LM3409: Unexpected Output Current

I've also confirmed that there is a measured 48.8 mOhms across the sense resistor using a relatively accurate, but not calibrated, ohm-meter. I also measured the voltage on both sides of the sense resistor and let the oscilloscope do the math (don't have a floating, diff probe) and I read about 940mV across the sense resistor during the peak current just after it turns on. Again, this happens whether the LM3409 EN pin is GND or 3.3V, which is really weird.

I suspected my power supply so I used a different supply, without current limit. It's can do 15V 13.5A and it collapses when the output load is connected. So something is definitely not correct with the current limiting.

Hello James,

The initial post with the scope shot I can understand. That is likely just the device going into low power shutdown mode which usually happens in about 10ms of EN being pulled low. That is easy to fix, you can just drive the UVLO pin rather than the EN pin. That will prevent the device from shutting down and IADJ will stay high and VCC will remain alive.

As for tying EN low and the device is still behaving strangely that sounds like some other issue, maybe noise? These devices can have problems if the DAP is not soldered down properly and/or the layout is poor. Layout is critical in switching regulators and if that is a possible issue I would be happy to check your layout and comment because you are correct, VCC should be 0V with EN held continuously low and since VCC is the gate drive source the FET should have no way to turn on.

Regards,

Clint

Thanks Clint,

The first scope shot above is showing how I was controlling the IADJ pin relative to EN pin.  I've since just enabled the device and left it on for now to make things a bit easier.

I switched to using the actual device I intend (peltier) and that solved the power supply browning out.  I was using a 0.1 ohm resistor before thinking that the current limit of the LM3409 would drive it fine.  But I guess it doesn't like that low of load.  I would have expected it to limit current though and not take down the power supply.

Switching to the peltier and now the external power supply doesn't collapse.  However, I still see ~4.85A through the Peltier no matter what voltage is on IADJ pin.  I ramp the voltage from 240mV down to 0mV over the course of 10 seconds and I don't see any change in the output current.  Isn't that supposed to be an analog dim pin?

So I can now see that I hit the limit as based on the sense resistor value -- so that is a step forward.  Now I need to understand why IADJ pin isn't working as I expect it.

As for layout, attached is a screen shot.  Pin 1 of the LM3409 (U1) is the top-left as displayed:

James.

Hello James,

Did you measure the IADJ voltage directly at the pin? Is it possible that it has a solder bridge to another pin? Also, where exactly is C2 grounded? If the voltage directly at the IADJ pin is lowered (below 2.4V) the output current should be lowered.

I do have some concern about the layout though, other than wondering where C2 is grounded. Your component placement is very good, but I am not sure about the grounding. Having a ground return path that follows the forward switching current paths is crucial. Otherwise you will create large radiating loops that can cause all kinds of issues. Is there a ground plane on another layer? If so I would at a minimum put some vias next to the ground points for C5 and D1. If not I would highly recommend adding a ground plane. If that isn't possible I would at least go ahead and flood the rest of the top plane unused space with ground.

Regards,

Clint

Thanks for the feedback Clint.

As for pins shorted together, there are no adjacent pins that have a hard short between them--I measured that.  And visually the soldering looks OK.

I measured IADJ across C2, which is very close to the device.  So I think that should be fine.

And yes, I should have mentioned, this is a 4-layer board and there is a solid ground plane underneath this part.  I create a void in the ground plane directly underneath the switching node copper pour on the top layer to reduce its capacitance.  As for return path ground loop, I tried to keep that entirely on the top layer with only 1-connection point to the internal layer.

Apart from changes to the layout for future designs, do you have any ideas on how to troubleshoot that the output current doesn't appear to be following the voltage on the IADJ pin?  This is functionality that we need--in fact, it was one of the reasons we chose this part over others that only had PWM dimming.