TMS320F28075: TMS320F28075 ADC modules seems doesn't work ，what happened？

Hi,

We have assigned your query to the expert. Please expect response only by the first week of January '22 due to holidays.

Thanks,

Aditya

Hi Jason,

Can you explain a little more about the glitch/noisy signal?   Is it possible that this glitch is a continuous over-voltage (greater than 0.3V over VDDA) or under-voltage (less than 0.3V below VSSA) event on one of the ADC input pins? 

Does this issue resolve on a given device if the glitch is removed, or is the device permanently damaged? You might check for permanent damage on the ADC pins by checking the static leakage of the ADC pins (apply a voltage within the range of VSSA to VDDA: the current into or out of the pin should be less than a couple uA...if you see mA in or out of the pin it is a strong indication that the pin was damaged by an over or under voltage event or ESD event).  

Hi Devin,

thank you for your reply. I did find some glitch on one of the ADCA's input pins, it influence all the ADC result, All the 3 ADC modules' result in this chip turn out to be wrong. I will check the glitch's voltage again, to see if it over-voltage or under-voltage.
if the glitch removed, all the ADC modules can work well again, it seems the device doesn't have permanently damaged.

Hi Jason,

Ok, let us know what the glitch looks like.  

It may also be relevant to note what the driver for that ADC pin looks like (e.g. an op-amp with 200 ohms of series resistance, a voltage divider, etc) so we can estimate how much current is injected into the device during the event.  

Hi Devin
first let me explain what happened in those days.
i have two boards.
one board's performance is as i descripted before.
The difference i can only find is some glitch on ADC input pins.
The glitch or noisy comes from another power supply in the board( This power supply don't power the TMS320F28075 and completely isolated from TMS320F28075). if another power supply's output is short. i can get some glitch on ADC input pins.

maybe these glitch make the ADC goes into the situation i descripted before.

Today, I get another board, this board's performance is as following:
if i make the same short in the board. i still can find some glitch from the PIN49, ADCBINB3 as follow:

but the 28075's ADC still work, and the ADC result is wrong. for example
if i apply 1.64V in PIN40 ADCINA3, I can get a result about 1773 or 1774 in the ADCresult register, but if i removed the short, the result is 2033 as expect.
the wave of this pin as follow:

all these input pin have an R-C filter ，

the Resistor's value is 1k ohm and the capacitor's value is 100nF.

Hi Jason,

What frequency do you sample the ADC inputs?  What is the frequency of the glitch/noise? Is the glitch/noise synchronous to the ADC sampling (if perhaps the glitch is ePWM driven and the same ePWM is triggering the ADC sampling).   

That glitch doesn't seem to be too bad (it is short and the magnitude is not less than 0.3V below VSSA or >0.3V above VDDA).  What, if anything, does the glitch look like on the channel you are sampling (A3)?  Is any other channel sampled on ADC-A before or after A3? 

Any observable glitch on VDDA or VREFHI?  

Hi Devin

I find that the glitch synchronous with ADC converter, the glitch will happen just before program enter into the ADC end interrupt, when I change the ADC converter frequency, the glitch's frequency change with ADC converter. but i can't find any signal change with my ADC converter. it is confused. I used a GPIO to index my program enter into ADC end interrupt. I stopped the GPIO, and still can find the glitch's frequency change with my ADC frequency. The tms320f28075 has three ADC modules. those modules will sample and converter simultaneously. Before the A3 channel, the ADC A2 will sample and converter, and after the A3 channel , A4 channel will sample and converter. No glitch can be seen on VDDA and VREFHI.

Hi Jason,

Does the glitch still exist if the ADC is not sampling?  For instance, is the glitch caused by ePWM switching and actuating some power stage, which is also synchronous with the ADC sampling because the ePWM is the trigger for ADC sampling? 

The most sensitive point in the ADC conversion would be right before the S+H period ends, as the end of the S+H period is the instant that the ADC samples the voltage.  Any glitch on the input that doesn't fully settle before the end of the S+H period could affect the results.   You might try: increasing (or decreasing) the S+H duration to try and move the glitch away from the end of the S+H period.  You can also try reducing the 1k resistor to something more like 10 ohms: this will allow the ADC input to recover faster in case of a glitch.  You can also reduce the capacitance to something closer to 200-300pF as this will further speed up recovery time.  

Note: that 1kohm + 100nF is a very high impedance for the ADC input.  This could work out well if you are using a charge-sharing input design (see: https://www.ti.com/lit/an/spracv0/spracv0.pdf) but probably won't work well if you are sampling fast (for ADC input drive stage designs for fast sampling see: https://www.ti.com/lit/an/spract6/spract6.pdf)    

• Hi Devin
  if the ADC not sampling, the glitch will disappeared. To avoid the power stage's distribution, my ADC is trigger by Timer not PWM. I tried to change the trigger time. still can see glitch at the time when ADC is sampling and convert. I will read the document you provided, maybe it is the reason.