• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 48 subscribers
  • Views 871 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.

DAC3283 3rd harmonic issue

Luke Stevens
Prodigy 20 points

Other Parts Discussed in Thread: DAC3283, DAC3283EVM, ADS42LB69

We're currently in the lab testing out a new design which has several DAC3283's and we're seeing an issue with the 3rd harmonic. 

We're sampling around 20 MHz, and outputting a single tones at various frequencies of 1.25, 2.5 and 5 MHz and we're seeing about 50 dBc SFDR due to the 3rd harmonic. 

Our configuration is as described in the data sheet for a 4:1 output, IOUTFS = 20 mA. 

We bypassed the output filer and mixer on the EVM and even though the demo samples at a much higher rate(307MHz x2 interp) we see similar performance. 

Any ideas what could cause this? 

  • 0
    TI__Guru* 83185 points

    Hello Luke,

    Let us first verify your observation in our lab with our DAC3283EVM.

    I have a couple of thoughts regarding the HD3 issue. Would you be able to try the following?

    1. The output transformer on the DAC3283 EVM is the ADT4-1T, which covers the 1dB bandwidth of 14MHz to 500MHz. Would you be able to sample the DAC at higher speed and try to measure the DAC performance at higher output frequency to see if performance improves? If you need to test out the DAC at lower than 10MHz BW, you may try the wider bandwidth version of ADT4-1WT (note the additional of 'w' in the part number) or use ADT4-6T.

    2. Could you try to optimize your spectrum analyzer's attenuation setting so that you are not overdriving the analyzer front-end? Try to increase the attenuation in 5dB steps to see if HD3 improves. You can refer to the following app note by Agilent for more detail:

    http://cp.literature.agilent.com/litweb/pdf/5952-0292.pdf

    We will investigate this in parallel to see our measurement results. Please also update us on the two approaches that I mentioned above. Thanks.

    -Kang

  • 0 in reply to Kang Hsia
    TI__Guru* 83185 points

    Hello Luke,

    Also, a capture of the spectrum analyzer plot would help with the troubleshooting process as well. Thanks.

    -Kang

  • 0 in reply to Kang Hsia
    Prodigy 20 points

    1. The output transformer on the DAC3283 EVM is the ADT4-1T, which covers the 1dB bandwidth of 14MHz to 500MHz. Would you be able to sample the DAC at higher speed and try to measure the DAC performance at higher output frequency to see if performance improves? If you need to test out the DAC at lower than 10MHz BW, you may try the wider bandwidth version of ADT4-1WT (note the additional of 'w' in the part number) or use ADT4-6T.

    When using the EVM we have been sampling at 307 MHz which is the fixed sample rate on the EVM. The HD3 issue at low output frequencies we see on our card, which is sampling at 20 MHz, is also present on the EVM. The HD3 issue seems to improve as your output frequency goes higher. 

    We adjusted the output gain via the SPI registers to see if we were clipping on the output. The result was a linear drop in the fundamental and HD3. 

    2. Could you try to optimize your spectrum analyzer's attenuation setting so that you are not overdriving the analyzer front-end? Try to increase the attenuation in 5dB steps to see if HD3 improves. You can refer to the following app note by Agilent for more detail:

    This was something I thought might be an issue already. I've increased the attenuation from 0, 5 to 10 dB and did not see an improvement. We also have the ADS42LB69 on the same card. I took the output of the DAC A(DAC B looks the same) and cabled it to one of the ADC inputs. This confirmed the measurement we took with the spectrum analyzer. 

    We have changed the output network of 1 channel to use a 1:1 configuration and saw an improvement in SFDR. 

    Below is a list of SFDR at 3 different output frequencies on our card. Note: HD3 is always the highest spur. 

    Output Frequency = 1.25 MHz

    • 4:1 Configuration, SFDR = 52 dBc
    • 1:1 Configuration, SFDR = 73 dBc

    Output Frequency = 2.5 MHz

    • 4:1 Configuration, SFDR = 76.5 dBc
    • 1:1 Configuration, SFDR = 82 dBc

    Output Frequency = 5.0 MHz

    • 4:1 Configuration, SFDR = 85 dBc
    • 1:1 Configuration, SFDR = 88 dBc

    The behavior of HD3 is consistent between our board and the demo board; however our performance is better across the board. 

    I can have plots ready tomorrow.


  • 0 in reply to Luke Stevens
    TI__Guru* 83185 points

    Case update and summary:

    The rise in HD3 distortion level at low frequency spectrum (i.e. 1.25MHz) is due to the transformer bandwidth and also the spectrum analyzer input bandwidth limitation.

    When the customer compare the result between their PCB board and the DAC3283 EVM, they noticed that the DAC3283EVM yields worse performance than their board. The DAC3283EVM default transformer is ADT4-1T, which covers the bandwidth of 9MHz to 625MHz. This transformer increases the HD3 distortion at low frequency below 9MHz due to the impedance variation of the transformer at this region. After changing the transformer to ADT4-6T with range of 60kHz to 300MHz, the HD3 distortion level improved to 90+dBc level.

    When measuring the DAC distortion at low frequency range, the spectrum analyzer's input structure is important. The customer was able to overcome the measurement limitation on the spectrum analyzer side by setting their spectrum analyzer with DC coupling setting. Further adjustments of DC coupling caps were made to improve the measurement result. 

    TI lab confirms the result with R&S FSEB30 spectrum analyzer with default DC coupled analyzer input. The distortion performance of the analyzer itself is confirmed by injecting a high quality 1MHz source from the HP8644B signal generator with 1MHz BPF. HD2 and HD3 distortions at 2MHz and 3MHz are in the 90+dBc range.

    TI will continue to work with the customer on this case offline to provide further support if needed.

    -Kang