I have been setting up a low distortion sine wave generator based on this design from LT (http://www.linear.com/solutions/1623):

enter image description here

For resistors R1 and R2 I am using a ganged pot so that both change together, however, I have found that when I change the resistance of my ganged pot the amplitude also changes which is not what I expected. Other simpler Wien bridge type circuits that I have seen use the same set up for changing the frequency so I'm a bit stumped as to why the amplitude changes in this circuit provided by LT. To make sure the ganged pot wasn't doing anything strange I replaced it with different values of fixed resistance and I still see the amplitude change with the frequency.

Can anyone offer some insight into why I'm seeing this behaviour?


  • \$\begingroup\$ are you using a frequency near the GBW limit of the amplifier? \$\endgroup\$ – Neil_UK Dec 12 '16 at 17:11
  • \$\begingroup\$ Is your actual implementation exactly as portrayed in the LT schematic? Have you got supply decoupling capacitors? Are values the same and does it use exactly the same devices as specified? What variation in amplitude are you seeing and what nominal amplitude are you seeing? What supply voltages are you using? Are your pots wired so that as one gets smaller in value the other gets bigger in value? \$\endgroup\$ – Andy aka Dec 12 '16 at 17:51
  • \$\begingroup\$ I would hope that I'm not at the GBW limit. I'm only wanting frequencies from about 500Hz up to 5kHz. \$\endgroup\$ – CakeOrDeath Dec 13 '16 at 9:51
  • \$\begingroup\$ Couple of changes in the circuit as I couldn't find some of the exact components. R1 and R2 are a ganged 10k pot (to give more frequency range), LT1004 has been changed to LT1634 (precision 1.25V shunt voltage), 2N4338 has been changed to PF5102 and at the moment I have a pot 5.6k resistor just so I can play around with the gain (although I have changed this to a fixed resistance and still see the same issue). Supply voltage is 15V. Pots are wired so that the third terminal on both is grounded so they should change together in the same direction. No decoupling caps yet but can try adding them. \$\endgroup\$ – CakeOrDeath Dec 13 '16 at 10:00
  • \$\begingroup\$ So what I see is that in the frequency range 500 - 2000Hz the Vpk value is around 2V (there is some slight drift of around +/-0.2V when changing the value of the dual ganged pot but I think this might be due to the fact that the pot isn't the best quality). Over 2kHz the Vpk starts to drop off and ends up at around 1V when I get to 5kHz. \$\endgroup\$ – CakeOrDeath Dec 14 '16 at 11:37

It is the task of the LT1055 opamp to (nearly) act as peak rectifier. However, because the rectified voltage must be able to change in both directions (control function) there must be the possibility for (slowly) decreasing of the peak value (discharging of the feedback cap). That is the function of the 22k parallel resistor. Hence, the whole rectifying circuit has a fixed time constant which must be at least 10 times the oscillation period (rule of thumb).

As a consequence, the time for discharging the feedback capacitor depends on the oscillation period (when the diode is off). That means: The rectified mean voltage which controls the FET resistance - and, thus, the oscillation condition (and the oscillation amplitude) - is NOT independent on frequency.

With other words: To meet the oscillation condition (opamp gain of "3") the FET resistance must always have the same value. Hence, the rectified voltage must have the same fixed value (control voltage). But this voltage depends somewhat on the oscillation period (more time for discharging the cap for lower frequencies). Thus, the control voltage lowers for smaller frequencies. This "drift" of the rectified voltage is automatically compensated by a larger oscillation amplitude.

  • \$\begingroup\$ Hi LvW, thanks for your answer. I'm not sure I completely follow. In the section "But this voltage depends somewhat on the oscillation period (more time for discharging the cap for lower frequencies). Thus, the control voltage lowers for smaller frequencies." Did you mean "Thus, the control voltage lowers for HIGHER frequencies."? If this is an issue with the rectifying part of the circuit, can I simply change the value of the 15uF cap so that the Vpk of the sine wave is constant over the frequency range that I am interested in (0.5-5kHz)? \$\endgroup\$ – CakeOrDeath Dec 14 '16 at 11:41
  • \$\begingroup\$ You have a kind of one-way rectification. Hence, the cap ischarged for the positive half wave (diode is on) - and it iwill be slightly discharged because of the parallel resitor. If the period is greater there will be more time for the discharging process. This changes the mean voltage for controlling the FET. If you can live with a sinus signal that has a bit more THD (distortions) I would recommend classical diode stabilization (antiparallel), which is NOT sensitive to frequency variations. \$\endgroup\$ – LvW Dec 14 '16 at 12:53
  • \$\begingroup\$ I would rather have as little THD as possible (hence why I went with this circuit) but it is probably worth measuring this for both set ups just to see if how much THD increases. \$\endgroup\$ – CakeOrDeath Dec 14 '16 at 13:57
  • \$\begingroup\$ Did you take other oscillator configurations (other than WIEN) also into consideration (single-element tunable)? For some topologies you have a filtered signal available - and perhaps you do not need no extra amplitude-control circuitry. \$\endgroup\$ – LvW Dec 14 '16 at 15:00
  • \$\begingroup\$ For example, see here: researchgate.net/publication/… \$\endgroup\$ – LvW Dec 14 '16 at 15:04

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.