MSP430FR6045: USS-related error messages and inquiries about abnormal values in UPS, DNS

Hi,

1. It might caused by an inappropriate threshold settings. Could you send some screenshots about your configuration page, advanced configuration page and the ADC capture page with captured data. 

2. Error code 103: something wrong with your configurations. 

Error code 135: The algorithm will generate this error code 135 when it detects the dTOF value is beyond the limit by the max sample shift value. You should set the max sample shift value a little higher than the maximum dTOF value on your pipe.

Best regards,

Cash Hao

Here are my ADC capture pictures and parameters.

Even when testing under the same conditions, there are times when the waveform is good and there are times when it is bad.

Could there be another external factor?

Hi,

Can you also get a screen shot of the ADC capture when it is bad?

You can try the below settings.

Numbers of pulses change to 12 or 16. 

Signal sampling frequency to 3600

Envelope crossing threshold to 20

Best regards,

Cash Hao

You can also follow this guide for setting a proper configurations. 

https://dev.ti.com/tirex/explore/node?node=AP3BtYCoe9uXZzBARBtGhA__IOGqZri__LATEST

This is the result of setting the environment value and testing it as you said.

Also, this is the result of sometimes bad ADC captures when I tested it before.

thank you

Ok, try with the new settings first. 

About the bad ADC captures, it might caused by the low ADC sampling frequency. It shouldn't affect the absTOF measurement. 

Best regards,

Cash Hao

I think we've figured out the cause. After a certain period of time, the size of the ADC Capture changed (the value also changes when Zero Flow is maintained or the flow rate is changed). In other words, cycle slip seems to have occurred. But I don't know why ADC Capture has changed.

In addition, the following message occurred when water flowed during the test.

However, the water inside the flow meter is supposed to flow from bottom to top. Can that happen?

Thank you.

Also, in Zero Flow, the absolute ToF value changes with time. I wonder if this is also happening because the size of ADC Capture is changed.

Hi,

Jung Jayden said:

the size of the ADC Capture changed

The size of the ADC capture shouldn't change. Can you use the "continuous capture" function to record the ADC capture for a long time until it records this issue?

Jung Jayden said:

In addition, the following message occurred when water flowed during the test.

These messages should happen when there is NO water in pipe, right?

Jung Jayden said:

Also, in Zero Flow, the absolute ToF value changes with time. I wonder if this is also happening because the size of ADC Capture is changed.

It might cause by the environment is changing slowly. Such as temperature. Or the transducer need more time to be stable. 

Best regards,

Cash Hao

Continuous Capture was done for about an hour and a half.

The first and the last were graphed.

But I don't know how to upload the CSV file.
Can you send me a mail?

thank you

Sure. cash-hao@ti.com

Yesterday I sent the ADC Continuous Capture File to the email you uploaded.

If you haven't checked, please check.

thank you

Hi,

I received the data and will need some time to analyze it. Hopefully, I will get something by the end of this week. 

Best regards,

Cash Hao

OK! Thank you for your support!

And I have something to ask you. I wonder if the length of the line from the transducer to the MCU in the pipe is affected by other values such as ADC. If there is an effect, I would like to know what the minimumlength is.

Thank you!!!!

Hi,

We do not have an exact data of the length from the transducer to the MCU. However, it is preferred to keep it as shorter as you can. 

Best regards,

Cash Hao

I was experimenting with  strap on transducers and ran into the exact same situation as you are dealing with. Random errors and data jumping around exactly like you are dealing with. I spent hours experimenting but without access to the source code I finally gave up. I hope you are able to figure this out.

Hi,

I took three signals in the ADC capture data, and find the peak value of each lobe using three-point parabolic interpolation. The peak value of each lobe showed below. 

Obviously, the maximum peak of the signal is changed during time. We do not know what caused it yet, but let's continue first. 

Use the peak value of the second lobe divide the maximum peak value for all three signals. 

data1, 152.639/1034.34 = 0.148; data2, 107.323/691.823 = 0.155; data3, 96.1524/646.648 = 0.149.

These data do not change much even the maximum peak value changes hugely. Based on this, you can set the envelope crossing threshold near 0.15 to keep tracking the second lobe at these circumstances. That means the change on the maximum peak value will not affect the absTOF result. The cycle slip issue is not caused by the change on the signal. 

Back to what may cause the change on the signal. I also plot the max ADC value of each ADC capture and showed below. 

It is continuous declining. Not a sudden drop. I would say it may mostly caused by the environment change. Maybe the Installation method of the pipe/transducer. Or some other thing. You can change your pipe/transducer to test it again. 

