Hello,
I should adjust the output voltage of my power supply very fast - the transient should settle down within 1us! The VID control of LM10011 is what I want but it looks this device has quite a delay as below. Could you suggest any other DAC device that can support my requirement?
Thank you a lot!
Hi Katlynne,
Thank you for suggestion. It looks DAC53204 has quite a long settling time when it's compared to my customer's requirement.
1.If this long settling time comes from output buffer, do you think that an unbuffered DAC such as DAC8830 can implement the power supply with <1us of transient time if it's accompanied with an external high speed amp?
2.I'm looking at OPA855 (GBW 8000MHz) or BUF802 (GBW 3100MHz). It looks BUF802 has additional features and I do not know the exact target application for this device. Could you suggest a more suitable device for the output buffer?
3. In addition, could you advise for how to estimate the proper GBW for the given settling time? For example, if I need 10us of DAC output settling time, how much of GBW of output buffer OPAMP should be guaranteed?
Hi Ella,
If the customer truly needs less than one us then I have some additional comments. The settling time of the DAC8830 is given in the datasheet as 1us to settle within 1/2 LSB. 1//2 LSB would be 36uV for a 5V reference. If their requirements for settling are not that strict then the settling time would end up being lower. There is also the time required to update the DAC. It takes 16 clocks to update the output, so if the max clock speed of 50MHz is used then it would take around 300ns to update the output. Then the settling time of the high speed amp would come into play.
I can loop in our high speed DAC team for some more recommendations if you are worried that the above solution would not meet the customer's requirements.
It looks like the BUF802 has a high offset voltage which wouldn't be ideal for this application. The offset voltage of the OPA855 is also large compared to the step size of the DAC8830 (16-bits). If the customer does not need high resolution then this is not an issue. The LM10001 that you mentioned before only had 6-bits of resolution.
For large signal steps (100mV or larger) you would want to look at the slew rate of the op amp. The slew rate is usually specified from the time it takes from 10% to 90% of the final value. The settling time would be on top of that. The OPA855 shows the full transient response for a 2V step (slew rate + settling):
The GBW comes into play for small signal steps.
GBW is the gain bandwidth product. The resulting bandwidth (BW) is GBW/Gain:
You can see in the plot that the that the bandwidth is reduced to ~400MHz when the gain is 20V/V. For this application the customer would likely use a gain of 1, so the bandwidth would be the full 8GHz. The small signal rise time can be estimated as (0.35)/BW. For a 1us rise time and a gain of 1, you'd need a GBW of about 350kHz. You should add plenty of error to this, but this amp is likely much higher than the customer would need. 1MHz is probably enough, but you also need to make sure the settling time of the op amp is also low so you are well within the customer requirements. The OPA189 looks like a good choice. It has low offset voltage and has fast transient response times. Let me know if you have additional questions.
Best,
Katlynne
