LM4132AMF-3.3 vreference output appears to be outside specifications

Hi Fausto,

I will look into this and get back to you. How many units have you looked at? How many are showing this behavior?

Chris

Chris:

thank you for your fast reply
We have tried about 3 units
They all show the same behaviour
I am making the measruments as accurate as possible
I only have this reference circuit installed on the board to avoid anyother source of error

Thanks

Fausto

Hi Fausto,

Are you mounting these units on boards yourself? It sounds like you might be if the reference is the only unit mounted on the board. There is an effect known as thermal hysteresis, which sounds like is the cause of the shift you are seeing. Thermal hysteresis is the change in the room temperature reference voltage level after temperature excursions, which could be attributed to ambient temperature, or even the soldering of the unit.
The units are 100% tested to meet the initial accuracy tolerance spec per datasheet, prior to any temperature excursion seen after shipping out of the factory. The initial accuracy is guaranteed at room temperature, but is the initial accuracy. Temperature excursions and board stress will affect the part.

In regards to board stresses, flex in the PCB can transfer to the package and affect the die. To minimize this effect, it is recommended to place voltage references near the edges of a PCB, where the mechanical stresses are the least (highest at the center). Adding a U-shape cutout around the part will also help reduce mechanical stresses.

How are you mounting the unit on your board? What does the board look like, and where is the unit in relation to the board?

I hope this information is helpful.

Chris

Chris:

The overall size of theboard is 6.5" x 6.5", 0.063" thickness

The regulator section is loacated about 1" away from the edge of the board

The board is just mounted on stand offs over our test bench (not installed in the final assembly)

I do not have a "U" cutout. I will add this cutout on my next rev

Does this shift from mechanical stress only affects the initial accuracy?

What about the long time stability? (this is my main concern)

is there any relation between mechanical stress vs Vout error that you can provide? on which direction?

Thank you

Fausto Bartra

Chris:

We also tried adding stresst to the board by flexing it, but it had no effect on the output voltage
I think this is a different problem
Can you email me detail information on how this device can be tested to match your specifications

Thank you

Fausto Bartra

Chris:

is there an eval board for this part (or similar)?

Thanks

Fausto

Hi Fausto,

I do not believe TI has any evaluation boards for its voltage references.

What is the procedure you are following to solder the unit down?  

What is the measured value of each unit?  What is the integration/aperture time of the DMM?  Do you have another meter you can use? Have you looked at voltage references before without issue? 

Chris

Chris:

these prototype boards are hand assembled
Can you provide us with a procedure to properly assemble and test these devices?

Thank you

Fausto Bartra

Hi Fausto,

I don't have a specific guideline, but as long as you can stay within the maximum ratings in the datasheet on page 3, you should be okay. For hand assembly that is likely the line for less than 10 seconds at 260C.
Are there any markings on the part you can read?

Chris:

I can see "R4EA"..

Thanks

Fausto

Hi Fausto,

Thanks for getting the markings.

It's hard to say what exactly is the source of the problem you are seeing.  A couple of ideas for you.

1) Have you tried other meters?  Also, like I mentioned above checked the integration/aperture time of the meter and tried increasing it?  

2) Are you certain the meter is accurate and calibrated?

3)You could try loading it on purpose and seeing what affect that has, to see if there is excessive drop in your measurement setup, which may be amplified by loading.

Hope that helps.

Chris

Chris:

To solve this problem I actually replaced the TI part with LT1790ACS6-3.3 frrom Linear technology and it work the first time no ptoblems

Thanks for your help

Fausto Bartra