I came across this column in talkingelectronics.com where the author discusses small signal and medium signal amplifiers. This medium signal term is one that I’ve never heard used before, and I’ve read a lot of literature that uses the small and large signal terms. I’ve heard the small signal term applied when an incremental term is shown, such as when talking about the small signal current gain, hfe. This is applied when the signal is a minor part of the DC operating point, or even when the signal is a major part, as long as it’s a part, not the whole.
The large signal term is used when talking about the absolute value such as when talking about hFE. This means the transistor’s large signal current gain is the ratio of the total collector current divided by the total base current. So 1 milliamp Ic divided by 10 microamps Ib gives a large signal current gain of 100.
Another term he uses is “the capacitor transfers…” Here is part of his paragraph:
Both stages in Fig 62 produce a high gain but the final gain will depend on the amount of energy each capacitor will transfer.
For instance, the 22k will pull the 10u high but the 47k discharges the 10u and so it will be partially charged for the next cycle. This means the energy transfer will only be equivalent to a load resistor of 47k.
The “load resistor of 47k” doesn’t agree with what I’ve learned. The 100k and 47k bias resistors are a voltage divider and also have a Thevenin equivalent resistance and voltage source. The Thevenin equivalent resistance of the pair is the same as if the two resistors are in parallel, or Rth = 31.97k. The Thevenin equivalent voltage source is 3.197 volts. These are part of the load that the transistor sees. And the author forgot that the transistors are operating in the linear part, so the bases of the transistors act like a load resistance. The base of the second transistor loads the first transistor, so the load is even lower than the Rth. In this case, the transistor’s base is probably less than a thousand ohms. This means the 47k is insignificant.
He talks about “discharge and partial charge..” of the 10u coupling capacitor. From what I understand, the 10u capacitor has a very low impedance at audio frequencies and to the AC signal it is as if the capacitor is not there, it is only blocking the DC from one stage to the next. Because the AC voltage across the capacitor’s very low impedance is very low, the “charge and discharge” he talks about does not seem to matter (the capacitor is initially charged when the amp is powered up, and discharged when it’s powered down). For example, we will assume that the second stage has a voltage gain of 100. Therefore the peak-to-peak voltage at the base must be less than the supply voltage (10 volts) divided by 100, or 1/10 of a volt. Any higher voltage and the signal at the second collector would be clipped and distorted. The voltage across the 10 uF coupling capacitor is a very small fraction of that 1/10 volt.
Another point he doesn’t talk about is the value of some of the electrolytic capacitors used for bypassing (I did later find where he discusses bypassing capacitors). One example is the bypass capacitor across the emitter resistor. In the text below Fig.61b he calls the emitter resistor “negative feedback”, but negative feedback in its true sense is taking some of the output of an amplifier and feeding it back into the input in the opposite polarity so that it reduces the gain of the amplifier. Putting the resistor in the emitter is not an example of this.
From what I’ve read, the value of the emitter bypass capacitor should be ten times higher than that required for its reactance to match the emitter resistor at the lowest frequency the amp is designed to pass. For example, the amp has a 220 ohm emitter resistor, and is designed to pass 32 Hz. The reactance of a 22 uF capacitor is approximately 220 ohms at 32 Hz, so a 220 uF capacitor, ten times higher, should be used. If instead the 22 uF capacitor was used, then it would have 220 ohms reactance at 32 Hz, and that in parallel with the 220 ohm emitter resistor would cause the signal to be rolled off (amplified less) at low frequencies. This capacitor should be large for good low frequency response. But in some cases, the designer may want low frequencies attenuated, so the capacitor can be a smaller value.
He then goes on to explain where he believe is is necessary to match the output of an amplifier to the load of the following stage.
Background In the early days of electronics, when single transistors were used to amplify a signal, typically audio, the stages of amplification were coupled with matching transformers, so that the higher impedance of the output would get matched to the lower impedance of the following stage. But matching transformers were expensive, were optimized for a limited part of the audio frequencies, and prone to picking up hum and other magnetic interference, so they were replaced with R-C coupling, or resistance-capacitance coupling. The electrolytic capacitors could be made large enough to easily couple the lowest audio frequencies to the next stage. The next stage could have its input impedance increased by adding the emitter resistor, with some sacrifice in stage gain. But germanium and then silicon transistors were cheaper, and it was cost effective to add another transistor than try to match the stages with transformers.
