Other Parts Discussed in Thread: ADS6149
Having developed a working data acquisition board based on the ADS6149 chip, I'm trying to completely characterize its performance using both dynamic (FFT->SFDR/SNR) and static (histogram->INL/DNL) tests. So far, the performance is very near the datasheet level, however there are some issues that maybe someone can clarify:
1. For some reason I cannot get the ADC to output full-scale codes. If I overdrive the input, the output is a clipped sine with codes between about +/-8160 rather than +/-8192. Even with inputs as small as +0.5dBFS or as large as +5dBFS no codes exceeding 8163 were ever seen. The ramp test mode works perfectly, so this is not a signal integrity issue. Any ideas?
2. The built-in gain feature looks like it is a purely digital one. With gain of +6dB there are exactly 8192 missing codes (all even codes are missing), with gain of +0.5dB each 17th code is always missing. Also the SNR/SFDR measurements taken with gain enabled correspond to the behaviour of purely digital gain (SNR = SNR0-G, and the harmonics drop as expected with the drop in signal level and not more). Am I correct and the programmable gain is simply a digital multiplier?
3. When trying to measure INL/DNL with the sine histogram test, all is well until the converter starts clipping. Up to -0.1dBFS the measured INL/DNL are perfectly in-line with the datasheet values. When I overdrive the converter by as much as 1dB, at high input frequencies (above 30MHz) the measured INL looks very irregular with peaks of up to 20LSB. This is visible even on the histogram itself - some values (not necessarily near full scale) occur a lot often than others. Again, this occurs only after overdriving the chip, so i know it is not a software bug.
Really looking forward to some insights,
Greg
