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.

LM5118 unexpected restart, due to VCC ringing?

Other Parts Discussed in Thread: LM5118, LM5118EVAL

Hello,

I am seeing some erratic behavior with the LM5118. 

When I step up the load, the output voltage shifts downward by ~10%.  What appears to be happening is the boost part of the circuit isn't engaging properly.  When I ramp down the current, as some point, the boost circuitry starts to become active and output voltage begins to rise before it unexpected shuts down and begins the soft start process.  At the end of that process, Vout is back to nominal.

The current limit appears to not be exceeded.  What I found is that ringing on the VCC pin increases (to the point where it dips below 3.5V, for less than 100ns) shortly before the shutdown.  My question is - 1) can ringing on the VCC pin causes a VCC UVLO condition?  (Note that no external supply is driving VCCX at this point)

It appears that it does cause the shutdown, because when I drive VCCX with an external supply at 10V, the shutdown doesn't occur, but also, Vout stays at the nominal voltage (with boost circuitry working properly) after stepping up the load.  Could the VCC ringing be causing that too?  Note that the VCC pin has a capacitor connected to it in accordance with datasheet guidelines, and the maximum threshold voltages of the FETs are 5V.


Thanks.

  • John,

    Yes, VCC below the VCC UV threshold will discharge SS and cause an automatic restart. the converter,  The extreme ringing you are experiencing on VCC is difficult to understand.  If you are using a good quality ceramic cap and locating it close the pins of the LM5118, the VCC voltage should not dip by volts when the capacitor sources current to the gate drivers. What is the length of leads between the capacitor and the IC?

  • I was wondering if lead length might be a factor, but it still seems like a lot of ringing.  The lead length from IC to capacitor is ~.35 inches.  Capacitors connected to pins 17 and 18 are basically located between that VCC cap and the IC, thus helping to contribute to that distance.

    The input current to the circuit is increasing as the boost circuitry is starting to become active.  Also at that time, the ringing on the input voltage node and the switch nodes increase by large margins (more so on the switch nodes).  I'm wondering now if that ringing is somehow propagating through the IC to the VCC pin, and perhaps implementing RC snubbers on the diodes would help to resolve the VCC ringing.

    Could this VCC ringing also be a potential cause for what I'm seeing after stepping up the load (i.e., voltage stays about 10% lower because boost circuitry isn't engage)?  I've seen that behavior on multiple IC's.

    Thanks.

  • It definitely sounds like noise coupling from SW pins is a factor in the problem.  The Vin ringing would not couple through the part to the VCC pin because the VCC capacitor is much lower impedance than the internal regulator devices.   There will be some PSRR feed-through the regulator but not volts in amplitude.  

    The key is to get stable voltages on the input bin and bias supplies (VCC) .  Ceramic bypass capacitors need to be near the IC.  You might check the LM5118EVAL design on www.ti.com for a bill of materials and board layout. 

  • Thanks, David.  I appear to still be getting cases, not necessarily coinciding with an unexpected shutdown, where the boost part of the circuit doesn't seem to be consistently engaging (causing Vout to settle about 10% lower than expected after a step increase in load).  Could either of these two situations below have anything to do with that?  It is worth noting that both of these situations seem to be getting violated in the demo board design (I explain my reasoning below for each), so perhaps neither is significant.

    1)  VCC regulator current limit being exceeded.  Granted, I still see this behavior when driving VCCX with another source, but using equation 25 in the datasheet, I get more than the specified 35mA limit when combining the gate charges of the two FETs (the equation is a bit unclear with its use of "Qc" but I'm assuming it's calling for the sum of the gate charges of the buck and boost FETs).  What are the consequences of exceeding that limit?  On the demo board, Qc at Vgs = 10 for the selected FET is 68nC.  With Fsw of 300kHz, and total Qg of 136nC for the two FETs, that's an IVcc of over 40mA (not counting Iload_internal).

    2)  (This doesn't seem applicable because it only seems related to the buck function, but in case it is significant...) The bootstrap capacitor is too small according to the quick start excel sheet calculations for the gate charge of the buck FET.  Demo board FET Qg at VGS=10 is 68nC typical, which the spreadsheet then calls for .19uF of capacitance, yet the demo board only uses .1 uF.

    Thanks.

  • David,

    I have discovered that I don't see this condition when I drive VCCX with a higher voltage (~10V).  Perhaps driving it to that higher voltage offsets coupling to the VCC pin, or something about the FETs I've chosen or other components requires a larger VCC/VCCX drive.  I realize this issue has become sort of nebulous, but any additional insight you could provide would be appreciated.

    Thanks,

    John