Under which conditions does NPN astable multivibrator start oscillating?

[king.oslo]

Hello there,

I've been trying to prototype an astable multivibrator all day. My schematic is attached. I've followed this introduction: http://www.electronics-tutorials.ws/waveforms/astable.html I've searched google, and it seems everyone whom is talking about the circuit "assumes" that it will start oscillating, and only talks about the RC time constant, and how the nodes charge up/discharge as transistors open and close.

I've built it up on my breadboard. I've used BC368 NPN transistors and 1n caps. At first I thought there was a problem with base current (but I dont think that is the problem) I attempted to change the resistors to make it start oscillating at the desired frequency and current consumption. It starts oscillating with R1 = R4 = 47ohms (too high, draws .1amps) and R2 = R3 = 3K3. But if I increase R1 and R4, it wont start. It also starts oscillating at R2 = R3 = 10K and R1 = R4 = 1K (frequency too low)

I am starting to think that the value of R1 and R4 is of utmost importance to make the circuit start oscillating. Is this correct?
And more importantly, how can I make this circuit really unstable so that it start oscillating every time?   

Thank you for your time.

Kind regards,
Marius

[IanB]

In the simulated world, every device is ideal and identical. Both transistors are exactly the same, each 3k3 resistor is exactly 3300.00000000 ohms, and each 1n capacitor is 0.00100000000 microfarads.

In real life, this isn't true. Everything has manufacturing tolerances and varies slightly.

Try making one of your 3k3 resistors 3k31 or 3310 ohms instead of exactly 3300 ohms. If your circuit is perfectly balanced at start up it doesn't know which way to go.

Also I think your power supply should be pure DC and not AC. Not sure why you have a sine function in there?

[king.oslo]

In the simulated world, every device is ideal and identical. Both transistors are exactly the same, each 3k3 resistor is exactly 3300.00000000 ohms, and each 1n capacitor is 0.00100000000 microfarads.

In real life, this isn't true. Everything has manufacturing tolerances and varies slightly.

Try making one of your 3k3 resistors 3k31 or 3310 ohms instead of exactly 3300 ohms. If your circuit is perfectly balanced at start up it doesn't know which way to go.

Also I think your power supply should be pure DC and not AC. Not sure why you have a sine function in there?

Dear IanB,

Thanks again for your reply.

The sine is there because in my desperation, I thought that real world power supplies were not perfect DC, so I introduced 1mV of ripple hoping it would start to oscillate then.

I changed the values of the resistors and capacitors to introduce errors. The circuit does not oscillate, not in LTSpice or on my breadboard.

Kind regards,
Marius

[IanB]

The circuit does not oscillate, not in LTSpice or on my breadboard.

That does not seem to make sense. Let me try.

[king.oslo]

The circuit does not oscillate, not in LTSpice or on my breadboard.

That does not seem to make sense. Let me try.

Thanks! m

[Wytnucls]

I have one on a breadboard here and it oscillates at 2Hz without any problems with 5VDC, starting oscillations at 2.5VDC.
At 5VDC, the circuit draws about 57mA max, with 2 LEDs to show the oscillation.
resistors: 47x2 8.2Kx2
Caps: 50V 47uFx2
Transistors: C3311x2

[IanB]

It looks like you have to use the startup option in LTspice to get the simulated circuit to start oscillating properly. The real circuit should not have a problem. You might want to double check that your transistors are good?

[king.oslo]

Neither of you used the same resistor and capacitor values as I. With certain values, it will oscillate, but not with most values I tried yesterday.

So I need to learn how to make it unstable. If you are able to simulate with the resistor values I suggested, that'd be great. R1, R4: 1K. R2, R3: 3k3. Caps: 1n.

Thanks.

Kind regards,
Marius

[BravoV]

Add 'startup' after the .tran 1s directive as Ian did at yours, that will make it more realistic to induce unbalance situation.

[king.oslo]

Add 'startup' after the .tran 1s directive as Ian did at yours, that will make it more realistic to induce unbalance situation.

Yes. Check out the attached image.

Thanks.M

[BravoV]

Try the attached LTSpice simulation file, it works flawlessly here.

Also your circuit connections is kinda weird and not common for transistor based multi-vibrator circuit, check yours against mine and IanB.

[grumpydoc]

Are you using the same two transistors as your previous thread? Have you considered that your experiments with these transistors might have <ahem> altered their characteristics? 

[king.oslo]

Try attached LTSpice simulation file, it works flawlessly here.

Also your circuit connections is kinda weird and not common for transistor based multi-vibrator circuit, check yours against mine and IanB.

