I have been tried to investigate and fix this problem by myself. I however cannot find the solution. Please give me some advises.

I used 3 traditional oscillator circuits with opamps (Wien bridge, phase-shift, and twin-T circuits). All of them work correctly when I measure with oscilloscope. I get the sinusoidal waves and can adjust their frequencies and gains. Everything is fine.

The problem is when I use spectrum analyzer (Rigol DS815). After the probe touches the circuit, my sinusoid is ruined (The harmonics rise up). I saw the change in oscilloscope at the same time too. Sometimes the signal disappears and I have to adjust gain to take it back.

In wien bridge, I varied frequencies from 500Hz to 300kHz and Tested with UA741, TL082. Still cannot work it out.

I am thinking that the problem is not about the oscillators, but how I use spectrum analyzer to probe it since the oscilloscope works fine.

  • \$\begingroup\$ Spectrum analyzers have 50 ohm input impedance which severely loads your DUT. These opamp are not meant to drive such a low impedance. \$\endgroup\$ – Mike Feb 23 '18 at 20:00
  • \$\begingroup\$ Can I find an opamp which can drive this or I have to do something with impedance. \$\endgroup\$ – pakornosky Feb 23 '18 at 20:12

Your spectrum analyzer has a 50 Ω input impedance, so it is loading the circuit and changing its behavior, unlike the oscilloscope probe which has a very high input impedance (intended to minimize disturbance to a complete, existing circuit no matter where you probe it).

To get consistent results, you should design your oscillator assuming a 50 Ω load, and test it in the presence of that load.

This may require placing an amplifier before the output, to produce a sufficient signal level at that impedance, or it may be only a matching network. An advantage of using an amplifier — called a buffer in this role — is that it also prevents the circuit from being affected by mismatches as much, because the the input to the amplifier is (approximately) unaffected by the situation at its output.

Then you can connect any of

  • your spectrum analyzer,
  • a 50 Ω dummy load (non-inductive resistor), optionally with oscilloscope probe in parallel, or
  • an arbitrary length of 50 Ω coax connected to another device with 50 Ω input

to that 50 Ω output.

This is the conventional approach for modular RF design. The impedance does not have to be 50 Ω but that is conventional for RF test equipment, and it also needs to match the characteristic impedance of the coaxial cable you are using.

  • \$\begingroup\$ So I need an impedance matching part between my oscillator and spectrum analyzer to convert my impedance to 50 Ohm. Let me know if I misunderstand your advises. \$\endgroup\$ – pakornosky Feb 23 '18 at 20:09
  • \$\begingroup\$ Read the answer carefully. Kevin says, "This may require placing an amplifier before the output." \$\endgroup\$ – Transistor Feb 23 '18 at 20:20
  • 4
    \$\begingroup\$ Another approach is just to place a 450 ohm resistor in series with the output right at the osc, and mentally add 20dB to whatever the analyser reads (or add a suitable offset in the instrument settings). This makes the input 500 ohms and 20dB less sensitive, which is probably fine. Other values can be used (4k95 for example makes a 5k input impedance and 40dB of losses). \$\endgroup\$ – Dan Mills Feb 23 '18 at 20:25
  • \$\begingroup\$ Thank you everyone. You all are very helpful. One day I will master it like you guys :) \$\endgroup\$ – pakornosky Feb 23 '18 at 21:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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