# Some advice needed on oscillator circuit and amplifier

Discussion in 'The Projects Forum' started by The_Bagel_Guy, Nov 3, 2013.

1. ### The_Bagel_Guy Thread Starter New Member

Nov 3, 2013
7
0
Hey guys,
Just a brief intro...
At the moment I'm (trying) to do my final year project for my mechanical engineering masters. As such my knowledge of electronics is somewhat "fuzzy" to say the least and I need a little help and guidance. I've been assigned a project designing a magnetic bearing system which while interesting is a little beyond the scope of my basic electronics knowledge.

At the moment I'm trying to get my head around the oscillator circuit and amplifier. I've done the circuit for the oscillator which "should" oscillate at 50kHz, Image below if all has worked well. I'm connecting the +ve and -ve terminals of the op-amp to +5v and ground respectively and the non inverting input of the first op-amp will be +2.5V. So this should be oscillating between 0 and 5V at close to 50kHz.

The Op-amp is an TLV2474 from RS
http://uk.rs-online.com/web/p/operational-amplifiers/6612230/
Resistor values I've chosen are
R=4,700Ohms
Rf=27,000 Ohms + Variable resistor in series to allow for some adjustment
Rg=6,800 Ohms

Rf/Rg=1 to get gain to unity

Capacitor

I'm fairly comfortable with this part, I'll be hooking it up to an oscilloscope tomorrow to check it works.

What I'm not sure is how to convert the very small signal coming out of the oscillator to a true bi-directional signal strong enough to drive my sensor.

My plan at the moment is to use a Darlington pair to amplify my signal and allow me to draw more current without distorting the sinewave, then put a capacitor on the output (1 micro Farad) to block the DC offset. By my understanding this will give me the AC signal shifted down to +12V to -12V as long as I do the Darlington pair part right is this correct?

My second question, how do I actually use the Darlington pair to amplify but not distort my signal? Should I be looking at something different for amplifying the signal?

Sorry if this seems very simple but I'm looking for someone with know how to point out if I've made any blindingly obvious mistakes this far.

Thanks for any help!

Last edited: Nov 3, 2013
2. ### LvW Well-Known Member

Jun 13, 2013
760
103
Hello, The Bagel Guy, welcome to the board.

May I ask you why do you expect a "very small signal coming out of the oscillator"?
What is "very small"?
More than that, your pdf attachement shows the opamp data sheet but not your circuit. How does it look like?

3. ### The_Bagel_Guy Thread Starter New Member

Nov 3, 2013
7
0
LvW, sorry about the image of the circuit being missing. This is the link to it on photobucket:
http://s1091.photobucket.com/user/The_Bagel_Guy/media/DSCF1258_zps8a6b5c61.jpg.html?sort=3&o=0
And the real life version:
http://i1091.photobucket.com/albums/i387/The_Bagel_Guy/DSCF12591_zps06efa43a.jpg

I'm not completely sure on how to analyse the current in a circuit such as this, so I've assumed that the values of the resistors will result in a very low current in the order of a few mA. I've been working under the assumption I need a very small current in the oscillator to produce a clean sine wave. I've tried to analyse the system from a controls point of view and set the phase lag of each low pass filter to -45 degrees. Obviously the diagram I've drawn is missing the blocking capacitor and the Darlington pair. I hope this helps.

4. ### MikeML AAC Fanatic!

Oct 2, 2009
5,445
1,073
With this type of oscillator, you take the output at one of the OPamps; not from the RC network as you show it...

If you do, you probably wont need a buffer...

5. ### The_Bagel_Guy Thread Starter New Member

Nov 3, 2013
7
0
Mike, why does the output have to come off the output of the op-amp and not after the final RC? Will my current output give me poor results?

I've used the circuit from a Texas Instruments document, page 36 figure 7:
http://www.ti.com/sc/docs/apps/msp/journal/aug2000/aug_07.pdf

Cheers,

Chris

6. ### MikeML AAC Fanatic!

Oct 2, 2009
5,445
1,073
Here is something I posted on another site. Look at the outputs from the various opamps. Any of those are a low-output-impedance takeoff that is limited only by the opamp's ability to source/sink current. Note the slight reduction in output amplitude as you move down the stages...

If the amplitude is sufficient, and you dont need the lowest possible distortion, you don't have to add the fifth stage...

• ###### BubbaB.jpg
File size:
132.8 KB
Views:
48
Last edited: Nov 3, 2013
7. ### LvW Well-Known Member

Jun 13, 2013
760
103
Bagel Guy, before selecting one of the various outputs it would be interesting to know if your oscillator is oscillating at all!
What kind of amplitude stabilization do you intend to use?

