Common Emitter Amplifier Questions

[sa230e]

Hey,

I'm a real electronics noob. I was really into it a couple years ago and I sort of put it on the back burner and now I've been slowly getting back into it - starting with the basics: simple amplifier circuits. I have some NPN BJTs and I want to use them to amplify a guitar signal and I've been reading tutorials and simulating some basic circuits in LTSpice and I have some questions.

Fig 1:

So I have this common emitter amplifier pictured (fig 1) and it works just as expected. The quiescent collector voltage is 6v (1/2 of my Vcc) it has a voltage gain of 5. I'm feeding it a sine wave with amplitude 1v p-p and I get a sine wave 5v p-p at the collector. So far so good.

Fig 2:

So I increase the input signal to 2v p-p (fig 2) and it starts to clip asymetrically. I thought transistors were supposed to clip symmetrically. The bottom half of the signal clips at around 2v but the top half is undistorted.

Fig 3:

When I increase the input signal amplitude a lot, to 10v p-p (fig 3) the output starts to get REALLY ugly. The top half of the signal clips at 12v just like I would expect but the bottom half of the signal gets these "humps". I've circled them in red. What is going on here? I can't imagine it would sound good. Also the "width" of the waveform seems to be distended, if that makes any sense.

Fig 4:

Now if I bypass the emitter resistor (fig 4), the "humps" go away (and the voltage gain shoots way up as expected) and the shape of the waveform changes even more. It looks a square wave with a very large duty cycle. I guess that would make sense if it was clipping very assymetrically but my bias point doesn't change. It stays centre biased regardless of the amplitude.

I was hoping to design a simple amplifier with a few cascaded BJT common emitter stages to get plenty of gain and distortion but I didn't expect the transistors to behave this way. I actually built a small amplifier with JFETS (1 stage common source unbypassed feeding into a JFET mu-amp) and I didn't get this effect. It clipped nicely and I was actually surprised at how good it sounded. Am I doing something wrong here? I admit I'm a newb and I've had a harder time wrapping my head around how a BJT works.

Thanks in advance for your help.

[Enzo]

Your bias is not the voltage on the base, it is the voltage on the base with respect to the emitter.   At idle, what voltage is on the emitter?

[sa230e]

Your bias is not the voltage on the base, it is the voltage on the base with respect to the emitter.   At idle, what voltage is on the emitter?

Quiescent emitter voltage is 1.2v. The tutorial said it should be around 10-15% of Vcc.

Also, isn't the voltage of the base relative to the emitter constant (~0.7v) since it's a diode?

[Loudthud]

In figure 3 the low impedance of the generator is feeding signal through the base-emitter junction (which is acting like a diode) and pulling up on the emitter resistor. This subtracts from the voltage available to the collector resistor. The transistor is saturated which means the collector to base junction is also forward biased so signal also feeds directly from base to collector without the usual phase inversion. C1 takes on a charge different for the quiescent no signal condition which makes the output wave shape change over the first few cycles. To minimize this effect you need a resistor in series with the signal on one side of the capacitor. Solid state circuits sometimes act badly when severly overdriven. The usual remedy (bandaide) is diode clipping to limit signal amplitude.

[sa230e]

In figure 3 the low impedance of the generator is feeding signal through the base-emitter junction (which is acting like a diode) and pulling up on the emitter resistor. This subtracts from the voltage available to the collector resistor. The transistor is saturated which means the collector to base junction is also forward biased so signal also feeds directly from base to collector without the usual phase inversion. C1 takes on a charge different for the quiescent no signal condition which makes the output wave shape change over the first few cycles. To minimize this effect you need a resistor in series with the signal on one side of the capacitor. Solid state circuits sometimes act badly when severly overdriven. The usual remedy (bandaide) is diode clipping to limit signal amplitude.

Thanks.  :tu: So how would you go about calculating the value of the series resistor? Does it matter which side of the capacitor it goes on? Does the resistor affect the gain of the stage?

