I'm looking for tips on assembling a voltage regulator circuit to provide voltage to a sensor.

I need to regulate the output voltage to between 4.99 and 5.01 VDC. The input will be a 9V battery. The load will be steady state and under 10mA current. I need a circuit that will self regulate to the proper output voltage without outside input (no human fiddling with a POT after initial assembly), and give proper output voltage even as the 9V battery ages and drops in voltage. Duty cycle will be short (a few minutes) so efficiency isn't a concern.

This adjustable voltage regulator based on an LM317 looks great, but in the real world will the output voltage actually drop by a significant amount as the 9V battery depleats and looses voltage? Also is this regulator good enough to output between 4.99 and 5.01 reliably? I'm open to any and all suggestions.

 

The chip will hold within 1/5th of 1% but you're going to have to build an adjustable circuit to get the initial accuracy.

 

This is not a task for an LM317. The output voltage typically drifts far too much and it consumes way to much power. You need a precision ultra low dropout regulator for your application.

 

Good point. The datasheet lists a maximum time drift of 1% per thousand hours. The 317 will have to be checked for calibration every 100 to 200 hours.

 

Thanks guys for the excellent info, I will look for what you suggest. Do you have a favorite or known-good model?

This sensor is part of a diagnostic tool that will only see about 5 minutes of usage per use, maybe an hour or two total on-time during the life of the tool in most cases. Does this lessen the concern for the 1% drift per thousand hours, and power consumption?

 

maybe an hour or two total on-time during the life of the tool in most cases. Does this lessen the concern for the 1% drift per thousand hours, and power consumption?

Do the math. 2 hours times 1%/1000 hours = 0.02% drift with time.

 

Thanks, I wasn't sure if it was that strait forward. .02% is tolerable.

But perhaps this is more along the lines of what Evil suggested: http://www.ti.com/lit/ds/symlink/lp2986.pdf

Thanks again for the info. I know sometimes what's on paper isn't reflected in the real world and I appreciate you guys helping with real world info.

 

But perhaps this is more along the lines of what Evil suggested: http://www.ti.com/lit/ds/symlink/lp2986.pdf

All in all, a good chip, but you will still have to adjust the initial voltage and nothing is said about drift per time on the datasheet.

 

You want a voltage reference chip. There are plenty of chips out there for that task.

http://www.mouser.com/Search/Refine.aspx?N=1720874

 

I agree with spinnaker. Pick a voltage reference with a 5V output and the accuracy you need.

 

Wow! There are about 4000 of those within .2% in initial accuracy.

 

I've used a voltage reference chip with a current amplifier in the past... This was used to excite load cells to get an accurate conversion on an ADC chip.




I strongly suggest you study the PDFs I've attached for you.

 

Thank you very much, I will read the PDF's.

If it turns out that the current requirement for my sensor falls within the current capability of the voltage reference device (<10mA for the LT1021), is there any downside to using the reference device alone without a current amplifier?

 

Wow! There are about 4000 of those within .2% in initial accuracy.

The reason places like mouser have that wonderful parametric search tool.

 

Thank you very much, I will read the PDF's.

If it turns out that the current requirement for my sensor falls within the current capability of the voltage reference device (<10mA for the LT1021), is there any downside to using the reference device alone without a current amplifier?

If the reference has to provide the current then it will heat up some from the I*Vdrop across the reference which could change the voltage slightly.
Otherwise you don't need the amplifier.

 

Those PDF's were a good read, thanks! Using a good Fluke meter I determined my sensor only draws 2.7uA, much less than I thought, so I should be able to use a reference IC without a current amplifier. Maybe that Figure 10 circuit above will do the trick, time to build one and find out.

 

Those PDF's were a good read, thanks! Using a good Fluke meter I determined my sensor only draws 2.7uA, much less than I thought, so I should be able to use a reference IC without a current amplifier. Maybe that Figure 10 circuit above will do the trick, time to build one and find out.

Crutschow's right, you'd only need the current amplifier if the reference (as per specs) cannot deliver enough current for your application.
Also, no voltage reference is completely immune to changes in temperature. Although some chips are more stable than others, temp effects can only be minimized and never completely eliminated. And temperature variations can arise either from the power being dissipated by the device itself or from changes in ambient temperatures. One trick to stabilize temperature in a circuit is to place the circuit itself inside a small "oven" that will keep it hot well above ambient temperature and well below its operating parameters... say about 50°C (122 °F). Crystal oscillators are particularly prone to variations in frequency due to temperature changes, and one way to work around that problem (when one needs high frequency stability) is to used so called OCXO's, which stand for "Oven Controlled Crystal Oscillator".

 

In the "Figure 10" schematic above, what does it mean when they have the diode connecting to the center of R2? I've never seen that before.

 

In the "Figure 10" schematic above, what does it mean when they have the diode connecting to the center of R2? I've never seen that before.

Part of temperature compensation, the diode is used as a temperature sensor. Give this EDN article a read. They explain the principal well and also mention Linear Technologies. You will see how it works in the scheme of things.

Ron

 

Thanks, the article mentions using a diode connected bipolar transistor instead of a real diode, but the schematic explicitly shows model 1N4148 diode. What I'm confused about is where does the 1N4148 diode physically connect to R2?