If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

• TI Thinks Resolved

# INA326: INAMP Output Sticks High to Rail Voltage

Part Number: INA326

I have the following circuit without R20, the 10K resistor to ground.  Occasionally the output upon power sticks to the 5V rail.  I've never had this problem before, but use this circuit at much higher gains.  The gain is set at 11.1X.

The input is from a Hall sensor, but substituting a 10K resistive pot produces the same problem.  Can someone help me understand why this is happening and what I can do to fix it?

• Hi Marvin,

Q: I have the following circuit without R20, the 10K resistor to ground.  Occasionally the output upon power sticks to the 5V rail.

You have DC bias at Vin- at approx 2.5V. Your DC bais at Vin+ should be also near 2.5V + signal. What is your DC bias at Vin+ when Vout = 5V.

The only scenario that Vout is close to 5V, assume you have correct input common voltage and deltaVin = Vin+ - Vin- = 0.450V.

/cfs-file/__key/communityserver-discussions-components-files/14/5367.INA326-E2E-09242020.TSC

What happened if the output power is railed at 5Vdc at the output? Do you have power cycle it in order to reset it? Are there any transient input events occurred when the Vout is stuck at 5V?

Q: The input is from a Hall sensor, but substituting a 10K resistive pot produces the same problem.  Can someone help me understand why this is happening and what I can do to fix it?

What is the DC bias at Vin+ when you substitute R20 in 10 kohm as shown in your image? If there is no DC bias, the circuit will not work. Vout = Gain(Vin+ - Vin-) and the Gain=11.111 V/V.

Best,

Raymond

• In reply to Raymond Zhang1:

The Vcc is from USB and about 5.197 VDC.

The input at the + pin is about 2.821 VDC.

The — pin offset is set to 2.601.

That yields 2.448 VDC at the output pin.

When the chip sticks high, it happens when power is first applied (i.e., when the USB cable is plugged in).  When the chip sticks high the output pin is 5.19 VDC.

• In reply to Marvin De Beque:

Hi Marvin,

Q: That yields 2.448 VDC at the output pin.

This seems to be working properly with Vin+ = 2.821V and Vin- = 2.601V.

Q: When the chip sticks high, it happens when power is first applied (i.e., when the USB cable is plugged in).  When the chip sticks high the output pin is 5.19 VDC.

With USB cable is plugged in, it is difficult to tell how the power and signal pins are connected in sequences. You may have bounce and debounce power on/off issues when USB plug is inserted. Say one moment the power pin is connected, then next msecs, the Vcc pin is disconnected, then Vcc pin is reconnected again. So I am unable to explain what is really happening during the USB plug-in process.

If you place a toggle switch at +5V pin on the USB power cable, your issue will be likely resolved. With the switch in off position, plug in the USB cable, then turn on the toggle switch and apply power to your INA326 circuit. If you still see issues, please let us know.

Best,

Raymond

• In reply to Raymond Zhang1:

The USB power is not the issue.  We've used the INA326 in multiple designs for amplifying strain gauge bridges.  We've sold well over 100 of these designs with not one failure like the one we are seeing now.

The only difference between the circuit shown and our strain gauge amps is gain.  This circuit is set for a gain of 11.1.  The strain gauge amps have gains ranging from 1,111 to 2,260, all running from USB power.

The circuit is essentially the same as the one shown, but the gain resistors are different—RG1 is 3.6K or 2K, and R14 is 2M  or 2.26M.  C23 is changed to 47pF.

This leads me to believe that the INA326 does not perform well with low gain, but I can't see why.

• In reply to Marvin De Beque:

Hi Marvin,

Q: I have the following circuit without R20, the 10K resistor to ground.  Occasionally the output upon power sticks to the 5V rail.

If plug in the USB power on then off 10 times with 30-60 seconds apart, how many times Vout is stuck at 5V rail. Does the Vout=5V have anything to do with how fast or slow that USB power cable is engaged in the power socket.  What is load at the output of INA326? I assumed that this is undamaged part.