[J M Fahey]

99.9% of people who write thousands of pages about tube vs transistor clipping HAVE NO CLUE.

Even BIG names, GURUS, etc.

When explaining things they DRAW nice little squarewaves and call them SS, then nice little rounded top and bottom sinewaves and call them Tube.

BOTH are wrong.  :trouble

The only way to know is to SCOPE or its poor cousin: simulate.

Thanks God we have Loudthud here who belongs to the 0.01% who actually builds, scopes and posts.   :dbtu:

And now you who at least simulates, so you are in the top 0.1%

Yup, that´s how a single transistor stage clips when heavily overdriven.

What loudthud said is correct, once the transistor saturates, it becomes just a closed switch, collector gets practically shorted to emitter, so by definition both carry same voltage or signal, and whatever appears at the emitter will be at the collector.

So if driven by a low enough impedance generator, signal at the base will also appear, unchanged, at the emitter and the collector.

So while the transistor can handle it, the collector signal will be that at base or emitter (they track each other) inverted and amplified.
When saturated beyond that signal positive peaks will appear at the emitter/base/collector (when heavily saturated they become all the same) so also at the output, without phase inversion and unamplified.

Those are the little humps you see at the bottom.

And yes, they sound annoying buzzy. 

Try it yourself.

Add a series resistor between generator and base, vary its value between, say, 470 ohms and 22K and check its effect on the waveshape.

One small redeeming factor that you didn't consider is that if you have a preamp , powered by, say, 12V as you show, and made out of cascaded simple stages as shown, most a stage will receive is what the earlier one can supply  so probably 10Vpp (or even less) available will not make it act that funny.

One problem with simulation is that you can set it with "impossible"  parameters, in this case 12Vpp or more generator signal, zero impedance generators, etc.

Go one step ahead and add another stage after that one, drive it and post waveforms.

Things will start to get somewhat more reasonable.

Also remember that the first stage, which in theory could be driven with a powerful generator as supplied within the simulator, in practice will get a guitar signal, which by definition is weak (1Vpp tops, and that with heavy strumming) and quite high impedance.

[Loudthud]

Thanks to JM for the kind words!

I don't know of any equasion to calculate the base resistor. Since most equasions make the assumption that the circuit is acting as a linear system, they stop working when things become nonlinear. LTSpice takes many nonlinearities into account so I would expect it to do a good job in this case. The easiest thing to do is try a resistor in simulation and adjust as necessary until the desired result is obtained. The resistor will change the gain somewhat along with the bandwidth and it shouldn't matter much which side of the cap it is on.

[Roly]

I'll just add that the effect that you have circled in red is called "collector following" - as said above, when the transistor is saturated fully on the Collector will then simply follow the Base drive, the whole acting as a simple resistive voltage divider across the supply, driven at the midpoint by the Base signal, i.e. the Base is connected to the Collector and follows it.

Just a word of caution; sims like LTSpice are great these days, but they still are mathematical simulations, not the real thing and depend heavily on the quality of the device models used, particularly when it comes to modeling non-linear (overload) behaviour.  The ultimate "simulation" is to mock it up on the bench when every factor in the device "model" is taken care of because it's the real thing.

Also; sims can only only show you what is "on circuit" and an amp that sims very nicely may have horrible hum if you are unwise enough to place the mains switch right next to the input socket (as I just witnessed on a build).  With some circuits such as Radio Frequency power amplifiers "off circuit" stray capacitances and inductances can become so significant they can make a nonsense of a simulation.  Guitar amps are pretty tolerant but there are still limits to the liberties you can take with "off circuit" physical layout and construction techniques.

Engineers have a saying - "throw it up against the universe (i.e. build it) and see what sticks".

[sa230e]

99.9% of people who write thousands of pages about tube vs transistor clipping HAVE NO CLUE.

Even BIG names, GURUS, etc.

