Author Topic: Failire modes for linear and switching regulators (Read 2931 times)

Pack34 · « **on:** December 27, 2017, 10:19:09 pm »

Is there anything besides a short or overcurrent event that could cause a regulator to output 0V?

I've heard in side conversations that some designs would have a power board that would have a bad regulator. However, at worst everything is sized for 200% maximum expected output. Some where its rated for 1A and the load is only a couple mA.

I'm extremely concerned with the assembly quality as I've been seeing a good amount of shorts and solderballing from the assembler but that doesn't mean that there isn't room to make the design more robust.

I've never had a regulator not come up before this assembler. Is there any type of protection circuit, or soft-starting I can do to mitigate this?

T3sl4co1l · « **Reply #1 on:** December 27, 2017, 11:46:02 pm »

Quote from: Pack34 on December 27, 2017, 10:19:09 pm

Is there anything besides a short or overcurrent event that could cause a regulator to output 0V?

Sure! Uh... fire? Gamma ray scalpel?

Or good old, not applying any input?

Quote

I've heard in side conversations that some designs would have a power board that would have a bad regulator. However, at worst everything is sized for 200% maximum expected output. Some where its rated for 1A and the load is only a couple mA.

I'm extremely concerned with the assembly quality as I've been seeing a good amount of shorts and solderballing from the assembler but that doesn't mean that there isn't room to make the design more robust.

I've never had a regulator not come up before this assembler. Is there any type of protection circuit, or soft-starting I can do to mitigate this?

Well, nothing can mitigate truly shitty assembly, but if you're suspecting that initial startup is blowing parts as a result of bad soldering, you can add a current limiter that simply won't deliver more current than the regulator can safely handle (through any combination of pins, if applicable?).

Tim

TimNJ · « **Reply #2 on:** December 28, 2017, 03:42:28 pm »

Are you confident these are assembly/manufacturing quality issues? If so, not to state the obvious, but can't you rectify this situation by talking to your assembler? Are you wave soldering? You might want to look at some general guidelines for component clearances to reduce the risk of bridging components. In general, stay away from 0603s and lower, if you can.

http://www.seeedstudio.com/blog/2017/05/15/component-layout-considerations/#43_Wave_Soldering

What topology are you using? Power rating? Controller?

Most modern controllers have lots of built in protection features, but there are still some traps. Excessive ringing can damage MOSFETs. When using a half-bridge converter, if your gate drives overlap (somehow) you'll blow them out. If this is an off-line switcher, an improperly sized PFC FET can blow up on startup if the inrush current is too high.

But, of course, tough to know without knowing the specifics of your design.

Pack34 · « **Reply #3 on:** December 28, 2017, 05:09:19 pm »

Quote from: TimNJ on December 28, 2017, 03:42:28 pm

Are you confident these are assembly/manufacturing quality issues? If so, not to state the obvious, but can't you rectify this situation by talking to your assembler? Are you wave soldering? You might want to look at some general guidelines for component clearances to reduce the risk of bridging components. In general, stay away from 0603s and lower, if you can.

http://www.seeedstudio.com/blog/2017/05/15/component-layout-considerations/#43_Wave_Soldering

What topology are you using? Power rating? Controller?

Most modern controllers have lots of built in protection features, but there are still some traps. Excessive ringing can damage MOSFETs. When using a half-bridge converter, if your gate drives overlap (somehow) you'll blow them out. If this is an off-line switcher, an improperly sized PFC FET can blow up on startup if the inrush current is too high.

But, of course, tough to know without knowing the specifics of your design.

Pack34 · « **Reply #4 on:** December 28, 2017, 05:16:40 pm »

Quote from: TimNJ on December 28, 2017, 03:42:28 pm

Are you confident these are assembly/manufacturing quality issues? If so, not to state the obvious, but can't you rectify this situation by talking to your assembler? Are you wave soldering? You might want to look at some general guidelines for component clearances to reduce the risk of bridging components. In general, stay away from 0603s and lower, if you can.

http://www.seeedstudio.com/blog/2017/05/15/component-layout-considerations/#43_Wave_Soldering

What topology are you using? Power rating? Controller?

