DAC8871: Is this a Major Carry glitch?

Ted H

Part Number: DAC8871
Other Parts Discussed in Thread: OPA4277, OPA2277

I am witnessing an interesting phenomenon on the output of the DAC8871. I am aware of the major carry glitch phenomenon, but what I am seeing far exceeds what is characterized in the datasheet and seems to behave differently. I am by no means pushing the envelope of speed for this device. The clock rate is 1.68 MHz. VREFH=+10V. VREFL=-10V. Referring to the attached images, there are several items of note:

1. On a modest 0.156V step (128 LSB's) in output voltage, the 'glitch' swings roughly 7V (+5.5V, -1.5V). See IMAGE_01.

2. Rather than normal ringing or even a capacitive voltage swing, the 'glitch appears to ramp linearly for the majority of its duration. See IMAGE_02.

3. The duration of the 'glitch' lasts through 11 clock cycles (~6.6 usec). See IMAGE_03.

4. On a ramp sequence between the reference voltages the 'glitches' appear at regular intervals corresponding to code changes of 16 (4 bits). See IMAGE_04.

5. Most interesting of all is that the magnitude of these 'glitches' is far greater when the code is ramping down then when ramping up. See IMAGE_05.

IMAGE_01:

IMAGE_02:

IMAGE_03:

IMAGE_04:

IMAGE_05:

over 7 years ago

0 Paul_Frost over 7 years ago

TI__Guru 53673 points

Hi Ted,

This is very strange behavior. I suspect that this is some issue with the reference, causing the reference to change when the code is changing. Do you think you could share your schematic? A useful observation would be monitor the VREF+ and VREF- during a glitch event as well.

Let me know,
Paul

0 Ted H over 7 years ago in reply to Paul_Frost

Intellectual 275 points

Let me know where I can email my schematic and I will share it with you.

0 Paul_Frost over 7 years ago in reply to Ted H

TI__Guru 53673 points

I do not see any obvious issue with the schematic. Let me know when you get a chance to measure the references. In addition, can you measure the response on the output of the DAC directly, as well as the output of the DAC's buffer?

0 Ted H over 7 years ago in reply to Paul_Frost

Intellectual 275 points

VREFH looks relatively clean.

VREFL exhibits similar behavior to the DAC output. The amplitude of the glitch appears greater on the VREFL pin.

Here is the DAC8871 output vs VREFL.

Here is the buffer output vs VREFL

0 Paul_Frost over 7 years ago in reply to Ted H

TI__Guru 53673 points

Hi Ted,

I think you have a stability problem with the AMP03. While filtering the reference input may have some improvement, I think implementing a unity gain buffer for the reference inputs is a better solution. Adding buffers for both the VREFH and VREFL would allow you to use the VREF sense pins, as well as adding a stabilization capacitor to the outputs. Take a look at this TI Design for some ideas on the buffer configuration.

As far as the cause of the glitches, I think you look at the DAC8871 PDS, Figure 27, you will see that from code to code there can be changes in current of a few hundred micro-amps. I think it is these dynamic changes that the AMP03 is struggling to support.

Let me know if this helps.

Thanks!

0 Ted H over 7 years ago in reply to Paul_Frost

Intellectual 275 points

Thanks for the feedback. Looking at the TI Design you referenced above, I have a comment and a question.
I presume the non-inverting input on U7B is tied to ground.

If I implement the reference topology of figure 3, is there any harm in adding an RC LPF on the VREFx FORCE input as shown in Figure 40 of the DAC8871 datasheet?

0 Paul_Frost over 7 years ago in reply to Ted H

TI__Guru 53673 points

Hi Ted,

Yes, the non-inverting input should be grounded. You should be okay to put the same filter for the topology in figure 3. You should also be ok omitting U7A in this design, as your reference should be okay without it.

Thanks!

0 Ted H over 7 years ago in reply to Paul_Frost

Intellectual 275 points

Is there any issue with using a single OPA4277 IC rather than two OPA2277 ICs for the VREFx input circuit?

0 Paul_Frost over 7 years ago in reply to Ted H

TI__Guru 53673 points

No, I think that would be a good solution.

0 Ted H over 7 years ago in reply to Paul_Frost

Intellectual 275 points

Great. Thank you for your assistance.

Data converters

Data converters forum

DAC8871: Is this a Major Carry glitch?