Fluke 287 and 289  with a Σ-△ ADC rather than Integrated ADC?

Higher modules with RMS  feature Fluke exclusively cusommized SL10327 with △-Σ ADC such as LTC2415, LTC2435, etc. Now I wonder cheap ones like Fluke 15B and 17B applied double integral DMM chips.

Many reckon that △-ΣADC has a weak linear performance, not suitable for high ends, while there is a piece of 24-bit △-Σ ADC LTC2440 inside a Fluke 8808A five and half. I found no signs of integrated ADC from Agilent 34410, nor from Agilent AD637, rather I found a Xilinx XC3S200 FPGA, an 8-bit 80Msps AD9283- 80 and a 10-bit 20Msps AD9200.

From Fluke 289 teardown photos we can find LTC2415, 24 bit △-Σ ADC. So are we seeing a trend of Delta-Sigma replacing the dual-slope ADC? Well, quite likely if you please. Fluke 189 has a LTC2435 20 bit △-Σ ADC.

As the Agilent 34410A nowadays is geared with AD9283 and an 8bit 80MSPS 10bit 20MSPS AD9200, the performance is far higher than the previous generation 34401A:

In terms of DC performance changes: from 6 readings / second to 1,000 on six and half; from 0 readings / second to 10,000 on five and half; from 1,000 readings / second to 50,000 on four and half.

In terms of AC performance change: From 50 readings / second to 500 on six and half; In terms of trigger response: From <1 m sec to <1 u sec; and the trigger jitter: From <500 u sec to <1 u sec.

However according to datasheet of FS9721 that three pins like FTA, FTB, and FTC are obviously Σ-△-type only. Some companies simply started a business selling half billion electrical weighing equipments with Σ-△ADC inside. As for  double intergrated, even solutions for 40000 counts and 80000 counts are using it. Do notice that Fluke sells way less than several million DMMs--almost 10% of the latter solution sales.

It’d be a dream for double intergrated ADC to make such a fast reading in 34410A. The charge balance ADC of 34401A has always been between the ADC and Σ-△ things. And I reckon (SAR)ADS in 80196 is nothing but an auxiliary to make it up for the speed limit.

Thus finally I conclude the linear of Σ-△ is more than okey considering current double integral chip is much worse, even with a high-performance integration capacitor is way far from enough. Although we have to admit in most cases a three and half handles 90% of daily applications, thus such pursuit of precision would be a luxury, isn’t it?