I would like to know about AD628's Gain error.
D/S revG P3 say Gain Error is +/- 0.1%
Figure14 says max about 10ppm.
These Gain Errors are same meaning?
This spec has a big margin ?
I heard back from our test department. It turns out that the test in figure 14 is measured with the second amplifier A2 in a gain of 10 so the overall gain of the AD628 is 1. They calibrate the gain of 10 stage so the measured error is caused by the integrated resistors around A1.
However, the person who made figure 14 made a math error when they were converting from percent to ppm, so it is off by a factor of 100. Therefore, instead of the numbers being about 5ppm in the figure, they should have been about 500ppm (which is 0.05%).
Thanks for finding this error for us.
Trust the ±0.1% guaranteed number in the spec table on page 3. Graphs in the Typical Performance Characteristics section of the datasheet were typical at the time of initial characterization and those distributions can shift over time with process shifts, etc. Guaranteed limits are valid for every part that comes out of the factory.
I don't have access to the characterization data for this product, so I can't tell you for sure what figure 14 is referring to. I have asked one of our characterization engineers for help with this. Because AD628 has a default gain of 0.1 and figure 14 says "+1 Gain Error," I suspect that figure 14 is measuring only the gain error of output amplifier A2 as a voltage follower, ignoring errors from A1 and resistors.
Thank you very much for your prompt support.
I'm waiting for the information about figure 14.
I'm sorry that I couldn't reply soon.
Gain error spec "0.01% typ" is for A1.
Figure 14 is total gain error (A1X0.1 and A2X1000 ) as Gain=100, that is 0.05% typ.
Thank you very much for you support.
Sorry if I wasn't clear enough the first time. You're exactly right about the Gain Error in the specifications table.
But Figure 14 is in a gain of +1 (A1 is x0.1 and A2 is x10). The Figure 14 distribution is centered at about 0.05%, but in cases like this we can adjust the trim program until the distribution is centered at zero, which would bring the ±1 sigma typical to about 0.01%.
In all cases, we guarantee that the Gain Error from A1 will be less than 0.1%, and every part tested in Figure 14 meets this specification. Designs and error budgets should not rely on the typical numbers, as typicals can shift over time and process variations.
Sorry for my misreading.
Fig 14 is at Gain 1 (A1 is x0.1 and A2 is x10). but scale is 100ppm but not ppm.
A1 gain error is guaranteed under 0.1%.
A total gain error will be controlled within 0.1%.
Thank you very much for your nice support.
Retrieving data ...