Can't get higher frequency in Colpitts oscillator

[kamilkarp]

Hello!
I want to get around 120-150MHz in my oscillator. Unfortunatelly, I can't get frequencies above 70MHz. With that circuit attached I'm getting around 65MHz. Everytime I change coil or capacitors(tried many combinations of 60-200nH coils and 2p-68p capacitors). Can someone explain me, why I can't get much higher frequencies? What's wrong in my circuit?

Can someone explain how to calculate proper values of resistors/capacitors/coil or provide a book name, which describes it in a simple way?

In LTspice I can get frequencies 70-170MHz, but in reality, when I solder circuit I can't.

Thanks in advance.

 

[andre_luis]

Aren't you aware that the transistor you are using is suited for applications in the audio range? You need select another one able to deal with RF applications. Perhaps the 2N2222 which is present on LTSpice library could do the job, but I admit that did not check its bandwidth, so just make a try.

 

[d123]

Hi,

It's a good idea to check datasheet to simulation realities sometimes based on all the important circuit parameters required. e.g. Although 99% faithful, I can simulate 5V devices at 20V supply and it gives results as if they'd work, not always but sometimes, the simulator appears to ignore certain realities a user needs to try to remember and contrast.

It might be a wildly wrong theory but I sometimes wonder if the 2N2222A and 2N2907A BJTs simulate so well and can make any circuit work, it seems, if it isn't related to the (ideal?) model somehow giving ideal seeming results.

 

[BigBoss]

-BC546B is not a RF transistor, it will not oscillate
-Feedback Capacitor values are wrong
This oscillator is wrong designed.

 

[andre_luis]

but I sometimes wonder if the 2N2222A and 2N2907A BJTs simulate so well and can make any circuit work, it seems, if it isn't related to the (ideal?) model somehow giving ideal seeming results.

It seems like not all the same parameters are filled in the spice models of transistors designed for quite different applications, so I presume that the transistors designed to deal with higher frequencies have their model parameters optimized to these bandwidth. Anyway, as you, I feel that in the range of a houndred megahertz we could not expect on simulation an accurate behavior compared to the actual circuit assembled with real components.

 

[Audioguru]

My FM transmitter oscillates perfectly at 100MHz using a 2N3904 audio transistor. Of course I did not build it on solderless breadboard that has additional stray capacitance and inductance all over the place.
Oh, you didn't build it but you are simply simulating it? My FM transmitter simulates perfectly with its audio transistor oscillating at 100MHz.

 

[d123]

Hi,

Apart from changing the BJT to a more suitable one (maybe the 2n2222 would work because it says 250MHz and 300MHz in the TO-18 and the TO-92 datasheets I have - That's derating enough frequency-wise is it?) - and I've never needed to do this myself - an expert here will be able to tell you if dead bug/pin-to-pin soldering might help for prototyping when a board that should work doesn't in the real world. Saves on 1 pcb until you're sure the circuit is right and it's easier to add/remove components! As I said, I do not know if that's any use as a method of troubleshooting this issue or not, or in fact unnecessary as it sounds like the choice of BJT is the culprit.

- - - Updated - - -

Hi,

Of course I did not build it on solderless breadboard that has additional stray capacitance and inductance all over the place.
Oh, you didn't build it but you are simply simulating it? My FM transmitter simulates perfectly with its audio transistor oscillating at 100MHz.

It helps to read to the end of post #1, the next-to-last line. Could you post the schematic for your fantastic FM transmitter instead of telling us about it? I think that's what the OP would appreciate more than a pointless qualitative description of a working version.

 

[andre_luis]

@d123, I did some simulations here with the BC548 and 2N3904 and was able to confirm a very coherent operation with the one reported by OP, which was in the range of few over 70MHz. Since I did not do any calculation to optimize the bias current for the maximum value of the product gain x frequency, it would not be surprising that this oscillation value could even increase a few by changing values of some components.

When switching the original BC548B through 2N2222 the oscillation frequency decayed, which at first sight might sound strange, but the new frequency was approximately 50MHz, took by transient and FFT analysis, which leads to the possibility that the above circuit was originaly designed to operate in overtone (we can see peaks multiple of 50MHz).



Regarding the higher oscillation frequency unexpectedly obtained with the less suited transistor, I think this could be due to the higher h_fe of the BC548, but it is just a guess, it is really strange (or the parameters of the LTspice NPN library models are only truthful for the application range to which they were designed ? whatever...).

Anyway, as I am not familiar in this area of RF, something that caught my attention was the lack of a shock inductor in the transistor collector, as well as the fact that the above circuit is slightly different from some topologies called Colpits, but I may be completely wrong in my assumptions.

 

[Audioguru]

I didn't post the schematic and photo of the stripboard layout for my FM transmitter because it is a different version of a Colpitts oscillator and it has been all over the web for the last 13 years. Here it is and here is the RF part tweaked for the best 100MHz sinewave and simulated:

 

[d123]

@d123, I did some simulations here with the BC548 and 2N3904 and was able to confirm a very coherent operation with the one reported by OP, which was in the range of few over 70MHz. Since I did not do any calculation to optimize the bias current for the maximum value of the product gain x frequency, it would not be surprising that this oscillation value could even increase a few by changing values of some components.

When switching the original BC548B through 2N2222 the oscillation frequency decayed, which at first sight might sound strange, but the new frequency was approximately 50MHz, took by transient and FFT analysis, which leads to the possibility that the above circuit was originaly designed to operate in overtone (we can see peaks multiple of 50MHz).

Regarding the higher oscillation frequency unexpectedly obtained with the less suited transistor, I think this could be due to the higher h_fe of the BC548, but it is just a guess, it is really strange (or the parameters of the LTspice NPN library models are only truthful for the application range to which they were designed ? whatever...).

Hi andre,

I see, I see... Do I sound convincing enough...or is it obvious I'm "empty answering" to save face because most of that knowledge is beyond me!

Seriously, I'm glad you said that because the thought about BC5xx beta passed through my mind but felt like an intuitive yet incongruous link/assumption as to simulation results. The BC5xx and BC8xx datasheet I have says fT 100MHz - at least 150MHz lower than the 2N2222A sheets I have. Go figure, unless your theory about beta is right on that point. Total switching characteristics for 2N2222 says 320nS, which would mean about 3MHz. Surprisingly, the BCxxx datasheet hasn't got those figures so I can't see how they compare to 2n22 fT (without searching online for another, I mean).

I still think that some of the older models must be quite polished and perfected by now, but/and perhaps they are a little too good and in that way lead to over-optimistic simulation results. Simulating is an excellent guide when taken warily and with a pinch of salt at times, in my case at least (functional circuits that can't simulate and vice versa, and my limited experience means some simulations are unintentionally misleading).

What I would wonder, and Audioguru will obviously know this, and I don't mean to be rude to the original oscillator designer - it's a doubt based on my limited experience and certainly the none I have with Colpitts oscillators beyond reading about them ages ago - is if the original circuit that presumably was sourced from some online resource or other is a circuit that works...just...and then under one set of operating conditions amd maybe even at only one miserably narrow temperature band (25ºC tested and guaranteed style!), etc...but is maybe not a bona fide repeatable and reliable design (and you can kind of guess that the extent of thorough testing was until a simulation worked and that's good enough, but not then checking across temp range, fault conditions, and all that stuff) - like the things I may design: for a one-off maybe, you'll get away with it but for production - you must be joking, in a literal sense.

Thanks for your answer and the time you spent checking the circuit, really helpful.