LM5146-Q1: Device and MOSFET getting hot with no load

Hello

Please check your component values against the quick start spreadsheet tool on the TI product page for this device; just to be sure.

Also, please be sure that all components are in place, especially C563 capacitor.

Please provide waveforms of the SW, HO, and LO signals in one scope shot.

Also, a view of your PCB would be helpful.

You may try to shorting out R1235 with zero Ohms and test.

Also, test with pin #8 grounded.

Have you tried it with some small load??

I will assign this to our controller engineer for further help.

Thanks

Hi Frank

Below is a scope capture with the SW, HO and LO signals, this is under load, approx 800mA.

CH1 - SW

CH2 - HO

CH4 - LO

Shorting the boost resistor made no discernible difference to the traces, grounding SYNCIN gave an improvement to the jitter in the signals.

The PCB layout is as shown below, there is a solid PCB ground plane under the circuitry on the top and bottom sides.

Attached is the excel spreadsheet for the compensation values, I had to PDF this for it to be accepted.

Ideally we would like to test using the same inductor and MOSFETs as the LM5146 EVM which looks nice and smooth however these parts all appear to be end of life now, do you have suggested alternatives we could try?  Then we can set a higher inductance and lower frequency.  We are also considering a schottky diode on the bottom side FET.

LM5146-Q1 Quickstart Tool r2 - 48Vin 12Vout 8A 550kHz.pdf

Please let me know if you have any additional questions or require clarification on any items.

Regards

Malcolm

Hey guys,

Just wanted to add on and follow the same query as I am getting the same waveform for the high side and my IC is reaching 55C under a very small load of dc motor drawing 400mA. My output is 18V  tho.

I saw my waveform and heating is reduced if the voltage is dropped to below 35V, hey Malcolm do you get the same thing too?

Also for a resistive load I get good waveform perfect squarewave but on an inductive load like a dc motor the waveform returns.

Also Frank why should pin 8 SYNCIN be grounded what effect will be see if it's floating.

Naman

Hi Malcolm,

These waveform show discontinuous conduction mode (DCM) and look pretty good to me.

When the inductor current falls to 0A, both FETs are turned OFF, and the SW node L-C rings as expected.

The HO-pulldown is SW node, so when SW rings and HS FET is OFF then HO should ring with it.

Your waveforms in the first post looks like the HS FET did not turn on properly after discontinuous conduction but here it looks better. Maybe we were seeing noise from SYNC-IN triggering a bad clock pulse. That small overshoot is an imperfection but a boot resistor should help, also could be related to probing.

Layout looks good, LO trace is a bit long but LO noise is ok from recent waveform. Maybe the most concerning thing is how well PGND is grounded with that small via.

What do these gate drive signals look like at full load? Also can you line up the HO and SW scope channels?

Hi Naman,

Go ahead and post your waveforms in a related question (yellow button up top), I think your inductive load is a separate beast.

Grounding SYNC-IN will prevent noise from triggering an random clock pulse and messing with stability.

-Orlando

Hi Orlando

Thank you for the feedback.  I'll take some new scope traces and provide these tomorrow.

I had grounded the sync in pin as Frank suggested to enable DCM, I'll try the same with the EVM tomorrow..  Without the sync pin grounded I don't get the same waveforms as the EVM which is shown below (long ground leads, hence the ringing).

The EVM documentation doesn't show the switching waveforms for each of the operating modes, is there a TI app note that covers this or do you have another reference?

With regards to the temperature rise we are seeing on the LM5146 and MOSFETS should we look to use a top side MOSFET with a lower gate capacitance or reduce our operating frequency and inductor?  How about a schottky diode on the lower MOSFET?  Given our 12V output voltage will feeding this through a schottky diode to VCC provide a significant improvement in LM5146 thermals?

Regards

Malcolm

Hey Orlando, 

Thanks for taking up my query.

The two waveforms I have the one with a good square wave is a resistive pump load which is consuming around 6A and running at 24V, whereas the second image with the Hi side doing the funky thing is with a motor running at 24V and consuming 400mA. 

Hey Malcolm, 

Thanks for adjusting my question with your 

Naman

Thanks a lot, Orlando, fixed my issues it's perfect now. 

Naman