DAC5687 spurious performance issue

Mikhail Matusov

Other Parts Discussed in Thread: DAC5687, DAC3283

Hello,

We have a problem with a batch of cards using DAC5687. They don't meet our spurious performance spec with the output signal at -40 dBm (which roughly corresponds to -40 dBFS). The problem seems to be with the DAC itself, which demonstrates poor SFDR performance under certain conditions. The spurs we see don't seem to be the ones predicted in the datasheet. I have attached 4 plots demonstating the problem. They are taken at two frequencies 1 MHz apart and at two different output levels. The strongest spurs we see all seem to be aliases of odd harmonics as follows:

Fs - 3*f
Fs - 5*f
2*Fs - 7*f

The DAC is run in the interpolation mode:

Fs = 280.6272 MHz. Intepolation factor is set to 4 and modulation to Fs/4. The fine tuning NCO is not used. The PLL is not used either. Input data is complex non-interleaved, supplied by GC5016.

Any comments will be highly appreciated.

Thanks,

Mikhail

over 14 years ago

0 Kang Hsia over 14 years ago

TI__Guru* 78424 points

Hello,

I see the observation here is that as the fundamental power decreased from -20dBFS to -40dBFS, the actual 3rd and 5th order distortion power do not decrease proportionally. Since the distortion powers do not decrease, the actual SFDR performance in dBc worsens.

I was able to duplicate your setup in the lab and saw the same thing. According to design, this observation is within expectation. The reason being the DAC is not an entire linear circuit when compared to an amplifier. For the DAC, even though the fundamental power drops to very low power, there are still bits switching to create the waveform, and the distortions of these switches would remain.

Therefore, in your setup, even though the power has dropped to certain level, the inherent current steering switches still causes the remaining distortion.

-KH

0 Mikhail Matusov over 14 years ago in reply to Kang Hsia

Intellectual 350 points

Kang,

Thank you for your reply. I didn't expect the SFDR to stay the same as the carrier drops, but expected a little better. What's the expected variation between different parts/batches? Some of our boards show a few dB better performance than others. Also, do you think there might be something we can do to get a couple of dBs better results?

/Mikhail

0 Kang Hsia over 14 years ago in reply to Mikhail Matusov

TI__Guru* 78424 points

Hello Mikhail,

The few dB variation in performance is reasonable variation between parts/batches. Since these are distortions at the DAC's analog output, the DAC's setting may not be able to improve the performance any better. We do have other DACs such as the DAC3283 that are designed for higher linearity and better performance.

-KH

0 Mikhail Matusov over 14 years ago in reply to Kang Hsia

Intellectual 350 points

Hello Kang,

Would it be possible for you to repeat this experiment with the DAC3283 and post the plots here?

Thanks,

/Mikhail

0 Kang Hsia over 14 years ago in reply to Mikhail Matusov

TI__Guru* 78424 points

Hi Mikhail,

Please see attached measurement and the DAC3283 EVM setup. As expected, the performance of the DAC3283 is better than the DAC5687. Keep in mind this is a typical performance measured at 65Mhz IF at room temperature.

-KH

DAC3283 SFDR.ppt

0 Kang Hsia over 14 years ago in reply to Kang Hsia

TI__Guru* 78424 points

DAC3283 EVM configuration file attached.

0 Mikhail Matusov over 14 years ago in reply to Kang Hsia

Intellectual 350 points

Hi Kang,

Thanks a lot for your help!

/Mikhail

Data converters

Data converters forum

DAC5687 spurious performance issue