OPT3101: Weird fluctuation of I/Q-Phase in illumination xtalk calibration

Hi Moses,

This twisted curve is not expected. You mentioned the possibility of overheating during soldering. Are there any visual signs of damage on boards 2 and 3? Are these boards all identical? Can you think of any parameters that may have changed in between runs of these tests?

Thank you,

Brent Elliott

I've confirmed with hardware guy. There is no visual damage on the PCB. The PCBs are made by PCB factory in Japan with high quality production and the temperature is strictly controlled. So we believe there is no issue with the overheating problem.

The temperature setting of thermo-chamber was -10~40℃, which I believe there is no temperature damage to the chip. And there is no difference on parameters setting and hardware settings among those four prototypes.
The measurement of no.1 and 2 are done at same time, same chamber. And the measurement of no. 3 and 4 are done at same time, same chamber, too.

So I would like to know, is there any possible issue on the OPT3101 internally, such as broken ADC or other internal defect?

Hi Moses,

Let me look into this. I will try to get a reply by the end of today, latest tomorrow.

Thank you,

Brent Elliott

Hi Moses,

Can you explain the procedure for using the thermo-chamber? How fast are you heating/cooling the device? Are the separate sweeps on device 2 and 3 both heating/cooling the device or both? There could be a hysteresis effect.

Thank you,

Brent Elliott

I've done experiment of hysteresis effect.

The results show that there is no obvious hysteresis effect, cause I_PHASE_XTALK was stable, and Q_PHASE_XTALK remain problematic.

(Left: No.3; Right: No.4)
(Blue: 40℃～-10℃ descent; Orange: -10℃～40℃ ascent)

I would like to know:

1. Are twisted curve of Q_PHASE_XTALK acceptable according to the experience of TI?
   Is there any legacy data from TI's development stage shows that I_PHASE_XTALK or Q_PHASE_XTALK could be or never twisted?
   Or is there some experience form customers shows that even with twisted curve still produce acceptable performance?

2. Assume that Q_PHASE_XTALK's twisted data is forced to be used as calibration data, how much impact would it do to the performance?
   If the impact is very little, then we will consider to use the twisted data.

Hi Moses,

It is possible that I_Phase is effected less by temperature than Q_Phase, so hysteresis may not show up on it as much as it does for Q_Phase. Can you let me know the rate at which you are heating/cooling in the chamber? Are you able to slow down or speed up the heating/cooling rate to see how it impacts the curve? Based on the latest plots that you sent, it seems like hysteresis may be happening. Is there any difference in the setup between the latest plots you sent and the previous plots (the ones that looked really bad)?

Thank you,

Brent Elliott

The heating/colling rate of when hysteresis effect was the fastest rate that our chamber can do.

The thermo-chamber is capable to do in following heating/colling rate:

-40℃~+150℃ within 70mins.
+20℃~-40℃ within 60mins.

Except the hysteresis effect experiment, we do all the measurement of xtalk in following procedure:

1. Set the desired temperature and wait until it reaches.

2. Wait another 3 mins after it reach the desired temperature.

3. Start measurement.

4. Raise the setting temperature for 5 degrees and back to step 1.

The results of hysteresis effect experiment is, indeed, different on Q_PHASE_XTALK when heating and cooling. But even in the exactly same procedure, same day, without moving any parts, the Q_PHASE_XTALK could be different as I've shown the following pictures before. So this is not convincing and we have no resource to do further experiment with hysteresis effect.

Even we know there is hysteresis effect and assume that we avoided the hysteresis effect, how to make the curve of Q_PHASE_XTALK not twisted?

I just need to know, is slightly twisted curve normal? What issue could be considered in hardware level if it is not normal? (cause we got no issue when calibrating the OPT3101EVM with modified laser diode using the same procedure)

Can the twisted curve be explained as the nature characteristic of OPT3101 or the photodiode?

It will be appreciated if my questions are answered without diverging the issue.

Hi Moses,

Let me get back to you on this tomorrow.

Thank you,

Brent Elliott

Hi, hope you doing well.

I think I forgot to share some information.

The reason I focus on the hardware instead of the procedure is because we use exactly the same procedure to calibrate since the beginning.
When it was OPT3101EVM with modified laser diode, the I/Q-Phase vs temp graph were as the following.

(Left No.1; Right: No.2 )

They were beautiful, straight line. 

But when it comes to our re-designed board, some of them were not straight:

Hi Moses,

I just sent you an email. Are you able to share the schematic and layout of the PCB so I can check if any design guidelines were overlooked?

Thank you,

Brent Elliott

Thanks for following up the issues.

I'm waiting for our hardware person to response, we will make further contact when he is good to share the details of hardware.

Meanwhile, I have done the experiment of hysteresis effect. It did produce different results.

The left pair of graph is done by increasing the temperature, and the right one is done by decreasing the temperature.
(Temperature settings of thermo-chamber: -10, 0, 10, 20, 30, 40C)

The amplitude when Tx is covered is shown by blue dots and line, and the scale of Y-axis is shown at left side.
The I/Q-Phase are shown by orange dots and line, and the scale of Y-axis is shown at right side (plus/minus reversed).
Amplitude, I-Phase, Q-Phase are measured with xtalk data and xtalk temperature compensation data written into register.

The amplitude should keep at low value, but it almost did not go below 20 codes which a value below 10 should be observed at room temperature.
I've keep the scale of Y-axis all the same, so we can compare the magnitude.
And we can see that I-Phase did not change too much, but Q-Phase was changing and the way it changes affected the amplitude so they changed exactly in same way. The rate of the change is way too much to ensure the precision of distance measurement.
Also, the results when increasing/decreasing of temperature are different.

So we have the following issues:

1. It could not keep the amp below 10 even after calibrated in same room temperature.

2. The fluctuation of q-phase is too much that causes the amp goes high with it which will reduce the precision of measurement a lot.

3. The hysteresis effect did happen? Or the result just changes every time...

Regarding the issues, what could be considered?

Hi Moses,

Thank you for the response. I won't be able to reply to this today but I will get back to you on Monday.

Thank you,

Brent Elliott