3
\$\begingroup\$

I will use the adjustable LT3015 as an example. http://cds.linear.com/docs/en/datasheet/3015fb.pdf

You can set the output voltage based on the equation Vout=-1.22(1+R2/R1) + R2*Iadj where Iadj is 30nA nominally.

Now, the datasheet shows nominal R1 values for different output voltages. Most of these nominal R1 values are around 12K, giving us a resistor divider current of ~100uA. It also mentions that the R1 value should be less than 50K to help keep R2 value low enough so as to minimize errors from the adj pin current.

My question is, is there a specific reason they chose the nominal R1 values to have a 100uA divider current? Is there anything stopping me from using much lower values, lets say in the hundreds of ohms?

I understand that will increase the quiescent current(correct me if I'm wrong), but if I don't care about that, does it matter? Are there any pros and cons regarding transients or efficiency depending on what my resistor divider current is?

In their table of output voltage resistor divider values, for -5V they have R1 as 12.1Kohm and R2 as 37.4Kohm. Why not R1 as 1Kohm and R2 as 3.09Kohm?

Is this something I should contact LT about or is there an actual reasoning. Looking at other regulators, I don't think it specifies either.

\$\endgroup\$
5
\$\begingroup\$

A lower resistance will result in a higher wasted power which will require a higher wattage rating in the resistors. Too high a resistance and you may have issues with tolerance, and too high a resistance in parallel with the high Impedance input of the feedback pin can result in the resulting voltage being wrong.

The input is often one side of an op amp or comparator block.

\$\endgroup\$
  • \$\begingroup\$ Regarding wasted power. If the resistor value is 12K, you have 1.22/12K for a current of 100uA. You end up getting around 123uW. If you used 1Kohm, you'd get like 1.2mW. Those values don't seem that significant to me if you were using a .1W resistor or something. At least not significant enough to drive a design change. And based on your points about a high resistance resulting in tolerance issues, it seems that it's better to aim for lower values rather than higher values. \$\endgroup\$ – Dmoles Aug 2 '16 at 23:32
  • \$\begingroup\$ Let's ask the question a different way: what do you gain by using smaller resistances? I think there may very well be valid reasons to do so. At the end of the day, if you don't mind wasting a little more power, dissipating a little more heat, and using up more of the maximum supply current, then I think you are good to go. \$\endgroup\$ – kjgregory Aug 3 '16 at 1:11
  • 2
    \$\begingroup\$ @Dmoles it could be significant if you are building a battery powered design that, say, spends most of its time in sleep mode. In that case, 100 uA could be 100 times what the rest of the circuit draws. \$\endgroup\$ – alex.forencich Aug 3 '16 at 17:25
5
\$\begingroup\$

Passerby has given a good answer for your specific regulator, I'm going to answer (partially) this comment:

Looking at other regulators, I don't think it specifies either.

Yes, it sometimes does, in particular, it does for the venerable LM317, or its better-behaved sibling, the LM117.

enter image description here

Every example circuit you see will show R1 equal to either 120 or 240 ohms, which is a suspiciously low value for a simple voltage feedback divider. A lot of current is wasted in R1, so why is it so low?

The LM117, being a floating regulator, requires a certain amount of current flowing in order to regulate. The datasheet I'm looking at now, from National Semiconductor (Requiescat In Pace) states a minimum of 5 mA.

By a coincidence (not really, as you will see), VREF (always 1.25), divided by the resistance 240 ohms, turns out to be 5.2 mA. This is just above the guaranteed minimum current for the LM117 to regulate.

\$\endgroup\$
  • \$\begingroup\$ That's only required if the load is small, no? If I out is over 10 mA, it regulates fine with larger value resistors. \$\endgroup\$ – Passerby Aug 3 '16 at 0:21
  • \$\begingroup\$ @Passerby Yeah, at least that's what I've always believed. Should be fine if you can guarantee a minimum load, or if you don't care about regulation for small loads. \$\endgroup\$ – pipe Aug 3 '16 at 1:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.