Then the opamp IC came along, with the 741 becoming the post popular. Dozens of transistors could be included to make a high gain audio amplifier in a single package. Later, higher performance opamps came on the market, with 2 or 4 opamps in a single package, such as the 1458 and LM324. For the last few decades the audio designer has used opamps in place of transistors for most of the gain in the audio equipment. But I digress.
Back to his eBook. In any case, he dwells on this subject of matching the stages for maximum transfer of signal. This is a good idea when dealing with RF, especially VHF and above because of the need for power output. But for audio it is a non-issue. Nowadays it is most common to find the output of a stage is much lower impedance than the following input. This keeps the wide bandwidth of the signal, and has other benefits such as with noise. In fact, audio designers have to be concerned about having too wide a bandwidth which can cause instability if an amplifier is loaded with a capacitive load.
Referring to his figures 56 through 61b. He calls this a “Bridge”; I’ve never heard it called that before; I’ve heard the term voltage divider used. He shows the base bias resistors Ra and Rb. He states that in order to keep them from loading down the previous stage, “.. they have to be as high as possible.” And he uses 1M and 470k in his examples. In the textbooks, they teach us that these resistors are chosen to dominate the impedance looking into the stage’s input, since the emitter degeneration resistor causes the base to have a higher impedance. The values chosen are typically 100k and 10k for Ra and Rb respectively. Why so much lower than what he says is necessary? Because these resistors have to keep the bias voltage steady at temperature extremes, especially at high temps where the transistor’s gain and leakage can become excessive. Using the high values he says to use can become a problem at higher temperatures. If you want to test it, put your amplifier in your car on a warm sunny day and see how well it works when the sun gets the amplifier hot to the touch.
Another one of his statements I take issue with:
However the capacitor on the input will produce losses from one stage to the other and the capacitor on the output will reduce the gain of this stage.
The capacitor is there to block the DC current from going outside of the stage. At audio frequencies the capacitor has a very low impedance and there is almost no signal lost when it goes through the capacitor from one stage to the next. Perhaps he is experimenting with 25 year old capacitors that are marked 10 uF but have dried up inside so that their actual capacitance is almost nothing and the signal loss across this capacitor is excessive. If I found that my meter measured a volt or more of audio frequencies on one end of the capacitor, but a much smaller amount on the other end, I would immediately know that this capacitor is bad and needs to be replaced.
In other parts of his “eBooks” he again goes into this distractive mode where he dwells on something that seems to be a minor point, and makes a mountain out of a molehill, so to speak. Yet, as in the above example, he skips over other important points. He has a very large amount of good schematics and literature, but it’s my opinion that he has a way to go before he can say that it will make a good text for learning about electronics.
One suggestion I have for anyone who wants to learn about electronics is to buy the book “The Art of Electronics” by Horowitz and Hill (search Amazon for it). This book is expensive, so you could try to find it in a library, but it’s not the kind of text you can borrow for a few weeks and get everything you need from it. It really needs to be studied on and off for a length of time to get its full potential, which is a wealth of information. I use it as a reference for much of my electronics design. I have not noticed a single instance in the “eBooks” where the author quotes any electronics reference material. He implies that his “eBooks” and website make a good tutorial for learning electronics. I think it would be a good idea if he could point the reader to other references so the reader can get more and broader information on this subject.
Update May 13 – I continued to read other parts of his columns. I came across this item in his digital column:
How can you tell a DIGITAL CIRCUIT from an ANALOGUE CIRCUIT?
1. Absence of capacitors. There are NO capacitors in a DIGITAL CIRCUIT.
2. A switch or push-button will be activating the circuit.
3. The circuit will be driving a DIGITAL or ON – OFF item such as a relay or globe.
Well, I can think of a digital circuit that uses capacitors. A counter circuit is preceded by a Schmitt trigger to eliminate noise and make a fast risetime pulse. This is then coupled to the counter circuit through a small value coupling capacitor. For number 2, my counter example above is activated by a photocell that puts out a pulse every time a light beam is broken. No switches or pushbuttons are used. For number 3, my example counts the pulses and outputs the total to a numerical display. No relays or globes (lamps) are used. So far he has struck out on all three of his narrow-minded definitions.
So far, I get the feeling more and more as I read that a neophyte might feel he has been misled after reading his “eBooks” and then later learning the other side of the story from other textbooks.