Solutions for LPC1768 ADC noise caused by debugging?

[josephmurray]

I am using the Keil uVision5 IDE with the ULINKpro Debug and Trace Adaptor connected to the MCB1700 Evaluation Board. (Target MCU is the NXP LPC1768)

It has been discovered that when in a debug session, analog to digital  conversions are affected with additional noise.

This has been revealed by duplicating the ADC result to the DAC and scoping the output.

Does anyone have a solution/suggestion(s) in regards to the additional noise on ADC samples caused by JTAG/Trace during a debug session?

The NXP Application Note AN10974 mentions this problem without offering a solution - does this mean there is no solution?

[craigmcqueenir]

I've been having this issue, and it's taken a while to try to figure out if it was an issue with the microprocessor or with my firmware.

That app note AN10974 is a good discovery. I think that information should be also in the errata for the processor, because it's a significant limitation of the processor that has caused me to spend some significant engineering time looking into it. I have the errata sheet ES_LPC176x, and item "ADC.3: Noise caused by nearby I/O pin switching activity or board design/layout could cause the ADC conversion results to be above the expected range." sounds potentially similar. However, in my case, I'm seeing glitches down to zero or towards zero, rather than glitches towards 0xFFF.

It seems to affect mostly AD0[3], on P0.26 (pin 6), which is right next to the SWCLK signal for JTAG on pin 5. To a lesser degree it affects AD0[2], on P0.25 (pin 7). However, if I put an oscilloscope on the pins, I see no visible glitches, so I guess cross-talk must be happening within the micro.

[lpcxpresso_supp]

You might find the analysis of LPC1768 + ADC done by the mbed team of interest : Getting the best ADC performance from mbed | mbed 

Regards,

LPCXpresso Support

[josephmurray]

Cheers, I've had a look at this.

Filtering and grounding techniques are implemented but I still find the best fix is in not using the debugging tool.

[Carlos_Mendoza · Accepted solution]

Hi Joseph,

Unfortunately the additional noise introduced by the JTAG during a debug session can't be avoided, one workaround would be to avoid the usage of the debugger during the ADC measurement testing stage of you project and instead save the measured data for further analysis.

Hope it helps!

Best Regards,
Carlos Mendoza
Technical Support Engineer

[josephmurray]

Thanks for your reply. 

I have come to this realisation, I might try using an external ADC.