ADS1263: System Noise Analysis

Thank you for the prompt reply to my question. I will checkout the calculator tool and the noise document. I will leave this thread open as I may come back later with more questions regarding the system noise.

- Steven

Hi Steven Park,

Understood, this thread will stay open for ~30 days. If it closes though you can always start a new one

-Bryan

Hello Bryan,

(could be related to https://e2e.ti.com/support/data-converters-group/data-converters/f/data-converters-forum/1278052/ads1263-ads1263-3-wire-rtd-low-side-reference-noise-source)

I ended up ordering an EVM to compare the measurements between our product and EVM measurement. I'm seeing about 500uV of RMS noise at 38400SPS and increases as the voltage increases. The RMS noise increase as the input voltage increase. I've used two AA Batteries to supply the input voltage. 

- ADC setting

     - continuous measurements and read by command

     - Input mux: AIN0 and AINCOM, (AINCOM shorted to REFOUT)

     - Internal Reference is enabled

     - 38400 SPS

- I've calculated the external amplifier noise spec

- I've measured the noise level in AVDD and DVDD power supply line

- The above measurements are made using the EVM.

Below is the schematic of our ADC, the input goes through a buffer amplifier and resistor network to step down the voltage. This is done to provide wider voltage range.

To compare the voltage between ADS1263 on our board and on EVM, I jumped wires from ADC1 and REFOUT net and fed them to EVM input. I also disconnected on board power supplies on EVM and used 5Viso on our board to power the EVM ADC. I was getting expected noise data 100uV RMS noise on EVM while measuring 500uV RMS noise on our board.

At this point, I suspect the noise could be related to how I'm measuring the voltage using my python script or how I'm analyzing the data.

The python script sends START command, and then starts collecting data by sending READ command. It collects 8192 samples and perform standard deviation calculation on those samples.

I'd like to know if you have any suggestions on how to tackle this issue or your thoughts on where could that noise factor coming from.

Thanks,

Steven

Hi Steven Park,

How are the batteries connected to the ADC? Is the battery voltage floating, or is it referenced to some voltage e.g. REFOUT?

And to confirm, when you apply the battery voltage to the EVM, you are seeing 500uV RMS of noise at 38400SPS? What gain are you using?

Why are you using a very high offset, high noise amplifier (LM2902) as buffer for a 32-bit ADC? This basically defeats the purpose of using such a high-resolution ADC

Can you send me a picture of how you have the two boards jumpered together? And please confirm that under these circumstances you are able to see 100uV RMS noise

If you think there is some issue with the data capture, you can always pull the data off of the EVM and calculate the RMS value directly using the raw data. This will help you determine if you are calculating the noise the same way as the EVM. Also, can you confirm which version of the EVM you are using? We recently released a new version in the last six months so I want to be sure I know which one you have. Is it this one: 

-Bryan

Hello Bryan,

I was not involved in the process of picking LM2902, so I don't have the answer to why this specific amplifier was chosen.

Here's some pictures of the test setup

1. Jumpers circled orange are removed and AVDD and DVDD are shorted together

2. R5 and R21 are removed and shorted

3. C4 and C21 are removed

4. On board Isolated 5V line is jumped to AVDD OUT on EVM

5. On board GND is jumped to GND on EVM

6. On board AIN0 (Note: Net name on schematic is ADC1) is jumped to AIN of EVM

7. On board AINCOM (Note: AINCOM is shorted to REFOUT on our board refer to the schematic in previous post) is jumped to AINCOM of EVM

8. Battery Ground is connected to Terminal block of our device which connects to the on-board GND mentioned in step 5

9. Positive of the battery connects to the Terminal block of our device which connects to the non-inverting input of the Buffer amplifier through a resistor. Net name A/DIN1 in schematic.

Two separate measurements at different voltage level using EVM

Corresponding measurements mad on our Device

The measurements were not synced, I measured on EVM then measured on our device.

Thanks,

Steven

Hi Steven Park,

In the GUI images you sent, the ADC is running at 2.5 SPS, but in your previous threads you said the ADC was running at 38400 SPS - can you confirm? The noise looks pretty good for 38400 SPS, but not great if you are running at 2.5 SPS. Also, it would take a very long to generate 8192 samples at 2.5 SPS

Also can you post the data you are taking from your board, instead of the screenshots from Excel?

