• State Resolved
  • Locked Locked
  • Replies 15 replies
  • Subscribers 46 subscribers
  • Views 749 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

DAC82002: DAC82002 exhibits high output noise

Zhenrong Liu
Prodigy 20 points
Part Number: DAC82002
Other Parts Discussed in Thread: DAC8830, ADS127L21, OPA2210, OPA227

Tool/software:

In my design, I need two 16-bit DACs. I have been using DAC8830 before, which has excellent noise performance. The new design uses DAC82002 as dual output to save one DAC chip.

But during testing, I found that the output noise of DAC82002 was much higher than that of DAC8830.

My reference voltage(LM399 with good filtering and buffering) has a very low noise of 0.15ppmRMS. It was measured using ADS127L21 with 160000-OSR and sinc3-filter in 16MHz high-speed mode. And I averaged the results 10 times, so the actual data rate is 5sps.

Because my PCB is compatible with both DAC8830 and DAC82002, I have tested them both. When using DAC8830, the output noise is approximately the same as the reference voltage noise, which is also about 0.15ppmRMS. However, when using DAC82002, the output noise increased tenfold to 1.5ppmRMS, I can't figure out the reason.

DAC8830 and DAC82002 are tested whit the same reference voltage and power supply, and even similar PCB layout. The voltage reference is buffered to drive a 10uF X7R capacitor, and the REF pin and VDD pin of the chip are decoupled by a 100nF X7R capacitor. The power supply is 5V for analog circuits, and the ground is an entire plane referenced by the analog circuit. DAC8830 performs well in this design, so I believe my circuit design and PCB layout are fine.

So the question is why DAC82002 exhibits such high output noise, and is it not suitable for such low-noise(several uVs level) design? Or are there any key points in the application design of DAC82002 that I am not clear about?

Best regards.

  • 0
    Prodigy 20 points

    By the way, the SPI interface of the DAC is designed to be tri-state, and the DAC will not receive any SPI signals(CLK or DIN)when CS is inactive. So I'm sure the noise doesn't come from the digital feedthrough.

  • 0 in reply to Zhenrong Liu
    TI__Mastermind 42163 points

    Hi Zhenrong,

    Can you share your schematic? Are there any loads connected to the DAC outputs? Are the two DACs using the same output buffer?

    Best,

    Katlynne Jones

  • 0 in reply to Katlynne Jones
    Prodigy 20 points

    Hi Katlynne

    This is the schematic of my DAC output section.

    The REF_7V is from LM399's output buffer, and the DAC_VO outputs to the input buffer of ADC with out any other load.

    Best regards.

  • 0 in reply to Katlynne Jones
    Prodigy 20 points

    Sorry, I forgot to label the values of R3 and R4 in the schematic. They are 30ohm and 390ohm, respectively.

  • 0 in reply to Zhenrong Liu
    TI__Mastermind 42163 points

    Hi Zhenrong,

    Thanks for sharing. I don't immediately see anything wrong with your schematic. Just to clarify, does the DAC8830 in your circuit also use the OPA2210 and OPA227 buffers? 

    Are you taking the measurement at a specific output code? Or is the noise seen across all codes? Can you repeat the measurement directly at the DAC82002 output? It should have a very low 0.1uVpp DC noise. 

    Best,

    Katlynne Jones

  • 0 in reply to Katlynne Jones
    Prodigy 20 points

    Yes, DAC8830 uses the same buffers, the only difference is that two DAC8830s replace one DAC82002.

    I tested different output codes. Noise is not fixed, but proportional to the output code. It looks like it amplifies the noise of the reference.

    I don't think the noise is related to the buffer, because DAC8830 works very well.

    Directly measuring the output of DAC82002 requires flying wires on the PCB, which may affect the testing. I will try it later.

    Best regards.

  • 0 in reply to Zhenrong Liu
    TI__Mastermind 42163 points

    Hi Zhenrong,

    Let me check with the deign team to see if they have any other thoughts.

    Best,

    Katlynne Jones

  • 0 in reply to Katlynne Jones
    Prodigy 20 points

    Hi Katlynne

    Can you verify this issue on the EVM-kit? I don't have it.

    You should have a very-low-noise reference source and a precision voltmeter.

    Best regards.

  • 0 in reply to Zhenrong Liu
    TI__Mastermind 42163 points

    Hi Zhenrong,

    Yes. I also have the design team comparing the behavior of both devices in simulation. That should give you a better idea of if the extra noise is device related or not. Please give a few days for them to be able to complete the sim. 

    Best,

    Katlynne Jones

  • 0 in reply to Katlynne Jones
    TI__Mastermind 42163 points

    Hi Zhenrong,

    Is this low frequency noise or wider band noise?

    Best,

    Katlynne Jones

  • 0 in reply to Katlynne Jones
    TI__Mastermind 42163 points

    Hi Zhenrong,

    In addition, have you measured the output noise of the reference of both circuits to make sure they are actually contributing the same noise. The fact that the DAC82002 output noise is increasing with code does indicate to me that the reference could be the problem. The DAC82002 is a string DAC. Beyond the 1/f noise, the device would have a flat 10nV/rtHz noise density as we just have the resistor noise from the string. 

    Best,

    Katlynne Jones

  • 0 in reply to Katlynne Jones
    Prodigy 20 points

    Hi Katlynne

    I tested the reference voltage input to the DAC82002 and it was indeed the source of noise. However, the same reference voltage(using the same circuit) behaves well on the DAC8830s. So I think it should be the REF port of DAC82002 that causes the abnormality of the reference voltage buffer.

    In order to suppress noise as much as possible, the reference voltage buffer is designed to have a very small bandwidth. This means that it has a slow speed of Load regulation, which can cause problems on some chips.

    So I tried to increase the bandwidth and response speed of the reference buffer. As shown in the figure below, R9 decreases to 30 ohms, C5 decreases to 100nF, and C6 decreases to 1uF. Then everything became okayRelaxed

    The problem was solved, but I don't understand why. Technically, R2R DAC is just a static resistor string, so it is a constant load for the reference buffer when the output code is unchanged. It should not and will not absorb charge from the REF pin, why does it require load regulation of reference voltage? Why the DAC8830 works well in this situation? It's so confusing.

    Best regards.

  • 0 in reply to Katlynne Jones
    Prodigy 20 points

    Hi Katlynne

    It's a 0.1Hz~10Hz noise.

  • 0 in reply to Zhenrong Liu
    TI__Mastermind 42163 points

    Hi Zhenrong,

    I'm glad to hear that you've got to the bottom of your issue. 

    An R-2R DAC is not a static string (a string DAC is a static string). You can see in this figure from the datasheet that the switches are connected to either VREF or ground meaning that the reference impedance will be changing. 

    Here is a figure of the DAC82002 reference current changing over code for two different reference voltages. 

    The DAC82002 has a lower reference impedance than the DAC8830, so this could be the reason the same reference buffer configuration was not working. 

    Best,

    Katlynne Jones

  • 0 in reply to Katlynne Jones
    Prodigy 20 points

    Hi Katlynne

    Yes, the R-2R DAC is not a static string. But I think it's only true when the output code changes. When the output code is unchanged, the state combination of the switches is unchanged, so the impedance is also unchanged. In my case, noise occurs when the output code is not changed.

    Although it doesn't make sense, fortunately I am an engineer rather than a chip designer, so I don't have to dig deeper into it. Perhaps there are some unknown secrets to the internal design of DAC82002, which only its designers can answer.Stuck out tongue winking eye

    Anyway, the problem has been solved. Thanks for your great help.

    Best regards.