I see that your circuit is similar to the figure 1 below. Please provide me with the above information, and I will look into it.

Best,

Raymond

• In reply to Raymond Zhang1:

The degree of repeatability varies with board.  We have produced 24 of them so far.  We have a number of boards with no apparent issue and some with the issue that is persistent.  I've got three boards on the bench right now that seem to work more often than fail.

At first there was no known cause for the failure.  I've tried replacing chips and in some cases that seems to have helped, but not consistently and I doubt that it is bad or damaged chips.

The load on the chip is 3K.  There is a 1K in-line resistor and a 2K resistor to ground to reduce the output range to the A/D input.

• In reply to Marvin De Beque:

Hi Marvin,

The degree of repeatability varies with board.  We have produced 24 of them so far. and some with the issue that is persistent.

I am interested in the ones that have persistent issues, where Vout is stuck to +5V when you engage the USB power plug. I'd like to know the the "failure" rate when these parts are engaged in USB power. How many units do you see with this particular behavior?

What USB power modules are you using? Do you have scope that can capture the USB's voltage and current profile during the plug-in period? I am interested the board and USB socket that have consistent issues. Here is what you have to do.

1. Set the data collection period up to 5 seconds or so (the data collection time has to be greater than Vout is stuck to 5V event).

2. Set a triggled signal via INA326's Vcc power pin (say channel 1, trigger the voltage at approx. 1V ). If you have a low current probe, place it in channel 2. I am interested the voltage and current transients when USB power plug is engaged to the USB socket. Please send me the captured scope shots during power-on event, when Vout is stuck to 5Vdc at its output. If you have 4 channel scope, I will be interested to see Vin+ and Vin- input signals as well.

3. I assumed that all USB 5V socket are operational, and no significantly high frequency noises with in-spec dc regulations. If you are not sure, please capture 5Vdc signal, and check for voltage ripple, high frequency noises. I assumed that each USB port can handle up to 0.5A . Do you know that the %load_regulation on your +5V USB adapter.

I want you to know that this may take several days to get resolved. Meantime, I am trying to understand what you have in your configuration and setup that may cause the issue.

Best,

Raymond

• In reply to Raymond Zhang1:

Thank you!  I'll look into getting the scope plots for you.  Unfortunately, all I have are two analog channels at my bench.

• In reply to Marvin De Beque:

Hi Marvin,

I am going reassign this inquiry to the application engineer who is specialized in INA326 product.

If you have other questions, please let me know.

Best,

Raymond

• In reply to Raymond Zhang1:

Hi Marvin,

Sorry to hear of your recent challenge with the INA326, and thanks for being a long time user.  I do have a theory of what might be contributing, and it would be great if you could do an experiment to confirm.

On page 9 of the datasheet, you'll find some guidelines on how much current can be safely pushed through the gain set resistors.  This is due to the fact that these currents have to come from the charge pumps internal to the device, used to boost the supplies inside the part to get the very useful rail-to-rail performance.  Figure 1 attempts to guide the user in useful R1/R2 values for typical gains.  For instance, for a gain of 10 it suggests 40k & 200k.  In a gain of 10, the maximum input differential voltage would be expected to be 0.5V (5V/10), and hence 0.5V/40k=12.5uA... the same max current value described on page 9 of the datasheet.

However, during startup, by the fact that the output is railing (and there is a Vdd/2 bias applied to the input ramping with the supply), I believe the differential input voltage is likely to be much larger than 0.5V.  That means, at least momentarily, a larger current is demanded through the gain set resistors.  I'm theorizing that this large current demand potentially collapses the internal charge pump, and it probably doesn't recover until a power cycle where it get a new shot at delivering the always slightly different requested current.

What I'd like you to do is increase the values of R1 and R2 by at least a factor of 5X to see if that resolves the issue.  I suspect that momentarily you have as much as 2.5V across the inputs, and dividing by 12.5uA yields a minimum resistor value of 200k (and scale other gain set resistor 5X as well, and adjust the cap accordingly).  I look forward to your results.

Thanks,
Scott