Noise on Emitter Follower

[fatalelieberi]

Hi,

So, I'm learning emitter follower by building a circuit that is described in Learning the Art of Electronic book in the Lab 3 chapter. In the book, there's an emitter follower circuit to demonstrate the clipping effect and how to remedy it. The circuit is attached below. However, there's a note in the book that a base resistor is needed to omit any oscillation. I was very skeptical when I read that note, so I tried to make the circuit without the base resistor.

At first, the circuit was fine. No noise was found. However, it started to pickup noise randomly. Sometimes it was free from noise, the other time, the noise came. I also tried to reduce the noise using the suggestion from the book, such as shortening the wires, and putting some bypass capacitors. However, I found out that the noise came when the output wires (the red and black wire with alligator clips) of the signal generator were moved apart. You can see the picture in the attachment.

For the cause, I suspect that when I moved the wires apart, the wires formed some kind of antenna and it picked up radio waves because when I zoomed in, it had 50MHz sine wave with 1.5 max amplitude in the noise. I researched that 50MHz band is used for amateur radio. However, is my guess correct? Is there a way to confirm it?

Also, when I put the resistor into the base, that phenomena didn't happen. What I don't understand is, how does the resistor suppress the noise?

[rs20]

I'm sure T3sl4co1l has explained this in the past on another thread, but I can't remember the details. But essentially, I think your BJT, parasitics of the BJT, and maybe the inductance of the loop from the siggen  have created an oscillator. Adding the resistor damps the oscillator to the extent that it stops oscillating.

Also, just for the sake of clear communication, what you have there is undesired oscillation. The word "noise" is normally reserved for broadband noise, even if they look similar on a zoomed-out oscilloscope.

You can also experiment with adding a decoupling capacitor to the power supply of your circuit, close to the emitter follower (i.e., right on the veroboard). I think the inductance of the power supply might also contribute.

[fatalelieberi]

I'm sure T3sl4co1l has explained this in the past on another thread, but I can't remember the details. But essentially, I think your BJT, parasitics of the BJT, and maybe the inductance of the loop from the siggen  have created an oscillator. Adding the resistor damps the oscillator to the extent that it stops oscillating.

So, it is not because it's picking radio waves? I thought I'm creating an antenna accidentally.

You can also experiment with adding a decoupling capacitor to the power supply of your circuit, close to the emitter follower (i.e., right on the veroboard). I think the inductance of the power supply might also contribute.

Tried that, too, but it got worse even though I have put it next to the BJT. The amplitude doubled from 1.5V to 3V.

Also, just for the sake of clear communication, what you have there is undesired oscillation. The word "noise" is normally reserved for broadband noise, even if they look similar on a zoomed-out oscilloscope.

My bad, I didn't know that. I'll refer this sort of thing as oscillation then.

[mrkev]

You actually did make an antena. IMHO you are not recieving, but broadcasting... 50Mhz is probably close to the frequency limits of your BJT and that's where all the parasitic stuff matters.
Emitter follower works based on negative feedback. If you go up with base voltage, transistor will conduct more and that causes rise of the voltage over your 3k3 resistor - but that actively reduces B-E voltage, which acts against the original cause.

Now oscillations happend when you have positive feedback.  And when you get close to maximum frequency of the transitor, those things are possible, because parasitic capacity causes delay. Also, all those long wires have parasitic inductance. You can have something like 800nH there, which has impedance of 250R (and your circuit is no longer emitter follower, but something rather complex).
What probably happends is that rise in the current causes magnetic field in that "antenna", which is picked up by the base, which causes delayed rise to the current and so on.
It could also be, that you have Clapp oscillator, where most of the capacitors are in fact parasitic capacitors of the device...

I would try something like 100pF between Base and Collector.

[danadak]

This might help.


http://audioworkshop.org/downloads/AMPLIFIERS_OSCILLATION_BJT_CIRCUITS.pdf


http://www.qsl.net/kl7jef/Experiment%202-The%20Emitter-Follower%20Amplifier.pdf


The negative R view -

https://books.google.com/books?id=xYKGAAAAQBAJ&pg=PA221&lpg=PA221&dq=emitter+follower+oscillation&source=bl&ots=jgaT_2BK3J&sig=jGWfwVPE_w8FdcVXTxOcPMT9nYk&hl=en&sa=X&ved=0ahUKEwjr-taH4ZLWAhVFjFQKHSztC6I4ChDoAQhXMAw#v=onepage&q=emitter%20follower%20oscillation&f=false


Regards, Dana.

