Discrete 38khz astable oscillator drifting to 50Khz

[gphil5]

Hello all,

I've breadboarded a simple astable multivibrator to generate 38kHz (identical to http://upload.wikimedia.org/wikipedia/commons/6/6a/Transistor_Multivibrator.svg) using 2N3904s.

Upon power up, it's a solid 38Khz. After 60 seconds, it's drifted up to ~44Khz... and stabilizes to ~52Khz after several more minutes. I've yet to see this drift mentioned on the multitude of astable multivibrator documents I've read, and was curious if this was common, expected, or if there is something I can do to minimize the drift I'm experiencing.

Thanks in advance.

Phil

 

[NascentOxygen]

Hi gphil5. http://img96.imageshack.us/img96/5725/red5e5etimes5e5e45e5e25.gif

The first thing to rule out is power-supply drift. What are you using for your supply, and are you monitoring it to ensure it is constant and stable?

What value are the components you are using?

 

[gphil5]

NascentOxygen,

I've used both both a decent lab supply, and a cheap ebay one... same thing.

However, I hooked it up to a 9V battery after your response... no drift. Since this will be ultimately powered by battery, I guess the issue is resolved.

However... can you explain more about how a power-supply drift impacts the frequency of such a circuit? I can see it impacting the output amplitude... and even subtle frequency shifts over <1s, but I don't follow why even if my supply wasn't smooth, how I experienced a 5 minute steady ~40% rise in expected frequency.

If relevant, components are carbon film resistors & ceramic disc caps.

Thanks,

Phil

 

[NascentOxygen]

Was the lab power supply also 9v?

What value Rs and Cs is your multivibrator using?

 

[meBigGuy]

Has to be power supply drift or heating. A 30% change is pretty drastic. Your component selection could be affecting performance also. Really high resistances or really low resistances could increase sensitivity.

The circuit architecture is sensitive to supply voltage. The supply voltage affects the time to charge/discharge the capacitors through their resistors. But 30% seems large to me.

 

[Baluncore]

I would heat the individual components momentarily to test for thermally sensitivity.
Start with the ceramic caps, they can do funny things when they warm up.

 

[AlephZero]

Your circuit diagram doesn't show any component values, so the obvious question is whether the currents through the resistors are consistent with what the transistors can handle - sufficient base current to switch them fully, and not too much collector current to overheat them.

Aside from that, a 30% change in frequency caused be the static components doesn't seem very plausible, for your oscillator frequency. (If the frequency were 0.1Hz or 100 Mhz, that would be a different story!)

You don't say what transistor type you are using, but there is a "gotcha" with this simple circuit. When the circuit switches, you are applying a reverse bias to switch the transistor off, which is comparable with the power supply voltage (9V). But most low power general purpose transistors only have a reverse bias rating of about 5V, and beyond that the base-emitter junction can act as a zener diode and start to leak current.

See the section on astable oscillators here: http://www.sentex.ca/~mec1995/tutorial/xtor/xtor7/xtor7.html

This doesn't stop the oscillator working, but it does mess up the frequency stability. I suspect this is happening to one or both of your transistors, and the effect is temperature dependent.

 

[gphil5]

AlephZero...

Using labels in the above link:
R1=R4=470
R2=R3=1.8k
C1=C2=0.01uF
Q1=Q1=2N3904

I understand I'll need some trim pots to fine tune, but was just wanting to get the basics working before proceeding. I don't completely follow the R2/R1 relationship compared to Hfe. R1=150 - 470 work, with changes to the rise time on the output... which is the best range for R1?

====

Now, heating was indeed a culprit... 1/4W resistors were warm to the touch. Jumping up to 1/2W appears to have solved quite a bit. Any comments on values posted above are still welcome to fully understand.

 

[AlephZero]

R2 and R3 set the frequency.
For a good rise time, you want R1 and R4 to small be compared with R2 and R3.

But if you make the ratio of R2/R1 too big, the transistor can't fully turn on, because there is not enough base current through R2 to match the collector current through R1. For your 2N3904, the minimum hfe is 30 according to http://www.datasheets360.com/part/detail/2n3904/-4945907495188929865 so you must have R2/R1 < 30. Choosing R2/R1 = 10 is a common rule of thumb.

Instead of using higher power resistors, you would probably be better increasing all the resistor values to reduce the currents and power dissipation, and reducing the capacitor to keep the same frequency. Try something like C1 = C2 = 0.001 uF, R2 = R3 = 18k, R1 = R4 = 1.8k.

Or even R2 = R3 about 100k, R1 = R4 = 10k, and C1 = C2 = 0.22nF. (choose R2 and R3 to give the frequency you want).

The hfe of the transistors will probably be higher for small collector currents, which tend to improve the rise time. You can easily find alternative transistors with much bigger hfe, for example BC108C has hfe > 400 at 2mA collector current.

If you want to reduce the rise time even more, try some of the other circuits in my web link.

