Post Go back to editing

Design help

Category: Hardware
Product Number: LTC6930

I am replacing a now obsolete EPSON CMOS oscillator that served as the primary clock in a dual axis stepper motor design for over 15 years. The primary clock frequency is 32.768kHz --- we wish to use the Linear Tech LTC6930 to divide down its internal 4.192Khz by 128 and use the the resulting 32.768kHz as our new primary clock signal.

We will be driving a pair of AD9833 DDS waveform generators which operate as stepper clocks for 2 independent stepper drivers.

1. To your knowledge, will the LTC6930 operate with the AD9833 in a single 5V supply environment? 

2. Will the LTC6930 drive two AD9833 devices directly without issue, or should I add a series resistor from the LTC6930 output to each AD9833 input?

3. If resistors are required, what would the appropriate value be --- (470 Ohm, 1K, 10K, etc;) ?

4. The approximate distance from the primary clock output to the AD9833s is 1.48 inches for the farthest unit and 0.498 inches for the closest unit with lots of ground plane all around.

Any suggestions or recommendations greatly appreciated .

Best regards,

Randy

  • Hi Randy,

    I am not familiar with AD9833, however, according to the datasheet, LTC6930 output is compatible with the MCLK input of AD9833.  Since the interface between AD9833 and LTC6930 is CMOS logic. I recommend placing a 30 Ohm resistor closer to the LTC6930 between the output of the LTC6930 and the input of the AD9833

     If synchronization is important between motor drivers, I recommend giving attention to the layout planning. If not, these distances are fairly small for 32KHz CMOS signals, There should not be a problem with the propagation of the signals. Depending on the input impedance of the motor drivers, a 1:2 buffer might be needed to drive multiple drivers from AD9833. 

    One recommendation from my side would be about clock accuracy. What are your clock performance requirements? Did you compare the EPSON CMOS oscillator performance and LTC6930 performance? 

    LTC6930 is a silicon oscillator with a fairly good frequency accuracy and temperature coefficient, However, it might not catch the performance of a good crystal oscillator.  I added the clock accuracy vs temperature graph.

    Thanks,

    Emrecan

  • Hi Emrecan-

    30 Ohm resistor closer to the LTC6930 between the output of the LTC6930 and the input of the LTC6930

    May I assume you meant the input of the AD9833?

    There should be a problem with the propagation of the signals

    Is that "should NOT be a problem"?

    The steppers operate independently, so sync is not an issue.

    The device is in a laboratory environment at room temperature, so we are in the "sweet spot" according to the temperature graph.

    Below is a schematic based on your suggestion. Please verify that I understand your meaning.

    Thanks for responding,

    Randy

  • Hi Randy,

    My apologies for the typo. I corrected them. You are right. 

    Your diagram looks good. It works as expected. However, I would remove the RA and RB and fan out the output signal after a series resistor. I added a diagram to explain.

    Thanks,

    Emrecan

  • Emrecan, Unless you are matching the trace length of MCLKA and MCLKB, it is better to use two separate resistors as originally shown. (And even if you do match lengths, it is still better, although the difference is minor.)