OPA376: OPA376 issue in high temperature

Hi Nathan,

Nathan Ding said:

Personally I think it may be related to common-mode input range, because the negative input voltage is lower than reference ground of OPA376, and there is an extra voltage drop of positive input. But in d/s, I only find the common-mode  voltage range in 25C, could it become very low when temperature rising

I think you are right. The common mode input voltage is -20mV which may become a problem at elevated temperatures.

nathan_opa376.TSC

One remedy is to move the ground reference of OPA376:

This would force the common mode input voltage of OPA376 to be positive.

Kai

HI Nathan,

I agree that the issue may be related to the input common-mode voltage range being negative below (V-).

Although, the device specifies common-mode range (V-)-0.1 and (V+)+0.1V at 25C room temperature, the tested condition for common-mode rejection is from (V-) < VCM < (V+)-1.3V.  The datasheet offers typical performance over temperature for the common-mode above (V-), and as you have mention, it degrades at the higher temperatures.

CMRR over temperature for (V-) < VCM < (V+)-1.3V:

  I agree with Kai, raising the common-mode will be the best solution.

Thank you and Regards,

Luis

Hi Kai Luis,

Thanks for feedback.

Basically I guess it is related to common-mode input voltage, but I am confused that why the voltage drop in high temperature is so large? up to 1V or even larger. Based on the CMRR vs. Temperature data, the CMRR only reduces 10dB in 100C compared to 25C. 

It is possible to cause such a lager voltage drop in 90C because of common-mode voltage range or CMRR? Do we have the data about common-mode voltage range in high temperature? Or if there is any other possible reason? Thanks.

Regards,

Nathan

Hi Nathan,

keep in mind that many specifications like the input offset voltage (and figure 7 as well) are only valid at a common mode input voltage of Vs / 2. This is very often overlooked. At Vs / 2 the OPAmp works idealy. But at different common mode input voltages the situation may look totally different. See figure 23.

A specified common mode input voltage range of 0.1V beyond the supply voltages does not mean that the OPAmp is achieving its highest precision there. It may only mean that the OPAmp does not show phase reversal or something like that. But it doesn't say anything about the input offset voltage.

The specification of common mode rejection is more useful. If the common mode rejection is only specified for a common mode input voltage from 0V on (and not from -0.1V on !), then read this as: "If you want highest precision, then don't use a common mode input voltage below 0V". So, using a common mode input voltage of -20mV would even not be a good idea at 25°C, the temperature at which the common mode rejection is specified. And using a common mode input voltage of -20mV at 90°C would be a double mistake.

The big change of output voltage is mainly caused by the very high gain of your circuit (82V/V). So, even a small change in the input stage appears at the output of OPAmp highly amplified. 1V output voltage change means an input offset voltage change of merely 12mV which looks normal when the input voltage begins to leave the linear operating range of OPAmp.

If you want to run the OPAmp with a common mode input voltage of -20mV at 90°C then you have to choose an OPAmp which is specified to work properly at these conditions. And the OPA376 is not guaranteed to properly work under these conditions.

Kai

Hi Kai, Thank you,

Hi Nathan, 

The OPA376 datasheet VCM specification only offers assurances for VCM above the negative voltage, best bet is to modify the circuit to raise the common-mode voltage to a voltage above (V-) . 

The OPA396 or OPA391 are other precision amplifiers  offering a CMRR specification extending the range to (V-) -0.1V.  The CMRR is 90-dB minimum over the extended TA = -40C to +125C temperature range.  Although, these devices offer lower bandwidth than the OPAx376.

Thank you and Regards,

Luis

Hi Luis,

Does OPA391 have DBV package samples? I only find the DCK package on ti.com, but customer need DBV package...

Or is there other devices that is DBV package and can also meet the requirements above? Thanks.

Regards,

Nathan

Hi Marek, Luis,

But OPA325 and OPA392 doesn't give the value when Vcm < V(-), 

OPA392 also doesn't give the CMRR when Vcm < V(-). 

Do we have any other devices? Thanks.

Regards,

Nathan

HI Nathan,

You are correct on the OPA392, this is why I did not suggest the OPA392.

However, the OPA325 does support VCM < V(-) , in the minimum range (V-) - 0.1V or -100mV below the (V-) negative supply. It offers a specification over the temperature range -40C to +125C. This should cover the -20mV common-mode condition from the Customer.

Kindly let me know if I misunderstand the question.

