PGA300: Calculable criteria of temperature compensation coefficients

I couldn’t attached examples files, so I upload these files again.

Hello,

Were there any other setup differences between the successful calibrations and the unsuccessful calibrations? Can you try a higher value for the PGAIN? For the successful calibration, can you please share the results from the EEPROM Config Digital page?

Regards,

1)There are no setup differences in success and fail. When I set success and fail products, there internal conditions were same, and also there were in the same temperature chamber simultaneously.

2)Yes, I can try to a higher value for the PGAIN. However, I’m concerning about not obtaining temperature compensation coefficients. Because I have an experience that temperature compensation coefficients are not obtained, when I set higher PGAIN value (= 16) than current value (= 13.33). I don’t know why that is happen. But PGAIN is set as current value (= 13.33), the probability of obtaining these compensation coefficients is higher than PGAIN value is 16.First image is the setting data that I set PGAIN value 16.

3)Second images are the reference of test data including EEPROM Config Digital page. Left side and right side images are successful calibration and unsuccessful calibration, respectively.

Sincely,

Hello,

Thank you for the detailed information. Do you notice any change in the frequency of failed calibrations if you disable the filter?

Regards,

Hello,

No I don't notice anthings. If you need more information about this problem, let me know.

Sincely,

Hello,

I will do some testing with manually entering data on my end to try to replicate the issue and better understand it. In the meantime, can you also do a test, and try entering the exact same measured voltage in the DAC calibration section for all 3 temperatures on the failed calibration and see if that affects the GUI's ability to generate coefficients? You can manually enter the same ADC values from your previous calibration to keep everything else the same.

Regards,

Hello,

In the same ADC input values, the results of calculation for compensation was not available.

When I changed ADC values, the results were turned false to success as follows.

Regards,

Hello,

What are your DAC calibration settings and measurements for these results? Is it the same for both of the examples?

Regards,

Hello,

The DAC calibration setting and result are as follows.

The measurement sensor ranges have 3 settings that are 1K, 3K, and 9K.

We secure the pressure characteristic to change P_GAIN.

Except for changing P_GAIN, oter settings are same.

The data we posted Nov 6, 2019 2:40 AM is 9K range (P_GAIN:13.33).

The data we posted Dec 13, 2019 4:15 AM is 1K range (P_GAIN:40).

Regards,

Hello,

I've been able to replicate your findings in the GUI on my end as well. I will need to discuss with some colleagues to see why this might be occurring. There is generally measure of the quality of the fit for the calibration once it is calculated, and it will show an error if the fit is not good enough, but the differences between the two examples is very small, so it's a little confusing as to why it will not go through in this case.

Regards,

Hello,

I would like to know which parameter causes this error, and also want to understand this error's threshold.

Regards,

Hello,

I have to discuss this with the GUI designer to understand what is happening with the error calculation and why such small changes to the inputs will result in an improper output.

Regards,

Hello,

My apologies for the delay. I am working on this, and I hope to have an update for you by the end of the week.

Regards,

Hello,

I am currently in contact with the GUI designer and waiting for a response on the implementation of the fit error calculation and its effect on the calibration coefficient outputs. I will get back to you as soon as I have a response.

Regards,

Hello,

I am still working with the GUI designer to get a response for you. I hope to have another update for you by Wednesday, the 5th.

Regards,

Hello,

Can you please share the GUI version that you are using? You can see it in the Help->About menu.

I still have not gotten word from our GUI expert, but I will let you know as soon as I do.

In the meantime, I did some additional testing, and it appears that the coefficient calculation fails most often when the TADC data for all 3 temperatures are close together. Even if 2 of the temperatures show very similar data but the third is very different it will most likely fail the calculation.

Are you using an external temperature sensor, or the internal temperature sensor? If you are using an external temperature sensor, I suggest as a workaround to increase the TGAIN so that the different temperatures produce larger differences in the TADC codes. If you are using the internal temperature sensor it is usually recommended to use a TGAIN of 5, so one option would be to use a larger temperature range. You could adjust the gain higher, but it is a large jump between the gain of 5 and the gain of 20, so it may run into issues of clipping the ADC input.

Regards,

Hello,

GUI  version that I use is 3.5.1.0.

I hope that the GUI expert will give you information, and you'll get back to me with certain answer as soon as pollible.  

I used external temperature sensor. Because of specification of external temperature sensor, TGAIN of 1.33 is adopted for our product. If GAIN value set higher than now, it's input range becomes over specification.

Regards,

Hello,

Can you adjust the current output to the temperature sensor and the gain to allow for a larger difference between the temperature codes from the TADC? It does seem for the moment that the problem is caused generally when the TADC codes for different temperatures are too close together, based on the testing that I have done.

For the moment just for testing purposes it will be fine if the input is clipped on one end or the other. I would just like to see if the calibration completes for a wider range of temperature codes.

Regards,

Hello,

I will take into consideration about changing gain.

If I set TGAIN value higher than current value, I think the permitted value to ste is only 5. Isn't it true?

I think the judgement value must be existed to distinguish success or false in the calculation. Please expose the judgement and let me know. 

Regards,  

Can you clarify what the STE is that you're referring to? The gain can be set to any value, since the TGAIN output will never overdrive the input to the TADC. If the gain is too high you will just end up clipping the signal, but it cannot damage anything.

Regards,

Hello,

Could you tell me the STE? I don't know it. I recognize that the TGAIN value is set from pull-down menu only. Could the TGAIN value be set any value by another method? If it's possible, I would like to know the method.

When will you tell me the judgement value that is used for distinguishing success and fail?

Regards,

Hello,

I don't understand what STE means. I'm not sure what you're referring to. Can you please clarify what you're asking about?

The TGAIN only has a few values that you can select. Those are the values from the dropdown. All I meant, is that you can use any of those values safely, and you won't damage anything.

The GUI was developed by an outside service, and I am still waiting to get some confirmation about the coefficient calculation error at this time.

Regards,

Hello,

I received your reply at Feb 21, 2020 4:42PM that's actually includes STE, so I don't understand what it mean. Threfore, I asked you a question about STE.

I understand TGAIN is only set from the dropdown. The acuired data, that is got with TGAIN value 5, would be saturated.

Please follow the developer and request when will you return the answer.

Regards,

Hello,

I asked because of your message on February 18th:

Hello,

I'm sorry for misswriting. I actually wanted to write "set". Incidentally, how is the progress status of the judgement value?

Regards,  

Hello,

Ok, that makes sense. I'm sorry for the misunderstanding. I am still waiting on a response from the team that worked on the GUI.

Regards,