[StillTrying]

What are the supply voltages, are you now using +15V and -15V for the supply, which is not as in the 1st image.
Just to be different,  I'll say it might be the power supply oscillating.

[fatalelieberi]

What are the supply voltages, are you now using +15V and -15V for the supply, which is not as in the 1st image.
Just to be different,  I'll say it might be the power supply oscillating.

Yes, I did change it by adding the -15V. Sorry, I forgot to mention it, my bad

This might help.


http://audioworkshop.org/downloads/AMPLIFIERS_OSCILLATION_BJT_CIRCUITS.pdf


http://www.qsl.net/kl7jef/Experiment%202-The%20Emitter-Follower%20Amplifier.pdf


The negative R view -

https://books.google.com/books?id=xYKGAAAAQBAJ&pg=PA221&lpg=PA221&dq=emitter+follower+oscillation&source=bl&ots=jgaT_2BK3J&sig=jGWfwVPE_w8FdcVXTxOcPMT9nYk&hl=en&sa=X&ved=0ahUKEwjr-taH4ZLWAhVFjFQKHSztC6I4ChDoAQhXMAw#v=onepage&q=emitter%20follower%20oscillation&f=false


Regards, Dana.

Thanks a lot for the links. It is very helpful and providing me with new insights, although I haven't finished reading it.

Also, I tried to replicate this oscillation just now to see if some remedies suggested can help to prevent it. However, something very interesting happened (or didn't happen) as the oscillation just vanished! I've removed the base resistor and then moved the wires apart, but it couldn't be found. I even used longer wires to make the oscillation easily take place. But, no, I couldn't reproduce it at all. I'm very sure this isn't caused by me, meddling something in the circuit because I left it overnight and no one touched it all day. So, basically, it is still the same circuit that I used when the oscillation occurred. However, this makes my guess that it picks up radio waves stronger because I do notice that there are illegal radio transmitters in the neighborhood. Sometimes, I can pick up strange music in the background when I'm turning on my old radio, even though it is tuned correctly. In spite of it, I'll try again to reproduce the oscillation. Let's see if the "antenna hypothesis" can be broken or can be confirmed.

[matseng]

Starting at 7:15 in this video might be relevant to this. A ferrite bead on the leg of the transistor kills unwanted high-frequency oscillations.

https://youtu.be/81C4IfONt3o

w2aew to the rescue again.... ;-)

[Kevin.D]

Emitter followers can form a Colpitts type oscillator if there is some stray inductance present in the base drive and some capacitance on the output, oscillation can be sensitive to the stray output capacitance value (your manipulation of the output wires) and can cease at higher values, the best assurance of stability though is addition of base resistor as you already discovered.
Here's a link to a video by (devttys0) with a nice little example.


regards .

[David Hess]

Here is a paper which discusses the math and solutions from the dark ages before the Internet.  I have it on paper so here is a scan since I have never been able to find it available online.  Excerpt:

At high frequencies, an R-C emitter load is transformed by beta(f) to appear at the transistor base as a negative resistance in series with a capacitor.  Both the resistance and the capacitance are frequency dependent.

When an inductive source impedance is placed at the base, it results in oscillation if the external resistance of the base circuit is small enough.  Then the total resistance in the loop is zero or negative at the frequency at which XL = XC.

MOSFETs can suffer from the same problem breaking into a howl powerful enough to ensure destruction in short order.  Tetrode vacuum tubes suffer from a similar problem with negative grid resistance at low plate voltages resulting in "snivets" named for the resulting display on a television CRT. 

[Audioguru]

An emitter follower oscillates when it is missing a nearby supply bypass capacitor. All electronic circuits try to oscillate when they are built on a solderless breadboard with many long messy wires all over the place as antennas and with capacitance between them and capacitance between all the rows of contacts and the messy wires.
Make it compactly on a pcb or soldered on a compact stripboard layout with a shielded input cable and a nearby supply bypass capacitor and it will work perfectly.

