I have been looking for some time for a good design for a ~10MHz sine wave generator, with variable frequency from about 5-20MHz. The max required peak to peak amplitude is 10V. Many suggestions on the internet include using DDS such as AD9833, however, that feels like overkill for what I need. I have also seen a similar question to this, however, their application requires an extremely wideband (1-200MHz).

I am using this as a Local Oscillator, which I am mixing with an error signal.

Preferably, I would like to use a simple, cheap circuit design involving op-amps, with <5% distortion. However if there is a better design, I am all ears.

  • \$\begingroup\$ AD9833 can only go up to 12.5MHz, so that's not really overkill. In fact, it's less than what you describe you need. 20MHz is getting awfully high, so when you find a design make sure to be very aware of RF artifacts such as parasitic capacitance and inductance around your board. \$\endgroup\$
    – DSWG
    Commented Sep 11, 2018 at 0:24
  • 1
    \$\begingroup\$ Thanks for your advice - I meant overkill in the sense that I would need to program and control it from a PC, whereas for my application a passive sine wave generator would be more suitable. \$\endgroup\$ Commented Sep 11, 2018 at 0:58
  • \$\begingroup\$ No PC needed. A microcontroller is all you would need, and there are pre-existing libraries available for Arduino specifically for it. Super easy. But yeah - your call, not mine. \$\endgroup\$
    – DSWG
    Commented Sep 11, 2018 at 1:34
  • 2
    \$\begingroup\$ Your specs can be expanded please, I think you want a VCO and does the 3rd harmonic content cause that much phase error, if the harmonics are filtered out in the mixer? It may be smarter to use a PLL chip. Pls improve specs and system requirements. \$\endgroup\$ Commented Sep 11, 2018 at 2:31
  • 1
    \$\begingroup\$ What tuning resolution? What jitter or phase noise? What harmonics can you tolerate? What about residual AM? Residual PM (aka phase noise)? \$\endgroup\$ Commented Sep 11, 2018 at 3:15

1 Answer 1


enter image description here

@rocketracket The frequency will change either by shunt attenuating or Series R. both resulting in attenuation and frequency shift. But the FET AC drop distorts low freq. sine. while the Pot does not.

I still don't know what this is for and it is better with a PLL. If you value your time. Contact me.

For 1~200MHz, custom video design with current control is needed.

The 199 vs 200 difference inversely controls the Q and so 198 is faster but starts to clip. This was designed for 5V R-R output. The Cap can be tied to 5V or 10V. This gives a step pulse to one CAP to kick start Oscillator otherwise it takes a long time to grow to full swing.

This is not a custom design service site, but this design is custom from me.

Obvious choices are Op Amp or video amp with BW > 30MHz. e.g. LT6236CS6IC OPAMP GP 215MHZ RRO TSOT23-6

It regulates by going into saturation which reduces the gain to zero. Since gain or Pot R controls frequency, and amplitude of the integrator, the output adjusts to peak to peak swing. Naturally, a quick change in R changes gain momentarily over a 4:1 range, but the sine Vpp regulates according to matched feedback with <1% R mismatch gain for loop bandwidth or Q or clipping tradeoffs.

It is not intended for rapid FM but for pure sine as the Q is quite high for pure sine.

The other approaches

  • Take any simple Relaxation triangle VCO and convert to sine , with bipolar triangle, using back-back LED or Zeners and R to regulate smooth linear to sine gain reduction curve after 2/3Vpp.
  • any VCO into a counter with R,2R,4R,8R DAC or use 1 or 2 Johnson Counters with sine function weighted resistors. Cheap and Dirty Compute R values from Sine in Excel.
  • reverse engineer the FM sine gen, in a old Wavetek , HP or sim. Sig Gen,

  • Use 400MHz VCO and down convert

  • 1
    \$\begingroup\$ Thanks again for your help. This worked very nicely for my use. \$\endgroup\$ Commented Sep 12, 2018 at 21:01
  • \$\begingroup\$ you are welcome , This is an old problem yet there may be better ways, yet no one had any better ideas or approved of this one. You can add complementary NPN/PNP inside an OpAmp feedback loop then add 50 ohms series too for 22Vpp from 24V then 12V bias \$\endgroup\$ Commented Sep 12, 2018 at 21:26
  • \$\begingroup\$ Another quick question - is there a way to easily change this to 10Vpp instead of 5V? I'm thinking of adding another simple amplifier after this to double the amplitude. p.s. in my design, I already have +/-15V and +/-5V rails used for other components, so using those voltages instead of +2.5V would be even better. \$\endgroup\$ Commented Sep 12, 2018 at 22:08
  • \$\begingroup\$ Then use +/-5 with Vref=0 to get 10Vpp and +5V on cap with RRO OA. It always works to max supply voltage of output with Rail to Rail Output types. Then Vref is always midpoint. Fast changes have need time to stabilize \$\endgroup\$ Commented Sep 13, 2018 at 10:32

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.