LM5116 Switching Noise

Cam,

I looks like there is jitter in your waveform.  Until that is cleared up, you will have a problem with switching noise.  Either the layout a not what it should be, especially the grounding or the compensation is not quite right. If you used the calculator you should be ok in compensation but it still might need some adjustment.   One thing to try right away is adding resistance to R34 and R25.  Try something like 4 or 5 ohms and see if that helps.

What switching frequency are you using?  The time base is cut off in the scope shot.  Are your grounds well though out or were they just dropped into a ground plane?  Also, what is the resistor network attached to VCCX?  If VOUT is 15 volts, a much better low impedance way to supply VCCX would add a 3.3V zener diode in series with the trace from VCCX to VOUT.

Regards,

Ron Crews

Hi Ronald.  Thanks for your reply.

Switching frequency is 250kHz.  The resistor network was originally designed to drop VOUT of 18V down to 15 for VCCX.  That has since changed, and VOUT is now supposed to be 15V.  That resistor network is currently unpopulated.  I missed that in my first scan of changes I see.  Having said that, the zener idea is a far better idea than what I originally had.

All grounding goes to a ground plane.

Great tips so far.  Thanks.

Cam,

I think what I would do next is clean up the switching waveform.  It is generating noise which is hard to suppress.  Try looking at the comp network and be sure your Gnd lead on the Current Sense negative pin runs directly to the current sense resistor and not to a ground plane at the IC or some other ground line.  That is frequently an issue.

Currently, the current sense negative pin is indeed through a ground plane.  I can try an run some wire directly from pin 13 to the current sense resistor.

Are there any other guilty parties that shouldn't be run through a ground plane?

Cam,

Technically all the analog ground pins should be connected together and then connected to the power ground or the ground of the input capacitor.  Analog grounds are things like the timing resistor, the soft-start capacitor, UVLO and the like.  The current sense pin is very sensitive since this is a current mode part. Just unsolder the pin, carefully lift up up and solder a small wire on and run to the current sense resistor's negative side.  That may solve your problem.

Regards,

Ron Crews

Hi Ron.

I spent a few weeks and researched some alternate components, and better techniques of layouts for the board.

I added low ESR caps and faster switching, low capacitive FETs and diodes.  I removed the ground plane entirely and now run individual grounds from the ground of the input cap to the passives on the IC and to switching diode, output caps, etc.

The result of this work reduced switching noise to less than 10mV.  So thank you for your input.

A secondary issue has arrisen from this new board.  whilst my switching spike noise is only 10mV, I'm struggling with voltage ripple as my current increases.

I've used the calculator to help me select components and comp loop values, but at 375mA, my ripple is above 150mV.  My output capacitance is 100uF.

I'm worried that my gate signals aren't being generated properly.  Above I have pictures of gate signal that is nice and square.  Now I have jittery signals that are not square at all.

I have attached pictures to help me describe my issues.

HS gate at pin:

It looks like I'm having issues attaching pictures other than this one.  Please email me at cbuskell@leespecialties for the remaining pictures at layouts

I have the HS and HS gate signals at pin and FET, switching node, output at inductor and capacitor.

Thanks for your time and understanding