[SOLVED] LM2941 Low dropout regulator - ESR of bypass capacitor

[Zaphappy]

I'm using the LM2941 with the recommended minimum 22uF. Datasheet says the the ESR is critical (.01-5ohms).

I used a tantalum and used no resistance in series and the regulator died (After much use though) Seems the 1.27V reff is now .9V

My question: Why is this value so critical and how do I test to determine if I have enough series R without a trial by smoke ?

 

[betwixt]

You need an ESR meter to measure the equivalent series resistance. There are many references on here if you search for them.

I'm a bit worried that you seem to think you should add a resistor. That isn't how it works, the ESR is internal to the capacitor and ideally should be zero, it never is because of imperfections in the capacitor structure but a lower ESR is better than a high one. If the ESR in the capacitor you used was causing a problem it would be seen as oscillation on the regulator output and would be fixed by replacing the capacitor. It's more likely the LM2941 is just faulty.

Brian.

 

[FvM]

FYI (from LM2941 datasheet):

A Tantalum capacitor with a minimum capacitance value of 22 μF, and ESR in the range of 0.01Ω to 5Ω, is required at the output pin for loop stability. It must be located less than 1 cm from the device. There is no limitation on any additional capacitance.

Alternately, a high quality X5R/X7R 22 μF ceramic capacitor may be used for the output capacitor only if an appropriate value of series resistance is added to simulate the ESR requirement.
The ceramic capacitor selection must include an appropriate voltage de-rating of the capacitance value due to the applied output voltage. The series resistor (for ESR simulation)
should be in the range of 0.1Ω to 1.0Ω.

I'm not aware of a tantalum capacitor with an ESR outside the required range. Ceramic capacitors have possibly lower ESR and can cause instability. Even with unsuitable bypass capacitors, the device would unlikely be damaged.

 

[LvW]

I'm a bit worried that you seem to think you should add a resistor. That isn't how it works, the ESR is internal to the capacitor and ideally should be zero, it never is because of imperfections in the capacitor structure but a lower ESR is better than a high one.
Brian.

This statement regarding the ESR may be true for many applications - however, for LDO`s it is NOT.
A finite ESR in combination with the corresponding capacitance forms a zero in the feedback path that improves stability.
As a reference, I quote from Ron Mancini`s columne "Understanding linear regulators":
"If Cout has a large equivalent series resistance, Resr, a pole and zero form in the feedbacl loop. Such a combination is always more stable than a single pole"

 

[Zaphappy]

Thanks for the explanation LvW - The datasheet was very clear that I needed the range of ESR but I did not listen. It also recommends using a resistor in series to account for a low ESR part.

I'm still not entirely certain it failed due to low ESR cap but I'm replace and try again. Betwixt you may be correct about a faulty part.. I hope.

Thank you all.

- - - Updated - - -

So After all that I measured the ESR of my cap and it's .3-.2 ohms at 1K - 10Khz ??

Does anyone (Besides Betwixt) see a problem with adding a little R in series ?

LvW - Can I put the scope across the REF and look for instabilities ?

 

[mtwieg]

It's doubtful that instabilities actually destroyed the IC directly, but it may have contributed. In the past I've seen LDOs become unstable when not enough capacitance is on the output, and in addition to the output oscillation sometimes the actual DC output voltage will be too high as well. Not sure how that occurs, but that could definitely cause a failure in certain cases.

And to accurately measure capacitor ESR that low, you need a pretty good LCR meter. But in any case it's very likely that the ESR of any tantalum will be fine for this regulator. So yes you should look at the output and ref pin to see if it's unstable or not. Make sure you do this with different load currents applied.

 

[LvW]

Can I put the scope across the REF and look for instabilities ?

Of course, you can. However, for my opinion it is better to analyze the loop gain first using a simulation program.
Thus, you get a picture and a feeling about the influence of a realistic lossy capacitor on stability.

 

[betwixt]

My apologies for misleading you, I should read the data sheet before commenting!

I assumed the capacitor in question was on the input pin of the LM2941 but I can now see the quote refers to the one at the output.

Brian.

 

[crutschow]

How much current are you drawing from the regulator and what are the input and output voltages?

 

[goldsmith]

It's doubtful that instabilities actually destroyed the IC directly

Hit to the point !

 

[Zaphappy]

Thank you for all the great information and suggestions.. I did fix the problem and it was not the capacitor.. I had wrongly put a series
current limiting resistor in series with the IRF540 gate. Once I lowered this the circuit worked great... Woops..

 

[goldsmith]

I had wrongly put a series
current limiting resistor in series with the IRF540 gate. Once I lowered this the circuit worked great... Woops..

Hi again
Yes , if you have used that resistor in series with gate ( with high value ) thus your mosfet has been worked in linear region which means high value of dissipations .
Best Wishes
Goldsmith

 

[biff44]

As some said, a capacitor with the wrong esr value will cause oscillations due to inadequate feedback phase margin. It is best to not use this type of LDO, as there are newer style LDOs that do not care about the ESR of the caps used. However if you need to use the LM2941, you should pay attention to the data sheet recomendations!!!