When explaining things they DRAW nice little squarewaves and call them SS, then nice little rounded top and bottom sinewaves and call them Tube.

BOTH are wrong.  :trouble

Interesting. Actually that's another topic I was getting to: what the are differences (if any) in the way semiconductors clip. But I think that topic deserves it's own thread.

The only way to know is to SCOPE or its poor cousin: simulate.

Thanks God we have Loudthud here who belongs to the 0.01% who actually builds, scopes and posts.   :dbtu:

And now you who at least simulates, so you are in the top 0.1%

I'm not going to lie, I don't simulate for the love of objectivity. I simulate because none of the things I build would work if I didn't. I'm just not anywhere near the level where I can design a circuit on paper and know how it will behave. If I disprove any "myths" to myself or anyone else, it's by accident. I'm probably labouring under as many naive and misguided assumptions as anybody.

Also, I'm kind of lazy. It's easier to throw something together in a simulator than on the bench. So I like knowing if something won't work before I go to the trouble of building it.

Yup, that´s how a single transistor stage clips when heavily overdriven.

What loudthud said is correct, once the transistor saturates, it becomes just a closed switch, collector gets practically shorted to emitter, so by definition both carry same voltage or signal, and whatever appears at the emitter will be at the collector.

So if driven by a low enough impedance generator, signal at the base will also appear, unchanged, at the emitter and the collector.

So while the transistor can handle it, the collector signal will be that at base or emitter (they track each other) inverted and amplified.
When saturated beyond that signal positive peaks will appear at the emitter/base/collector (when heavily saturated they become all the same) so also at the output, without phase inversion and unamplified.

Those are the little humps you see at the bottom.

Gotcha. That makes sense.

And yes, they sound annoying buzzy. 

Try it yourself.

Add a series resistor between generator and base, vary its value between, say, 470 ohms and 22K and check its effect on the waveshape.

Thanks, I'll try it out.

One small redeeming factor that you didn't consider is that if you have a preamp , powered by, say, 12V as you show, and made out of cascaded simple stages as shown, most a stage will receive is what the earlier one can supply  so probably 10Vpp (or even less) available will not make it act that funny.

Yeah that's what I was sort of trying to simulate. All I wanted was to overdrive the transistor. I didn't think it would make a difference if I did it with signal generator or another stage. It was just easier to set up the signal generator for 10v p-p than to build another stage.

One problem with simulation is that you can set it with "impossible"  parameters, in this case 12Vpp or more generator signal, zero impedance generators, etc.

Go one step ahead and add another stage after that one, drive it and post waveforms.

Things will start to get somewhat more reasonable.

Also remember that the first stage, which in theory could be driven with a powerful generator as supplied within the simulator, in practice will get a guitar signal, which by definition is weak (1Vpp tops, and that with heavy strumming) and quite high impedance.

Actually I didn't mention I was planning on using a JFET for the first stage to present a high input impedance to the guitar and avoid "tone-sucking".

[sa230e]

I'll just add that the effect that you have circled in red is called "collector following" - as said above, when the transistor is saturated fully on the Collector will then simply follow the Base drive, the whole acting as a simple resistive voltage divider across the supply, driven at the midpoint by the Base signal, i.e. the Base is connected to the Collector and follows it.

Thanks. Never heard the term before. You'd think it would be mentioned more often. Maybe with the prevalence of cheap op-amps designers don't even consider using BJTs for high gain amplification anymore?

Just a word of caution; sims like LTSpice are great these days, but they still are mathematical simulations, not the real thing and depend heavily on the quality of the device models used, particularly when it comes to modeling non-linear (overload) behaviour.  The ultimate "simulation" is to mock it up on the bench when every factor in the device "model" is taken care of because it's the real thing.

Yeah but it's the best I can do right now. I don't own a scope and can't really justify the cost so LTSPICE is a HUGE help for me. Generally, if something I design doesn't work in a simulator it doesn't work on the bench either so it saves a lot of time and effort.

