that is used to generate the constant time if is Icoll. See the plot on pg. 13 of the datasheet. The internal moset is on when the PowerMOS is on. During this time all of the current will be shunted to ground. The voltage across the cap should be "zero" and should not be charging. During the PowerMos off time, the internal mosfet is off allowing the cap to charge until it reaches the internal reference voltage. Then the internal logic switches the internal mosfet and the PowerMOS on. Since you have a constant current on a cap, the charge time is constant, so the PowerMOS off time is constant, hence the name.

If your cap is charging during the on time (when the mosfet is on) then it means one of three probable things.
1. The logic controlling the gate is incorrect.
2. Icoll is too high.
3. The resistance of the mosfet is too high.