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.

  • Resolved

ADS1256 Crystal oscilator start-up problems

Intellectual 560 points

Replies: 8

Views: 6117

Hello,

I'm using the ADS1256 in my design and I have used crystal to generate ADC's clock.

I followed recommendations from data sheet and evaluation module for clock generation. I have used 7.68 MHz crystal with same properties and parallel capacitors of 18 pF as like as crystal used in EVM. My problem is in fact that oscillator will not start every time when I turn on board power supply. One of ten times it fails.

Start up time of 3.3V power supply for digital side and +5V of analogue side is less than 10 ms and both power supplies turn on and off correctly  (no spikes, voltage drops,..).

I checked load capacitance for my crystal and crystal used in EVM. Both have 18pF load capacitance.

I assuming that parallel capacitors of 18 pF used in EVM (also used in my design) are not enough to start up oscilator to oscillate.

According to definition: CL=(C1*C2)/(C1+C2) +Cstray. For C1=C2=18pF CL equals 9pF+Cstray where Cstray can be between 3 to 5 pF. So, the maximum value for CL will be 14 pF in best case. Manufacturers of crystals recommend CL to be as close to possible to Cl of crystal.

Can you comment this problem?

Thanks in advace.

Josko Marasovic

   

  

 

 

  • Hi Josko,

    I attached an app note on using a crystal oscillator with the MSC1210. Most of the info in the application note does not pertain to design with the ADS1256, however, there is one section you should look into; Section 8.3: Crystal Start up issues. Here we discuss crystal start up issues similar to what you are describing you see with the ADC. We explained how to perform a test to see the stability of your start up circuit. You essentially will need to sweep a resistance in series with the crystal to see how the start up behavior reacts. In our testing, I believe we used a POT and slowly increased the resistance. From there, you can decide if the capacitance values you have in place are "safe" in the system or "risky." Taking a quick glance at your description of your circuit, my guess is if you decrease your C1 and C2 capacitors to maybe 5pF, you will have more success.

    Regards,

    Tony Calabria

    SBAA123_Using_Crystal_Oscillators_with_MSC12xx.pdf
  • In reply to Tony Calabria:

    Hello Tony,

    Thanks for useful application note.

    But I am still confused with yours application engineers choice of Cx1 and Cx2 capacitors in ADS1256 evaluation module.

    They have used 18pF with crystal which has load capacitance CL=18pF. So, they have 9 pF from serial combination of Cx1 and Cx2, and 3~5pF form stray capacitance (According to equation 3 from attached application note).

    In , best case they have total load capacitance of 14pF which is 4 pF less than crystal’s CL. Why the haven’t used 22pF instead of 18 pF or maybe 27 pF?  

    Please, find bellow oscilograms of start up and steady state of crystal oscillator with different values of Cx capacitors. For all values of Cx, start up seems ok. I had 18pF by default when modules came from reflow soldering. 

    Why they woudn't oscillate first time on power supply up when they came from reflow soldering, I don't know.  

    Best regards,

    Josko Marasovic

  • In reply to Josko Marasovic:

    Hi Josko,

    There are additional capacitance that you may not necessarily be accounting for in your design. You are going to have a couple of areas that stray capacitance can come from; a differential capacitance between the two XTAL pins, a capacitance from each XTAL pin to ground, and board capacitance coming from the layout and trace widths. This can easily be larger than the 3-5pF which you are stating accounts for the stray capacitance. For this reason, we recommend going with smaller capacitors, 10 - 15pF would work best. You mention that you have trouble with your crystal starting up the first time after the power supply comes up. After the first time, does it start up problem free? If so, your start up problem may be due to too large of capacitance, maybe the board needs to be cleaned as flux may be causing a problem across the crystal, maybe adding a little heat to the board (components heating) take the crystal a little time to start up. 