Best regards,

Cash Hao

This may help, I am using homemade transducers that mount on the outside of 2" Schedule 40 pipe. I have good results when water temp is 20C. When the water warms up to around 35C from the being continuously pumped my signals degrades and I get the same results that you are experiencing. The signal attenuates from the PVC plastic being softer and absorbing more of the signal. I can increase the gain but the errors 135 and 103 become numerous and I also get the TOF large jumps happening. I have fiddled with every setting I can find with no solution. What we really need is a very good explanation of how the software works and access to the source code.

Hi Crisler,

Well, the clamp-on meter is a different application from the residential meter.  We have little knowledge on the clamp-on meter. We have one application note that related to this clamp-on meter application, but it is not including the temperature test. https://www.ti.com/lit/an/slaa949/slaa949.pdf

About how the software works, we only have the explanation such as this one for now in chapter 2 https://www.ti.com/lit/ug/tidudn0/tidudn0.pdf

Best regards,

Cash Hao

My point is that when the signal becomes attenuated by the damping of the softer plastic when warm, the signal seems to degrade and the errors codes(103 & 135) become more frequent. This leads me to believe the problem has something to do with poor signal quality. Or some noise distortion that the software isn't designed to filter out. I understand that there isn't much information regarding the clamp on transducers. It might be possible to modify the software but without access and a good understanding we are all simply tweaking parameters and hoping it works. And when it doesn't, as in this case, the road dead ends.

What would really be helpful is an explanation/tutorial with real example detailing capturing the data and then running the FFT algorithms on that data to get a TOF value. This is a complicated process and needs extensive information not a vague overview.  

Was the software written by a third party? It seems like a black box which works until it doesn't and at that point variables(parameters)  are adjusted and everyone hopes for the best.

There is a way to enable a DSP filter if you think it's required for your product. It uses more memory but it may be useful in your application. Also, if your amplitude is changing over time, you may need to enable or make your own gain adjustment algorithm so the amplitude remains fairly constant.

The TI library is not running any FFTs to get TOF. Delta TOF is calculated by simply doing a cosine interpolation between a few multiply-accumulates to estimate where the max cross-correlation term would be, hence the inclusion of the LEA. Their methodology is explained in depth with references to white papers, some freely available, in Chapter 2 of TIDUES5A, the "Optimized Ultrasonic Sensing Metrology Reference Design for Water Flow Measurement". Their measurement methodology can be easily validated using Excel after reading the white papers the implementation is based on.

There was a link circulating on this forum which was a request link for the USS Library source code, but no one ever approved the requests of the few people who found it. For ultimate freedom, you'll need to write your own library. All of the registers controlling the sequencer and SDHS are documented in the device family User Guide. For the LEA functions, use the DSP library. You'd have to implement the cosine interpolation yourself but it's not impossible.

Hi Hao

A question arose during testing while changing the sample.
I wonder if the ADC or the signal itself is affected by the area of the reflector that reflects the ultrasonic wave.

Thank you

Hi,

Jung Jayden said:

I wonder if the ADC or the signal itself is affected by the area of the reflector that reflects the ultrasonic wave.

It could affect the signal. What kind of pipe you are using now? Is it a U-type pipe?

Best regards,

Cash Hao

Hi Crisler,

This library is written by TI. Sorry, for the bad experience of using this library. We are trying to fulfil more training materials about how to use this library. 

Best regards,

Cash Hao

Hi, hao

A flexible hose like the picture below is connected to the pipe and used. Will using a U-type pipe or another pipe have any impact?

And it was tested in environment 1 in the figure below, and it is being tested by changing to environment 2. Could there be another problem?

And in response to the previous question, do you recommend a large reflector?

thank you

Hi,

I think the flexible hoses should be fine if you keep a proper distance between the pipe and the bending. However, I never used this kind of flexible hoses to connect the pipe. So I am not sure if it will cause other issues.

By what I mean about the U-type pipe, it is showed below. The reflector size of this kind of pipe is not the larger the better. You may need to test about it. 

Best regards,

Cash Hao

If the U-Type you're talking about is a reflective type, you're right.
Can I understand that a wide reflector is recommended?

Also, I'm still testing, changing the environment and pipes, but the same thing happens. Are there any other additional solutions? please...

thank you

Hi, How

I have a question. When cycle slip occurs, is it okay to replace the value with the previous value except for that value?

Thank you

Hi，

Could you keep record the ADC continuous capture data including the data is normal and the cycle slips happens?

Jung Jayden said:

When cycle slip occurs, is it okay to replace the value with the previous value except for that value?

It is okay to do that. But more commonly, you can directly drop the cycle slips data and wait for another valid data. 

Best regards,

Cash Hao

Hi,

Thank you for your reply.

I sended the file to the your e-mail!

