LM3429: LM3429 running hot

Hi,

The datasheet (page 20) states: A minimum on-time must be maintained in order for PWM dimming to operate in the linear region of its transfer function. etc...
Further on an equation is given which is applicable to boost converters.
Clint explained the reason for this. It has nothing to do with the problem.

A schematic and a heating pin point would be very helpfull indeed.

Kind regards

Hello Philip,

The minimum pulse on time is really just to maintain control during PWM dimming. If nDIM is pulled low the device should stop switching and dissipate very little power. Do you know what exact component(s) are getting hot when nDIM is pulled low? Have you verified nDIM is lower than the minimum threshold? Do you have a schematic and BOM you could share? 

Thanks,

Clint

The circuit is more or less as suggested by Webench. The IC itself is the hot part. nDim is driven directly from the output of a PIC, and sits at about 85 m V above ground (with about 20mV of noise superimposed). There is no evidence of any switching going on while in the off state, so I'm assuming that it is turning off ok. It has survived being switched on for 24h continuously, so it seems to be relatively benign, but I would still rather know why it's hot in case it has a longer term effect that will bite us later.

Phil

Hello Phil,

Sorry for the slow response, I have been out for two weeks for the holidays.

I don't see anything concerning that should make it run hot when nDIM is low so my suspicion would be nDIM itself. Do you have the ability to look at nDIM with an o-scope? Usually nDIM is used as an input UVLO with resistors which provides hysteresis. My concern with driving it directly is the maybe the signal is too fast which could cause it to ring below ground, the control signal is pulling it below ground, or the signal is too slow passing the threshold too slowly and causing an on/off oscillation.

I would try implementing on/off control as shown in figure 24 of the datasheet as it is used for PWM dimming and see if that solves the issue if you see any anomalies on the nDIM waveform.

Thanks,

Clint

Hi Clint

I'm not using it in PWM mode - it's either on or off. I have the ability to PWM it to reduce the current if the internal temperature gets silly, but I've never seen it happen in testing, except when I cooked it to make sure the temperature mitigation measures actually work.

I have looked at all the pins on the scope, and can't see anything unusual.The scope is digital, with a 100MHz bandwidth, so not ideal, but I would have expected to see something if it was big enough to cross the input threshold at the frequencies that are around on the board.

If it was only one sample, I would suspect that something was broken, but all 56 that we've built so far do the same thing.

Regards

Phil

Hello Phil,

This is an odd one. Does the nDIM signal look the same on the stoplight section? What are the LED string voltages for each? I still worry about noise on nDIM as I have seen noise on that type of pin couple into other things internally with other ICs in the past which is why I suggested the circuit as used for PWM dimming as it gives hysteresis and some noise immunity. Perhaps just adding a small capacitor from nDIM to ground could tell us something if things change, it would be an easy thing to try.

In either case I am going to try and contact the systems engineer for the device to see if there is anything I should know or might be missing. Worst case, if you would like, is you could do a friend request so we can do private messages. That way I can provide an e-mail address and an address if you would like to send me a board to check out.

Thanks,

Clint

Hello Phil,

The systems engineer took a look at the schematic and didn't see anything either, there are also no known issues with nDIM control of this device. Is it possible some external circuitry such as the "fault detect" that we cannot fully see is pulling on some pin beyond abs max? If nDIM is low the device stops switching and the only power dissipation should be non switching Iq times the input voltage unless some pin has been stressed.

Could you perhaps provide the exact differences between the two circuits?

Thanks,

Clint

Here's the DI driver for comparison.

Regards

Phil

Thanks Phil,

Give me a little time to compare these closely. In the mean time is there anything else on the board that could be generating the heat, perhaps a component below the IC on the backside of the board? Is there any way to measure the current the circuit and the IC itself is drawing? Is there any possibility of a manufacturing issue such that the IC DAP is not properly connected? 50C while off will not hurt anything, but I don't see where that power could be coming from unless it is not the power but the ability to dissipate the little power there is easily.

Thanks,

Clint

Well spotted! I can only claim incpompetence. Here's channel 1 for comparison. Channel 2 is the one with the issue. I will measure the currents and get back to you.

I don't think it's a manufacturing fault, as they all seem to do it (we've built over 50 so far). During EMC immunity testing, we had one powered up for nearly 10 hours continuously, with no apparent ill effects, and all of the functions were lit for the duration of the test. The PCB layouts for both LM3429s are very similar, with as much copper land as I could fit in, and plenty of thermal vias through to the back side of the board, whic is mostly ground plane (and no components - it fits against the inside face of the lamp housing).

Regards

Phil

Hello Phil,

I see nothing wrong with the schematic. Have you been able to measure the current draw at VIN? The quiescent current will cause some device heating but it's not likely to be that much. Still it would be good to know what the current draw is at the input and if it is a concerning amount or not.

Thanks,

Clint