The output data of the ADC chip is incorrect no matter how I process it. Please help.

shijizai

The output data of the ADC chip is incorrect no matter how I process it. Please help. [Copy link]

 

The ADC chip used is TI's ADS7054 (the input is differential input and the output is single-ended output). The signal I want to measure is a differential signal, and the range of V+ - V- is +2.1~-2.1, but the value output after ADS conversion exceeds this range.

The following two figures are the differential signal of the signal source and the output voltage conversion of ADS7054 (screenshots taken from the manual)

The figure above is the measured differential signal source. The voltage is expressed in degrees per second. The maximum speed is 300 degrees per second, which means that the maximum value of V- is -0.05V and the maximum value of V+ is +2.05V.

///////////////////////////////////////////////////// ///////////////////////////////////////////////////// ///////////////////////////////////////////////////// //////////////////////////////////////////////

The above figure shows the output signal conversion relationship of the ADC chip. The reference voltage Vref is 3.3V. If the output of the chip is a hexadecimal number Sig_out,

Then when V+ - V- >0, the voltage value represented is V = Sig_out*2*3.3/2^14;

Then when V+ - V- <0, the voltage value V = (Sig_out-0x2000)*2*3.3/2^14;

Then, based on the voltage value and the input and output rules of the differential signal source (the first figure, OUT+, OUT-), we can get the angular velocity (degrees per second) represented by this voltage: W = V/2/0.0035;

According to this idea, the final angular velocity W should be within the standard angular velocity range of the signal source, that is, it should not exceed 300 degrees per second. However, in fact, Sig_out has the phenomenon of outputting the full scale 1FFF. At this time, the voltage is 3.3V, which exceeds the range of +2.1~-2.1, and the converted angular velocity is also greater than 300 degrees per second. There are also other cases where it exceeds 2.1V. This result does not correspond to the actual situation. What is wrong? Looking forward to the experts passing by!

Your answer to this question will be greatly appreciated!

Jacktang

Is it just a matter of calculating the angular velocity, and is it related to the formula?

You can verify whether other ADC conversions such as voltage also have this error

shijizai

Jacktang posted on 2021-3-18 14:29 Is it just a problem of angular velocity calculation? Is it related to the formula? Can you verify whether other ADC conversions such as voltage also have this error?

I didn't expect this, I'll give it a try

okhxyyo

Has the landlord's problem been solved?

shijizai

okhxyyo posted on 2021-3-19 10:24 Has the original poster solved the problem?

No

okhxyyo

Shijizai posted on 2021-3-19 15:24 No

You can ask @chunyang @littleshrimp @ w494143467 @ freebsder

shijizai

Jacktang posted on 2021-3-18 14:29 Is it just a problem of angular velocity calculation? Is it related to the formula? Can you verify whether other ADC conversions such as voltage also have this error?

I tried it, the input voltage of the ADC chip should be a differential signal, but when I actually tried it, V+ and V- were both connected to GND, and the chip output was 0; V+ was connected to 3.3V, V- was connected to GND, and the chip was full scale 1FFF; then I tried to connect V+ to the positive pole of the voltage source, and V- to the negative pole of the voltage source, and kept adjusting the voltage of the voltage source. I found that when the voltage was less than 1V, the ADC chip output was 0, and when it was greater than 1V, the ADC chip output was full scale 1FFF. I don't know if this situation is caused by the differential signal being terminated with a DC signal, but the manual says that the output of the ADC is V+ - V-, so if this is the case, it should have little to do with differential or not.

Compare it with the original measurement situation. When the differential signal source to be measured is connected to V+ and V-, the ADC signal will not only have two values 0 and 1FFF, but also other values. However, the value at this time will exceed the output range of the differential signal source after conversion, so it is also wrong.

