What's wrong with this Colpitts oscillator?

Thread Starter

Elerion

Joined Sep 11, 2017
125
Hello everyone.
I know what Colpitts oscillators are, and I've used them many times.
But I've never designed them from scratch.
So today I've been simulating (and building) Colpitts using BJT, in common emitter, common collector and common base.
They all oscillate. That's easy to do. So is getting the (more or less) right frequency.

But I get distorted outputs, mainly because transistors go into cutt-off or saturation, and I'm not able to solve it.
After two or three hours searching, reading, and some more time of trials, I finally decided to ask for help.
Could anyone help me see what am I doing wrong, why is this happening and how to solve it?
I just want to fully understand what I'm doing.
I've selected one of the oscillators (a common emitter one) and attached it.

Thank you!
 

Attachments

ian field

Joined Oct 27, 2012
6,536
Hello everyone.
I know what Colpitts oscillators are, and I've used them many times.
But I've never designed them from scratch.
So today I've been simulating (and building) Colpitts using BJT, in common emitter, common collector and common base.
They all oscillate. That's easy to do. So is getting the (more or less) right frequency.

But I get distorted outputs, mainly because transistors go into cutt-off or saturation, and I'm not able to solve it.
After two or three hours searching, reading, and some more time of trials, I finally decided to ask for help.
Could anyone help me see what am I doing wrong, why is this happening and how to solve it?
I just want to fully understand what I'm doing.

I've selected one of the oscillators (a common emitter one) and attached it.

Thank you!

Why does it have to be Colpitts? Recently I've been playing with the Butler oscillator. Its low loading on the LC tank, uses a single untapped inductor and usually produces a reasonable sine wave.
 

Thread Starter

Elerion

Joined Sep 11, 2017
125
Why does it have to be Colpitts?
Because I want to understand this particular circuit.
I'm not really trying to build something with it.

Don't over drive the Base.
E
Thank you. I suppose it works better because the added base resistor adds damping.
But most of the circuits I see around don't use this resistors, and they suppose to work fine.
(I particulary like the attached image, comparing the different approaches).
All design rules around just say "get enough gain" and "f=1/(2pi*sqrt(LC))".
 

Attachments

ebp

Joined Feb 8, 2018
2,332
I outside of my areas here, but ...
looks to me like simply too much gain; can you lower the AC gain by reducing C1 fairly dramatically?
 

Thread Starter

Elerion

Joined Sep 11, 2017
125
can you lower the AC gain by reducing C1 fairly dramatically?
Yes, it helps too. Thanks.

Using the series resistor, shown in ericgibbs circuit, I managed decent sine output from the common collector colpitts.
But there's still a current spike which I don't know why is happening.
I attached a screen capture which I hope will be helpful to see what I talk about.
Does anyone know what is causing it?
 

Attachments

Thread Starter

Elerion

Joined Sep 11, 2017
125
That is a nice implementation too.

Now, to get a little bit slower frequency, I want to use smaller caps and bigger inductor (that I have available).
30 pF and 750 uH.
This means that the reactance is much higher.
Now the oscillators don't work at all, or do badly.

What kind of adjustments are required to be made when changing the reactance of the tank elements?
Any tip?

AFAIK, feedback ratio is the ratio of the capacitors, so it remains the same as before (as both capacitors are the same value).
To start oscillations, gm * R(collector) > feedback_ratio.
 
Last edited:

Thread Starter

Elerion

Joined Sep 11, 2017
125
Sure!
I uploaded the file containing all my circuits, grouped by configuration. Just ignore the JFET one, on the top-right.

The issues now are getting the common emitters to work with the 30pF/750uH tank elements, and getting common collector configuration to avoid the current spike (when using 1000p/10uH, because using 30pF/750uH lead to a heavily clipped output), as explained in my post #6.

Thank you!
 

Attachments

Danko

Joined Nov 22, 2017
1,829
Hello everyone.
I know what Colpitts oscillators are, and I've used them many times.
But I've never designed them from scratch.
So today I've been simulating (and building) Colpitts using BJT, in common emitter, common collector and common base.
They all oscillate. That's easy to do. So is getting the (more or less) right frequency.

But I get distorted outputs, mainly because transistors go into cutt-off or saturation, and I'm not able to solve it.
After two or three hours searching, reading, and some more time of trials, I finally decided to ask for help.
Could anyone help me see what am I doing wrong, why is this happening and how to solve it?
I just want to fully understand what I'm doing.
I've selected one of the oscillators (a common emitter one) and attached it.

Thank you!
Look at post #23 , maybe it will useful to you.
EDIT: Colpitts component values proportions
Before design you need to know frequency F, resistance of load R_load and voltage of power supply V_ps.
Reactance X_L1 = X_C3 + X_C1 ≈ R_load / (5 .. 7) R_load usually is about 1kOhm
Capacitance C1 ≈ 2 * C3 * (V_ps - 1) for V_ps ≥ 1.5V
Capacitance C4 ≈ (10 .. 20) * C3
Resistance R2 ≈ (1.5 .. 2) * X_C1
Resistance R5 should be as low as possible, but keep sine form without distortion
Resistance R3 should be as high, as possible, but keep sine amplitude (peak-peak) on R_load close to 2 * V_ps
For example see picture below.
 
Last edited:

MisterBill2

Joined Jan 23, 2018
18,167
I outside of my areas here, but ...
looks to me like simply too much gain; can you lower the AC gain by reducing C1 fairly dramatically?
It certainly DOES look like excessive gain. Adding a resistor in series with the emitter is one way to get some negative feedback which will tend to reduce the amplitude. But way back in history there was a difference in that the two capacitors in series that provided the feedback point were not of equal value. Then somehow somebody decided that they should be equal, and since then many sources specify that they should be equal. Most ARRL publications show this, and it has been incorrect for about 50 years. In the very first drawing, C2 and C3 appear to be the same value, but if they are not the same then the feedback can be less and at some point the waveform should be undistorted.
 

