DSP56F807PY80E: ADC reading differs from expected 2048 counts at 1.6 V input Hello NXP Community, I am troubleshooting an ADC offset issue on a board using the DSP56F807PY80E. The following voltages are measured at the device pins: Pins Function Measured voltage 102–109 ANA0–ANA7 inputs 1.60 V 99 and 110 VREF and VREF2 3.20 V 88, 100 and 111 VDDA 3.30 V For a single-ended conversion with no offset correction, I expect approximately 2048 counts, since the input voltage is half the reference voltage: ADC count = (1.60 / 3.20) × 4096 = 2048 However, the ADC readings differ from this expected value. Could someone please advise: Which ADC configuration and offset registers should I check? Could the difference be related to how the ADC result register is read or shifted? What is the recommended procedure to reset or reinitialize the ADC and verify its offset? Any guidance or example initialization code would be appreciated. Thank you for your support. Re: DSP56F807PY80E: ADC reading differs from expected 2048 counts at 1.6 V input Hello @Hari0369 ,
Thanks for the detailed information.
Your calculation is correct for an ideal 12-bit conversion. With a 1.60 V input and a 3.20 V reference, the expected result is about 2048 counts.
The first thing I suggest checking is how the ADC result is read.
On the DSP56F807, the 12-bit ADC result is stored in ADRSLT bits 14–3. With no offset applied, 2048 counts would appear as about 0x4000 in the raw ADRSLT register.
So please shift the raw result right by 3 bits before comparing it with 2048:
ADC value = ADRSLTn >> 3
For your test condition, you should see approximately:
Raw ADRSLT value : 0x4000
After >> 3 : 0x0800 = 2048
I also suggest checking the following ADC settings:
- Check ADLST1/ADLST2 and make sure the correct analog input is selected. - Check ADCR1.CHNCFG and make sure single-ended operation is selected. - Check ADSDIS and make sure the sample slot you are using is enabled. - Check the corresponding ADOFS register. For this test, please set the offset to 0x0000. - Check ADCR2 and make sure the ADC clock and sampling configuration are correct. - Make sure the ADRSLT register you read matches the sample slot you configured.
For a clean test, I suggest using only one ADC channel.
First, stop the ADC and wait until ADSTAT.CIP indicates that the ADC is idle. Then disable the unused sample slots and configure one known channel. Set the corresponding ADOFS register to 0x0000 and enable only the sample slot you want to test.
After that, start one conversion and save the raw ADRSLT value. Then shift this value right by 3 bits and compare the result with the expected ADC count.
If the result is still different from expected, could you please send us the following values?
- ADCR1 - ADCR2 - ADLST1/ADLST2 - ADSDIS - The corresponding ADOFS value - The raw ADRSLT value
These values should help us determine whether the difference comes from the result format, ADC configuration, offset setting, or the analog input path.
Hope it helps.
BR
Celeste Re: DSP56F807PY80E: ADC reading differs from expected 2048 counts at 1.6 V input Hi @Celeste_Liu, Thank you for sharing the details and troubleshooting suggestions. I have used the same firmware to program approximately 500 DSP controllers. Most units operate as expected, but around 10 units show ADC readings that differ from the expected count under the same test conditions. I understand that the datasheet specifies an ADC offset voltage range of −90 mV to +10 mV, with a typical value of −25 mV. Could you please clarify whether resetting or reinitializing the DSP or ADC could bring the reading closer to the expected 2048 counts at a 1.60 V input and 3.20 V reference? If so, could you please share the recommended procedure? Thank you for your support. Re: DSP56F807PY80E: ADC reading differs from expected 2048 counts at 1.6 V input Hello @Hari0369 ,
Thanks for the additional information.
Reinitializing the ADC can help rule out an ADC configuration or state issue, but it normally will not change the inherent ADC offset of the device.
The datasheet specifies an ADC offset voltage from -90 mV to +10 mV, with -25 mV typical. With VREF = 3.20 V, 1 LSB is approximately 0.781 mV, so some difference from the ideal 2048 counts may be expected.
For the affected units, I suggest the following check:
Stop the ADC and wait until ADSTAT.CIP indicates that the ADC is idle.
Reinitialize the ADC with the same configuration used on the normal units.
Make sure the corresponding ADOFS register is set to 0x0000.
Perform several conversions and compare the averaged result before and after reinitialization.
Repeat the test after a complete power cycle.
If the same units consistently show the same offset, please send us the actual ADC counts from one normal unit and one affected unit at 1.60 V input and 3.20 V reference. We can then compare the difference with the ADC specification.
BR
Celeste
View full article