About the SFDR and THD specification of AD3551R and LTC1668

Hi Benuyan,

We have not concluded the evaluation of the AD3551R above 100 kHz. The AD3551R is capable of generating tones up to half the update rate but overall performance depends on the amplifier, filter and configuration of the DAC.

From the amplifiers we have tested, the AD8065 would be close to meeting the 90 dBc THD requirement at 1 MHz. We are still testing other options aiming at improving DC and AC performance.

The main difference in performance between AD3551R and LTC1668 is the noise floor: 20 pA/√Hz vs 50 pA/√Hz. Our preliminary measurements show that the AD3551R would be on par with the LTC1668 in terms of THD at 1 MHz. However there are several adjustments we are testing to improve performance. We aim at publishing an application note with recommendations when the work is concluded.

Best regards,

Antonio

Hi, Will the reference design of LTC2757/LTC2758 using composite amplifier suit for 1MHz, >90dB THD design for this AD3551R chip?

I don't think so. The amplifiers in that design are way too slow, so the output would just oscillate trying to follow the DAC. The topology is probably not adequate since Iout has to dump current into a virtual ground at 2.5V.

I would recommend a two-stage composite amplifier with ADA4807 as first stage and LT1363 as second stage.

I don't think so. The amplifiers in that design are way too slow, so the output would just oscillate trying to follow the DAC. The topology is probably not adequate since Iout has to dump current into a virtual ground at 2.5V.

I would recommend a two-stage composite amplifier with ADA4807 as first stage and LT1363 as second stage.

Thank you for your support! I compare this two chips, ADA4807 and LT1363,

https://www.analog.com/cn/parametricsearch/11091#/sort=2687,asc&p4502=50M|4G&d=5047|4501|4089|4095|2687|4502|300|4101|4102|4124|4108|2839|2840|s3|s16|s5&p2687=1&sel=row144|row219

ADA4807 is superior to all specifications of LT1363 except for the Vs Span MAX. Could we use ADA4807 alone?

The output of AD3551R is used to drive a Howland VCCS current source injecting ~1mA current in to biological tissues. So the output of AD3551R should be precision, and the span of voltage output can be left to the enhanced howland current source.

How about LTC6226/LTC6227?

I am a little bit puzzled on the selection of DAC output driver buffers for AD3551R, may be I should stick with your upcoming reference designs which features a low distortion solution. Or, could you tell me some design considerations in selecting the right buffers for AD35531R?

The frequency range is 1KHz~1MHz, so maybe some high-end audio amplifiers could do this?

Yes, you may use ADA4807 or ADA4805 if the voltage span is less than 10V. Note that current sourcing / sinking capability is exhausted as the output voltage gets close to the rails. It should be no problem if the Howland amplifier has high input impedance.

The LTC6226 is a good candidate as well; it has enough slew rate but perhaps too much bandwidth. The only limitation is that it cannot support the +/- 10V range.

GBWP and Slew Rate are relevant if you plan on using the DAC to generate fast settling pulses. Working with harmonic signals you should focus on closed-loop gain and THD. You just need enough bandwidth to play the highest tone with suitable amplitude and phase deviation.

With 20 pA/√Hz DAC current noise, you get 30 nV/√Hz with RFB1 and 60 nV/√Hz with RFB2, so don't worry too much about the amplifier intrinsic noise.