REF3030: temperature drift correlation

Dear Diego,

sorry for highjacking this topic/discussion, but may i ask you on this:

The temperature coefficient is determined using a box method, therefore the max could appear at any temperature.

While most parts will have a similar shape, we do not guarantee that the max will be at any specific temperature.

Is this true for temperature drift information of LDOs too? I mean the background to it beeing the method used?

And additionally, is it true for basically all temperature drift data f.e. for amplifiers also?

Thanks in advance, best regards

Michael

Hi Michael,

I'll answer this in place for Diego. From my understanding that is true because if it's not a EC spec then its typical and not guaranteed. But for more accurate answer or confirmation, I would suggest you to ask this in the LDO subforum.

-MZ

Hi Michael,

From my understanding LDOs probably do use the box method when calculating the temperature coefficient for their accuracy calculation as it is the most common method. This temperature coefficient is then used in the accuracy calculation as shown in following app note:

http://www.ti.com/lit/an/slva079/slva079.pdf

A good introduction to temperature drift/coefficient is this following document:

http://www.ti.com.cn/cn/lit/an/slyt183/slyt183.pdf

Typically the calculation from the app note linked above is included in most datasheets for voltage references.

The difference between temperature drift in LDO’s and VRefs is that LDOs don’t guarantee a max and only give a typical. This might be because LDO’s such as the REF1117 aren’t meant to be used as references but as clean DC/DC converters. In newer datasheets the temperature drift of the REF1117 doesn’t exist and it’s just bundled into accuracy. If you are trying to characterize the max then that is very application specific for LDOs, you might have to prototype and test since a simulation might not handle it. There is no method to take the typical and convert it to a max.

“I really need to know more on this whole temperature drift topic and understand more on how to "read" the datasheets information regarding it. What can i expect from the graphs, which technology should be used/preferred, which technology does provide the smallest spread (drift wise) by design, etc.”

Any graph is just a typical graph and that means the curve is not guaranteed. For example in the TL431 datasheet in section 7.14 figure 1 it shows 3 typical lines for Vref vs temperature. All 3 lines meet the EC table specs but their slope varies positive or negative for the same device. This is because temperature drift is a non-linear function and we cannot guarantee where in the function the device will be. This is also the reason the box method is used compared to other methods such slope and butterfly.

www.ti.com/.../tl431.pdf

The technology that provides the smallest drift for voltage references will be a series Vref such as a REF50xx family. But it has limitations of input voltage and output current. This is compared to a shunt ref such as the TL431 that has no input voltage limitation or a LDO that has a very high output current.

Temperature drift/temperature coefficients come in many different formats that can get confusing. For LDOs you will only see accuracy. In Vrefs you will see ppm/C which can be converted to a %. In amplifiers you see uV/C which looks the same as ppm/C but uV/C is over the whole voltage range while ppm/C is a percent of the output voltage. For example a variable output voltage reference will have a higher uV/C at 5V than at 1.25V but the same ppm/C.

In terms of design for smallest drift that is complex. While we do give a block diagram of our devices, the internals are much more complex and that data is not shared.  All designs are optimized for different results.

“Where does the price difference come from, the trimming of REG1117?

Does this trimming influence the voltage drift?

Does figure 6 of LM1117 datasheet Show "the same thing" as REG1117 datasheet page 7 bottom left graph?

Is there something like an overview/comparison of different Regulators (preferably fixed 3.3V) which is showing (initial) accuracy, temperature influence and such?”

My take on these in reference to Vref is that the 2 devices are very different in the side even if the block diagrams are the same. They were developed by 2 different competitors that were then acquired by TI.

Overview comparison can be done on TI.com

http://www.ti.com/lsds/ti/power-management/single-channel-ldo-products.page#p634min=-47;3.3&p634max=3.3;125&p1154=3.3;3.3

For example this link has all the fixed 3.3 LDO sorted by increasing accuracy.

Temperature drift is very complicated and trimming could be a factor but typically trimming is focused on getting as close to the fixed output voltage as possible at very specific conditions. In Vrefs the trimming will get the Vref to a set value i.e. 2.5V then they are tested and then sorted into different categories.

I hope this was helpful.

-MZ

Hi MZ,

yes, this helps me a lot, thanks for all this info and hints.

I understand now more on the backgrounds of it because i'm kind of new to this Overall temperature-drift Topic.

Take care

Michael