OPA2140: Maximum load / sink currents - Output voltage accuracy

Hi Frederic,

how precise is your +5V reference and what precision do you need for the -5V reference?

Kai

Hello Raymond, 

thank you for your quick feedback.

I think I need to be a little more explicit

Regarding the first question

I've seen curve on figure 31, showing that the typical power consumption is lower at low temperature than at high temperature. As a consequence, I consider that the minimum guaranteed load current is lower than 20mA (taking into account the sink curve). 

I assume in worst case this value may be even lower due to process variations in the design of the component, let say maybe classically +/-20% (probably linked to the wafer thickness variation), leading to a minimum worst case to 16mA. 

Regarding thermal aspect, the component external temperature is estimated to 105°C (including calculator internal over-heating), giving some margin regarding thermal limitation. And as spotted previously the current limitation is higher at high temperature (estimated worst case at 40mA * 0.8 = 32mA @ 125°C, twice the worst case estimated at low temperature).

Thus, based on that analysis, I think that a load current below 16mA could be acceptable (with some margin if possible)

does it make sense to you ?

Regarding the second question

let me see in detail what you've send. I may come back to you if needed.

My question was also regarding any output resistors to be considered ?

Best regard,

Hello kai,

the precision of the voltage reference is less +/-0.25% (aging and temperature including)

the goal of my questions is to :

- check that the buffer / inverter realized with OPA2140 drives enough current (duplication of the buffer if needed)

- calculate the worst case tolerance of both +5V and -5V bufferized voltage 

- compare with solution based on LDO which leads to +/-3% of tolerance (probably with an higher output current)

Both solutions will compared in terms of board surface, performance, recurring cost

best regards

Hello frederic,

The figure 31 in OPA2140 datasheet is referring to short-circuit current for both sourcing and sinking vs. temperature. The short-circuit current the op amp can source or sink at a given temperature is for momentary events, and it is not designed for constant operation. In addition, that plot is generate with Vs=+/-18V.

Please let me know your +/-5V current source/sink requirements operated at 105C. What is your low operating temperature range, since you also look at near -50C range in Figure 31?

I am not certain that using op amp as LDO regulator is good options for your application, especially when you source/sink large amount of current in this part.  Initially, I thought that you are using the Op Amp as voltage references. Please let me know what types of loads you are driving for the application. 

In LDO, the power dissipation is (Vin-Vout)*Iout. In Op Amp, the output current is directly source/sink from rails. So if large amount current is source/sink at the load, more heat is dissipated at Op Amp IC (V+(Iout + Iq) for sourcing). Since this is not power amplifier, the excess amount of heat will put too much thermal stress inside of die. 

Regards,

Raymond

Hello,

I agree that opamp is not a LDO... (obvioulsy)

LDO has a tolerance of +/-3% and I'm seeking better accuracies (LDO is also a solution that is currently evaluated)

Regarding the global function itself : we use opamp saturation to perform rate and amplitude limitations. As a consequence, tolerance on opamp power supplies directly affect the tolerance on the global function itself. Thus we would like to minimize, as much as feasible, the power supply tolerance for the only components involved in this function.

main board power supplies is +/-12V 

exact reference used in design : OPA2140AIDGKT ==> VSSOP ==> ROJA = 180°C/W

The proposal is to use a +5V voltage reference as opamp input OPA2140, 

- one opamp in follower => +5V buffered reference

- one opamp in inverter (with #100kohms 0.1% resistors) => -5V buffered reference

These two +/-5V buffered reference are used to power supply 3 opamps (taken in 2 OPA2140 packaging containing 2 opamps each) leading to a worst case consumption of 4x2.7mA (all temp) = 10.8mA with opamp output current to be added (currently >20kohms i.e. < 250uA on each opamp output, eventuallty could even be minimized. Thus estimated to worst case 11.8mA, potentially permanent.

low temperature range : -40°C 70°C (external) + internal over-heating max 105°C

for follower (5V at opamp output) :

- quiescent power max : (12V - (-12V)) x 2.7mA = 64.8mW per opamp

- power due to load max : (12V - (5V)) x 12mA = 84mW per opamp

- total power for 1 opamp in OPA2140 = 150mW

- total power for 2 opamp in OPA2140 = 300mW

- die overheating : 300mW x 180°C/W = 54°C 

- with maximum external temperature : 105°C ==> 156°C... (ok worst case)

And also potentially a reliability impact to be assessed due to component stress

I assume the calculation may be quite similar for the inverter 

do these calculations make sense to you ?

In such solution, one of the key parameter is the OPA2140 quiescent current max 2.7mA, especially for the opamps we want to power supply

Do you see other components than OPA2140 that could be used for this application, components rail-to-rail (to limit the voltage drop in saturation) but with a lower maximum quiescent current ?

best regards,

Frédéric.

Hello Frédéric,

Thanks for the additional information and analysis.

I am concerned about the design margin, since OPA2140 will operate at elevated temperature continuously at approx. 105C. 

The low operating temperature at -40 is fine, since operating environment is approx. 70C and the op amp is not likely seen the low temp limit (unless it is soaking temperature conditions). 

I am recommending two options instead of OPA2140 per your application. OPA2192 seems to have better specification than OPA2140. 

Regards,

Raymond