I am trying to build a function generator for sinusoidal waveform with frequencies ranging from 1 Khz to 100 Mhz with amplitude of about 4 Vp-p to 10 Vp-p. I have been looking at AD9951 DDS that takes 400 Mhz clock input and provides a maximum output sinewave of 100 Mhz. The IC is powered by 1.8V and takes 3.3V only for DVDD_I/O (pin 43). The datasheet is not detailed enough to touch upon a few point I need clarification with.

  1. I am looking at a 3.3V, 400 Mhz oscillator (SG3225VAN 400.000000M-KEGA3). The OSC and complementary_OSC (pins 9 and 8) on AD9951 takes 1.8V, is it safe to use a voltage divider at the output of the 400 Mhz oscillator to bring down peak-to-peak voltage to about 1.8V. So using 220R and 68R brings the voltage down to 0.78Vpeak (0.78 x 2 = 1.56Vp-p). Is this approach correct or is there a better way to do this?
  2. Since AD9951 powers up at 1.8V, I assume, the DAC (pins 20,21) would also output close to 1.8V?
  3. As I need about 10Vp-p, how to efficiently amplify the ~1.8V DAC signal to 10Vp-p? I have been looking at RF transformers and dedicated ICs such as Mav-11+ which unfortunately cannot be shipped to Australia. Can I get a recommendation on how I can effectively amplify the sinusoidal signal at high frequency of 100 - 200 Mhz?

Edit The load is going to be simple electrical circuits like opamp analysis, transistor biasing and also providing a signal to biosensors and changed the maximum generated frequency to 100 Mhz.


1 Answer 1


First issue I see with your proposal is, the AD9951 is capable of 400 MSps. That means 200 MHz is just at the Nyquist frequency for this chip. You'll notice that no performance metrics are given in the datasheet for output frequencies above 160 MHz. If you want good performance (low harmonic content, etc.) you're probably better off finding a different chip with a higher sample rate, using some kind of mixing scheme to multiply up a lower frequency output from a DDS system to 200 MHz, or using an analog oscillator at 200 MHz.

Second issue, using a transformer to bring up the output amplitude from ~ 1.8 V (peak-peak) to 10 V means the DSP output has to provide all of the power to the load. If your load is 50 ohms, that means your DSP will will be looking into something that looks like ~10 ohms, which it's probably not designed to do. You'll want to use some kind of amplifier here instead of a transformer.

Since AD9951 powers up at 1.8V, I assume, the DAC (pins 20,21) would also output close to 1.8V?

See Figure 2 in the datasheet. The output is a differential current sink, and you "must terminate outputs to AVDD". On page 4 the full scale current range is given as between 5 and 15 mA. The voltage you get from that will depend on what resistance you pull up the outputs with.

This also means with your idea of using a transformer to step up the output voltage amplitude, you'll only be able to deliver ~1-3 mA into the 10 V load.

  • \$\begingroup\$ This is constrictive. I did not notice 160 MHz SFDR. But if I intend to design a system with say 150 MHz, that would be fine right? \$\endgroup\$ May 21, 2020 at 7:24
  • \$\begingroup\$ Can I pick your brain for the second point you made about 50 ohms. I know 50 ohm is the internal resistance of a function generator which is fixed but I do not intend to do any antenna impedance matching. A function generator is mostly used in high Z mode. With this in mind, the output impedance from an amplifier would also be high Z which shouldn't be a issue right? \$\endgroup\$ May 21, 2020 at 7:31
  • \$\begingroup\$ @DavidNorman, what is your actual load? Please edit your question to add this information. IME at 200 MHz, 50 ohm outputs are much more common than high impedance. \$\endgroup\$
    – The Photon
    May 21, 2020 at 13:44
  • \$\begingroup\$ load description added. \$\endgroup\$ May 21, 2020 at 18:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.