NI PXIe-4467, ADC INL issue - site to site deviation.

[Venkat056]

Hi Guys,

I am currently testing a 62.5uV LSB ADC and facing issues with the INL parameter. During debugging, I observed the following:

Test Setup:

1. The hardware is configured for multisite testing:

   • Site0: AO0 (4467)
   • Site1: AO1 (4467)

2. A 200mHz ramp is being forced to the DUT from the AO positive terminal for INL testing.

3. The 4467 is set to the 10V range.

4. The 4467 is operating in pseudo-differential mode. The AO negative terminal is left open, as it is internally connected to GND through a 50-ohm resistor.

5. An amplifier is present in the input path.

Issues Observed:

1. Ramp Voltage < 5V (Amplifier acts as a buffer):

   • Site-to-site deviation: 0.2 LSB

2. Ramp Voltage between 5V and 10V (Amplifier gain decreases):

   • Site-to-site deviation: 0.5 LSB

Observations:

1. The same behavior is observed across three different PCB boards with identical circuitry.

2. Swapping the 4467 channels (AO1 to Site0, AO0 to Site1) yields similar results, indicating no hardware-related issue.

3. The test program has been verified and is functioning correctly.

Suspect:

• The instrument appears to be the main suspect. Testing with another 4467 unit produced similar results.

Queries:

1. Could the way we are handling the instrument be causing this issue, or is there another factor?

2. Why is a 0.5 LSB deviation observed at 10V, while only 0.2 LSB is seen at 5V?

3. My requirement is to achieve a site-to-site deviation of less than 10 milli-LSB. Is this achievable?

Thank you,

Venkat.

[santo_13]

---

@Venkat056 wrote:

1. An amplifier is present in the input path.

---

Why are you not suspecting the amplifier inducing the error since it is uncalibrated and in the signal path?

Please share a simplified electrical signal path diagram of the 4467 and DUT.

[Venkat056]

Hi Santhosh,

Thanks for your reply.

Why are you not suspecting the amplifier inducing the error since it is uncalibrated and in the signal path?

The AO negative of 4467 connected to gnd and observed there is only 0.2LSB deviation ( without gnd as mentioned earlier 0.5LSB observed) and the both sites input is calibrated by program.

Please share a simplified electrical signal path diagram of the 4467 and DUT.

Thank you,

Venkat

[santo_13]

---

@Venkat056 wrote:

Hi Santhosh,

Thanks for your reply.

Why are you not suspecting the amplifier inducing the error since it is uncalibrated and in the signal path?

The AO negative of 4467 connected to gnd and observed there is only 0.2LSB deviation ( without gnd as mentioned earlier 0.5LSB observed) and the both sites input is calibrated by program.

Please share a simplified electrical signal path diagram of the 4467 and DUT.

Thank you,

Venkat

---

Is there a reason you still choose to leave the AO- open on the load board side instead of grounding it?

Analog Output - NI

Since your amplifier is ground referenced, it is recommended to use differential to avoid ground loops.

What happens when you bypass your on-board circuit and connect 4467 direct to the DUT?

Additional point to note is that Delta-sigma converters are not the best choice to generate ramp signals due to the filters at the front end.

Why not use the pure-tone feature of 4467 to estimate INL and DNL using Histogram technique?

[Venkat056]

Hi Santhosh,

Is there a reason you still choose to leave the AO- open on the load board side instead of grounding it?

My PCB does not have a provision to connect to GND. The only way to connect a ground is by using an external wire from AO negative to GND.

What happens when you bypass your on-board circuit and connect 4467 direct to the DUT?

I can apply a maximum of 5V to the DUT, which aligns with its specified voltage limit. I observed a site-to-site deviation of 0.2 LSB, consistent with the behavior seen in the on-board circuit.

Why not use the pure-tone feature of 4467 to estimate INL and DNL using Histogram technique?

As per requirements, INL must be characterized using the ramp method only.

Thank you,

Venkat.