CCS/TMS320F280049: F28004x ADCOSDETECT

Ricky Zhang

Part Number: TMS320F280049
Other Parts Discussed in Thread: C2000WARE

Tool/software: Code Composer Studio

Hi, Devin,

Thanks for your response.

Customer added the register definition in header file by themselves and made the debugging worked, while you do have an update on the C2000WARE update?

In the other hand, they met new problem when using this feature as the ADC result seems to be incorrect when it's enabled.

Channel A9 is left unconnected and will be sampled. When ADCOSDETECT is configured to 3/4/5/6/7, the ADCRESULT will be 1591/2259/0/3868/0, where pull down seems to work, while pull up and resistor divider seems to get smaller results.

Could you please take a look at this issue and suggest where the problem is?

Best Regards,

Ricky Zhang

over 6 years ago

0 Devin Cottier over 6 years ago

TI__Guru 60865 points

Hi Ricky,

Channel 8 and 9 are connected to VREFLO, not unconnected:

0 Ricky Zhang over 6 years ago in reply to Devin Cottier

TI__Genius 9520 points

Devin,

Thanks for your quick response, while it seems to be Soprano analog sub-system where channel 8 and 9 are connected to Vreflo, however, my customer is using F28004x device, so could you please confirm if it's same on both families?

Best Regards,

Ricky Zhang

0 Ricky Zhang over 6 years ago in reply to Devin Cottier

TI__Genius 9520 points

Devin,
Any ideas please?

0 Devin Cottier over 6 years ago in reply to Ricky Zhang

TI__Guru 60865 points

Hi Ricky,

Is this on the 100-pin package or a package w/o A9 pinned out?

S+H = 10 sysclk cycles = 100ns is too small of a S+H window for a 5k source impedance (see TRM "Choosing an Acquisition Window Duration"):

tau = (5k + 500)*12.5pF + 5k*(0pF + 8.1pF) = 109ns
k = ln(4096/1) - ln(8.1pF / 12.5pF) = 8.3 + 0.4 = 8.7
(assuming 1 LSB error is fine)

So settling time = 109ns*8.7 = ~950ns, so you'd want ACQPS to be at least 94 = 95 SYSCLK cycles.

0 Ricky Zhang over 6 years ago in reply to Devin Cottier

TI__Genius 9520 points

Devin,

Thank you for response.

Yes, customer is using a 100-pin package. I'm asking customer to trying larger ACQPS and will let you know if there're any feedbacks.

Best Regards,

Ricky Zhang

0 Devin Cottier over 6 years ago in reply to Ricky Zhang

TI__Guru 60865 points

Hi Ricky,

Any update from the customer here? If not I'll assume that the S+H duration was the issue.

0 Ricky Zhang over 6 years ago in reply to Devin Cottier

TI__Genius 9520 points

Devin,

Sorry for late update.

Customer feedback to us that they fixed this issue by following your advice to enlarge the S+H duration.

Thanks for your following-up and let's close this post now.

Best Regards,

Ricky Zhang

C2000™︎ microcontrollers

C2000 microcontrollers forum

CCS/TMS320F280049: F28004x ADCOSDETECT