Most modern controllers have lots of built in protection features, but there are still some traps. Excessive ringing can damage MOSFETs. When using a half-bridge converter, if your gate drives overlap (somehow) you'll blow them out. If this is an off-line switcher, an improperly sized PFC FET can blow up on startup if the inrush current is too high.

But, of course, tough to know without knowing the specifics of your design.

Can you go into more detail about the ringing? I'm using a wall-wart and I can see the 60Hz oscillation on that rail. Could this cause issues?

TimNJ · « **Reply #5 on:** December 28, 2017, 06:10:22 pm »

Sure. I'm not the most knowledgeable but...

Many power converter circuits use inductors, as you know. When you try to switch an inductive load, the parasitic capacitance of the FET and the circuit inductance can form a high frequency resonant tank. Thus, when you switch a MOSFET (off, in particular) you can get a significant amount of ringing on the rising edge. If you are trying to switch 50V using a 60V FET, you can easily blow out a FET if the ring voltage exceeds 60V even for just a few nanoseconds. To mitigate this, people commonly use RCD clamps.

That said, it looks like you are talking mostly about small local regulators. I should've picked up on that. Since there are so many, does any one regulator seem to be most troublesome? Again, I'm a bit confused because this largely seems like a workmanship problem. What do the failure modes of these regulators matter if you can just prevent them from failing by improving build quality? Not trying to sound snarky; I'm just confused on the issue.

Under load, it is not uncommon to see some ripple on the output Switch-mode AC-DC converters work by first converting AC to a high DC voltage. Ideally, that voltage is ripple-free, but since the primary-side bulk capacitance is finite, there will always be some ripple riding on top of it, which gets transferred to the output. These are mainly off-line switcher issues though.

Pack34 · « **Reply #6 on:** December 28, 2017, 06:24:34 pm »

The most common issues seem to revolve around the LT1930 and LP2951. Neither have any real load. The LP2951 is basically a reference voltage used to translate the 0-3.3V clock signal to go from -5V to 8V for the sensor. The LT1930 is in a SEPIC configuration to provide a smooth 12V power rail to the sensor. The sensor itself only consumes a total of 4mW.

I could definitely just replace everything with whatever is most current but an issue really needs to be found before haphazardly making changes.

Neomys Sapiens · « **Reply #7 on:** December 29, 2017, 07:22:35 pm »

There are a lot of possible failure modes and the assorted root causes for them. When I look at your list of components, it seems to ba a rather complex PS. Yet you seem to focus on the assembly. What about the layout? I once did a board with a safety-related comparator circuit with multiple, totally independent channels. We thought we checked everything. But when fully populated, a random and always changing set of channels would not work. Until we shielded the inductors of the channel-specific DC/DC converters. It was inductive coupling that caused some of them to fail starting properly.
But if you had similar power units made before and problems seem to be connected with the assembly shop, why not take it from them? Or has parts sourcing changed too (maybe done by the assembler?)?
If they bought fakes, that might be the reason why circuits known to run do not perform anymore on this special board.


EEVblog® Main Site	EEVblog® on Youtube	EEVblog® on Twitter	EEVblog® on Facebook	EEVblog® on Odysee

EEVblog® Electronics Community Forum

Author Topic: Failire modes for linear and switching regulators (Read 2931 times)

Pack34

Failire modes for linear and switching regulators

T3sl4co1l

Re: Failire modes for linear and switching regulators

TimNJ

Re: Failire modes for linear and switching regulators

Pack34

Re: Failire modes for linear and switching regulators

Pack34

Re: Failire modes for linear and switching regulators

TimNJ

Re: Failire modes for linear and switching regulators

Pack34

Re: Failire modes for linear and switching regulators

Neomys Sapiens

Re: Failire modes for linear and switching regulators

Share me