[David Hess]

An emitter follower oscillates when it is missing a nearby supply bypass capacitor.

...

Make it compactly on a pcb or soldered on a compact stripboard layout with a shielded input cable and a nearby supply bypass capacitor and it will work perfectly.

In the example I gave where the emitter load is reflected as a negative impedance, adding or improving the supply decoupling makes the situation worse.  The oscillation can be stopped by adding the snubber in series with the collector supply however this is not as efficient as placing it in the base or emitter and it reduces the output compliance although this may be acceptable.

[fatalelieberi]

So, after experimenting for a while, I managed to replicate (sort of) yesterday's result but with lower frequency. I got 1-1.3MHz instead of 50MHz. I did it by using bigger base resistor. It started to oscillate when I used a 10K. Also, the bigger the resistance, the bigger the oscillation amplitude as seen in the attachment. I think it was caused by the resistance value that is too big that it cannot dissipate the oscillation. Is it correct? Additionally, the phase seemed to shift, too, as the resistance get bigger.

Also, I've tried some suggestions by adding decoupling capacitor across its power supply (+15V and -15V). However, it didn't work, no noticeable improvement, if any.  I also shortened some wires to reduce the inductance, but still no result. The only way that's going to work is using smaller resistor although I have a need to use bigger resistors (the next lab task from the book, measuring Zin and Zout).

However, I forgot to tell you guys, that I actually use a virtual ground using an LM324N, TIP41C, and TIP42C as can be seen in the attachment. Perhaps this circuit could cause this oscillation?

[David Hess]

I do not see any decoupling on your virtual ground circuit at its input or output.  Is it being used to split 30 volts into +/-15 volts for the emitter follower?

I can see the 100kHz test signal and 2.5us/div but is that high frequency oscillation being displayed as aliasing?  If so it may be a much higher frequency.

[Mechatrommer]

However, I forgot to tell you guys, that I actually use a virtual ground using an LM324N, TIP41C, and TIP42C as can be seen in the attachment. Perhaps this circuit could cause this oscillation?

whats the point of virtual gnd if you still feeding and bjt with ±voltage?, what is V+? whats the power supply for LM324? you dont give clear schematics, hence you dont get clear answer... emitter follower if done correctly will not have noise as you indicated. as many people said, its oscillation due to feedback line, and you dont show that clearly. could this cause oscillation? highly probably, but where? we dont know, since there is no clear and complete schematics...

[T3sl4co1l]

If it is intermittent, then it is probably a nearby radiotransmitter.

Does its amplitude vary, perhaps in the same way that the loudness of speech varies?  If so, it could very well be a nearby amateur radio operator.

This is a very common shortcoming of breadboarded circuits.  When I went to college, the labs were all beset by about 100mV of the college radio station (~100MHz FM).  Your only reprieve was setting the oscilloscope bandwidth to 20MHz, but this only helped so much (the filter is not very sharp, so the attenuation at 100MHz is modest).

At another lab I've worked, there was an intermittent ~30MHz tone.  It was steady in amplitude and frequency, not modulated very much.  It would come and go in the 10s of seconds range.  Made some measurements kind of annoying to do...

Here in my own lab, I see about even representation from several radio bands: shortwave (mostly nearby switching power supplies -- not my own, probably neighbors), FM BCB, and 400-500MHz (cell, ISM, and others) being the strongest peaks on the spectrum.  They aren't usually annoying (under 10mV).

As the realty people say, location location location....

Tim

[fatalelieberi]

Hi all,

Sorry for late response. I've been busy at work and only now I get the chance to update my progress. So, yesterday I found out the problem. It was the virtual ground circuit. One of the resistors was blown. As soon as I replaced it, the oscillation disappeared. I think I shorted the output of the virtual ground to -15V when measuring the waveform through the oscilloscope and function generator ground wire (I didn't know at that time that my function generator was earthed). Thanks for your guys help!

[exe]

The problem of emitter follower is related to the transistor itself, not to opamp compensation?