Help with Hartley Oscillator

Discussion in 'Homework Help' started by Symmetry, Sep 20, 2013.

1. Symmetry Thread Starter New Member

Sep 20, 2013
7
0
I'm hoping someone can help me with this.
I am attempting to simulate the attached hartley oscillator in multisim. I am required to modify the circuit to produce a 1MHz waveform with a 100mV amplitude.
I know that the tank circuit dictates the frequency using (f = 1 / 2π √LC). However which components in this circuit control the amplitude? In what way does it accomplish this?
I have done a lot of research. I have found most of the online tutorials seem to be targetted either above my level, or are too vague. I am now at the point in my research where i have specific questiosn, and need specific answers to further my knowledge.
I'd appreciate any help anyone can offer.

File size:
73 KB
Views:
198
2. RichardO Well-Known Member

May 4, 2013
1,618
552
A hint. What loop gain is needed to oscillate?

3. t_n_k AAC Fanatic!

Mar 6, 2009
5,448
784
Not convinced the tank as per the schematic is tuned for 1MHz. Check the values. Are the two inductors L2 & L3 meant to be mutually coupled?

4. Symmetry Thread Starter New Member

Sep 20, 2013
7
0
T_N_K - Yes, sorry, the values are just the ones initially provided by the circuit. I've adjusted them in the simulator. No, there is no requirement for the inductors to be mutually coupled.

RichardO - This is the barkhausen criteria, right? This is my understanding so far, plz correct as needed...
The loop gain has to be slightly higher than attenuation (B). Now assuming i use the same size inductors in my tank circuit, B = L2/L1, so therefore attenutation is 1, and gain will have to be slightly higher than one.

And yet other sources i've read have stated that gain should just be one. Not sure what to make of that.

5. t_n_k AAC Fanatic!

Mar 6, 2009
5,448
784
The loop gain should presumably be unity. Otherwise oscillations may not be sustained. If the gain is slightly greater than unity, then for a perfect (linear) amplifier the oscillations would increase in magnitude rather than be maintained at a steady state value. One might argue the start-up condition requires a greater than unity gain with the gain then fortuitously settling to a unity value once steady-state oscillating conditions are achieved. I'm not convinced of that argument, valid as it may be.

In a practical BJT amplifier the device may be operating in a highly non linear mode [e.g. class C like] with the collector current comprising pulses of current which excite the tank circuit at the correct frequency. The non-linearity may also contribute to the effective amplitude control in which the device bias point self adjusts (perhaps near base / emitter cut-off) to achieve a stable output amplitude.

Some tweaking is therefore usually the norm in achieving the desired operating conditions in practice. Directly hitting on 100mV output (say) would be part of the tweaking process. The secret is to work out what one needs to tweak.

6. Symmetry Thread Starter New Member

Sep 20, 2013
7
0
Thats the part i'm struggling with. I've been tweaking components but it's producing an erratic output. Do you have any advice on what i need to tweak, and why?

My thoughts so far...
* If the loop has a gain of 1, and i need 100mV, then i suppose the input would need to be 100mV. So i thought tweaking the voltage divider to bias the base differently, but alas no luck.
* As the feedback comes from the tank, maybe the values in the tank need to be altered (whilst still maintaing a frequency of 1Mhz). I've run into more problems there. While doing this does change the amplitude, it gives me a weird frequency output i.e, the freq input to the base is 1Mhz, but the frequency at the output is a lot less (depending on the values). I cant even begin to guess why that would be.

I feel i am close to cracking this, i just need a couple of gentle nudges for it to all click together.

Last edited: Sep 22, 2013
7. Symmetry Thread Starter New Member

Sep 20, 2013
7
0
I have further experimented with the values in the tnak circuit, and i have found a combination which does give me the desired 100mV. Only now i've lost my 1MHz frequency altogether. I've doubled checked the values in the formula, they are correct and should work.
I'm missing something critical.

Mar 6, 2009
5,448
784

9. Symmetry Thread Starter New Member

Sep 20, 2013
7
0
In the first file (hartley1), the values in the tank circuit calculate out to 1MHz using f = 1 / 2π √LC. This seems to be proven by 'probe2'. Yet the output reads as 500kHz. I can't explain this.

In the second file (hartley2), the values still calculate to 1MHz, yet the simulation disagrees. Can other values in the circuit effect this? At least this time the output amplitude is correct.

File size:
162.8 KB
Views:
85
File size:
159.7 KB
Views:
66
10. t_n_k AAC Fanatic!

Mar 6, 2009
5,448
784
I'm not sure of the particular issues in your simulation.

In the first instance I'm wondering why you chose an ideal BJT device model. Do you know what the gm value is in this case?

Also the large coupling caps are atypical at 1 Mhz.

As for achieving 100mV output - perhaps a coupled winding might be the answer.

Last edited: Sep 22, 2013
11. Symmetry Thread Starter New Member

Sep 20, 2013
7
0
I didnt choose these values, so much as based it on an existing schematic. I dont know what the gm value is.
*Do you think i'd be better off with a particular BJT, rather than an ideal one?
*Also the large coupling caps (C1 and C2?); what values are typical for 1MHz? Can you suggest a resource?

12. t_n_k AAC Fanatic!

Mar 6, 2009
5,448
784
For the time being I'd lose the 2.5mH collector choke. It will play a part in causing spurious resonance with the 1uF collector coupling capacitor. Consider what the resonant frequency of 2.5mH inductance with a 1uF capacitor would be. Yep 3.18kHz.

Replace the 2.5mH with a suitable load resistance. This means you would need to know the mean collector current. Consider how Re influences the loop gain.

Consider drawing the equivalent a.c. circuit of what you are simulating. Don't go down the simulator route until you can do an a.c. analysis. Simulation with an actual BJT model will possibly alert you to likely non linear behaviour. At 1MHz your coupling caps can probably be 10nF without any issues.

I expect getting an oscillator to run at 100mV output without some transformer action or voltage division will be troublesome.

13. Symmetry Thread Starter New Member

Sep 20, 2013
7
0
Wow. I didn't even see that.

I briefly had a bit of a play with replacing the choke with a resistor, and the first thing i noticed is that i lose my sinusoidal output regardless of resistor size. I will do as you suggest, and do an AC analysis.

14. vk6zgo Active Member

Jul 21, 2012
677
85
Real oscillators start reliably due to the presence of some noise in the circuit,which is amplified,then that part of the noise which is at the resonant frequency is further amplified,fed back again,& the oscillator takes off!

In a simulation,the devices are perfect,so there is no initial noise,hence no signal amplified,no oscillation.
Simulators usually use some "fiddle" or other to make oscillation start.
Maybe the "fiddle" in your simulator is less effective than others.

P.S. For a good explanation of oscillator operation,borrow a copy of the ARRL Handbook from your local library.
There is a lot more to oscillators than meets the eye!

Last edited: Sep 29, 2013
15. LvW Active Member

Jun 13, 2013
674
100
I don`t think that during circuit simulation the missing "noise" would cause any problems. Resulting from the power switch-on transient there is always a sufficiently large spike that contains the wanted frequency component.
Thus, the oscillator will always start oscillating - if the oscillation condition is fulfilled.