We have not yet completed the comparison between the PGA-SAR system and the delta-sigma converter. In my last article (ADC Throughput Time: Comparison of SAR Converters and Delta-Sigma Converters), we compared the throughput times of the two systems. In the article, we concluded that the throughput times of the PGA-SAR system and the delta-sigma converter we studied were very close; 70 ksps and 24 ksps . In your email, you said that the difference is so small that it makes no difference. What is the best system? This evaluation looks like a tie, but what about accuracy?
 
You think of accuracy in terms of whether the system can (on average) produce the correct output value. You can best describe accuracy in terms of DC specifications such as offset, gain, linearity, etc. In this evaluation, you should use the minimum and/or maximum specifications of the system devices.
 
You can calculate the reference-to-input (RTI) offset error of the PGA-SAR using the following formula :
 
SYSTEM OFFSET-ERROR = PGA OFFSET-ERROR + (OPA OFFSET-ERROR + ADC OFFSET-ERROR )/ GAIN PGA
 
You can also use a similar method to calculate the RTI gain error and linearity error of the system.
 
In general, you can combine the uncorrelated DC errors, bias gain, offset, and linearity using the root sum square (RSS) calculation method. Note that you should attribute these errors to the system inputs. You can combine the DC errors to determine the total unadjusted error (TUE) of the system using the following formula:
 
TUE SYSTEM = (SYSTEM OFFSET-ERROR 2 + SYSTEM GAIN-ERROR 2 + SYSTEM LINEARITY-ERROR 2 )
 
In summary, the accuracy, or TUE, of the SAR-ADC gets worse as the gain increases. The significance of this may not be immediately apparent, but keep in mind that there are two factors at play. Increasing the PGA gain reduces the system input full-scale voltage range and the actual voltage LSB size for our 12 -bit system , while the absolute voltage error of the OPA and ADC decreases. Unfortunately, the PGA offset voltage error remains constant.
 
Let’s contrast this discussion with the TUE characteristics of delta-sigma converters. Many of you responded to emails during this EDN series with answers to the question I asked in my previous article, “Take a Risk and Get Rid of Those Bits!” (Which is best, PGA-SAR or delta-sigma converters?). Many readers felt that delta-sigma converters won this comparison.
 
Just like all other ADCs , delta-sigma converters produce offset, gain, and linearity DC errors. The key difference between process gain and delta-sigma converters for analog gain and PGA-SAR circuits is that offset errors are not multiplied by process gain. On the other hand, gain and linearity errors are inversely proportional to process gain increases. The end result is that TUE decreases as process gain increases. However, the FSR percentage TUE remains constant.
 
Sometimes, theory is hard to explain, but facts are easier to understand. To make the discussion easier, Table 1 lists the data of the top PGA-SAR system compared with the advanced delta-sigma converter.
 
Table 1 compares the baseline data with the TUE performance of the PGA-SAR system and the delta-sigma converter.
 
In Table 1 , the components of the PGA-SAR system are PGA116 (programmable gain amplifier), OPA351 (operational amplifier), and ADS7886 (12-bit ADC ). The ∆-∑ converter is ADS1158 (24-bit ADC).
 
In a specific accuracy evaluation, we found that neither system had 12 -bit accuracy level performance, but the PGA-SAR combination was generally more accurate than the delta-sigma converter. I doubt you found two systems with comparable throughput where the PGA-SAR system was generally less accurate than the delta-sigma converter over its full gain range . Check out the TI E2E TM community for other data converter discussions: http://e2e.ti.com/cn/forums/default.aspx?GroupID=10 .