Also; sims can only only show you what is "on circuit" and an amp that sims very nicely may have horrible hum if you are unwise enough to place the mains switch right next to the input socket (as I just witnessed on a build).  With some circuits such as Radio Frequency power amplifiers "off circuit" stray capacitances and inductances can become so significant they can make a nonsense of a simulation.  Guitar amps are pretty tolerant but there are still limits to the liberties you can take with "off circuit" physical layout and construction techniques.

Engineers have a saying - "throw it up against the universe (i.e. build it) and see what sticks".

I was planning on powering it with a battery or maybe a regulated wall wart so I don't think mains hum will be a huge issues. Point taken though. I have no interest at the moment of working with frequencies higher than audio frequency. I just want to build guitar gear.

[Enzo]

Maybe with the prevalence of cheap op-amps designers don't even consider using BJTs for high gain amplification anymore?

Not at all.  Look at any of the Peavey TransTube models.  A lot of their older model names now also have been done as TransTube.  Bandit for example.   Those preamps are all transistor.    "TransTube" is just their name for using transistors to try to get tube-ish sound.   They don't sound bad.

Scopes can cost a bit new - $300-400 for something basic.   I am not a fan of computer based test gear, but especially for low voltage stuff like you are considering, there are scope accessories for your computer that cost a lot less than a new scope.   But scopes also come up on Craigs List cheap and your local Amateur radio gathering - Hamfests - sometimes will have working scopes for next to nothing.   The surplus facility at my local university also has them now and then for $50-100.

I was planning on powering it with a battery or maybe a regulated wall wart so I don't think mains hum will be a huge issues.

No hum with a battery, but wall warts are notoriously poorly filtered.  If you plan to use one, you will want to add a filter circuit between it and your work.


A regulated wall wart might do for you, but if you plan to do any amount of work of this sort, you really ought to consider building or buying a bench power supply.   Typical would be a dual supply, adjustable from 0-20 or 0-30 or something like that.  Two independent supplies let you make positive and negative power rails for something.    A basic power supply with a stock LM317 adjustabl;e regulator does this nicely.

I am trying to attach an example of the Trabstube preamp:

[sa230e]

Well, it works. I was looking at the Big Muff Pi schematic and it used 10k for the series resistor so that's what I tried first and worked.



EDIT: However, when the bypass cap is removed the "collector following" returns and I had to up the value to 27k.

With no bypass cap and no series resistor the amplifier has a gain of 5. However when I add the 27k series resistor, the gain drops off quite substantially ( I measured it at 1.8 ). It seems like the higher the amplitude of your signal, the higher the value that resistor needs to be to keep the transistor from saturating BUT the higher the value of the resistor, the less voltage gain you get out. So they sort of work against each other.

So it looks like if you want lots of gain without saturating the transistor, you NEED the bypass capacitor.

Interesting.

[Roly]

One reason that "collector following" is not commonly known is that the audio world for the most part concentrates on linear operation, that clipping is avoided like the plague and so gets little attention from the Hi-Fi brigade; to them it's a "fault" condition, but guitar amps spend most of their working life in some sort of overdrive and non-linear operation is of much more interest to us.


These days I use LTSpice a lot despite having a workshop full of test gear, for similar reasons, but mainly because it's quicker and easier than "birdsnesting" a circuit on the bench.  All I'm saying here is that it's a first (and possibly second and third step) in the design process, but it's not the last word - that comes from a hardware build.

I only used the example of putting the mains switch next to the input because it was under my nose, but there are many others more subtle.  To take your point about using a battery; this is fine while the battery is fresh and has a low internal resistance/impedance, but as it runs down the internal resistance rises and this can produce instability such as "motorboating" (very low frequency blocking oscillation) in the powered circuit unless a "useless" 100uF cap is included across the supply to provide a low impedance path to ground for signals impressed on the supply rail.

