DAC8775: Calibration step

Hi Sirui,

Joe will review this and respond when he is back in the office after the US holiday.

Hi Paul,

Do you have any update?

BR

Sirui

Sirui,

It would be very helpful to have the reference voltage measurement along with this drift in the voltage. This way we could separate the reference error with the gain error of the DAC. Is the reference voltage the internal reference or an external reference?

Also, for these five cases that are shown, one seems to be for the EVM. What are the other four cases? It would be useful to describe what each of these cases are.

Hi Joe,

Thanks for the reply, and I will ask customer to test along with the reference.

In the curve:

• Blue: No calibration
• Red: first offset calibration then gain calibration
• Green: only offset calibration
• Purple: Only gain calibration
• Cyan: EVM

No matter which way they calibrate first, -10V and +10V point error don't change.

Sirui,

If you calculate the equivalent gain error (starting with the no-calibration data), the gain error is about 0.0185V out of 10V. This is a gain error of about 0.18%. In the specification for the DAC8775, the voltage output gain error should be within ±0.1%.

However, if they use the internal reference, the reference error is ±10mV on a 5V reference. This would be an additional error of ±0.2% that is not included in the gain error specification.

If they use the calibration to remove the reference error, I think that would bring the DAC gain error specification back to the 0.1% level.

One other thing that I would note is that both the Red and Purple lines seem to have a large offset error and still have a large gain error. How do they perform the calibration? Do they start with an offset calibration first or a gain calibration first?

Joseph Wu

Hi Joe,

Customer start with offset calibration on 0V, then gain calibration. I have one more question: is this DAC has non-linear region near 0 and full scale region? Could we avoid this by extending full scale range by 20%?

BR

Sirui

Sirui,

Ok, I'll need to take a quick look at the calibration order.

Because the output is bipolar instead of unipolar, the output will not have a zero-code error. A zero-code error comes from an output buffer not being able to drive the output to directly to 0V. However, that is generally not a problem with a bipolar output where the output buffer supply is larger than the output of the DAC buffer.

The one issue that may occur is at the output full scale or output negative scale. For the DAC8775, if the reference voltage is smaller than 5V then the output won't reach 10V or -10 when set to the positive or negative full scale. For example, if the reference is 4.99V, then the maximum output at full scale and negative full scale would be 9.98V and -9.98V.

Joseph Wu

Sirui,

And yes you can increase the range by 20% and scale back with calibration. 

Joseph Wu

Sirui,

For the calibration, it looks like they are running the offset calibration when the DAC code is 0000h instead of when the output is 0V. Can you just make sure they run the offset calibration when the DAC code is 8000h (and the output is 0V)?

After that, the only thing they need to verify is the reference voltage. I think that will likely be the largest source of the final error.

Joseph Wu

Hi Joe,

I'm waiting for customer's test result, and will get you updated once I have it.

You mentioned about unipolar range, and how do we solve the zero-code error? I have another customer using DAC8565 and output 14mV at 0000h DAC code. In the datasheet the amplifier is clamed to be rail to rail.

BR

Sirui

Sirui,

Even with a "rail-to-rail" amplifier, the output does not reach either rail because the output transistors cannot drive all the way to the rails. Depending on the opamp and the output load, the output will just get close to the rail.

To get to exactly to 0V for a unipolar output, there are only a couple of solutions. On solution would be to use a DAC that uses unipolar mode, but has a output buffer that has a bipolar supply. I believe that the DAC8775 does this already. If you use the DAC8775 in 0-10V range, and have the buck-boost set to ±15V, the DAC output should go to 0V. The DAC8760 is another similar device that supports bipolar supplies and can output a unipolar DAC voltage. Another solution would be to build some sort of level shifting circuit with an opamp and bipolar supplies. This would also require some gain adjustment because the DAC itself still has a unipolar supply.

Joseph Wu