-Bryan

Hello,

I confirm it is 38400, the interface config doesn't update when I change the MODE2 register.Also the capture duration was almost instant.

data, raw_datas 
3351375689,0.00026143500008402043  
3350560009,0.00013003200001548976  
3349261664,0.00012495900000430993  
3348791806,0.00012192099995900207  
3350779772,0.00012147699999331962  
3350180482,0.00012164299994310568  
3349977573,0.00011797699994531285  
3349511086,0.00020326999992903438  
3350583603,0.000148623999962183  
3349064147,0.00011955100001159735  
3349676919,0.00011586599998736347  
3350103633,0.00011573700010103494  
3350643599,0.00011116300004232471  
3350251938,0.00011177399994721782  
3350878192,0.00011001400002896844  
3349887916,0.00010903299994424742  
3348376550,0.00010971800008974242  
3350605849,0.00010784799997054506  
3350591693,0.00010921800003416138  
3349610182,0.00010827399989921105  
3349978248,0.00010747699991497939  
3349502996,0.00010786600000756152  
3349627709,0.0001101440000184084  
3349771969,0.00011179199998423428  
3350387435,0.0001115889999709907  
3350394177,0.00011227400000279886  
3349538726,0.00011306900000818132  
3350321373,0.00011395899991839542  
3350610569,0.00010929200004738959  
3351102672,0.00011262500004249887  
3349925667,0.00011177300007148006  
3350229692,0.00011068100002376013  
3350449454,0.0001090150000209178  
3350851900,0.0001101440000184084  
3349195600,0.0001091259999839167  
3350534393,0.00010982899993905448  
3351155254,0.0001094039999998131  
3350069253,0.00010895899993101921  
3349531983,0.00010982900005274132  
3351037282,0.00010964399996282737  
3350135317,0.00010860700001558143  
3348780346,0.00015988300003755285  
3349983641,0.00012529100001756888  
3350482486,0.00010836700005256716  
3350405637,0.00010684800008675666  
3350142732,0.0001073290000022098  
3349895332,0.00010857000006581075  
3348822140,0.00011055100003432017  
3350678654,0.0001101809999681791  
3351311648,0.00011023699994439085  
3350447431,0.0001097919999892838  
3349543443,0.000110291999931178  
3350040941,0.00011275500003193883  
3350908527,0.00011503300004278572  
3349699164,0.00011423599994486722  
3350415075,0.0001147169999740072  
3351089190,0.00011416200004532584  
3350192617,0.00011477299995021895  
3348794501,0.00011330999996062019  
3350130597,0.00011155099991810857  
3350271489,0.0001118659999974625  
3348998085,0.00011279199998170952  
3350893697,0.0001114219999180932  
3350421816,0.0001113099999656697  
3349008196,0.00011032899999463552  
3350197336,0.00011066199999731907  
3349846796,0.00011071800008721766  
3349839381,0.00010973700000249664  
3349904095,0.00010921900002358598  
3349726129,0.00011249599992879666  
3350566077,0.0001129949999949531  
3350009932,0.00010942100004740496  
3350684046,0.00010942099993371812  
3349507716,0.0001088479999680203  
3350353056,0.00010943999996015918  
3349680963,0.00010897799995746027  
3349641190,0.0001681049999433526  
3350141383,0.0001232910000226184  
3350315306,0.00011234800001602707  
3351346028,0.00011194000001069071  
3350590344,0.00011127300001589902  
3349525242,0.00011005099997873913  
3350878192,0.00010986699999193661  
3350554617,0.00010858899997856497  
3348802592,0.00010764399996787688  
3351346703,0.00010829200004991435  
3349103247,0.00010931100007383066  
3349463899,0.00010764399996787688  
3350167675,0.00010862499993891106  
3348755403,0.00010823599996001576  
3350224974,0.0001083480000261261  
3349480078,0.00010805099998378864  
3349430193,0.00010738499997842155  
3349243463,0.00010784799997054506  
3349952631,0.00010788499992031575  
3349991730,0.00010756999995464867  
3348515417,0.00010790299995733221  
3349805675,0.00010775499993087578  
3349230653,0.00010755100004189444  
3351606910,0.00010779199999433331  

