• State TI Thinks Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 1 answer
  • Subscribers 46 subscribers
  • Views 714 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

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.

LM111QML: Servo loop - Offset voltage test

Namasivayam T
Prodigy 10 points
Part Number: LM111QML
Other Parts Discussed in Thread: LM111

Tool/software:

Hi TI team,

We have been conducting offset voltage test for the LM111JG comparator using the Single Auxiliary op amp servo loop method in a bench setup. The testing is performed at ambient temperature. We have observed an irregularity during this bench testing, which is listed below: 

  • We have taken three sample LM111JG devices to perform the offset voltage test. The servo loop circuit used for this testing is attached below as Circuit 1. Here we force a voltage at the auxiliary op-amp's non-inverting terminal 1 V, which we refer to as the target voltage in the test. When measuring the offset voltage at VOS out (Servo out), we observed that the VOS voltage changes at different times. Although the measured values fall within the datasheet limits, the offset voltage is expected to remain at a certain level with only minor variations. In our case, we are observing differences ranging from 200 mV to 1 V when measured at different times. The observed differences are tabulated below for reference. Refer table 1.

We have double-checked the circuit and didn’t find any issues. Please help us to understand if this behavior is due to the nature of the device or the target voltage, we are providing is wrong because of that this deviation is happening - If so please explain us the solution for this and tell how the target voltage that need to be forced is determined.

Note: We couldn't find LM111QML, so we are using LM111JG for the bench testing. If you provide a solution with respect to LM111QML also that will be helpful.

Table 1: 

VOS_OUT data comparison  

Device 

Attempt 1 

Attempt 2 

DUT_1 

-860mV 

-100.2mV 

DUT_2 

-2.476V 

-1.814V 

DUT_3 

-3.817V 

-2.906V 

DUT_4 

-2.004V 

-1.423V 

 

  • 0
    TI__Guru 64046 points

    Hello Namasivayam,

    What are the results when "target voltage " is set to zero or 7.5V? To get the best results, you need to "slice" the output waveform through the center of the waveform (Vpu / 2).

    Also try tying the output emitter to V- instead of GND.

  • 0 in reply to Paul Grohe
    Prodigy 10 points
    TARGET VOS_OUT DUT_OUT_BEFORE  110k DUT_INV+ DUT_INV- AUX_INV+ AUX_INV-
    13.5V 4.08V 13.5V 4.12mV 100uV 13.5V 13.5V
    7.5V 1.78V 7.5V 1.15mV 120uV 7.5V 7.5V
    1.4V -2.91V 1.4V -2.38mV 182uV 1.4V 1.4V
    0V -14.9V 76mV -14.7mV 162uV -520uV 72mV

    I have tried with the different voltages. When the target is 0V my SERVO OUT is saturating and when I'm giving target near to VPU the output I got above 4V. But by the datasheet of LM111JG 4mv is the max input vos, so it is exceeding the limit. At last what I observe is in 1.4v and 7.5v I got some passing results. But these also not repeating in my case at different times the output is changing as I mentioned previously. As you told need to try by connecting the emitter to V-. If you got any clue from the results let me know in order to resolve this.

  • 0 in reply to Namasivayam T
    TI__Guru 64046 points

    Hello Namasivayam,

    Try removing C2, and possibly increasing or decreasing the C1 integrating capacitor. Let the integrator determine the pole.

    What does the output look like? Are the edges sharp or distorted/rounded off?

    The comparator output does NOT go linear -  but instead it goes into a "PWM" mode. This creates the "analog" equivalent with the integrator stage (U2). The simulation should slow down when the output starts banging around.

    If the waveform is distorted - as in trying to go too fast, the distortion will bias the output in one direction.

    Is this only by simulation - or have you actually built the circuit?

    If simulation, make sure whenever you are simulating an oscillator (which you are), that the "Zero Initial Values" is selected in the transient analysis so that the simulator actually tries to start from zero and does not try to find an initial "operating point".

    Can you post your circuit?

  • 0 in reply to Paul Grohe
    Prodigy 10 points

    Hi Paul, 

    I have bench tested the same circuit which I have attached before. With that I have got some results that are matching with the datasheet for the tests input offset voltage, input offset current and input bias current (I'm performing all these tests in servo loop). In the input offset current and input bias current test I have used 50kohm resistors in the place of R1 and R2, according to the test procedures, I have interchanged them respectively and got passing result as per the datasheet. And in all tests, I have maintained the target voltage as 1v. 

    So, my question is

    1. How I should properly choose my target voltage for the servo loop method of testing comparator that is open collector like LM111 and when emitter pin is grounded. If I use 0v, my servo VOS_OUT is saturating to 15v, that is the reason I went with 1v. Pls explain how this target voltage playing the role when I force it between U1 and U2.

    2. By changing the target voltage does my DUT load is changing because I have a pullup resistor present there between U1 and U2?

  • 0 in reply to Namasivayam T
    TI__Guru 64046 points

    Hello Namasivayam,

    I personally have not built this circuit, so I am going by educated guesswork....

    Remember that the "output" voltage is "theoretical" output voltage created at the inverting input of the integrator (U2).

    The actual comparator output  voltage is digital 0 to 15V, so setting the servo voltage near or below zero requires very narrow duty cycles. It cannot go below zero. So the closer you get to zero - the worse the error.

    My guess would be to set the servo voltage to the middle of the actual output swing, or +7.5V. Just be ware that the OPA has input limits of 2V from each rail, so don't set the servo voltage higher than ±13V.

    In the actual ATE, the input common mode is changed by shifting the DUT supply voltages. Keep the output swing constant.

    You *could* also tie the emitter to DUT V- for a full 30V output swing to get the full VCM range - but now edge symmetry becomes even a bigger error. In the real device, the output should not affect the offset.

    You want to keep the edges symmetrical. Try lowering the value of the pull-up resistor to make the rising edge faster. But that does increase power dissipation and VOL swing.

    If test speed is not an issue, then set the loop to be as slow as possible to minimize the effects of the edge rates.

    So I would recommend setting the servo voltage to +7.5V to "slice" the output evenly.