TMS320F28335: ADC issue

Strong ZHANG

Part Number: TMS320F28335

The customer find a ADC issue of 28335, they configure the ADC as below with 150MHz CPU frequency, which work in Simultaneous Sampling Mode and the sampling window is about 240 ns(ACQ_PS = 2), they find the sample value of A0 channel is about 550 when the input voltage is 0.3V, which should be about 410, they change the ACQ_PS to 4, then the sample value of A0 is correct.

SysCtrlRegs.HISPCP.all = 1; // HSPCLK = SYSCLKOUT/ADC_MODCLK 75Mhz

AdcRegs.ADCTRL1.bit.SEQ_CASC = 1; // 1 Cascaded mode

AdcRegs.ADCTRL3.bit.ADCCLKPS =0x03; // 6 HSPCLK/[12*(ADCTRL1[7] + 1)]

AdcRegs.ADCTRL1.bit.ACQ_PS = 2; //3*clock

AdcRegs.ADCTRL3.bit.SMODE_SEL = 0x1; //同步。

AdcRegs.ADCCHSELSEQ1.bit.CONV00 = 0x0;

AdcRegs.ADCCHSELSEQ1.bit.CONV01 = 0x1;

AdcRegs.ADCCHSELSEQ1.bit.CONV02 = 0x2;

AdcRegs.ADCCHSELSEQ1.bit.CONV03 = 0x3;

AdcRegs.ADCCHSELSEQ2.bit.CONV04 = 0x4;

AdcRegs.ADCCHSELSEQ2.bit.CONV05 = 0x5;

AdcRegs.ADCCHSELSEQ2.bit.CONV06 = 0x6;

AdcRegs.ADCCHSELSEQ2.bit.CONV07 = 0x7;

But TMS320F28335 DSC Silicon Errata mention that If the ADC is used with a sampling window greater than 160 ns, there is no issue. The sampling window the customer set is 240 ns, please help to clarify why it still have the issue?

cid:image001.jpg@01D2CFF4.554C9DA0

over 8 years ago

0 tlee over 8 years ago

TI__Guru 62975 points

Strong,

Does the converted signal have low source impedance? The ACQ_PS value can make a difference when the source signal is too weak to charge the internal S/H cap within the sampling window.

-Tommy

0 Strong ZHANG over 8 years ago in reply to tlee

TI__Expert 4490 points

Here is the schematic of AD sample to A0.

0 tlee over 8 years ago in reply to Strong ZHANG

TI__Guru 62975 points

Strong,

For some reason I can't see the attached image here. I'll look at your offline version.

-Tommy

0 tlee over 8 years ago in reply to tlee

TI__Guru 62975 points

Strong,

Can you provide some background on the system? Are other ADC channels used? Do they have identical buffering? Are they experiencing issues?

-Tommy

0 Strong ZHANG over 8 years ago in reply to tlee

TI__Expert 4490 points

Tommy,

Thanks for your reply.

they use all channels of 28335. also use the same buffer for other channel. Only A0 have this issue, which is the first conversion of the ADC, if change the A0 Channel to the second converstion with the same ACQ_PS , it will be ok.

and Not each the customer's board have this problem, the Failure rate is about 1/500.

0 tlee over 8 years ago in reply to Strong ZHANG

TI__Guru 62975 points

Strong,

When A0/B0 is re-ordered to CONV01, does the new channel pair for CONV00 experience distortion? I would expect that to be the case if it is related to the first sample-pair advisory.

And to confirm, B0 is not affected when it is sampled with A0 in CONV00?

-Tommy

0 Strong ZHANG over 8 years ago in reply to tlee

TI__Expert 4490 points

A0 is the first sample, and only A0 have this issue, other channels is ok.

0 tlee over 8 years ago in reply to Strong ZHANG

TI__Guru 62975 points

Strong,

Based on the fact that only A0 is problematic (even when re-ordering the sequence channels), I am not confident that this is the same issue as described in the advisory.

For the advisory issue, I would expect the first simultaneous conversion to be affected regardless of channel selection. For example, if you have this original sequence:

A0 + B0 (bad conversion)
A1 + B1 (good conversion)

The error should remain with the first conversion if the sequence channels are re-ordered:

A1 + B1 (bad conversion)
A0 + B0 (good conversion)

-Tommy

0 Strong ZHANG over 8 years ago in reply to tlee

TI__Expert 4490 points

here is the issue, this original sequence:
1.A0 + B0 (A0 bad conversion, B0 ok)
2.A1 + B1 (good conversion)

The error should remain with the first conversion if the sequence channels are re-ordered:
1.A1 + B1 (did not take a check about A1)
2.A0 + B0 (good conversion)
and if use Sequential sampling mode, all channels are ok with sampling window of about 240 ns(ACQ_PS = 2)

0 tlee over 8 years ago in reply to Strong ZHANG

TI__Guru 62975 points

Strong,

If we can confirm that A1 has error when it is converted first, then we can consider this to be a simultaneous first-conversion issue with more confidence. In that case we may treat this as a potential defect because the behavior does not match our expected >160ns guideline in the advisory.

-Tommy

0 Strong ZHANG over 8 years ago in reply to tlee

TI__Expert 4490 points

Tommy,

Aleady confirm with customer that A1 has error when it is converted first. Could you help to explain why?

0 Strong ZHANG over 8 years ago in reply to tlee

TI__Expert 4490 points

Tommy,

Already checked with customer that A1 has error when it is converted first, Do we have a format report to explain why has this problem when the sampling window is 240ns

0 tlee over 8 years ago in reply to Strong ZHANG

TI__Guru 62975 points

Strong,

The reported behavior is not matching expectations so we have no way of understanding the root cause until we can evaluate it.

-Tommy

C2000™︎ microcontrollers

C2000 microcontrollers forum

TMS320F28335: ADC issue