• State
  • Locked Locked
  • Replies 3 replies
  • Subscribers 62 subscribers
  • Views 328 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

LM3445 Trouble

Benny Smith
Prodigy 225 points
Other Parts Discussed in Thread: LM3445

I have built two four-channel, master-slave, test circuits around the LM3445 Triac-Dimmable Offline LED Driver.

The first version works pretty well, although the channels do not have equal power outputs.

Both versions are mostly surface-mount and use a common rectifier bridge to supply the master and the three slaves.  This first version was hand-built.

I can supply a schematic if you are interested.

The operation of this version is shown in the first attached photo, labeled Normal Operation.  The top trace is the drive signal to the gate of the switching MOSFET.  The bottom trace is the drain voltage waveform for the MOSFET.  You can see that the operation is continuous as the dimmer turns on and off in each line cycle.

The second, nearly identical version I built does not work well at all.  It exhibits a strange behavior that prohibits it from reaching full power.  I would have enclosed a scope photo to show this behavior, but T.I. e2e forum only allows one enclosure.  I can send you the second photo if you are interested.  Anyway,  when the dimmer is dialed up, the gate-drive signal becomes erratic and intermittent, and the drain voltage follows suit.  The result is no significant power output. This board was built by a professional PCB house.

The schematic for these two versions is identical.  The layouts are nearly identical.  Yet, buried somewhere in the physical board is a reason why the gate-drive voltage takes a holiday as the buck voltage rises with the dimmer.  The ramp voltage on the Coff capacitor ceases as well.  I cannot figure out the cause of this very different behavior for two nearly-identical circuits.  It seems that the constant-off-time has been arbitrarily extended so that very few "on" cycles actually occur.  I keep looking back and forth from the working version to the not-so-working version, trying to come up with something that I can do to flush out the problem.


I cannot settle on a final LM3445 design for production until I understand how to solve this problem and keep it from recurring.

Have you any experience with a four-channel version using the LM3445?

Thanks,

Benny Smith

  • 0
    Prodigy 225 points

    Here is the scope photo of the faulty operation.

    The top trace is the gate drive signal to the MOSFET switch.  The bottom trace is the drain voltage of that same MOSFET.

    Note that the MOSFET (hence, the buck regulator)  takes a holiday for a good deal of the cycle, especially while the AC dimmer is active.

    Something about the turn-on of the dimmer each cycle suppresses the LM3445 buck controller and effectively shuts it off for relatively long periods of time.  You can see that there are numerous points where the MOSFET is turned on and its drain voltage goes to ~ zero.  In the Normal operation scope photo, you see that the MOSFET operation is pretty much continuous.

    Benny Smith

  • 0 in reply to Benny Smith
    Prodigy 225 points

    Here is a scope photo showing how the Coff capacitor voltage behaves at the instant where the AC dimmer urns on in each cycle.

    The upper trace is the Vbuck voltage, at the output of the valley-fill section.

    The lower trace is the voltage across Coff.  This trace should be a continuous series of ramps separated by short flat sections as the constant-off-time function operates.  The big 10V pulse occurring at the instant that the AC dimmer turns on is mysterious and obviously related to the faulty operation of the overall circuit. It appears that the normal ramping operation across Coff has momentarily been level-shifted by 10V and then back to zero again.

    The next post will show how the "normal" circuit operates.

    Benny Smith

  • 0 in reply to Benny Smith
    Prodigy 225 points

    Here is a photo showing the "normal" circuit operation.

    The top trace is the voltage at Vbuck.  You can see the instant wherer the AC dimmer turns ON.

    The bottom trace is the voltage across Coff.  Notice that the ramp-space periodic waveform across Coff interrupted just BEFORE the arrival of the AC dimmer turn-on edge.

    Benny Smith