    Regards,

    Tony Calabria

  • In reply to Tony Calabria:

    Hi Tony,

    About crystals starting up the first time after the power supply comes up, yes I had problem with crystal start up first time whith each module (I had 10 PCB protoypes).  After the first start up, problem had occured mybe one or two time and after that I tried to stuck crystal on start-up but I didn't  suceed. Now, each crystal on prototype module starts every time when power supply comes up. I have to admit that I have never had probelm with crystal oscillators as like as this.

    There is on additional thing that I didn't mention. I have provided one additional parallel connection of clock source for ADS1256. It is 3.3V oscilator which is connected to the ADC's XTAL1 pin but it is not paleced of courese. Maybe these traces contribude to stray capactiance to being grather than 3-5 pF.

    Please, find below TOP and BOT layout of my PCB. Parallel capacittors Cx1 andCx2 are below crystal pads. 

     

    Btw, even it is not subject to the this topic, do you have any useful application note about op amps or In-amps input termination? I use INA326 amplifier in single supply 0-5V to measure RTD resistance in 4-w kelvin connection. I have RF filter on in-amp input but I haven't done any termination. Because of that, inputs of in-amp float when RTD isn't connected and ouptut of in-amp goes to VCC or GND.

     Best regards,

    Josko

  • In reply to Josko Marasovic:

    Hi Josko,

    Thanks you for your feedback.

    As for termination, I would recommend using a simple pull up or pull down resistor on the two RTD input lines (after the RTD but before the RC filter). If you would like to pull it to something other than VCC or GND, you can use a pull up resistor to a midpoint using a resistor divider. If you want to post a little snip of your schematic, I will take a look at it to let you know what I think.

    Regards,

    Tony

  • In reply to Tony Calabria:

    Hi Tony,

    Your Welcome for feedback.

    Please, find attached part of schematic for RTD input.

    I am looking forward to your opinion.

    Best regards,

    Josko

     8357.Pt-100 interface.pdf

     

     

  • In reply to Josko Marasovic:

    Hi Josko,

    I attached an example of how we design in protection circuitry in a thermocouple applications. As you can see, in this example, we use pull up resistors to protect the analog input. In your application, you are using an excitation current source, so if you decide to use pullup resistors, you will create a current divider between the PT100 and the pullup resistor.

    In your application, are you suggesting that you are going to disconnect the PT100 and the 4 connections to J2:A, J3:B, J3:C, J3:D? If so, then the INA326 is going to have floating inputs as you suggested and then looking at the worst case situation, the INA326 output will rail to either 5V or 0V (even if the gain is set to 36). Regardless of the additional ADC gain selected, the absolute maximum input spec is AVDD+0.3V so if the input is seeing 5V, sure the ADC will read back full scale, but you will not have to worry about damaging the ADC. One other thing to make sure is that the D17 diode you are using are not rated less than 1V. The reason being is that I believe your Iexct is 1mA and when it sets your bias across R43, you will have the V- node at 1V. You do not want this 1V to be fighting with the D17 diode turning on.

    Otherwise, I do not believe that a pull up resistor is the right solution for you. If you are really concerned with the floating INA inputs, you may have to include a switch of some sort.

    Regards,

    Tony Calabria

     

     

    sloa123-noise-immune-isolated thermocouple.pdf
  • In reply to Tony Calabria:

    Hi Tony,

    Thanks for overview.

    In my application, there is more than one Pt-100 input and not all of them will be used at the same time. There will be situations where we will use all Pt-100 inputs but it depends of application where we will use our DAQ card.

    In case where only one Pt-100 input will be used,  all other Pt-100 inputs will be unused which means open (floating) inputs of INA326. This is what concern me and why I asked about input termination. I will consider your solution with switches but if there is no any reason to damage ADC input and INA326 by itself when the INA326 output goes to either 0V or 5V I will leave inputs to float.

    I have made attention on stand-off voltage of protection diodes when I was choosing them. They have VRRM greater than 1V, but it seems that I forgot on reverse leakage current. All diodes have reverse leakage current about 2-4uA which probably contributes to error of current source.

    Best regards,

    Josko

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.