LM3409HV: Problem when recalculating values on LM3409 to decease curren

Hello Ramiro,

For the shutdown mode, you want to pull EN high and just PWM UVLO. You have the diode connected to UVLO correct, just need to disconnect the cathode from EN and pull EN high.

You cannot tell what the current is to the LED by measuring the forward voltage and interpret from the datasheet. I would place a small resistor in series with the LEDs and measure the voltage across it (make sure it can handle the power). Or, better yet, if you have a current probe to see the output current.

As for the start-up it looks like the LM3409 is hitting maximum off-time. This means the output voltage is not getting high enough to reach the off-time threshold. A diode and resistor can be added from VCC to get the LM3409 to start. The divided voltage with the diode needs to be above the off-time threshold (1.243V), see Figure 27 in the datsheet. If Roff is 5.6K ohm a 10K ohm for the Roff to Vcc along with the diode will avoid the maximum off-time issue.

It seems this is running fairly high switching frequency.

Best Regards,

Hi Irwin:

Thanks for your reply.

Honestly, I dont know what do you mean with "You have the diode connected to UVLO correct, just need to disconnect the cathode from EN and pull EN high." How is posible to have anode connected correct and need to disconnect the cathode? : My PWM signal is going to EN, when the PWM signal is low the diode pulls UVLO down and when it is high the diode is reverse biased and UVLO goes high . Thats the way was recommended to avoid the delay. Can you please clarify this?

On the other hand, I do not understand the off-time comment. As i commneted previously, this effect does not happen when I use Rsns under 0.06, So I do not understand why is happen whit Rsns over 0.06.

Hello Ramiro,

Before you do any changes can you check the output voltage with an oscilloscope when it is trying to turn-on (so you can actually see the output voltage amplitude after the first pulse.  It would be nice to see the Coff voltage as well.  This may not be an EN/UVLO PWM issue, if it is see below.

Okay, you have a diode between UVLO and PWM.  Disconnect EN and pull it high, you don't want to PWM EN with low frequency.  Just PWM the UVLO with the diode in place so UVLO still works.  If EN is held low long enough VCC discharges and goes into low power shutdown.  Just PWM UVLO with the diode and pull EN high.  You can also try increasing the capacitor on VCC if it is close to working as it is.  You can check this by monitoring VCC (with an oscilloscope) to see if it is discharging to it's UVLO threshold>

If Rsns is lower the current will be higher, even the first pulse which may be enough to pull the output high enough for the off-timer to work.  Can you take an oscilloscope and look at Vout when it is first turning on with multiple pulses?

The other thing to try is the diode and 10K ohm to the Coff connection to make sure there is enough voltage to trip the off-timer.  This will probably solve most of your issues.

Best Regards,

Hi again Irwin:

Thanks a lot. One of your tips gives me a solution for one of the issues. Changing the capacitor of Vcc from 1 us to 4.7 us  (Cf1 in my design) solve the initial delay that I had with the PCB without changes ( Rsns 0.05). I have make some test, and it works with or without  UVLO-PWM diode. Here is the oscilloscope graph. The initial delay has totally disappear.

BUT I still need to increase Rsns to decrease the current at the output. So my next step has been again increase Rsns. Just using 0.068 ohms, I have again an interrupted pulse, as follows

I dont know if this effect its the same tan before. If I understoood correctly this is because, as I have decrease the current, the output voltaje has also decreased. Is that correct? I dont know how can I change that, just with the design I have. I dont want to add a resistor and a diode between output and Coff, unless necessary because I have some PCBs already done. But anyway if it is the only option I Will try. 

What should I try to correct this?

Thanks again for your help in advance

Thank you very much

Hello Ramiro,

You need to look at the output voltage and see what it is getting to during the pulses and/or the voltage at Coff.

What is your output? The schematic only shows SAL1 but I don't know what it is connected to. Is there a capacitor?

Best Regards,

Thanks.

I have 2 strings in parallel of 3 leds in series. No capacitor or resistance

When there is a pulse, i have 8.80volts at the output (first graph). When the output pulse is interrupted, I have less, about 8 volts.

I Will check the V in Coff in both cases(in some hoirs), but what should i expect, and what should i look for un case V in Coff is less than threshold? Why is happen?

Hello Ramiro,

If V in Coff is cannot get to the trip threshold it will cause a maximum off-time of 300 us and it will try again. Without an output capacitor it should work. I don't understand why the voltage doesn't just go to the Vf of the LEDs. It's ramping up at the beginning of the PWM pulse which it shouldn't do if there is no output capacitor. It should just go to the Vf of the LEDs.

Best Regards,

Well, I should had said that these graphs are just before the inductor, not after that. But for sure there is no capacitor. There are directly 2 leds string s un parallel

Hello,

If this is before the inductor, basically Vin? Why is it dropping when the PWM pulse happens?

Best Regards,

Sorry I did a mistake, No, the graphs are after the inductor. I put them here again:

Rsns 0.05, Roff 5k6, Cf=4.7uF. Blue line before inductor, yellow line after inductor

Rsns 0.05, Roff 5k6, Cf=4.7uF. Blue line before inductor, yellow line after inductor, the first pulse of the train of pulses is about 90us, with 267 us until second pulse. this second pulse is 30 us

And here are joined Vout for Rsns 0.05 (yellow) and Rsns 0.06(blue).

I have measure Vcoff, and honestly I am not able to see more than a small peak, about 800 mV-1 V in both cases (even with Rsns =0.05, where I have correct output), and with a very noisy signal. Don't know why. Its what I have (very strange):

I have try to add a diode and a 10k resistor between Vin and Coff. this is the result. Honestly, dont know what is happen. Blue line output, yellow line Vcoff. 1.60V in the peaks, over the trheshold. But I dont know why this shape and Why is still not working

Hello Ramiro,

Are you supplying this 12V? If so why isn't the blue line at 12V when the LM3409 MOSFET is on? If they are at the same level you cannot build current in the inductor. Check your power source to make sure it is at 12V as well as Vin to the LM3409 and the CS resistor.

Best Regards,

thanks again for your mail.

Does not seems to me that Input and output voltaje are too close.

Here you have: Rsns 0.05 . Blue line Vin at CSP point (Rsns+Cin node). Yellow line the output (about 8.8). Vin is 12 V except a small drop when MOSFET is conducting

Then we have change Rsns by 0.06. Blue line Vin at CSP point (Rsns+Cin node). Yellow line the output (interrupted pulse). Vin is 12 V , obviousy there is no drop as there are not real conduction

Regards

Hi irwin

Does It means that RDSon should be included in the datasheet calculation? In other words, Icurrent fórmula depends on efficiency and Rsns. Should be Rsns then Rsns+RDSon when i calculate the current?

Moreover, ok, There is a Big drop  of V in MOSFET due to the MOSFET resistance. But why is this affecting the behaviour of my circuit xxxvwhen I increase Rsns? Because i guess that, as Rsns is increasing and the current decreasing, the MOSFET voltage drop become smaller.

Best regards