# [Help] Wein Bridge Noise/Distortion Reduction

#### Boplogz

Joined Oct 17, 2022
15
Hi Everyone,

Im starting to become more interested in circuit design. Im still a student and trying different things.

I have this circuit design (see image below) based on the Wein bridge on the internet. I had to use NE5534 OP-amp because Im needing a 100khz-300khz frequency and 9Vpp later.

The circuit output my needed frequency and amplitude. However, the output I produced seems distorted (see image below). May I ask the experts here on what should I do to make the output a better sine wave.

Thank you for your kind help.

#### Attachments

• 19.9 KB Views: 4

#### crutschow

Joined Mar 14, 2008
31,129
You need a gain control circuit to adjust the feedback gain and keep the output below the clipping point which can use an incandescent bulb, diodes, or a FET.
Here's an article on that.

#### Ian0

Joined Aug 7, 2020
6,714
If you use a rail-to-rail op-amp, you will get symmetrical clipping. If you then set the gain so that it is as close as possible to 3 (but not less than 3) you will improve things a lot. If the gain is too low it won't start.
It depends on how good a sinewave you need, but if it still isn't good enough you will need an amplitude control circuit as @crutschow says.
Here's another alternative
http://4tubes.com/3-BOOKS/BOOKS-LITERATURE/ENGLISH/Magazines/Wireless-World/Wien-bridge oscillator with low harmonic distortion.pdf
because filament lamps are getting tricky to get, and if you're outside the USA neither you nor your suppliers will have a clue what a #344 lamp is, because everywhere else they are just rated in voltage and current.
[EDIT] I've often wondered, but never tried, these lamps as they have a high resistance so it doesn't make the other resistances in the circuit too low, but with quite a high filament mass, the amplitude might take a week to settle. (Also wondered if they might make good ambient temperature sensors, but that's another matter entirely)

Last edited:

#### Bordodynov

Joined May 20, 2015
3,025
Hi Everyone,

Im starting to become more interested in circuit design. Im still a student and trying different things.

I have this circuit design (see image below) based on the Wein bridge on the internet. I had to use NE5534 OP-amp because Im needing a 100khz-300khz frequency and 9Vpp later.

View attachment 280185

The circuit output my needed frequency and amplitude. However, the output I produced seems distorted (see image below). May I ask the experts here on what should I do to make the output a better sine wave.

View attachment 280186

Thank you for your kind help.
You are using MULTISIM, and it tends to increase the time step (this is to speed up the simulation). Reduce the maximum time step. For example 1.e-8.

#### Bordodynov

Joined May 20, 2015
3,025
See

#### LvW

Joined Jun 13, 2013
1,586
The above shown method to control and stabilize the amplitude is a very good one (and , therefore, somewhat complicated).
The goal of such a method is to slightly reduce the opamps gain for rising amplitudes - before clipping occurs.
A simple - but effective - method is to place two antiparallel diodes across the feedback resistor R2.
You can improve this method when you divide this resistor R2 in two parts (not necessarily equal) - and place both diodes across one resistor only.

Interesting observation for all oscillators:
With the aim to make the output signal as clean as possible the circuit should be as linear as possible - and for this reason it is required that the circuit contains a (small) non-linearity. Sounds contradictory?

#### Bordodynov

Joined May 20, 2015
3,025
The above shown method to control and stabilize the amplitude is a very good one (and , therefore, somewhat complicated).
The goal of such a method is to slightly reduce the opamps gain for rising amplitudes - before clipping occurs.
A simple - but effective - method is to place two antiparallel diodes across the feedback resistor R2.
You can improve this method when you divide this resistor R2 in two parts (not necessarily equal) - and place both diodes across one resistor only.

Interesting observation for all oscillators:
With the aim to make the output signal as clean as possible the circuit should be as linear as possible - and for this reason it is required that the circuit contains a (small) non-linearity. Sounds contradictory?
Initially I tried to do exactly as you suggested, but it didn't work out so I switched to a circuit using AGC.

#### Ian0

Joined Aug 7, 2020
6,714
The above shown method to control and stabilize the amplitude is a very good one (and , therefore, somewhat complicated).
The goal of such a method is to slightly reduce the opamps gain for rising amplitudes - before clipping occurs.
A simple - but effective - method is to place two antiparallel diodes across the feedback resistor R2.
You can improve this method when you divide this resistor R2 in two parts (not necessarily equal) - and place both diodes across one resistor only.
No real difference between clipping the output with diode, and clipping the output by he supply rails.

#### LvW

Joined Jun 13, 2013
1,586
No real difference between clipping the output with diode, and clipping the output by he supply rails.
No difference? This is impossible.
In general: Diodes cause "soft clipping" and supply rails cause "hard clipping".
That is for sure and this is the fundamental difference.
This method is described in many books and has been proven in practice with good results.

However: Amplitude limitations with two diodes require a careful design:
(a) Without the diodes, the gain should be only slightly above the nomnal gain (perhaps A=+3.1....3.2).
(b) The effect of the diodes non-linearity should be as small as possible, but large enough to allow amplitude limitation befor hard clipping can occur. That is the reason that we should divide the feedbacl resistor in two parts.
(Sometimes even three resistors are used: Ra in series with Rb||Rc and two diodes across Rc only).

Ref (for example): Sergio Franco: "Design with operational amplifiers....", 2nd edition, page 457 (Fig. 10.3)

Last edited:

#### crutschow

Joined Mar 14, 2008
31,129
I've often wondered, but never tried, these lamps as they have a high resistance so it doesn't make the other resistances in the circuit too low, but with quite a high filament mass, the amplitude might take a week to settle. (Also wondered if they might make good ambient temperature sensors, but that's another matter entirely)
Below is the simulation of an oscillator with an emulated 10V, 144mW, 344 incandescent bulb to control the op amp gain.

The bulb resistance is determined by Vin and the equation for "R" shown below the bulb.
The equation was empirically derived to emulate the typical resistance versus voltage for an incandescent bulb (a 344 bulb here).

The diode, resistors, and capacitor generate a Vin DC voltage approximately equal to the AC RMS voltage at node B, since the equation for R requires a DC voltage.
(Note that the RC time-constant of the circuit is likely much shorter than the thermal time-constant of the actual bulb.
A longer time-constant would increase the time for the amplitude to stabilize.)

The output signal (green trace) starts out with the op amp saturating at the peaks, and then settles to a lower voltage with a good sinewave output as the bulb resistance increases to lower the loop gain.

(The bulb resistance here should not be appreciably affected by normal ambient temperature changes since it is operating at about 12% of its rated operating power, which should give a filament temperature well above ambient, even if it is well below its rated power).

Last edited:

#### LvW

Joined Jun 13, 2013
1,586
Note that the RC time-constant of the circuit is likely much shorter than the thermal time-constant of the actual bulb.
A longer time-constant would increase the time for the amplitude to stabilize.
Yes - and since everything in analog electronics is a trade-off - the disadvantage of a longer time for stabilizing the amplitude has to be compared with the advantage that the effect of "breathing" (a slight amplitude modulation) is reduced.

#### Dave Lowther

Joined Sep 8, 2016
165
I like this oscillator with a small bulb as the non-linear element. I use one for spot testing at 1 kHz and it has THD of 3 ppm (measured using 24 bit sound card and muli-instrument s/w) https://www.elektormagazine.com/files/attachment/326 I'm using the circuit in fig 2 with an NE5532.