Upper frequency limit of an oscillator

[hevans1944]

Therefore you need to decrease the base resistor's values or increase the collector resistor's values.

I would try both. Increase the value of the collector resistor so saturation occurs at something less than ten milliamps, say 1.5 mA and at V_{ce (sat)} = 250 mV. Decrease the base resistance to make sure the transistor is driven into saturation, V_be(sat) > 700 mV. Increase the capacitor values to compensate for the decreased value of base resistor. Try these values and see what happens: Rc = 3.3 kΩ, Rb = 2.7 kΩ, C = 0.89 nF. You could start out with C = 1 nF = 0.001 μF (a common value) to see if it oscillates (a bit on the low frequency side of your target frequency) and then introduce a variable resistance for Rb to raise the frequency to 300 kHz.

I am not a great fan of simulation, until I know that a circuit works with real components, and then the simulation can be used to gain some insight on what happens when component values are permutated. Monte Carlo analysis is good for finding out how a circuit behaves over extremes of component values, either temperature induced or simply caused by component value tolerances. LTSpice is one of the better simulator softwares available, but it may need a little "help" to get the circuit oscillating. IMO, the best "simulator" is a pair of transistors, four resistors, and two capacitors wired up properly and connected to a stable voltage supply to implement an astable multivibrator. Once you have that working, adjust component values for the desired effect.

An oscilloscope is an invaluable tool to use when "playing" with oscillators.

Maybe it is built on a solderless breadboard and its stray capacitance between all the rows of contacts and wires and inductance of its many long wires prevent the very low value capacitors and very high value base resistors from working properly. ...

I agree completely. Picofarads and solderless breadboards do not mix well for prototype circuits. Also, for frequencies much greater than a few kilohertz, it may be easier to buy or build a crystal oscillator and divide the frequency down with binary counters. RC circuits do not make oscillators with good frequency stability. OTOH, LC oscillators are usually adequate if designed to oscillate at sufficiently high frequencies and then divided down.

 

[Harald Kapp]

they have to be different for it to oscillate in LTSPICE

Not necessarily. You can simulate with equal values by using the .IC statement to force the oscillator into an asymmetric initial condition.
Otherwise you are right: A perfectly symmetrical multivibrator won't simulate as expected in any SPICE variant.

 

[Harald Kapp]

Try these values and see what happens:

The link in post #4 shows a table of suitable configurationjs for R and C which is at least a good starting point to trim the circuit to the expected frequency. The minimum capacitance used is 1 nF which will also make the circuit much less susceptible to stray capacitances in the pF range.

 

[Akshatha Venkatesh]

The BC846B has a DC current gain between 200 and 450.
The transit frequency (where gain passes below 1) is >=100 MHz, therefore 300 kHz should pose no problem.

Are you sure you have all connections correct?

Actually I think my calculations are okay for now , but since both the sides are balanced , as soon as I start up , I am not getting oscillations in the simulation for a certain microseconds and then it starts oscillating. I just want to know how to calculate the base resistor value so that one transistor definitely turns on first . For example , If I change one side of 140kohm to 150kohm, it works , but how do I get it in calculation.

 

[Harald Kapp]

Why do you insist on these high resistor values and small capacitances?

I am not getting oscillations in the simulation for a certain microseconds

That's natural, see my post #22.

 

[WHONOES]

If you insist on using such high value resistors, stray capacitance is going to have a substantial effect. Also, as has already been alluded to, Ft will also come into play. ac gain will diminish with increasing frequency. It could be that your circuit simply runs out of gain or that stray capacitance is swamping it. Try smaller value resistors and larger capacitors.