8. ### The_Bagel_Guy Thread Starter New Member

Nov 3, 2013
7
0
LvW, in all honestly I hadn't thought about amplitude stabilisation. In fact, I'm not entirely sure what it is or how to achieve it, but I will read up on it today!

9. ### MikeML AAC Fanatic!

Oct 2, 2009
5,445
1,073
Amplitude stabilization is automatic in that oscillator. Stage A has reserve gain, and is driven into clipping. Look at V(a) in the simulation. Note that each of the subsequent stages act as a low-pass filter, so the harmonic distortion is progressively lower as you go from V(a) to V(e)...

10. ### LvW Well-Known Member

Jun 13, 2013
760
103
Yes, that was the background of my question:
* At first, to hear if the OP knows about the amplitude problem in each oscillator,
* secondly, to hear if he intends to use another method than simple amplitude clipping (depends on distortion requirements).

11. ### The_Bagel_Guy Thread Starter New Member

Nov 3, 2013
7
0
I something even more puzzling now.

I tested my circuit on the oscilloscope today and it wasn't oscillating.

Removed the feedback path through R_G in my diagram, so no feedback from the final RC to the 1st stage, and it started oscillating. Removing this path was actually an accident as i was only trying to adjust R_G but now it's producing a nice clean sine wave at the right frequency between 0 and 5V.

I've been staring at the circuit for a while now, nothing is shorting, all seems to be as it should be.... How is it oscillating?

I'm going to do a second circuit on another breadboard to see what's going on.

12. ### LvW Well-Known Member

Jun 13, 2013
760
103
What is R_G ?
Please, can you show us again your diagram as a pdf- file with the selected values (R and C)?
Please note, that you have 3 identical (and decoupled) RC sections, but the last RC section is LOADED by the inverter stage and thus does not have a 45 deg phase shift at the same frequency. This means that the inverter gain should be larger than "-4".

EDIT: Just discovered that my comments apply to Fig. 9 of the TI document.
For Fig. 7 you have only two identical RC stages (decoupled) and one (slightly) loaded RC section. Thus, each of the section provides app. 60 deg phase shift and the inverter gain is slightly larger than (2)^3=8.

Question: Why don`t you use the circuit in Fig. 9 ? It is easier to design and needs only 2 opamps. The performance is comparable.

Last edited: Nov 5, 2013
13. ### MikeML AAC Fanatic!

Oct 2, 2009
5,445
1,073
The thing you built is called a Bubba Oscillator. Read section 8.4 of this TI App note.

What are you trying to drive with it? The external load must be connected only to an opamp output pin; never an opamp input pin (the RC network). Connecting the load there will instantly kill the oscillation.

The opamp outputs can only stand to drive a load resistance of >~500Ω. If it oscillates unloaded, and then quits when you load it, then the load you are connecting has too low a resistance and exceeds the current drive of the opamps.

Last edited: Nov 5, 2013
14. ### LvW Well-Known Member

Jun 13, 2013
760
103
Just to make sure that we all speak about the same circuit:
The oscillator called "BUBBA oscillator" is given in Fig. 9 of the papaer referenced by Bagel_Guy. However, he has mentioned that he is designing according to Fig. 7 of that paper.

15. ### MikeML AAC Fanatic!

Oct 2, 2009
5,445
1,073
The picture he posted way back at the beginning had four RC networks, so unless he has changed his mind, he should still be building a Bubba.

16. ### LvW Well-Known Member

Jun 13, 2013
760
103
Yes - it is a bit confusing. He should clarify.
I was referring to his last information in post#5.
My opinion: I never would use a circuit with 3 or even 4 RC sections.

17. ### MikeML AAC Fanatic!

Oct 2, 2009
5,445
1,073
Actually, the Bubba oscillator is great. It only needs a loop gain of >4 to oscillate. It is self-stabilizing due to clipping in the first stage. The harmonic distortion is progressively less as you go down the chain. It starts up fast, and is quite stable. It provides sine/cosine outputs automatically. What is not to like?

18. ### LvW Well-Known Member

Jun 13, 2013
760
103
I think, one of the oscillator circuits based on integrator stages does the same (quadrature outputs, if desired). And the distortion capabilities are comparable.

19. ### The_Bagel_Guy Thread Starter New Member

Nov 3, 2013
7
0
Hey, sorry to you both for not responding sooner.
It looks as if I am actually designing a bubba oscillator, I was aiming for the buffered oscillator 3 RC in figure 7 but added the 4th RC as I was under the impression it would give a cleaner waveform. In hindsight it looks I i have inadvertently turned it into the Bubba circuit in Figure 9!
Although from what you were saying above MikeML It seems the Bubba oscillator is a good choice!

Thanks for all the help both!

20. ### LvW Well-Known Member

Jun 13, 2013
760
103
Does it oscillate already (with sufficient THD) ?