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.

LM317A: Total accuracy over time and temperature

Part Number: LM317A
Other Parts Discussed in Thread: LM317

Hello,

We are considering using LM317A for an application that require good Vout accuracy of 2.5% over time and in temperature. 

Our application is 30V -> 24V with very small Vin range and Vout load varation (Iout = 10mA), and I plan to use 0.1% feedback resistors for feedback accuracy.

In the datasheet I see two dominant parameters for accuracy that are :

-Load regulation (1%)

-Long term stability (1%)

My questions are :

For load regulation, does it takes into account the variation of Iadj ? Can we improve that parameter using a buffer before ADJ input ?

Does long term stability takes into account load regulation or should I sum these two parameter for global accuray ?

Does long term stability gets worst than 1% beyond 1000 hrs ?

Finally, is the LM317A the best linear regulator for our application or do you have a more accurate device that would suit our Vin/Vout Iout needs ?

Thank you for your support.

Adrien 

  • Hi Adrien,

    If you need +/- 2.5% tolerance then I would probably move towards another solution.  The LM317 was first designed around 50 years ago, and while it is still a good general purpose linear regulator, it is not as well suited to modern tight tolerances like +/- 2.5%.  The LM317 datasheet has different values for various components that change the tolerance, based on your conditions.  Vref by itself is +/- 2% across temperature.  Long term stability can add 0.3% (typical) or 1% (maximum).  Load regulation and line regulation are separate from this, and can increase this tolerance still.  The feedback resistors will also increase the tolerance slightly.  You are looking at at DC tolerance, maximum, of around +/- 3-4% and that does not include any transients on the line or the load.

    Long term stability a common spec for voltage references but is not a standard industry spec for LDOs. Voltage references are designed, manufactured, and tested to minimize long term drift however this is rarely a key spec for LDOs as long as the output remains within the accuracy spec. That being said it is common for any voltage reference to change during the first few hours that it is powered on. Even for precision voltage references it takes between 100hrs-1000hrs from the first time it is powered for the voltage to settle (see the example plot from a voltage reference below) and an LDO’s internal reference will behave similarly. 

    Most of the DC tolerance in modern LDOs are contained in the reference voltage (the rest is in the offset voltage of the error amp, which usually is single digit mV's).  In cases like this I typically recommend a precision reference connected to our tracking LDO (TPS7B4253, just +/- 4mV of error maximum).  A "tracking LDO" is just a fancy term for an LDO where the reference voltage is applied externally, so you can attach a precision reference.  With a modern LDO like the TPS7B4253, you get modern protection features and significantly lower cost design solutions (those older LM317 designs need large aluminum electrolytic or dry tantalum capacitors for stability, as opposed to small cheap ceramic capacitors). In the case that a 24V precision reference does not exist, the TPS7B4253 conveniently has the FB pin for feedback resistors so we can use 0.1% tolerance resistors to set Vout.  So Vref = 8V, Rtop = 2k, Rbottom = 1k, and the maximum error due to the offset voltage is +/- 4mV * gain = 3 == +/- 12mV, or just 0.05% of the Vout = 24.  If you pick a precision reference of, say 0.5%, then you have a maximum DC error of 0.55%.  The precision reference datasheet can list the long term stability and as Vref dominates, you can see how easy it will be to meet 2.5% tolerance even if you need some transient performance.

    Let me know if this is something you wish to pursue and we can loop in the voltage reference team for a suggested precision reference.

    Thanks,

    Stephen

  • Thank you very much for your detailed answer.

    I'll check this solution