data, raw_datas 
2748688363,0.00016402400001425121  
2748770606,5.394600003683081e-05  
2750499711,6.250199999158212e-05  
2748844084,4.707599998710066e-05  
2749824922,4.5816000010745483e-05  
2749306528,4.5927000030587806e-05  
2749558648,4.5963999980358494e-05  
2749849190,4.5483000008061936e-05  
2749992103,4.4650000006640767e-05  
2748610166,4.416800004491961e-05  
2750136364,4.437199999074437e-05  
2749282259,4.470499999342792e-05  
2749099575,4.446399998414563e-05  
2748447704,4.529800003183482e-05  
2750449154,4.398300001184907e-05  
2749173053,4.426100002774547e-05  
2749520896,4.6408999992308964e-05  
2748434222,0.00010481900000058886  
2750116140,6.15760000073351e-05  
2749191254,4.766800003608296e-05  
2750418818,4.635399994867839e-05  
2748506352,4.5130999978937325e-05  
2749450114,4.5371999988219613e-05  
2748559609,4.496500002915127e-05  
2749883570,4.390899999862086e-05  
2748842062,4.4001999981446716e-05  
2749675269,4.3946000005234964e-05  
2749350345,4.381599995895158e-05  
2749748073,4.450099999075974e-05  
2748387035,4.437199999074437e-05  
2749965813,4.435300002114673e-05  
2749264732,4.442699997753152e-05  
2749900424,4.3500999993284495e-05  
2748412650,4.309399997737273e-05  
2750105353,4.346399998667039e-05  
2749158223,4.32229999773881e-05  
2749530334,4.3593999976110354e-05  
2749826944,4.377900000918089e-05  
2749098226,4.337199999326913e-05  
2750015023,4.35379999998986e-05  
2749411690,4.3317000006481976e-05  
2749464270,4.357599999593731e-05  
2748187496,4.316799999060095e-05  
2749773689,4.344600000649734e-05  
2748769932,4.329799998004091e-05  
2750459265,4.318699996019859e-05  
2748577135,4.348300001311145e-05  
2750147150,4.32229999773881e-05  
2748984975,4.327900001044327e-05  
2749529659,4.376099997216443e-05  
2748539384,4.3613000002551416e-05  
2750423536,4.413099998146208e-05  
2748765213,4.403899998806082e-05  
2750047380,4.4354000010571326e-05  
2748966774,4.583499998034313e-05  
2749531008,8.46140000021478e-05  
2749756162,5.1705000032598036e-05  
2749604486,4.544600000144783e-05  
2749752792,4.39279999682185e-05  
2748567024,4.3650000009165524e-05  
2750029853,4.3650000009165524e-05  
2749032837,4.361200001312682e-05  
2749152829,4.4149000018478546e-05  
2748699825,4.535400000804657e-05  
2750520609,0.00012085600002365027  
2748509049,8.342800003902084e-05  
2749626057,4.4891000015923055e-05  
2748291310,4.389000002902321e-05  
2749855257,4.4149000018478546e-05  
2748799593,6.17249999663727e-05  
2750737001,4.487200004632541e-05  
2749084071,4.376099997216443e-05  
2749405623,4.387199999200675e-05  
2748333106,4.3261000030270225e-05  
2749710997,4.344600000649734e-05  
2748484106,4.339100001971019e-05  
2749791216,4.3501999982709094e-05  
2748200978,4.31500000104279e-05  
2749934803,4.344600000649734e-05  
2748753079,4.327900001044327e-05  
2750148498,4.348299995626803e-05  
2748541407,4.3168999980025546e-05  
2750158610,4.309399997737273e-05  
2748666792,4.348299995626803e-05  
2749224960,4.357599999593731e-05  
2748345914,4.366799998933857e-05  
2750423536,4.329800003688433e-05  
2748795548,4.3612999945708e-05  
2749372591,4.333499998665502e-05  
2749460899,4.324200000382916e-05  
2749599767,4.3204999997215054e-05  
2750391179,4.352000001972556e-05  
2748687690,4.337199999326913e-05  
2749601116,4.357599999593731e-05  
2749140021,4.383500004223606e-05  
2749942218,4.34280000263243e-05  
2748255582,4.333500004349844e-05  
2750174114,4.32239999668127e-05  
2749215521,4.316799999060095e-05  
2750334552,4.344600000649734e-05  

Here are the raw data files. The first column represents the data, and the second column shows the time it took to read the data. As you suggested, I exported data from EVM software and run the same standard deviation function on them. It matches the number EVM presented, so I'd assume, the calculation done on the samples collected using our board is correct.

Thanks,

Steven

Hi Steven Park,

To be clear, this is only showing 101 data points, not 8192. Is that what you intended to show? You can attach an Excel file using the "Image/video/file" selection in the Insert dropdown menu. I wanted to look at your data in the time domain to see if there were any obvious signs of error, but that is difficult with only 100 points.

