Solving REF192 IC Problem for ADC, Instrumentation Amps

[j777]

Hello,

I am using a REF192 from Analog devices to provide a 2.5V reference to an ADC (Analog devices AD7718) and two instrumentation amps (Linear Tech. LT1167). The VIN pin is connected to a 5V supply and has a 10uF tantalum cap and a .1uF ceramic cap in parallel. The VOUT pin also has a 10uF tantalum cap and a .1uF ceramic cap in parallel. The SLEEP pin is tied to VIN. For some reason when the circuit initially gets power a clean 2.5V can be read on VOUT but after about a minute it rises to about 3V and remains there until the power is removed and reapplied. So far I replaced the REF192 but the problem still exists. What could be causing this?


Thanks

 

[NoTime]

From the doc it looks like you have it hooked up ok.
Try disconecting the output from the rest of the circuit and see if the problem still occurs. Something could be wrong elsewhere.

 

[berkeman]

Does the device get hot to the touch when this happens? If so, suspect a fault in the output circuit as NoTime suggests.

It sounds a bit like there is something going on with the SLEEP~ pin over time. Are you sure it is tied to a hard 5V? Have you probed the SLEEP~ pin when the output drifts up like that? What is the load that is tied to the output pin? If it can turn into a pullup for some reason, the Vref output will no longer be determining the output voltage. That is true of most power supply circuits -- they function as pullups to a pull-down load. If the load pulls the Vout up somehow, the power supply output has no capability to pull the Vout down into regulation.

 

[j777]

The device never got hot and the voltage on the sleep pin remained the same at all times. One of the instrumentation amps turned out to be the culprit. I've replaced it and so far the reference voltage has remained at 2.5V and everything seems to be working. As is the case with any problem like this I'd love to know what exactly was causing it within the instrumentation amp but for now I'm glad it's fixed.

Thanks berkeman for your explanation on the possibility of the output being pulled up by something; it's very likely that this is what was happening. Also thank you NoTime; I did as you suggested first and found that the problem was gone which prompted me to replace the instrumentation amp.

 

[berkeman]

As is the case with any problem like this I'd love to know what exactly was causing it within the instrumentation amp but for now I'm glad it's fixed.

Glad that you're past that problem. One possible explanation (and something to watch out for in general) would be that you might have hit the amp input pin at some point with enough of an electrostatic discharge (ESD) zap, that it latched up and damaged that input pin structure. You have to be careful when you work with open, powered-up circuitry like that. Normally a product is closed up and the circuit boards are protected, with explicit ESD protection on any IOs that are user-touchable. But when you work with an open circuit board, and it is powered up, then even a small ESD hit from your finger or a tool you are holding can cause a latchup of a device, which can cause it to soft fail (where it works again after a power cycle), damage itself so even a power cycle doesn't fix it, or just plain smoke (if it draws enough current from the power supply).

You can wear a grounding wrist strap (which should always be worn when working with regular production devices that will ship to customers), or in the R&D Lab environment, you can just be careful about grounding yourself periodically, and especially as you are about to start probing or touching an open circuit board. You can ground yourself by touching any grounded piece of metal, like an unpainted part of a metal chassis for an instrument that has a 3-prong power plug.

 

[NoTime]

Also thank you NoTime; I did as you suggested first and found that the problem was gone which prompted me to replace the instrumentation amp.

Glad to be of assistance