 

[Akshatha Venkatesh]

I would try both. Increase the value of the collector resistor so saturation occurs at something less than ten milliamps, say 1.5 mA and at V_{ce (sat)} = 250 mV. Decrease the base resistance to make sure the transistor is driven into saturation, V_be(sat) > 700 mV. Increase the capacitor values to compensate for the decreased value of base resistor. Try these values and see what happens: Rc = 3.3 kΩ, Rb = 2.7 kΩ, C = 0.89 nF. You could start out with C = 1 nF = 0.001 μF (a common value) to see if it oscillates (a bit on the low frequency side of your target frequency) and then introduce a variable resistance for Rb to raise the frequency to 300 kHz.

I am not a great fan of simulation, until I know that a circuit works with real components, and then the simulation can be used to gain some insight on what happens when component values are permutated. Monte Carlo analysis is good for finding out how a circuit behaves over extremes of component values, either temperature induced or simply caused by component value tolerances. LTSpice is one of the better simulator softwares available, but it may need a little "help" to get the circuit oscillating. IMO, the best "simulator" is a pair of transistors, four resistors, and two capacitors wired up properly and connected to a stable voltage supply to implement an astable multivibrator. Once you have that working, adjust component values for the desired effect.

An oscilloscope is an invaluable tool to use when "playing" with oscillators.


I agree completely. Picofarads and solderless breadboards do not mix well for prototype circuits. Also, for frequencies much greater than a few kilohertz, it may be easier to buy or build a crystal oscillator and divide the frequency down with binary counters. RC circuits do not make oscillators with good frequency stability. OTOH, LC oscillators are usually adequate if designed to oscillate at sufficiently high frequencies and then divided down.

Hi, I tried the values that you have mentioned in Ltspice. I'm not getting any oscillations, the output is a constant 0.7 V. The transistor that I'm using is BC846B.

 

[Audioguru]

The values in a simulation are identical and there is no noise to start oscillations. One resistor or capacitor must have a different value than the other one and the circuit might need to be "kicked" to get oscillations started.

 

[Harald Kapp]

One resistor or capacitor must have a different value than the other one

No. Using initial conditions (post #22) you get the simulation started even with identical components.

 

[Audioguru]

Most of us build a real circuit and don't know about in a simulation you can use the .IC statement where you can specify asymmetrical starting conditions which kicks it to begin oscillating.

 

[(*steve*)]

Most of us build a real circuit and don't know about in a simulation

I don't know about simulation

Fixed it for you.

 

[Harald Kapp]

Most of us build a real circuit

Many, if not most, of those who use simuators think it is a design or development tool. It isn''t. Design and development require circuit theory, knowledge of components and math. The simulator is for veriification. One needs to know how o use a simulator correctly to get meaningful results.
If a simulator is used without a solid background of circuit theory, there's little chance to understand and verify the vaildity of the simulation results.

 

[hevans1944]

Most of us build a real circuit and don't know about in a simulation you can use the .IC statement where you can specify asymmetrical starting conditions which kicks it to begin oscillating.

Hi, I tried the values that you have mentioned in Ltspice. I'm not getting any oscillations, the output is a constant 0.7 V. The transistor that I'm using is BC846B.

Reading this thread became so frustrating that I fired up LTSpice XVII and modeled the stupid oscillator myself. Like @Audioguru said: most of us build a real circuit. I have all the parts to do so, except for the two BC846B transistors, for which I would have to substitute 2N3904. The simulation does allow us to play "what if" games with component values, assuming we know WTF we are doing, which isn't always the case.

Anyhoo, I am glad I made the model because version of LTSpice IV on my computer was waaaay out of date. Just goes to show you what I think of circuit modeling as an electronics hobby... get real parts and assemble real circuits if you want to pursue electronics, whether professionally or as an amateur. Attached is a screenshot of the model results showing the oscillator perking along at about 260 kHz. Lowering the supply voltage from 9 V to 5 V and decreasing the base capacitors to 2700 Ω raises the frequency to the 300 kHz target frequency. I have no intention of actually building this circuit from real components, but the OP should.

 

[Audioguru]

When I fire up my LTspiceXVII it warns me that the version is four hundred plus days old. But the last time I updated it screwed up really badly but works fine the way it is now.