You are right (Silly me, I was using too large value for maximum timestep).

If I go out and buy 2N2222, will it will work out on my breadboard you think? After fiddling with my BC368 all day yesterday, I became paranoid

Kind regards,
Marius

[king.oslo]

Are you using the same two transistors as your previous thread? Have you considered that your experiments with these transistors might have <ahem> altered their characteristics? 

I thought about it, but concluded that I didn't think so because I never let their temperature rise beyond room temperature.

[grumpydoc]

I thought about it, but concluded that I didn't think so because I never let their temperature rise beyond room temperature.

Well, umm, let me quote you from your first post in your other thread,

Consequently, the transistors get very hot.

If it doesn't work I'd get some new transistors - pretty much any small, general purpose NPN transistor should do you so the 2N2222 will be fine.

[Wytnucls]

What frequency are you trying to achieve with that set-up and these small caps?
400 KHz?
Maybe 1nF is way too small for proper oscillations.
Try 1 uF,that should give you 210 Hz.

[king.oslo]

I want approx 300KHz.M

[6502nop]

^ Go back a few and re-read BravoV's response!

I copied your schem, zipped over to my breadboard and threw this together. I used, as you emphatically wanted, two 1K and two 3K3s, two 1nFs, and a couple of matched MPSA42s (a "lighter" version of 2N2222 compatibles). You didn't specify voltage, so I just used a 9V battery.

Worked like a champ. So, I came back to post my results, when I saw BravoV's post, and looked at what I copied. My schematic is exactly like his - I copied it wrong! You need to move those caps! They should bridge R1/2 and R3/4, not R1/3 and R2/4. In your circuit, the caps never get charged/discharged by the other side's transistor turning on/off. The caps are, essentially, just biasing the bases of their own transistor.

nop

[king.oslo]

In your circuit, the caps never get charged/discharged by the other side's transistor turning on/off. The caps are, essentially, just biasing the bases of their own transistor.

nop, is that true? I have been staring at my circuit, comparing it to both IanB and BravoV for a minute now, and I cannot see this. The only thing I see, is that R2 and R3 have swapped places relative to IanB and BravoV, but they are connected to the right node.M

[Wytnucls]

Your circuit is correct.

[6502nop]

*Blink*Blink*Blink*

Boy, do I feel like an idiot. You're right - node-wise, the circuits are identical, since it doesn't matter which 3K3 resistor is used, as long as the "tank" they create connects to the opposite base.

I never could see anything in those 3-D posters, either.

However, while re-breadboarding this thing, I did manage to get it NOT to oscillate by just connecting those caps how I thought I saw your diagram in the first place: as bias caps for their own transistor. While a breadboard mistake may explain why your circuit isn't behaving on the bench, it doesn't explain why your sim is doing the same thing. 

I've never used LTSpice, but I noticed that Oslo's diagram doesn't specify Farads (he just uses "1n1" and not "1n1F", or "1000pF" as BravoV does). Does LTSpice ignore this, as it knows that it's a cap?

In any event, my circuit runs fine with the values (and nodes!) given.

nop
Off to the optometrist...

Hey, Dave? How 'bout a Dunce Cap emoticon?

[sense2k]

Hello
I am just trying to get this to work. I tried WYTNUCLS circuit and I get stuck because I am not able to see the connections clearly - can someone breadboard this and take a clear picture?
I am just starting to play around with this stuff but I have not been able to get this to work
thanks to any who help me out

[vk6zgo]

Let me declare my bias at the outset--I hate LT Spice!
I am also,not at all fond of breadboards.

I suggest the OP build up the circuit "spiderweb" fashion,with proper soldered connections,checking once,twice,thrice that it is connected correctly.
Before doing this,check the transistors with a multimeter.

[vk6zgo]

Looking back I noticed this comment:-

Neither of you used the same resistor and capacitor values as I. With certain values, it will oscillate, but not with most values I tried yesterday.

Was this in the simulation,or with the real thing?

If it is the real circuit,you already know it is wired correctly.
It is possible,but unlikely,that the transistors are running out of gain at 300kHz & a supply of 5v.

Are you sure the 5v stays at that figure-----if the "turned on" transistor draws enough current to make it dip substantially,oscillation may not proceed.

[alsetalokin4017]

FWIW... I have set up the circuit using BC337-25 transistors and the OP's original R and C values (1k, 3.3k, 1nF) and it works fine at 5V, nice and symmetrical.

