# AD9166 EVB output is nonlinearity more than dozens of LSBs

The configure of ACE is as follows:

Use DPG Downloader to send DC value，everytime I change the Value. I use Tek 6Series oscilloscope at 50Ω load and 20MHz Bandwidth.Averge mode(2048 times)，and the results are mean value of oscilloscope. Oscilloscope is working at 2mV/div.

The test results:

 value results/mV 255 55.876 256 55.837 257 55.692 258 55.425 511 50.286 512 50.176 513 50.201 767 44.746 768 44.699 769 44.552 1023 38.617 1024 38.473 1025 38.301 1026 38.191 1027 38.177 1028 38.141 1029 38.058 1030 37.951

By using matlab the plot is ：

It can be seen that there are large step on 0x100, 0x200, 0x300, 0x400. Why? I guess that in other value such as 0x1000, 0x1100, et . al will also exist large step.

Can anyone give me some help?

• Hello,

I do not see a reason for the discontinuity inside the AD9166. Are you sure this isn't due to measurement range change on the oscilloscope, during testing?

It can be seen that there are large step on 0x100, 0x200, 0x300, 0x400

The plot suggest discontinuity near 0x500 and again near 0x780. But not 0x100, 0x200, etc.

Can anyone give me some help?

what are you looking to do? Any device would always exhibit some amount of non-linearity.

What is the target frequency range, in the end application?

Landsman

• I will use DC coupled pulse to change frequency continuously. We expact the DC pulse (period=1us, pulse width=100ns) step is about increasing 10 LSB every time.

The figure X of plot is decimal number. On 0x100、0x200、0x300、0x400(Hex) will step more than 0.1mV which is fast lagrer than one LSB.

Use the same test condiction to test AD9736, which is  no such problem.

• Assumed answered offline.  Please let us know if you need further assistance.