 

[hevans1944]

Many, if not most, of those who use simuators think it is a design or development tool. It isn''t. Design and development require circuit theory, knowledge of components and math. The simulator is for veriification. One needs to know how o use a simulator correctly to get meaningful results.
If a simulator is used without a solid background of circuit theory, there's little chance to understand and verify the vaildity of the simulation results.

I remember the origins of SPICE (Simulation Program with Integrated Circuit Emphasis) from the 1970s while pursuing an electrical engineering degree, part-time, while working full-time for the University of Dayton Research Institute. It came out of the semiconductor integrated circuits industry's requirements to model an IC before committing to a tape-out and masking for initial production.

At that time I was studying nodal analysis of circuits using differential equations numerically solved with the 4th-order Runge-Kutta approximation method, which was particularly suitable for implementation with a computer program. My professor was developing software, for us to use in his class, that was reminiscent of what SPICE turned out to be. He may even have "borrowed" some of the concepts from work published by University of California at Berkley...

SPICE was announced to the world ... in Waterloo, Canada at the Sixteenth Midwest Symposium on Circuit Theory on April 12, 1973. The paper was presented by none other than Professor Donald O. Pederson of the University of
California, Berkeley.

Back then, all IC designs were laid out on red Rubylith film, with apertures mechanically cut out (usually with an x-y plotter) and the red Rubylith peeled away manually from the transparent backing layer to allow reduction-imaging and subsequent exposure of a photo-lithographic resist coating spun onto the wafer. This was, and still is, a time-consuming, labor-intensive, expensive procedure, alleviated today only somewhat by CAD automation and laser exposure of artwork instead of mechanical plotting.

SPICE allowed the semiconductor wafer manufacturers to verify circuit design before committing to actual wafer fabrication. It wasn't a panacea for all problems, and you needed "big iron" (mainframe computers) to obtain simulation results in a reasonable period of time. The absolute luxury we experience today of personal computers running free SPICE software with a graphic user interface at gigahertz clock speeds and producing interactive results in near real time just didn't exist "back in the day." Most of us who had any access at all to SPICE-type software had to punch a Hollerith card deck and submit it to the computer priests in white coats, ensconced behind glass-walled air conditioned rooms. Then we waited days (sometimes weeks) to get a paper printout of results. Wash, rinse, and repeat cycles were impractical, so we learned to get it right the first time... or maybe the second or third time.

There are subtle physical design rules, that vary from one process to another, that must also be followed to obtain profitable yields from a wafer. But SPICE was a very important first step in the design verification process that allowed IC "building blocks" to be created and used over and over again to construct new integrated circuits. It was instrumental in getting us where we are today, but as @Harald Kapp pointed out, SPICE was never a design tool. You had to have the right chops to be an integrated circuit designer, and you didn't acquire them by becoming an expert on how to use SPICE.

 

[hevans1944]

When I fire up my LTspiceXVII it warns me that the version is four hundred plus days old. But the last time I updated it screwed up really badly but works fine the way it is now.

When I fired up my LTSpiceIV and tried to "Sync Release" from the Tools menu choice, it failed to do so. So I went to the website and downloaded the latest version of LTSpice, which happened to be LTSpice XVII. Easy peasy and I was up and running in no time. Had a few false starts (probably my fault) that locked up the program, but exiting the program from Task Manager solved those problems. Had to play a bit with the resistors and the .ic command to get the circuit to oscillate, but I didn't attempt to find out what was "wrong" with my "design" by doing any further analysis. I just farkled with the resistor values until I got close to the target frequency of 300 kHz and then called it quits. Played with the supply voltage, raising it from 5 V to 9 V, "just in case" I decide to breadboard the circuit, but I think I have better use for my time. Maybe next week I'll fire up the Texas Instruments TINA program and take it out for a spin... it's been a while since I installed that one too, so it probably could use an update.