I also tested with the only 2n2222s that I could find in my stash, a mismatched pair, one from Motorola the other from ST. Both metal cans. These sort-of worked, but touchy and not nice and symmetrical. (Touching the metal can with a finger killed the oscillation and required power-cycling to restart it.) The frequency was higher with these transistors, around 338 kHz.

MPSA18s did not work.

The scopeshot below is with the BC337-25 pair.

[rob77]

1nF caps are too small , if you need a reliable higher frequency oscillator, then choose a different circuit.
why 1nF might be way too small is obvious when you re-draw the circuit as attached - the astable multivibrator is in fact a 2 stage AC coupled amplifier with 100% feedback. (1nF might be too low to provide enough feedback to start oscillations - depends on the transistors, biasing...etc... of course)

[king.oslo]

Looking back I noticed this comment:-

Neither of you used the same resistor and capacitor values as I. With certain values, it will oscillate, but not with most values I tried yesterday.

Was this in the simulation,or with the real thing?

If it is the real circuit,you already know it is wired correctly.
It is possible,but unlikely,that the transistors are running out of gain at 300kHz & a supply of 5v.

Are you sure the 5v stays at that figure-----if the "turned on" transistor draws enough current to make it dip substantially,oscillation may not proceed.

Breadboard i think

[Paul Price]

Don't blame your astable problems on LT Spice or your breadboard or layout.

The problem with the OP original circuit is that his circuit is completely balanced and oscillation has a chance to be suppressed by this perfect balance, current flow is just too symmetrical at turn on.  In common use 50-yrs ago, the OP circuit shown is slightly incomplete because the emitters are not connected together to ground through a small valued resistor.

Connecting both emitters of the 2n2222's together and adding a resistor that is approx 1/10 of the collector resistors ensures a highly regenerative action even with the tinniest imbalance in collector current at powering on.

[Audioguru]

I think you zapped the 2N2222 transistors because in the first circuit the supply was too high at 12V. The reverse-biased emitter base was trying to go to -11.3V but the maximum allowed voltage is only -5V on its datasheet. The emitter base junctions probably had avalanche breakdown at about 6V which overheats the tiny junction that has no cooling. Then the transistors are ruined.

If you use a supply voltage higher than 5V with that simple circuit then you must add protection diodes.

[Paul Price]

I saw that the OP's original circuit was using a 5V(not powered with a 12Vsupply which would be a problem) with a tiny amount of AC ripple added in desperation to get it to work.

[vk6zgo]

I think you zapped the 2N2222 transistors because in the first circuit the supply was too high at 12V. The reverse-biased emitter base was trying to go to -11.3V but the maximum allowed voltage is only -5V on its datasheet. The emitter base junctions probably had avalanche breakdown at about 6V which overheats the tiny junction that has no cooling. Then the transistors are ruined.

If you use a supply voltage higher than 5V with that simple circuit then you must add protection diodes.

Pray tell me where the -11.3 V comes from?
The circuit runs from a +12V supply.
The emitters are at zero volts, the bases cannot go negative w.r.t them,hence no reverse bias.

Wait!--I see where you're coming from!
When the previously cut off transistor conducts,the capacitor is effectively inverted,presenting apptox -12v a the other transistor's base.
(It's many years since I had to analyse the operation of a simple two-transistor M/V)
I have seen many such multivibrators operate happily with +12V supplies in the days before +5V became so common,so there is maybe more to it.

[vk6zgo]

I'm thinking that the transistors we used had a higher base-emitter reverse voltage rating than the 2N2222,so we got away with it.

I learnt about astable M/Vs back in the tube days,& musing about that made me realise you were correct.

If it didn't happen the way you said,tube astables wouldn't work.
Tubes need to be biased hard negative to cut off,& of course,the voltage rating grid w.r.t. cathode,(negative or positive) is only limited by internal breakdown of the vacuum or the mounting insulation,so there is no problem.

[IconicPCB]

To guarantee start up, consider adding a couple of diodes, Google will help with circuit schematic.

[nwvlab]

FWIW... I have set up the circuit using BC337-25 transistors and the OP's original R and C values (1k, 3.3k, 1nF) and it works fine at 5V, nice and symmetrical.

I also tested with the only 2n2222s that I could find in my stash, a mismatched pair, one from Motorola the other from ST. Both metal cans. These sort-of worked, but touchy and not nice and symmetrical. (Touching the metal can with a finger killed the oscillation and required power-cycling to restart it.) The frequency was higher with these transistors, around 338 kHz.

MPSA18s did not work.

The scopeshot below is with the BC337-25 pair.

wow! What kind of caps are these? They seem to me glass diodes