This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

AFE4300 weigh measurement - drift

Other Parts Discussed in Thread: AFE4300

I am coding and trouble shooting a board with a AFE4300 on it.  I've found I have a lot of offset from the time I read the Zero value of the load cells to the time I read the "loaded" value.  from the time the power is applied and 1st reading is made, it's taking over 6 seconds for the channel to stabilize. 

This is with the channel shorted to it's IMH counter part (and 0.866V) bias applied (gain Rs = 5K = x41 on 1st stage).  with the inputs shorted, I should not see any affects of my external RC filters but I see the single channel drift from more negative toward zero regardless of the gain setting or DAC offset value.  

Is this internal drift if the WT input section to be expected?  if not, where should I look to speed up the process?

Thanks,

Marty

can't seem to include plot so here is (6 seconds) of the 64sps ADC data.

-3522
-3522
-3516
-3507
-3501
-3497
-3497
-3490
-3484
-3484
-3488
-3486
-3477
-3469
-3471
-3467
-3465
-3457
-3457
-3463
-3460
-3449
-3442
-3440
-3440
-3437
-3435
-3435
-3428
-3423
-3423
-3420
-3419
-3417
-3414
-3417
-3414
-3407
-3403
-3401
-3401
-3403
-3403
-3400
-3398
-3395
-3390
-3386
-3386
-3384
-3385
-3383
-3386
-3384
-3382
-3376
-3374
-3374
-3375
-3375
-3369
-3371
-3370
-3365
-3360
-3361
-3361
-3357
-3354
-3352
-3355
-3359
-3351
-3344
-3346
-3347
-3346
-3343
-3348
-3349
-3339
-3334
-3333
-3336
-3334
-3336
-3336
-3337
-3335
-3329
-3325
-3327
-3328
-3324
-3321
-3321
-3327
-3324
-3322
-3323
-3324
-3326
-3322
-3320
-3323
-3321
-3317
-3314
-3310
-3308
-3311
-3311
-3309
-3310
-3311
-3308
-3302
-3302
-3303
-3305
-3307
-3306
-3308
-3310
-3304
-3298
-3295
-3295
-3298
-3299
-3297
-3299
-3295
-3293
-3293
-3297
-3295
-3294
-3295
-3297
-3295
-3291
-3287
-3291
-3293
-3291
-3288
-3292
-3290
-3290
-3284
-3282
-3284
-3284
-3279
-3279
-3280
-3283
-3281
-3280
-3279
-3283
-3282
-3283
-3281
-3280
-3282
-3278
-3279
-3272
-3273
-3276
-3276
-3276
-3279
-3280
-3273
-3271
-3269
-3269
-3272
-3269
-3276
-3279
-3279
-3274
-3271
-3271
-3271
-3272
-3273
-3274
-3272
-3265
-3264
-3264
-3267
-3269
-3265
-3265
-3270
-3267
-3265
-3261
-3263
-3262
-3260
-3259
-3265
-3266
-3264
-3259
-3259
-3262
-3263
-3263
-3262
-3262
-3262
-3256
-3251
-3254
-3256
-3257
-3257
-3257
-3260
-3257
-3254
-3257
-3258
-3260
-3255
-3252
-3253
-3255
-3253
-3250
-3248
-3250
-3252
-3253
-3253
-3255
-3255
-3248
-3246
-3246
-3248
-3248
-3247
-3248
-3252
-3252
-3249
-3252
-3249
-3249
-3249
-3247
-3244
-3247
-3245
-3244
-3241
-3242
-3242
-3244
-3245
-3248
-3249
-3242
-3238
-3240
-3248
-3248
-3245
-3245
-3249
-3248
-3244
-3239
-3239
-3242
-3240
-3238
-3240
-3242
-3242
-3236
-3234
-3234
-3234
-3238
-3238
-3243
-3241
-3236
-3238
-3236
-3237
-3236
-3236
-3239
-3240
-3242
-3236
-3233
-3235
-3235
-3235
-3234
-3233
-3237
-3238
-3234
-3230
-3231
-3234
-3232
-3229
-3237
-3238
-3237
-3234
-3227
-3230
-3231
-3231
-3231
-3231
-3236
-3237
-3234
-3236
-3233
-3234
-3233
-3234
-3239
-3234
-3228
-3225
-3227
-3229
-3231
-3229
-3235
-3239
-3237
-3231
-3233
-3233
-3232
-3232
-3232
-3234
-3232
-3230
-3228
-3226
-3228
-3227
-3227
-3228
-3229
-3230
-3226
-3222
-3226
-3226
-3224
-3225
-3226
-3230
-3229
-3224
-3222
-3220
-3224
-3218
-3219
-3224
-3225
-3221
-3222
-3223
-3228
-3230
-3229
-3231
-3230
-3224
-3223
-3219
-3220
-3223
-3219
-3220
-3221



  • Hello Marty,

    Do you notice the drift across all the four WS input channels?
    What are the AFE register settings?
  • I only have one channel shorted but the the drifts effect on whole is why I started this experiment. is it possible it's only on one channel?
    here are all of my registers (after the test). the "x09" is the address (of Device control 1 in this case)
    reg adc x01=C1B0 x10=0
    reg wt x09=6005 x0D=6000 x0F=6
    reg bcm x0A=0 x0B=0 x0C=0 x0E=0
    reg misc x02=0 x03=FFFF x1A=0

    Each reading is done in "single conversion mode" where BIT 15 ( ADC_CONV_MODE) of AdcCntr1 (x01) is cleared and then set again to start the next reading.
    The Demo Kit needs to be powered before starting TI's AFE4300 Device GUI so I can not test this effect on it. Has anyone else notice this chip requiring a warm up period?
  • Do you notice the drift across all the four WS input channels?

    It would really help if you could show us a screenshot of the drift you are seeing.

    It sounds like you have an EVM kit. If so, you can try running our GUI with your same settings and seeing if you have the same problem. This will tell you if it is something in the hardware.
  • I can't seem to get your "paste from word" to work so here is the document direct

    I have plotted both the wakeup drift and an impulse response.  the impulse recovers quickly in about a second, much faster than the wake up drift.

    drift pic.doc 

  • I will short the other 3 channels and prove you with the results but it would help it you told me what your looking for so I can customize the experiment.
  • As a followup, I've removed the drift by running the AFE in continuous mode until I'm ready to make the measurement. This 2 to 4 second delay stabilizes the a shorted input channel. This implies I'm waiting for the rails (inside and out) to stabilize. looking at both of your reference designs (tidr119 and tidra02) there is a 49.9Ω resistor in the VLDO (pin 17). It's hard to see due to the poor quality of the PDF but it's R70 in both. 1.7V / 50Ω = 34 ma but the data sheet says the max should be no more than 20ma (page 3)??
    which is correct?
    This line also runs to the 4 load cells/bridges which add another 1.7ma each.
  • I see in the AFE4300 demo board there is also a R70 that is 50Ω but it is not populated.  this looks to me as if it's an error that keeps getting copied over and over. 3 times so far.  I have grow to expect better from TI.   It seems the knowledge base for this chip within TI is limited even though it's been out for a few years and has a demo board.  Beginners beware!

  • Marty,

    Sorry for the confusion, but the R70 resistor is something that is used internally. We have the footprint on there only so we can work with it.