This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LM3404: LM3404 with "Modified COT Application Circuit" in AN-1839

Part Number: LM3404

Hi,

we built a driver based on the AN-1839. We added transistor+resistor to the LM3404 to keep the current constant over input voltage changes.

This basciall works, but it turns out that not all drivers do. Our intense investigation did not lead to a clear conclusion so far. The problem might be that while the LEDs are shorted by the MOSFET (LEDs off) the voltage at the RON is about 750mV. By the datasheet, this should be enough to power up the regulator looking at the typ values for the min RON voltage. But could be also 1V.

If the regulator is powered, it does not start up normally. Instead it tries to power up every few seconds, but does only one ON-Time, then stops again.Seems the on-time is 5us or so. CS reaches 300mV.

Finally, we used the IC from the eval PCB: it also does work normally, but instead of making only one on-time, it makes many with the correct calculated on-time of 560ns. But then stops also after some 10s us.

The circuit does this even it totall isolated, only powered and output short. 

Vin=24V

Vout (to got to the error) = 0V, otherwise 9V.

I = 1A

L = 100uH

Ron = 100k

Rsns = 200mR

Everything else like on the eval PCB.

I would be happy about any suggestions.

If the "modified COT application circuit" is replaced by a simple RON resistor, then everything works normally.

Thanks,

Simon

  • Hello Simon,

    It is hard to say without seeing it myself or having more information such as scope shots and the full schematic.

    But one thing to keep in mind if you refer to the datasheet is that the device has a minimum output voltage versus a given switching frequency. With the standard RON circuit that is not really an issue as the switching frequency will just go way down when the output is shorted. But with the additional circuitry it will stay high so the minimum Vout will be limited. You could easily verify this by increasing the Rds(on) of the FET to meet the minimum Vout (or add some R in series with the FET). That would be my initial suspicion. There are graphs in the datasheet of Vout(min) versus switching frequency.

    Best regards,

    Clint

  • Hi Clint,

    Thanks for your answer. I totally understand that this is difficult to tell from the distance. I am rather looking for clues.

    That minimum output voltage might be the cause indeed.

    But i can't find any mention of a minimum voltage for the improved circuit. Rather the opposite: in the appnote it is said that the circuit is beneficial for the shunt dimming because of the reduced switching frequency.

    I also calculated the tables like in the appnote and it seems the short does not violate the 300ns:

    Can you tell me more about that please?

    I tried to put a diode in series with the 1k base resistor. That makes the regular try to start much more often, from once every few seconds to once every 10ms.

    Also swapping the short with a diode didn't make it work. Surely it changed to waveforms, but there was no constant 1A.

    What is inside at the Ron pin?  

    Any further info or clues are highly appreciated. Especially on the minimum voltage.

    Thanks,

    Simon

  • Hello Simon,

    The RON pin is just a comparator with a pulldown. The RC charges when the switch turns on and the switch will turn off when the capacitor has charged to 700mV or so. A pulldown discharges the capacitor when the switch turns off.

    As for the extra circuitry and the app note I think it maybe just isn't totally clearly written or spelled out. It is true that the frequency will go down with the extra circuitry, but it will not decrease nearly as much as without it. So I still believe it could be a Vo(min) versus switching frequency issue. The minimum ON time of 300ns can vary significantly over the input voltage range, temperature, and from part to part. The easiest way to test is to lower Fsw and/or increase the resistance of the shunt FET path.

    Best regards,

    Clint

  • Hi Clint,

    your feeling was definately right. I started to work after we put 2 S1B diodes in series with the shunt-MOSFET. One was not even enough.

    Frequency was already down at 250k, so not much to do there.

    Well, we will now fix it with a Z Diode. That is not the best option in terms of efficiency, but gives nice dimming.

    We improved on the control side and power up the regulator with the DIM pin only shortly before the shunt-MOSFET opens, so the power loss is

    very small.

    Thank you!

    Simon

  • Good to hear Simon. I will go ahead and close this thread (it will still be viewable and you can still post in it if needed).