LM1117: LM1117 worst case Vout specification

Hi Dave, 

Q1) Does the Vout min/max range specified on page 8 include all errors such as Temperature Stability and Long Term stability?

The min and max here are specified at 0 ≤ IOUT ≤ 800 mA, 6.5 V ≤ VIN ≤ 12 V, and -40 to 125 junction temperature. It's not including long term stability drift.

Q2) Recognizing that both the stated and graphically depicted temperature stability is "Typical", and thus unbounded, is the stated 0.5% error intended to describe the error span (e.g. +/- 0.25%) as one might infer when referring to Figure 6, or is it intended to describe a +/- 0.5% error relative 25C by not disclaiming 'unless otherwise specified, Tj = 25C'?

From my understanding, the temperature stability is probably applied to the internal reference, for this device, it would be the 1.25V bandgap reference. For the 5V output, there is a gain of 5/1.25 and this 0.5% will become 5/1.25 x 0.5% = 2% typically and this should include the whole temperature range. The reference temperature stability should be included in the min/max range specified on Page 8.

To calculate the worst-case Vout range, I would consider adding the long term stability one on top of the min/max limit. But this long term stability one is only typical and we can only estimate the worst case. 

Does it make sense?

Regards, 

Jason Song

Hi Jason,

Perhaps I should have added some assurance that the device was applied within operating limits, Vin = 11 V and Iout is approximately 25 mA.

The core of the question has to do with the manner in which the device is specified, primarily with regard to temperature and as I will try to elaborate below, the stability specification then also comes into question.

As I noted, the relevant sections of the specifications have not changed since TI acquired National Semiconductor (NSC), so I think that it is fair to take into consideration how this device was specified relative how NSC had specified other voltage regulators over the past 30+ years and how that may be relevant.

From what I can find, the LM1117 family was released in June of 1998 as a LDO replacement for the LM117 family.  The LM1117 family itself includes the LM1084, LM1085, and LM1086.

The LM117 specifies a min/max reference voltage range that applies across the full operating temperature range, and separately lists temperature stability as a percentage.  Other examples of that method of specifying temperature dependency include the LM150 and LM138.

The LM78xx, LP2950/1/2, etc. specify min/max Vout over the operating temperature range and provide an average temperature coefficient of output voltage.

The LM2935, LM2937, LM2940, etc. specify min/max Vout over the operating temperature range and include no additional specification line item regarding temperature performance.

In all of these examples, long term stability is specified as a separate parameter, but I haven't found a disclaimer that it is exclusive of any otherwise specified min/max Vout specification.

For the devices, such as the LM1117, which specify temperature stability as a percentage, it is not explicitly stated whether the intent is to provide helpful insight that the temperature error performance is typically better than implied by the min/max Vout specification, or if it is an additional parameter such as the typical long term percent stability parameter specified in the very next row of the specification table itself may be.

Given the consistency with which long term stability has been specified as a separate parameter coupled with the difference in how temperature stability is specified for the LM1117 (and a few others) it is reasonable to raise the question of intent as to whether either of these parameters are included within the separate min/max Vout range.  One can argue that if long term stability should be interpreted as a separate parameter, that temperature stability should be interpreted the same way.

I appreciate this may seem like too fine of a point to argue one way or the other, and that you and I can both have our own "understanding" as to what was "probably" intended by the original data sheet authors in ~1998, however I am still seeking an unequivical statement from TI to my original questions as written.

In 1998, James (Jim) Dreyfus in Santa Clara was the product engineer and it appears he survived the acquisition, if still with the combined company, he may be a good resource.

I'm unable to discuss in a public forum as to "why" this is important, but am willing to do so off-line as I would have been with TI's tech support who instead directed me to this forum.

Thanks,

--Dave

Hi Dave, 

Your feedback has been received, and thanks for sharing the knowledge of the history of the LM1117 family. Let me discuss with the team to see if we have better explanations on the temperature stability and long term stability. 

Regards, 
Jason Song

Hi Dave, 

I have also reached James Dreyfus to see if he could provide any inputs. I will keep you updated. 

Regards, 
Jason

Hi Dave, 

Sorry for the delay in response. I have received feedback from James who was in the team originally developed LM1117. According to him, 

“Temperature stability and long term stability would likely have come out of the electrical characterization over temperature and drift analysis during life tests, respectively, performed during product development. Could the behavior of the device have changed in the last 20 years?  Possibly. “

If this is true, then the temperature stability could be the drift on the internal reference and it should have been included in the output accuracy spec. 

We have also reached the voltage reference team who has more datasheet with temperature stability and long term stability data, according to them, when the two terms have been listed separately, those numbers have to be taken into account on top of the accuracy. 

The two sayings are contradicting to each other. But as the temperature and long-term stability data listed in the LM1117 datasheet were initially obtained with a smaller group of units and over the many years of production, those typical numbers have not been verified again. If the long term stability/temperature drift data are critical to your application. I would not recommend using the typical numbers listed in the datasheet regardless to do any worst-case estimation. 

If you would like to reach me directly for your application issue, please add me as a friend on E2E and we can then exchange direct emails. 

Regards, 

Jason Song