LM3409HVEVAL: Question regarding external PWM contol

Hello Rich,

The LM3409 is very versatile from a dimming standpoint, and you are most likely nearly correct in how it operates. Depending on how fast of a rise time you need for your current you may even be better off eliminating components for shunt FET dimming since it is such a fast device. I'll go through some of the dimming methods and let you decide or ask more questions:

1. In your case the EVM does not have extra circuitry for the COFF timer as explained in the latest datasheet. So when the shunt FET turns on the device will go into maximum off time/minimum on time mode because there is not enough output voltage to charge the COFF pin to 1.24V. So the current will not be zero but very low. This can cause delays in the current rise when the FET is turned off, but if timing requirements are not stringent it is a perfectly viable option.

2. If you add the off timer circuitry and match the inductor ripple depending on the shunt FET used it would operate like you are used to where the FET takes the 5A while it is on. This gives a much faster rise time for applications that require it.

3. EN/UVLO pin dimming. If the current rise times are fast enough (limited by (Vin-Vout)=Ldi/dt) you could eliminate shunt FET components altogether. This is a very fast device. I recommend the UVLO pin with a schottky diode, connect the anode to UVLO and the cathode to the PWM dim signal. I mention that because UVLO and EN act identically except EN will allow VCC to discharge and go into low power shutdown mode (startup delays). UVLO will stop all switching the same but will react instantly when pulled high no matter how long between pulses. You mentioned there could be some time between them.

So if you like to have no switching between pulses and the v=Ldi/dt is fast enough for you application you may want to save components and do it that way. Otherwise you have both options with shunt FET dimming if it works for what you need.

Best regards,

Clint

Clint, thank you for your detailed and prompt response.

To follow up on your approach of implementing PWM via the UVLO pin.  When you say "connect the anode to UVLO and the cathode to the PWM dim signal"  Is this the PWM2 signal that comes out of Q2 on the EVAL board? Or is this an external, say 0-5V signal that I would use to pull down the UVLO pin through a shottky diode?

And then with respect to minimizing components, are you saying that if I implement PWM via the UVLO pin I could remove all of the parts that support external PWM? (Such as Q2 and Q3 shown on the EVAL circuit). With this approach would the EN pin always be kept enabled?

Thanks again,

Rich

Hello Rich,

You would apply your own PWM signal to the schottky cathode. So when your signal is low it pulls UVLO low and it stops switching, then when your signal is high UVLO will go back high and it will start switching. So since you are not shunt FET dimming in that case you could remove R1, R2, R3, R4, C5, C9, Q2, and Q3 from the circuit. You would keep EN high always unless you need a low power shutdown feature.

Best regards,

Clint

Hi Clint:

So you are saying that driving UVLO with a PWM signal (through a diode) will switch as fast as (or nearly as fast as) a circuit implemented with external PWM? I've set my switching frequency at 400 kHz and would like to use a PWM frequency of 30 kHz. I also need to have some amount of linearity when my duty cycle is getting down around 1%. (I need to go 30kHz because this is part of an imaging system that uses very fast frame rates, if the PWM frequency is too low you will see the LED 'beat' with the camera frame rate.)

Rich

Hello Rich,

I see, more information helps for sure. 30kHz is pretty fast and using UVLO will not be as fast as the shunt FET dimming. If you need to get to 1% shunt FET dimming is your best bet. UVLO dimming can actually probably do it but the linearity will be worse. If that can be compensated for (it's an issue with any switching regulator) that is fine, otherwise I would stick with shunt FET dimming in your case. For the fastest and most linear response check the latest LM3409 datasheet for additional COFF circuitry that isn't included on the EVM. The TPS92515 datasheet has even a little more information (it is a very similar part, just integrated). When the shunt FET is on the output voltage is too low to charge the off timer, so adding the extra components avoids maximum off time and keeps the inductor current up for fast LED current rise times. Feel free to let me know if you have any questions on it.

Thanks,

Clint

Clint, thanks again for all the help. Looks like staying with external PWM suits my needs the best. But now I know.

Rich