When I checked the document(Optimized Ultrasonic Sensing Metrology Water Flow Measurement Reference Design (Rev. A)) additionally, I saw that there was Dead Time. Only the Dead Time values associated with DN-25 pipes are shown in the document. Is there a dead time for other pipe specifications?

Replied over the email.

Are the DSP filters you mention part of  * LIBRARY ALGORITHM BAND-PASS AND HILBERT COEFFICIENT related to Hilbert filter coefficients contained in USS_userConfig.h. How do those they work??

You can generated a band pass filter coefficients and use it in the library. It can filter the ADC data before it runs the algorithm.  

Okay, now to make them useful a few more question. When do we need them? What are we looking to filter and why?  How do you choose which one? How do we generate the coefficients?  Would a filter help out with this situation dealing with these error code we are experiencing in this post? Is there a write up on this?

Excellent reply, thank you. Very helpful.

Hi, 

I have an additional question.
There is a strange thing during calibration in USS Tool.

When you check the VFR mean value in the Waveform tab, the VFR value comes out well.
However, when the total function of volume flow rate is used, the value does not match.
I want to know if the total value is the calibrated value.

thank you

Hi,

I barely use this function. But I think since you have enable the advance calibration table, the total function of the volume flow rate is using the calibrated value. 

Best regards,

Cash Hao

Thank you for answer.

If so, how did you check when you measured the totalization value?

I think you have checked the value displayed on the LCD of the EVM Board. Is that right?

thank you

HI,

Actually, I barely check the totalization value. I would more focus on the dTOF and absTOF value. If those are correct than the volume flow rate and the total volume should not be an issue. 

Best regards,

Cash Hao

When the valve of the water supply is opened, the value rises or falls, although not as much as Cycle Slip.

Is there any way to solve it?

Hi,

Do you mean the dTOF goes up tp 10ns after the valve is open when the water flow rate should be zero?

Best regards

Cash Hao

Hi,

In terms of dToF, you are right.

At the moment of opening the valve at about 10L/h based on VFR, it sometimes rises and falls in hundred units.

Could it be mitigated by adjusting some parameters?

Thank you

HI,

Try with a lower setting of the USS_ALG_CORR_VAL_THRSH_CHK_FACT in the USS_userConfig.h might help in this condition. 

Best regards,

Cash Hao

Hello

I have an external crystal and DCO related question.

The link below explains that using an external crystal increases the current consumption but increases the accuracy.

How much accuracy does the use of an external crystal increase over DCO?
If you used DCO, can you give me the exact specs?

Link : https://e2e.ti.com/support/microcontrollers/msp-low-power-microcontrollers-group/msp430/f/msp-low-power-microcontroller-forum/763906/evm430-fr6047-using-built-in-clock-for-cpu/2824357?tisearch=e2e-sitesearch&keymatch=evm430%252520AND%252520fr6047%252520AND%252520dco#2824357

And I have a question about UPS DNS Gap in USS Tool.
Even if the GAP is set to 3000us, when the actual measurement is performed, about 3700us comes out. Is this correct?

thank you

Hi,

I do not think using the HFXT as the clock source for the MCLK instead of using the internal DCO as the source will help on the measurement accuracy. All the USS modules are clocking by the HSPLL which is sourcing by the USSXT. It is not relative to the HFXT.  

It is correct on the UPS DNS gap thing. The actually gap between the UPS and DNS should add the time of the initialization time for the USS module. 

Best regards,

Cash Hao

Hello 

thanks for the reply

1. When should "Abs ToF Additional Delay" be calculated? 
And how should I calculate "Abs ToF Additional Delay"?

2. When should the "Delta ToF Offset" be calculated? 
And how should The "Delta ToF Offset" be calculated? 

The "Calibration User's Guide" doesn't mention "Delta ToF Offset"
For the location of "Calibration User's Guide", refer to the following link.

<Link>

https://e2e.ti.com/support/microcontrollers/msp-low-power-microcontrollers-group/msp430/f/msp-low-power-microcontroller-forum/873589/evm430-fr6047-calibration-user-s-guide

HI,

1. In some situations, you might need the absTOF time of the first lobe of the signal. In our algorithm, the absTOF usually not locks at the first lobe. So you need to minus the time between the first lobe and the lock lobe using this parameter. 

2. If the dTOF of your meter is not zero at zero flow, and you have checked all the other things and nothing more can be done to solve this issue. Then you can use this parameter to make the dTOF to zero at zero flow on your meter. 

Best regards,

Cash Hao

hello

thank you for reply

but I don't quite understand about number 1. 

when is that situation?

Why does the absTOF have to be locked in the first lobe?

When they need to calculate the sound speed using the absTOF and the acoustic length。