The above two measurement situations have their own errors. I am a little confused. What went wrong?

shijizai

okhxyyo posted on 2021-3-19 15:30 You can ask @chunyang @littleshrimp @w494143467 @freebsder

Wow, you can do it this way. Just tag them in this post and they will be able to see it?

shijizai

@chunyang @littleshrimp @ w494143467 @ freebsder Guys, could you help me look into my problem? I'd be very grateful.

okhxyyo

Shijizai posted on 2021-3-19 15:35 Wow, you can do this, just tag them in this post, and they will see it?

Yes, they will receive an @ prompt.

okhxyyo

Shijizai posted on 2021-3-19 15:35 Wow, you can do this, just tag them in this post, and they will see it?

When posting on the forum, there is a function to invite netizens to answer. However, this function is not perfect yet. When it is perfected, it will recommend netizens who are enthusiastic about answering questions in the forum. The current recommendation is not very useful. You can directly @ the experts you know.

shijizai

okhxyyo posted on 2021-3-19 15:50 When posting on the forum, there is a function to invite netizens to answer. However, this function is not perfect yet. After it is perfected, it will be added in the invitation...

Um

littleshrimp

Is it because the sign bit is not understood? Normally, the output data needs to be shifted left by 2 bits and assigned to the ushort variable, and then divided by 4. Try connecting V- to the power supply V+ and ground to see if the result is 0x2000.

littleshrimp

littleshrimp posted on 2021-3-19 17:00 Is it because the sign bit is not understood? Normally, the output data needs to be shifted left by 2 bits and assigned to the ushort variable, and then divided by 4. Try connecting V- to the power supply V+ and grounding to see...

Wrong, short 16 bits signed

chunyang

I guess the original poster has misunderstood. For a differential input ADC, as long as the reference source is correctly selected, the input signal complies with the differential rule (it can be considered as positive and negative symmetry with GND, or no electrical connection with GND but the two lines have opposite phases and the same amplitude), and the range is within the ADC range, then the data output by the ADC should be a value between 0 and the maximum number. If the input is not a differential signal, such as the original poster using an ordinary signal source, the result is of no reference value.

If you want to use a common single-ended signal generator to test a differential input ADC, you can connect the generator output to the system ground and the positive input of the ADC, then use an op amp to build an inverter with a gain of 1, invert the output of the signal generator and connect it to the negative input of the ADC. At this time, the op amp needs to be powered by a positive and negative symmetrical power supply.

Jacktang

1. I suspect that the error in the actual test of the OP is large, because the measured voltage value is too small. When the general multimeter is tested, the value is small, but the error is the largest. Change to a high-precision measurement to verify.

2. There may be a problem with the circuit. Where does the reference voltage of the ADC come from? How is it given? Is it isolated? Is the AGND stable? These basic ADC conversion factors need to be eliminated by yourself first.

shijizai

littleshrimp posted on 2021-3-19 17:00 Did you not understand the sign bit? Normally, you need to shift the output data left by 2 bits and assign it to the ushort variable, and then divide it by 4. Try connecting V- to the power supply V+ and grounding it...

I have tried this. When V- is connected to the power supply and V+ is grounded, the output is 0x2000. When V- is grounded and V+ is connected to the power supply, the output is 1FFF. I also found that as long as the voltage exceeds 1V, the output is 1FFF.

shijizai

chunyang posted on 2021-3-19 17:54 I guess the original poster misunderstood. As long as the reference source is correctly selected for the differential input ADC, the input signal complies with the differential rule (it can be regarded as positive and negative symmetry with GND, or...

The signal output by my current differential signal source is biased to 1V instead of GND, so my ADC is only powered by a single power supply. Will this have a big impact on my results?

littleshrimp

[quote]Shijizai published on 2021-3-22 08:51 I have tried this. When V- is connected to the power supply and V+ is grounded, the result is 0x2000. When V- is grounded and V+ is connected to the power supply, the result is 1FFF. I also found that as long as the voltage exceeds 1V, the output will be 0x2000.

Is V- grounded, and does the output 1FFF when V+ reaches 1V? How is the 1V output generated? What is the voltage of the AVDD pin measured with a multimeter?

chunyang

The system reminds me that there is a reply, but I can't see it. Is it a cross-page bug? @ okhxyyo