Thank you,

Luis

Hi Luis,

Yes, I understand that OPA325 give CMRR at negative Vcm. But I have two questions:

1.Here I find there are two items about CMRR, from my understand the first test condition is V(-) - 0.1V < Vcm and TA=25C, and the second test condition is TA= -40C to 125C but Vcm=Vs/2. So it doesn't give the CMRR in negative Vcm and TA =125C

2.OPA325 doesn't give the Vos when Vcm < V(-)

Here is OPA391.

Thanks.

Regards,

Nathan

Hi Nathan,

Using the common-mode rejection ratio (CMRR) specification, one can estimate the change in the op-amp offset vs common-mode voltage change.

The OPA325 offset is typical ±40µV or ±150µV max, when common-mode is centered at the middle of the supplies, VCM= Vs/2.  The OPA325 worst case based om the maximum input offset drift at temperature +90C, is: 

150µV+7.5µV/C*(90C - 25C) = ±637.5µV offset at +90C (very worst case) where 150µV is the max offset, and 7.5µV/C is the maximum drift. 

(Edit: March 29, 2023, corrected calculation to account for offset drift at +90C). 

If we assume the device supply is +5V supply, and the common-mode required condition is -20mV, this corresponds to a common-mode change of -2.520V from the middle of the supplies.  The OPA325 minimum common-mode rejection over the -40C to +125C temperature range is 95-dB, or about ~17.8-uV/V. 

The change in offset with CM shift will be an estimated 17.8µV/V * -2.520V or -44.81µV offset change, worst case.  If we are conservative and take the max offset of ±637.5µV and add the 44.81µV offset shift with CM, you will end up with an estimated ~682µV offset at the amplifier input.  The customer circuit has a gain of ~82V/V, so this corresponds to an output error of 55mV reflected at the circuit output. 

Keep in mind, I used the min/max specs, the offset will be typically smaller than this estimate.  Also, Figure 3 shows the offset voltage vs common-mode voltage plot showing the typical offset @25C all the way to -100mV below the negative supply, where the offset is typically around ±25µV. Figure 19 shows the typical CMRR change vs temperature, where the CMRR seems very constant with temperature.

The OPA325 offers a much better CMRR spec of 95-dB (17.8-µV/V) minimum extended to (V-) - 0.1V over the temperature range, while the OPA376 has a much less common-mode rejection of ~70dB (316.22µV/V) typical at hot temperature. And, the OPA376 CMRR spec does not cover CM voltages below the negative supply.

Hence, the OPA325 should reduce the error significantly. 

Thank you and Regards,

Luis 

Hi Nathan,

The OPA325 worst case input offset based on the max offset drift at +90C, is:

150µV+7.5µV/C*(90C - 25C) = ±637.5µV offset at +90C (worst case) where 150µV is the max offset, and 7.5µV/C is the maximum drift. 

The offset change due to common-mode is -44.81µV offset change, worst case.  So the total input offset error is ~682µV at the amplifier input. The customer circuit has a gain of ~82V/V, so this corresponds to an output error of 55mV reflected at the circuit output. Most of the error is due to offset drift with temperature, and not common-mode in this case.

Regards,

Luis

Hi Luis,

Thanks for detailed explanation. I will apply for OPA325 samples for customer later.

So from my understand, this issue is caused by that OPA376 CMRR drops a lot with negative Vcm and high temperature? 

And more, I have another question, customer tried with TLV387 yesterday, they found that there are still a few problem at some sample points, the output voltage of 387 drops ~200mV(not as bad as OPA376, but still not perfect).

And then, customer try to change the reference ground of amplifier from the right -VIN1 to the left -VIN1, so that there is no negative voltage. By this way, customer finds that there isn't any issue no matter whether using OPA376 or TLV387 (but change ground reference method is only for test, because they don't have enough time to change layout now) 

So the issue is actually related to negative Vcm, but the question is that why it still doesn't work perfectly with TLV387? Do you have any suggestions? Thanks.

Here is the CMRR of TLV387 I find in D/S.

Let me know you need additional information.

Regards,

Nathan

HI Nathan,

We discussed the issue on the OPA376. The datasheet shows the CMRR changes from 90dB @25C to 60dB @150C, at hot temperature, where CMRR is specified above V(-) supply.  The datasheet offers no assurances for CMRR for voltage more negative than (V-). We recommended the OPA325. The OPA387 is a zero-drift (chopper) amplifier that was not discussed above.  

Since the error is rather large 200mV (2.43mV input referred); and the error occurs in some measurements/samples per the description, and the issue goes away while changing the grounding scheme, this may not be related only to the common-mode input voltage on the amplifier but potential external noise, ground shifting, settling or a combination of unrelated issues.

Let's continue the conversation via direct email.

Thank you,

Luis