There are quite a few hidden traps with circuit layout.  The circuit may show an 0.1uF bypass across an op-amp, but where exactly does it go w.r.t. the device?  If it is at the bench supply at the far end of a couple of one foot clipleads you may well get oscillations and transient instability on signal edges, however if it is placed "VHF style" right on the op-amp supply pins the same circuit may be quite docile.

A common mistake when building a valve amp is to place a valve too close to the power transformer.  The stray magnetic hum field can then directly modulate the electron stream in the valve and produce intractable hum until the amp is rebuilt with a better layout.

Another cause of "off circuit" problems are grounding arrangements.  The ideal is a star format single point earthing system, and in many cases you can get away with the "ground everything everywhere" to the chassis, but it is "getting away with it" and if you want to avoid earth loop induced problems you design the earth returns with the same care you design everything else.  Unless you specifically account for such "off circuit" loops etc., in your sim then it may not neatly translate into hardware.

Sims, like all our measuring instruments, have limitations (e.g. most multimeters will also respond to the presence of DC when trying to measure AC, say ripple on the supply rails) and give you nonsense results.  This doesn't make the meter useless, just that you need to be aware of its limitation and know how to work around it to get sensible results.

I am not saying that theory and practice don't agree, my mentor would say that "if you think theory and practice don't agree then you haven't applied enough theory", but some of that theory is obscure and "off circuit" and that experience teaches you technique which is the application of (more) theory in the actual construction phase.

This is why the circuit alone is seldom the entire story, that "ground" here may not actually be the same as "ground" over there, and the difference may cause obscure and frustrating problems.  Even in guitar amp builds overlooking impedances in common, particularly with ground returns, is a common cause of problems.

[sa230e]

With regards to "hidden traps", I'm aware of them but I guess I've been lucky enough not to run into them on the bench. So far I've designed and built one amplifier from scratch and was lucky enough to have to not need to do a lot of troubleshooting to get it to work. I designed it in LTSPICE and built it on a prototyping board. I even built a little enclosure for it.

I did have problems with oscillations initially but it turns out it was because I had the input jack wired backwards. I simply swapped the polarity and it worked without a hitch.

For the layout, the rule of thumb I followed was that I built the preamp on the right side of the board and the power amp on the left. It was powered off a 9v battery so there wasn't much of a power supply but I would have put that on the right as well, far as possible from the preamp. The idea was to keep the high impedance, high gain circuitry isolated as possible and I think it worked decently.

I designed it like that because of the reasons you noted. When I started in electronics I was only interested in making tube amps so I came across a lot of material about hum and interference, like how to twist the heater wires, adding a DC offset to the heater supply, orienting the output transformer 90 degrees from the power transformer and so on.

Admittedly, I used to be in the tubes good, transistors bad camp but I started building solid state amps to get experience and also because of cost. Building that little amp actually disabused me of much of my tube-snobbery. It actually sounded better than I expected. It had a nice bluesy tone which I used to associate with tube amps. I used to think transistors clipped really hard so any sold state amp I built would have a hard, heavy-metal kind of tone. So I was surprised when I started testing it with my cheap Fender Squire Strat that it sounded pretty anemic. I didn't know how much gain I needed and it turns out my little amp didn't have enough for my liking. I was pretty unhappy with it until I plugged a guitar with humbuckers into it. Then it sounded much hotter!

So now what I'm doing is basically the next revision of that design. I wanted to build it on the same principles but I want the new version with more gain and a little more tone control. Maybe I'll start a thread about it when I get the design fleshed out a little more.

[sa230e]

Not to belabor the point or anything but since we're on the subject of collector following, I'm guessing this would be "drain following"?



I won't bother any one for solutions since playing around in the simulator suggests the solution is the same: a gate resistor. Interestingly I did need a much larger resistor to get this circuit to play nice (I used 500k) but it works.