BR-549

Joined Sep 22, 2013
4,928
I was taught a Colpits advantage was swamping the tube with capacitance......so the warm up drift was small.

And the tap is not in the middle. More cap on grid side....more power out. More cap on plate side....more stability thru tuning range.
 

Thread Starter

Elerion

Joined Sep 11, 2017
125
Look at post #23 , maybe it will useful to you.
EDIT: Colpitts component values proportions
Doesn't work for me.
Freq = 1.5 MHz
L = 750 uH
Ct = 15 pF
As supply is 9V, C2 = 16 * C1
This leads to C1 = 16 pF and C2 = 256 pF
R1 = 2 * XC2 = 2 * 414 ~ 800 ohm
RL (R3) = 5 * 7kohm (which is the reactance of the inductor)
Without load, doesn't work either.
C3 = 10..20 * C1 (around 200pF, but in this case, I use a bigger cap; didn't work either case)

I sweeped R2 from 50k to 1Meg, and R4 from 1 ohm to 20k.
 

Attachments

Danko

Joined Nov 22, 2017
1,829
Doesn't work for me.
Freq = 1.5 MHz
L = 750 uH
Ct = 15 pF
As supply is 9V, C2 = 16 * C1
This leads to C1 = 16 pF and C2 = 256 pF
R1 = 2 * XC2 = 2 * 414 ~ 800 ohm
RL (R3) = 5 * 7kohm (which is the reactance of the inductor)
Without load, doesn't work either.
C3 = 10..20 * C1 (around 200pF, but in this case, I use a bigger cap; didn't work either case)

I sweeped R2 from 50k to 1Meg, and R4 from 1 ohm to 20k.
Did you execute instructions correctly?
"Reactance X_L1 = X_C3 + X_C1 ≈ R_load / (5 .. 7) R_load usually is about 1kOhm"
If you do not plan to take out high frequency energy from oscillator, then R_load is about 1kOm. It is inner load for HF energy, circulated in oscillator.
"About 1kOm" means 0.6kOhm...1.4kOhm, but not 35kOhm, because even 1.8kOhm already is "About 2kOhm".
If you want take out HF energy for antenna or for heating inductor, then you need calculate R_load with V_ps and required HF power. For example, it may be 300Ohm, 75Ohm, 22Ohm or so.
Edit:
In other words:
In step 1 we get out R_load <1kOhm or =1kOhm.
In step 2 we calculate X_L1 = X_C3 + X_C1 = R_load / (5 .. 7).
 
Last edited:

MisterBill2

Joined Jan 23, 2018
18,167
I was taught a Colpits advantage was swamping the tube with capacitance......so the warm up drift was small.

And the tap is not in the middle. More cap on grid side....more power out. More cap on plate side....more stability thru tuning range.
The "swamping" function came a bit later. AND, the use of the word "swamp" is such a lazy cop-out, used by a whole lot of people. I will leave it at that.
If the goal is stability and a pure sine wave then one set of values would be chosen, with lower feedback. But if fast starting and maximum output are the goals then more feedback would be used. The design is a trade-off, one mode or the other.
 

Bordodynov

Joined May 20, 2015
3,177
You greatly reduced the capacitance and increased the inductance. Therefore, for normal operation, it is necessary to reduce the input capacitance and increase the input impedance of the amplifier. To do this, I added another transistor, included in the scheme with a common collector. Also I added a parasitic capacitance of 0.5 pF to each node. To increase the input resistance of the divider, I increased the value of the resistors.
2018-03-20_14-35-34.png
 

Attachments

Thread Starter

Elerion

Joined Sep 11, 2017
125
Did you execute instructions correctly?
"Reactance X_L1 = X_C3 + X_C1 ≈ R_load / (5 .. 7) R_load usually is about 1kOhm"
Ok, that leads to:
RL = 1k
L=750uH -> f = 42.5 kHz
Ct = 19 nF
C1 = 20 nF
C2 = 320 nF

And it works with 100 kohm base resistor, 24 ohm emitter resistor and 12 ohm feedback resistor.
A little distorted on the bottom, but pretty decent.
And obviously (?) it works if feeding a larger resistor load.

Where did you take all those rules from? From experience?
This circuit is not exactly the typical colpitts oscillator seen in manuals.
If I place the inductor parallel to tank capacitors (between collector and ground, instead of to supply, as seen in many other circuits), it doesn't work (even placing some collector resistor, 10k).
Also, this circuit it doesn't use a voltage divider at the transistor base, just a single "pull-up" resistor.
Did you explore these alternatives?
 

Thread Starter

Elerion

Joined Sep 11, 2017
125
Therefore, for normal operation, it is necessary to reduce the input capacitance and increase the input impedance of the amplifier.
Why it that? Could you please briefly explain the relation, or recommend me any reading where I can learn about that.

To do this, I added another transistor, included in the scheme with a common collector.
This reduced input capacitance?
 

Thread Starter

Elerion

Joined Sep 11, 2017
125
Did you execute instructions correctly?
"Reactance X_L1 = X_C3 + X_C1 ≈ R_load / (5 .. 7) R_load usually is about 1kOhm"
Sorry for posting a third time, but I could not edit my post.
I made a mistake with "the inductor parallel to tank capacitors". Obviously that won't be a common base amplifier anymore, but more like a common collector.

When I asked "did you explore these alternatives?", I meant common collector and common emitter versions of colpitts oscillator.
Thank you, and sorry.
 
Top