Simple Sine Wave Oscillator Circuit I think this diagram is incorrect

[juming]

i was given this circuit diagram by one of my instructors this past winter (summer for you americans) so.. i tried 2 build it.
three times i tried, it didnt work. twice on a copper board, once on a projectboard.
the instructor & i checked them each over numerous times... he even tried to build it himself, but his didnt even work. so i was wondering, could there be something wrong with the circuit diagram?

if anyone knows their stuff regarding this (diagram is labelled "Sine Wave Oscillator and Buffer Amplifier"), please check it over for me. it's attached as sin_osc.zip

thanks in advance

-greg

 

[Averagesupernova]

I seriously doubt even if it did work you would get a true sine wave out of it. Most likely you would get a sawtooth wave if not something closer to a square wave. What are you using to check if it works? I assume you have a scope? I have seen circuits similar to this and they are usually referred to as multivibrator circuits and meant to generate square waves. This one is configured a bit differently though with one side of the multivibrator not being symetrical with the other side.

 

[turin]

I tried to build one of these multivibrators that was quite similar to this one, but I wound up having to change a few connections (based somewhat on reasonability and somewhat on trial and error). This is the third significantly different variation of the circuit that I've seen posted on the internet (and the most complicated). Here's what I can tell you right off the bat (it helps to know what part of the circuit is supposed to perform what function):

The two left-most transistors with their pull-ups and HP cross-connected bases compose the multivibrator, which, as averagesupernova points out, usually is intended for squarewave output. The collector of the transistor in the middle is the output of the multivibrator, and is apparently being used as a squarewave input into a filtered buffer which apparently is intended to derive the sinewave. I don't understand what the negative feedback (the emitter resistors at the bottom) is for. I'll look at this in more detail and edit.

What exactly do you mean when you say that you couldn't get it to work? Do you mean that it wouldn't do anything or that you didn't see anything like a sinewave?

EDIT:

My multivibrator circuit (corresponding to the two transistors on the left of your schematic) did not have the negative feedback resistors. In other words, the emitters were directly grounded. Also, my circuit has the base of one transistor connected to the collector of the other by both the cap and resistor. This is done in both directions similar to the circuit that you have, except that the emitter connections in your circuit would be collector connections in my circuit.

For the output buffer stage (the right-most transistor with its corresponding connections), I would think that there should be some LP filtering (to smooth down the sharp edges of the squarewave), but all I see is HP filtering.

That's all I've got for you.

 

[juming]

we tried testing it using two separate oscilloscopes...
there was absolutely nothing on the scope, just the flat line... no vibrations, no square wave, no sawtooth... just flat-line.

i know, it looks like it's supposed to output a square wave, but my instructor told me that the two crossover capacitors have something to do with transforming that into a very-near sine wave.
(i really have no idea myself.. am in 2nd year electronic engineering... they haven't taught us much electronics!)

:) thanks for all your info!

have you got circuits similar to this that DO work? if so, can you zip & attach them?

 

[turin]

So are these transistors supposed to undergo small-signal operation? The conventional multivibrator, as I understand it, is supposed to act highly nonlinearly, slamming up to saturation and then down to cutoff with negligible switching time.

After a closer look at the circuit, it does seem to be intended for the small signal operation. In that case, this is almost totally different than the multivibrator idea. The component values are rather critical. You need to know your β for each transistor and you should also know the actual values of the components, not just nominal.

Do you have a curve tracer? It's been a while since I did this sort of thing, so I don't remember if there is a remedy for sensitive component values other than just making sure you know what they are.

 

[juming]

(bear in mind that you're dealing with a newbie when it comes to this sort of stuff)

yes, the instructor said that the reliability of the components is also a great factor in this circuit... we tried swapping the 1k resistor from the emitter of the second transistor to a 1k2 - I'm not sure what that was supposed to do though :)

err... what's a curve tracer?
and how would one calculate/find out the β for each transistor?
& the difference between nominal & actual values of components?

(you guessed it.. they've taught us nothing yet)

 

[turin]

I'm real suspicious that you should swap the capacitor and resistor that are connected to the collector and emitter of the middle transistor. (It sure would help if these components were numbered.) The base needs to be at a particular value to maintain the quiescent point, and I don't like the looks of the asymmetrical configuration. The way it's configured, it looks like the transistor on the left should be about 1 V below the transistor on the right. Off the top of my head, I don't see how that could be a good thing.

At what points have you probed the circuit?

(you guessed it.. they've taught us nothing yet)

And they probably never will. Oh man! Have I been there before! You're dragging up old unpleasent memories of late nights of studying separated by long days in the lab with an instructor that acts more like a magician than an electrician. It seems like you are so on your own, especially when they force you to work in those damn coalition groups.

... we tried swapping the 1k resistor from the emitter of the second transistor to a 1k2 - I'm not sure what that was supposed to do though

That's called "tweaking." It's what all the great engineers do, don't you know. Engineering design is a lot of guess-and-check. That is one thing they will most likely never come out and admit to their undergrad engineering students, at least, not officially. I wouldn't be able to tell you at this point if that was a "directed" guess or just an act of desparation. I would give your instructor the benefit of the doubt, though.

... what's a curve tracer?

Your instructor should definitely be able to tell you that. If not, I would reconsider engineering schools. That is a rather important piece of design equipment for these rough discrete transistor topologies. Without one, you can basically forget about small signal circuit development (or shell out a lot of cash for highly toleranced components, in which case you might as well build an ASIC).

... how would one calculate/find out the β for each transistor?

That's what you need to curve tracer for. You put the transistors on the curve tracer to find their quiescent points and the β's at those points. Then, you select your collector, emitter, and base components to maintain the quiescent point. Finally, your β value will determine the crossover coupling that you will get. I didn't actually happen to look at the parasitic capacitances of these BC337's, but I'm sure it's far less than 100 nF, so you probably don't have to worry about that. But you definitely need to know where you stand with quiescent point and β value.

Oh, and even though you didn't ask, I'll go ahead and tell you anyway. β is the small signal gain from base to collector. Actually, it's been so long that I can't remember if that's current gain, voltage gain, transimpedance, or transconductance. I'm pretty sure it's voltage gain (the small signal component, so subtract the quiescent bias).

... the difference between nominal & actual values of components?

This is one of those things that distinguishes good engineers from the rest. As an engineer, one of your fundamental distinctions is 1) the realization that nothing is what it says it is or should be, and 2) you know how to make it work regardless of this problem and keep it transparent to your customer. Once you graduate, this is one of those distinctions that you will (hopefully) get paid for. It is quite trivial; the real trick is #2. As an example, there are 1 kΩ pull-ups. They may actually be 986 Ω and 1003 Ω. This is probably critical to your circuit since it relies on the crosscoupling of the collector being pulled up by a 986 Ω resistor and the collector being pulled up by a 1003 Ω resistor. On first order, this will negatively affect the quiescent point (which is semi-critical). But, on second order, it will affect the swing of the collector voltage, which will get amplified at the collector of the other transistor through its gain. I think you really need to get ahold of a curve tracer for this one, but I may just be paranoid. It has been a while, like I said.

 

[turin]

I Told You Wrong!

The β value is a LARGE SIGNAL value. It is the DC current gain from base to collector. You still use the curve tracer to determine it, but what I was describing was the h_fe parameter, which is also not a voltage gain, but a current gain. Man, how my knowledge has deteriorated.

 

[flexten]

Every one of these i got, the caps are always hooked to the collector on both bipolars.
So i think the feedback cap and resistor on the leftmost bipolar needs switch ?
I am not an EE so take this with ...

Best