Let me see if I can summarize what you have tried so far. Please confirm / deny if anything is incorrect:

• When you measure the batteries with your board, you see ~500uVRMS noise
  • What is the voltage you measure in this case? Four AA batteries in series e.g. ~6V? And this is stepped down by ~80% through your resistor divider to 1.25V?
• When you measure the batteries with our EVM, you also see ~500uVRMS noise
• When you apply the output of your amplifier circuit to the analog input on our EVM and the output of your power supply to the 5V supply on our EVM, you now read 100uVRMS noise. In this case, you are using our software to measure data
  • I assume the same ~6V input in all cases?
• You've confirmed that the data manipulation function is correct such that when you calculate the RMS noise on the same dataset that the GUI uses, you are getting the same result

If these things are true, then this tentatively rules out the amplifier + resistor divider, the software, and the power supply. The remaining circuitry is really just the ADC. Have you reviewed your layout to make sure it is clean?

Have you also tried using our software with your board to see if you get the same results for both? You can remove the termination resistors on the EVM and use J2 to connect into your board. Note that there is no level translator on the IO voltage coming from our controller card, so you should use a 3.3V interface

-Bryan

Hello Bryan,

I didn't realize the file I attached only had 100 lines. Here's the link to the google sheet

• When you measure the batteries with your board, you see ~500uVRMS noise
  • What is the voltage you measure in this case? Four AA batteries in series e.g. ~6V? And this is stepped down by ~80% through your resistor divider to 1.25V?
  • Two different voltage measured, two AA and four AA Batteries. For two AA batteries, around ~3.2V stepped down to ~0.6V and RMS noise around ~500uV.
  • Four AA Batteries, ~6.4V down to 1.32V and RMS noise around ~0.700uV

• When you measure the batteries with our EVM, you also see ~500uVRMS noise
  • I had two setup
    • 1. EVM standalone out-of-box, measuring two AA batteries: ~100uV RMS Noise according to the data sheet
    • 2. Jumped wire as explained in the previous posts, Two AA batteries and four AA batteries
      • ADC1 net and REFOUT net of our product jumped to AIN0 and REFOUT to EVM.
      • 5Viso and analog ground of our product connects to the EVM. EVM power supply jumpers disconnected
      • Battery positive connects to the non-inverting input of buffer amplifier
      • Battery negative connects to the Analog ground
      • EVM software is used to collect samples
      • This also showed ~100uV noise

• When you apply the output of your amplifier circuit to the analog input on our EVM and the output of your power supply to the 5V supply on our EVM, you now read 100uVRMS noise. In this case, you are using our software to measure data
  • I assume the same ~6V input in all cases?
  • Yes, this is correct and EVM always read ~100uV RMS noise regardless of settings and voltage

• You've confirmed that the data manipulation function is correct such that when you calculate the RMS noise on the same dataset that the GUI uses, you are getting the same result
  • Yes, this is also correct. I've exported the Captured data and run the same excel function to calculate the standard deviation.

I do believe our hardware engineer followed the Layout guideline, but if you could send or provide supported link/document I will delegate it to our hardware engineer.

I was working on the hardware modification to connect our hardware to the EVM software. I should be able to test that tomorrow. By termination resistor, do you mean the resistors connected in series between the ADS1263 and J2?

Thanks,

Steven

Hi Steven Park

Thanks for confirming. You said

Steven Park said:

Four AA Batteries, ~6.4V down to 1.32V and RMS noise around ~0.700uV

Should that have been 700uV i.e. the noise went up compared to the 0.6V input case?

Steven Park said:

By termination resistor, do you mean the resistors connected in series between the ADS1263 and J2?

Correct. At the very least you want to disconnect DOUT and DRDY from the ADC on the EVM, otherwise you will get conflicting data on the bus when you wire in your ADC / board

-Bryan

Also, anything Google docs related gets blocked by our firewall, can you just post the file to the e2e post?

Good morning Brian,

Yes, the RMS noise increased as the the input voltage increased

ADC Raw Data Using Python.xlsx

Here's the raw data files imported in Excel.

I'll keep you posted on the test.

Steven

I just finished jumping SPI lines from our board to EVM. I removed termination resistors on CS,DRDY and DOUT lines. DRDY on our board only has 5V net, but I was still able to read data from our ADS1263 using EVM software.

So basically, I'm using EVM software to read our ADS1263 by jumping wires from J2.

20231024-ADC-Samples.xlsx

The RMS noises are not exactly the same as when I used my python script to read, but exhibits the similar behaviour. ~400uV RMS noise at two AA batteries at the input. ~600uV RMS noise at four AA batteries at the input.

I guess, this implies that the noise could be coming from the ADC circuit itself.

Steven

Hi Steven Park,

Steven Park said:

I guess, this implies that the noise could be coming from the ADC circuit itself.

Yes this is what it seems like, since you have ruled out almost all other portions of the circuit

-Bryan