LM1117: When using at low temperature

Hi ACGUY, 

1: Does TI know a similar case?

I have seen other devices that run into oscillation at low temperatures. Most of those cases can be improved with different output capacitors. For Tantalum capacitor, its effective capacitance is usually lower at cold temp. An application needs to ensure the effective capacitance is higher than 10uF (not the capacitor labeled value) across all operating temperatures of the LDO. 

The output capacitor is critical in maintaining regulator stability and must meet the required conditions for both minimum amount of capacitance and equivalent series resistance (ESR). The minimum output capacitance required by the LM1117 is 10 µF, if a tantalum capacitor is used. Any increase of the output capacitance will merely improve the loop stability and transient response. The ESR of the output capacitor should range between 0.3 Ω to 22 Ω. In the case of the adjustable regulator, when the CADJ is used, a larger output capacitance (22-µF tantalum) is required.

2: Is adding a ceramic cap appropriate as a countermeasure?

Depending on the ceramic and tantalum capacitor used, the added ceramic will add capacitance and is likely to reduce the ESR. Again, the effective capacitance and the ESR needs to meet the datasheet requirements across temperatures. 

3: Which parameter of LM1117 is involved?

The output effective capacitance and the ESR. 

4: Will this happen even if I use an industrial Spec IC?

The LM1117 is an industrial spec IC as contrary to the automotive-grade IC that usually ending in "Q1" as the part number. Will you help to rephrase this question?

Regards, 
Jason Song

Hi Jason,

Thank you for reply.
The nominal value of the using tantalum cap is 100uF.
I measured the capacitance of tantalum cap at 0-degC ambient, the capacitance is 94.88uF.
We are in trouble because it does not violate the design requirements stated in the data sheet.

The using device is LM1117DTX-3.3(0 to 125-degC), it is not LM1117IDTX-3.3(-40 to 125-degC).

Please comment based on the above.

Questions;
1: Does TI know a similar case?
2: Is adding a ceramic cap appropriate as a countermeasure?
3: Which parameter of LM1117 is involved?
4: Will this happen even if I use an industrial Spec IC?

Best regards,
ACGUY

Hi ACGUY, 

While I am working on the answers, would you be considering any alternatives?  

Please note we have an improved device the TLV767 which is ceramic stable and a much better all-around performance device. 

Regards, 
Jason Song

Hi ACGUY,

The 100uF capacitor is likely to create a very high inrush current during startup, and we would need to confirm if the oscillation is linked with the LDO entering in and out from the current limit during the startup. Will you help to provide a scope shoot that clearly shows the Vin, Vout, and the input current?

Will you also help to provide the exact condition and scope shoot on the oscillation that clearly shows the amplitude and the frequency? 

It would also be helpful if you could provide the ESR and capacitance of the output capacitor you selected across frequency. 

Regards, 
Jason Song

Hi Jason,

Please see the attached file.
Green : Vout 3.3V
Purple : Vin 5V
Red : (None)
At low temperature (less than 10-degC), Vout of the LM1117 has osiration. However, after a while, the Vout stabilizes due to the self-heating of the LM1117.
Therefor, we thought that the feed back loop frequency of the LM1117 is decline at low temperature, we added 10uF-cera-cap for output tantalum-cap. As a result, the condition of output voltage improved.

Please give us your opinion on this consideration.

Regards,
ACGUY

Hi ACGUY, 

It seems to me that you may have noisy ground when taking the scope shot, as all three channels are having similar spikes. On top of the higher frequency spikes, I can see that the output does have some oscillation.

For the LM1117, due to the age of the device, we have very limited resources available. It's best for us to confirm this type of behavior in simulation, but we are unable to do so for LM1117 due to the lack of a design database. 

The datasheet is suggesting a 10uF capacitor, is it possible for your to try the one with the datasheet recommendation? If adding the ceramic capacitors makes the oscillation disappear, I would suggest doing the load transient check to confirm if it indeed solves the problem. More details can be found here. 

www.ti.com/.../snva364a.pdf

Regards, 
Jason Song

Hi angelo,

What I would like to know is whether the LM1117 has the property of becoming unstable at low temperatures.

Regards,
ACGUY

Hi ACGUY,

I am unaware of any situations where the LM1117 becomes unstable at low temperatures, but it looks like what is happening is that you are hitting the LDO with another transient before the LDO can stabilize from the previous one. Have you tried using a lower capacitance value tantalum? it may help you increase the loop bandwidth enough to improve the response time. 

My guess from the data you have shown here is that the low ESR from the ceramic is allowing the LDO to respond quicker, but it doesn't look like a stability problem it seems to be that the transients are happening when the loop is out of regulation still.

Regards,

Mark