TI's new ADS127L21 could be the next best precision ADC after ADI's AD4x30-24

[laichh]

Well, the typical INL of 0.4 ppm of FS not is really impressive in shine of ADI AD4x30-24 (0.1 ppm of FS typical), but at a fraction of what the ADI cost.



https://www.ti.com/product/ADS127L21

[ch_scr]

Looking at the datasheets, the zero / offset error vs temperature is very close between the two (looming physical limits probably) at typical 50nV/°C vs 35nV/°C for the AD4030.
More interesting maybe the gain error vs temperature: typical 0.5ppm/°C (1ppm max) with the TI part vs. 0.025 ppm/°C for AD.
Haven't looked further yet, but for DC precision that's quite a difference in temperature sensitivity there?

[miro123]

General - My observation for that past 5..7 years.
1. Nice to see development in this area.  - performance from major manufacture went up.
2. The two major manufacture have chosen different way - Analog/LT choose performance while TI - versatility and easy integration.
3. High performance components becomes difficult to handle for home hobbyist. The modern components are approaching the physicals limits. New package is needed
4. The border between different adc architecture is narrowed. - e.g. ads127l21 is multibit 3-th order sigma delta while AD4x30 got the oversampling principe from SD converters.
On topic
The two components are not comparable
  1. different architecture
  2. different performance parameters.
  3. different features for different market.
  4. Number A from manufacture A is not equal to number A from manufacture B . different manufacture measure the parameters on different way and conditions.
ADS12721 can be compared with already great AD7177-2

Personal thoughts
I don't care about gain error temperature drift r  typical 0.5ppm/°C (1ppm max) with the TI part vs. 0.025 ppm/°C for AD. I cannot design analogue frontend with lower than 0.5ppm/K drift.

[MasterT]

Personal thoughts
I don't care about gain error temperature drift r  typical 0.5ppm/°C (1ppm max) with the TI part vs. 0.025 ppm/°C for AD. I cannot design analogue frontend with lower than 0.5ppm/K drift.

 I think the same, both ADC unbuffered, so all those numbers INL, uV bla-bla has no practical meanings w/o analog front-end.

Get ad127-l-11 for evaluation,  now can't find OPA to interface w/o significant lose SNR or linearity or both. Even passive components seems play a big role.

 

[ch_scr]

(...)
Personal thoughts
I don't care about gain error temperature drift r  typical 0.5ppm/°C (1ppm max) with the TI part vs. 0.025 ppm/°C for AD. I cannot design analogue frontend with lower than 0.5ppm/K drift.

Sure, but if you do that, would you rather take an additional 0.5ppm/°C hit or an insignificant 0.025ppm/°C one? Even if you have 5ppm/°C frontend, an additional 10-20% hit is noticeable.
In general, to get any of these numbers, the rest of the system has to support it, quite the challenge no doubt!

[laichh]

Get ad127-l-11 for evaluation,  now can't find OPA to interface w/o significant lose SNR or linearity or both. Even passive components seems play a big role.

Try THS2630 + LT5400



https://www.ti.com/product/THS2630




https://www.analog.com/en/products/lt5400.html

[MasterT]

THS2630 is wrong choise:

...
Rail-to-rail   No

Regarding recommended by TI THS4551 (page 100 Figure 9-8. ADS127L21 Circuit Diagram) I see in DS
Input bias current drift (5)   3.3 (typ) nA/C, than alone produce x 1.5k = ~5 uV/C. And it's just "basic buffer" with input impedance 1.5k, to get it up at least to 10 MOhms,  5uV likely jump to 50

[iMo]

In past I did a simulation with a simple bootstrapped opamp (2 opapmps + 2 transistors + 2 zeners) with following 10:1 divider (LT5400 9+9+1+1k) feeding a differential ADC input. You would need a floating +15/-15V source. The simulation showed it could be applicable for ~7digits.
https://www.eevblog.com/forum/projects/simple-dc-afe-for-adc-chips-with-unipolar-diff-inputs/msg2143462/#msg2143462

[exe]

Only one channel