REF3233: Output changes significantly every time power up

Could have to do with the RC filter at EN-input. Remove it and check again...

I am seeing the same problem with a REF3230 (3.0V). The output is sometimes 2.980 and sometimes 3.030. Simply powering down and back up can come up with exactly 3.000. The datasheet talks about slow rise time power supply. I have a 5V supply with a smooth 10 ms rise time. I have the enable input running from a system reset which occurs after 30 ms. The reset signal is 3.3V logic.

I tried changing to the RC circuit mentioned in the datasheet for the Enable input (750k and 100 nF) with no difference in behavior. There is a 10 uF cap on the output and the load is only 100 uA (an ADC). I added a 10k resistor for the load and that seemed to reduce the probability of this occuring from about 30% of the times to maybe less than 5% but it is still happening.

What is going on?

shalom@carmelinst.com

Hi Shalom,

the chapter with the RC filtering in the datasheet of REF3230 sounds like a bad choke to me. The RC filter seems to be the fix of an internal bug. Tell me how many customers ever have a stable rise time of input voltage, with a monotonic rise from 0V on!! In reality the input voltage can rise in any way you can imagine and a voltage reference should always settle to the proper output voltage, finally, when the dropout voltage is exceeded. If not, or if crazy RC filtering and other crap is nedded to fullfill this, trash this part.

If you cannot switch to the REF3030 or REF3130, which seem to work better in this regard, take a 5V reset chip, not 3.3V! Connect it directly to the input of REF3230. Take a reset chip with a push-pull-output stage like the MAX809 and connect a 22k resistor from its output to 0V to make it work down to 0V. Assure yourself that there's no ringing on the reset output line when toggling high!

Now you must check whether this scheme also works with non monotonic rises of input voltage. So, try to simulate this with a slowly varrying input voltage, which you turn up and down all the time. Try to simulate all kinds of input voltage slopes. Finally, check this scheme also with fast-rising input voltages. (See my thread about the LM4132...)

If everything works fine, take a hair dryer, heat the chip up a bit and check the performance again. Repeat this again with two or three fresh REF3230.

Kai

Hi Shalom,

I have a few questions about your application. Does your application only have a 5V supply with a 10ms rise time or is there a faster ramp rate voltage on the board? The REF3230 is an older device and since then we have refreshed the device in the REF3430 learning from our experience from the REF3230. I'll look into the REF3430 as it is P2P with the REF3230 and from my experience it should not experience these issues.

If this does not work, Kai has recommended the REF30xx and REF31xx that are also great devices but they do have a different footprint.

Is there a particular specifications you need for your 3V voltage reference?

- Marcoo

I looked at the REF3430. It does not say any more that there is a maximum risetime of 2 ms on the power supply. However, there are still comments about supply risetime:

Page 15:
"Supply voltages below the specified levels can cause the REF34xx to momentarily draw currents greater than the typical quiescent current. Use a power supply with a fast rising edge and low output impedance to easily prevent this issue."

Page 18:
"In applications with slow-rising input voltage signals, the reference exhibits overshoot or other transient anomalies that appear on the output. These phenomena also appear during shutdown as the internal circuitry loses power."

My questions:
1. Comment on page 15 -- how much more current? In my application a few mA will not be a problem.
2. Comment on page 15 -- Does everything settle to the specified output voltage and normal current draw once the supply is in proper range?
3. Comment on page 18 -- How much overshoot? If the normal output is 3V and the input voltage is 5, can the output overshoot all the way to the 5V?

Hi Shalom,

1. The current should be around 0.5mA. This current is limited by the inrush current protection of the REF34xx so it doesn't go that high.
2. Assuming the enable pin is in the proper range, the output voltage should follow the turnon time in the EC table. The output voltage can settle before the supply current stabilizes.
3.The overshoot should not exceed 0.5V.

Your supply current seems to have a ramp up, how are you sourcing the voltage reference that there is a 10ms ramp up?

-Marcoo