 

[Averagesupernova]

Switching from 1/4 watt to 1/2 watt will not fix anything except the resistors will be able to dissipate more heat. It will not fix your drift problem. I suspect AlephZero has hit the nail on the head concerning the reverse bias of the base-emitter junction when it toggles. Try a couple of diodes in parallel with the base emitter junction of each transistor. Anodes to ground. Also, I would use a series resistor right at the base.

 

[NascentOxygen]

AlephZero...

Using labels in the above link:
R1=R4=470
R2=R3=1.8k
C1=C2=0.01uF
Q1=Q1=2N3904

I understand I'll need some trim pots to fine tune, but was just wanting to get the basics working before proceeding. I don't completely follow the R2/R1 relationship compared to Hfe. R1=150 - 470 work, with changes to the rise time on the output... which is the best range for R1?

====

Now, heating was indeed a culprit... 1/4W resistors were warm to the touch. Jumping up to 1/2W appears to have solved quite a bit. Any comments on values posted above are still welcome to fully understand.

Your resistors are smaller (in Ohms) than I'd be using, unless there were good reason. You haven't said what load you will be driving? What are you going to be using this for?

Generally, you shouldn't be able to feel any component getting warm, it's a poor design and a waste of energy. If this is to be battery operated, why flatten the battery a lot sooner than is necessary? Do you need the operation to be symmetrical, so the output has a 50% duty cycle?

You mention risetime...so you have been examining this on an oscilloscope? If so, do the signals at each base show, for exactly 50% of the cycle, a nice smooth exponential rise? The base waveforms should be identical, so if they are not, how do they differ?

 

[AlephZero]

You haven't said what load you will be driving? What are you going to be using this for?

As a general design guideline, an oscillator that works at low power, followed by an amplifier, is better behaved than an oscillator that works at high power.

 

[meBigGuy]

Increase the resistors by 10X, reduce cap by 10X, use good quality cap (at least X7R ceramic)

Maybe use higher gain transistor (BC548)

Change over to a 555? Or is this an astable multi exercise.

EDIT -- looks like AlephZero said all this already

 

[gphil5]

Thank everyone for the input... I'll be implementing a lot of the suggestions. I appreciate the input!

Your resistors are smaller (in Ohms) than I'd be using, unless there were good reason.

The only reason I used such small resistors was because of my on-hand capacitor selection. Ordering smaller ones as suggested...

You haven't said what load you will be driving? What are you going to be using this for?

As a general design guideline, an oscillator that works at low power, followed by an amplifier, is better behaved than an oscillator that works at high power.

This oscillator is being used to switch an IR LED driver circuit.

Change over to a 555? Or is this an astable multi exercise.

BJT-emphasized project... IC's are allowed, but are discouraged. So I'm going with the self-imposed "zero ICs" route.


Again everyone, I appreciate the input!

 

[AlephZero]

This oscillator is being used to switch an IR LED driver circuit.

So, you need to find the input impedance of the driver circuit, how much current it will take from the astable, and then make sure the collector currents in the astable are bigger than that by a reasonable margin (e.g. x10).

Unless you have a very high power LED, I would guess "how much input current does the LED driver need" would be measured in microamps, so a few mA of current in the astable would be plenty. That would make R1 and R4 a few k ohms, which about what everybody was suggesting already.

 

[AlephZero]

use good quality cap (at least X7R ceramic)

I agree with use a good quality cap, but I don't know why you suggested a ceramic.

My instinct would have been to use a film cap to get better temperature stability in a timing circuit. (IMO it's a shame hardly anybody makes polycarbonate film caps any more...)

 

[tfr000]

Are you going to use it for something that needs stability? There are probably better (i.e. easier) ways to get it.

 

[meBigGuy]

I agree with use a good quality cap, but I don't know why you suggested a ceramic.

My instinct would have been to use a film cap to get better temperature stability in a timing circuit. (IMO it's a shame hardly anybody makes polycarbonate film caps any more...)

I said *at least* x7r ceramic. Better capacitors are always better.

 

[willem2]

Another possible problem is that the base of a transistor can go negative after the other transistor starts conduction. The maximum base-emitter reverse voltage of the 2n3904 is 6V. A 9 V voltage source might not be enough to cause reverse breakdown of the base emittor junction, but a larger voltage will. This is mentioned in the wikipedia article about multivibrators.

I rather wish people wouldn't link to wikipedia pictures without contex.
If you don't want to link to an article, you can link to
http://en.wikipedia.org/wiki/File:Transistor_Multivibrator.svg

 

[NascentOxygen]

Upon power up, it's a solid 38Khz. After 60 seconds, it's drifted up to ~44Khz... and stabilizes to ~52Khz after several more minutes. I've yet to see this drift mentioned on the multitude of astable multivibrator documents I've read, and was curious if this was common, expected, or if there is something I can do to minimize the drift I'm experiencing.

I'm hoping you post a follow-up, to let us know whether the drift has been minimized, and what changes you made that fixed it.