• Locked Locked
  • Replies 5 replies
  • Subscribers 47 subscribers
  • Views 6397 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

ADS1258 harmonics and noise revisited

Eduardo Pellini
Prodigy 140 points
Other Parts Discussed in Thread: ADS1258, OPA2727, OPA727

Greetings to everybody,

 

We`re working with a ADS1258 for the sampling of three differential analog channels, rated from -2.5V to 2.5 V with a band from DC to around 2000 Hz.

The AD has a voltage reference circuit with 2V5 and -2V5, with good stability. There is a buffer circuit with a OPA2727 between the MUXOut and the AD inputs.

We stablished a good layout, with propper bypass capacitors for both the analog and digital world.

The data rate was set to 15123 Hz, with a delay of 2, resulting on a total of 12177Hz of sample frequency for the AD.

Scannig the three differential pairs, we collect samples at 4059 Hz, per channel.

We noted in our application that, for low voltage input signals, our measurements have some impropper interferences.

We connected all analog inputs in the mux  to ground and collected some samples for more than a minute. One of this signals if being show on Fig. 1.

Fig. 1 - AD readings for one of the channels. Vertical axis are AD units. Horizontal axis are samples (@4059 Hz of sampling frequency).

Of course, even with the channels being connected to ground, there are some offset voltages in the channel inputs, dut to the mux and buffer, resulting in readings with a mean of -416 AD units, with standard deviation of 19  AD units. Ok. For now everything seems fine. However....

We expected this signals to be a white noise, but when we analyse the noise power espectrum we saw harmonic components with substantial energy around 60 to 70 Hz and near 0.8KHz and 1.7 KHz. The spectrum for the above signal is being shown on Fig. 2.

Fig. 2 - Noise power spectrum. Note the harmonic interferences.

Where do these harmonic signals come from ? 

We tried a lot of different setups, with different power souces and line filters, but we could not get rid of those components ! 

Any help are greatly appreciated.

Thanks in advance for your help and attention.

Eduardo Pellini

 

  • TI__Guru**** 172846 points

    Hi Eduardo,

    Are you taking these measurements on the ADS1258 using our EVM or your own hardware?  We'd like to try and set up a similar test here to see if we see the same results.

  • in reply to Tom Hendrick
    Prodigy 140 points

    Hi Tom, we are taking these measurements directly from our hardware. We di not have a EVM board.

    In our hardware, from the sixteen available channels on the mux input, 10 were tied to ground. The other six were connected to signal source 1 and ground, signal source 2 and ground and signal source 3 and ground.

    I can send you schematics if you want.

    Thank you

    Eduardo

  • in reply to Eduardo Pellini
    TI__Guru**** 172846 points

    Hi Eduardo,

    My apologies here - it seems like I've overlooked your follow up regarding your measurements, if you can, please do post the ADC potion of your schematic and we'll take a look at it.

  • in reply to Tom Hendrick
    Prodigy 140 points

    Hi Tom, sorry by the delay. We changed priorities among other projects.

    Now returning to the original problem.

    Here is the schematic for the AD (Fig. 1), Reference generator circuit (Fig. 2) and MuxOut/ADIn buffer and driving circuit (Fig. 3).

    Fig. 1 - AD 1258 Schematic - Nowadays, only the 6 first channels are being acquired in differential configuration (AIN0-AIN1, AIN2-AIN3 and AIN4-AIN5). The other diferential inputs, despibe being energized by the same signals, are not selected.

     

    Fig. 2 - The 2V5 and -2V5 Voltage Reference generator.

     

    Fig. 3 - AD muxout/adcin driving circuit.

     

    We don't know for sure if unused analog input should be left unconnected, connected to the input signals (as depicted on Fig.1) ou connected to ground, positive or negative reference.

    Anyway, we don't know why the power spectrum of the input noise shows these low frequency intereferences.

    Any help are greatly appreciatted.

    Eduardo

     

  • in reply to Eduardo Pellini
    TI__Mastermind 43915 points

    Hi Eduardo,

    I powered up our EVM board here and took a look at the frequency spectrum with the inputs shorted to see if anything stood out similar to what you see. Our noise floor looks fairly flat without any spurious frequencies. I attached a photo below:

    Looking at your original post, the frequencies are showing up around 60Hz, 120Hz, 300Hz, 800Hz, and about 1.7KHz. The 60Hz and possibly the 120Hz may be a result of the power supply coupling into the signal. When you short the two inputs together, are you tying them to a common mode point? and if so, what is the source of the common mode point?

    The 300Hz, 800Hz and 1.7KHz do not appear in our frequency domain graph. Do you have a switcher or charge pump on board to create the negative supplies or any high voltage supplies? In order to try and isolate where this is coming from, I would recommend try tying the inputs together directly at the ADCINN and ADCINP inputs to bypass the OPA727 driving amplifiers. If the frequencies still appear, it may be that it is coupling through the reference or the power supply.

    Regards,

    Tony Calabria