Now that we all realize that this is an example of a SIMPLE voltage regulator, why would we not go out of our way to make a complicated voltage regulator, with even better 'regulation'? To save time, I might point that a number of SIMPLE voltage regulators may protect circuits from each other, better than one master regulator feeding several channels of audio, doing different things.
The wiki article makes a clear distinction between a regulator, where there is an error amp element that compares ref with output and REGULATES the pass element to maintain constant Vout, and an open loop circuit like Joshua's that does not regulate.
jd
Now, many of us have known about and designed more complex voltage regulators, why not use them in all audio circuits?
First there is the single added transistor as a gain element with a Zener reference, popular in the 1960's. This did lower the output impedance, but it could be noisy, due to the Zener diode noise.
Then there was the first series of IC based positive and negative regulators, the LM104 and the LM105, designed at National Semiconductor by Bob Widlar. These are long forgotten, but are really excellent regulators based on IC design. Many 'lab' supplies used them, back in the late 60's and early 70's. However, these were not to 3 terminal type regulators, popular today, but multi-pin designs that required and external output transistor. However, they did allow for RC bypassing of the Zener reference, and therefore can be very quiet. These designs also allowed for variable voltage adjust and other features. I have used a set of these power supplies since 1974, for many of my prototypes over the decades, and can still use these supplies today, if necessary.
Why then, are these devices 'obsolete' and 3 terminal devices mostly used today? The answer is: Convenience. 3 terminal devices are easier to wire up. Of course, tradeoffs follow, such as noise, etc. Still, 3 terminal devices can still be very useful.
First there is the single added transistor as a gain element with a Zener reference, popular in the 1960's. This did lower the output impedance, but it could be noisy, due to the Zener diode noise.
Then there was the first series of IC based positive and negative regulators, the LM104 and the LM105, designed at National Semiconductor by Bob Widlar. These are long forgotten, but are really excellent regulators based on IC design. Many 'lab' supplies used them, back in the late 60's and early 70's. However, these were not to 3 terminal type regulators, popular today, but multi-pin designs that required and external output transistor. However, they did allow for RC bypassing of the Zener reference, and therefore can be very quiet. These designs also allowed for variable voltage adjust and other features. I have used a set of these power supplies since 1974, for many of my prototypes over the decades, and can still use these supplies today, if necessary.
Why then, are these devices 'obsolete' and 3 terminal devices mostly used today? The answer is: Convenience. 3 terminal devices are easier to wire up. Of course, tradeoffs follow, such as noise, etc. Still, 3 terminal devices can still be very useful.
Now, many of us have known about and designed more complex voltage regulators, why not use them in all audio circuits?
First there is the single added transistor as a gain element with a Zener reference, popular in the 1960's. This did lower the output impedance, but it could be noisy, due to the Zener diode noise.
Then there was the first series of IC based positive and negative regulators, the LM104 and the LM105, designed at National Semiconductor by Bob Widlar. These are long forgotten, but are really excellent regulators based on IC design. Many 'lab' supplies used them, back in the late 60's and early 70's. However, these were not to 3 terminal type regulators, popular today, but multi-pin designs that required and external output transistor. However, they did allow for RC bypassing of the Zener reference, and therefore can be very quiet. These designs also allowed for variable voltage adjust and other features. I have used a set of these power supplies since 1974, for many of my prototypes over the decades, and can still use these supplies today, if necessary.
Why then, are these devices 'obsolete' and 3 terminal devices mostly used today? The answer is: Convenience. 3 terminal devices are easier to wire up. Of course, tradeoffs follow, such as noise, etc. Still, 3 terminal devices can still be very useful.
First there is the single added transistor as a gain element with a Zener reference, popular in the 1960's. This did lower the output impedance, but it could be noisy, due to the Zener diode noise.
Then there was the first series of IC based positive and negative regulators, the LM104 and the LM105, designed at National Semiconductor by Bob Widlar. These are long forgotten, but are really excellent regulators based on IC design. Many 'lab' supplies used them, back in the late 60's and early 70's. However, these were not to 3 terminal type regulators, popular today, but multi-pin designs that required and external output transistor. However, they did allow for RC bypassing of the Zener reference, and therefore can be very quiet. These designs also allowed for variable voltage adjust and other features. I have used a set of these power supplies since 1974, for many of my prototypes over the decades, and can still use these supplies today, if necessary.
Why then, are these devices 'obsolete' and 3 terminal devices mostly used today? The answer is: Convenience. 3 terminal devices are easier to wire up. Of course, tradeoffs follow, such as noise, etc. Still, 3 terminal devices can still be very useful.
The simple linear voltage regulator has been shown. How can it be optimized or improved? There are several variations that are better in some design situations.
For example, what if you are making a small amplifier, without a regulated power supply, and you would like to efficiently regulate, or at least filter much of the residual ripple on the power supply from the input and voltage driver stages? Well, for a relatively efficient filter, you can remove the Zener diode completely, and make a Cap multiplier, instead. Carefully done, you can lose a minimum of volts across the output device so that you can smooth out the power supply ripple, yet lose only a volt or two. This is for when you have only a single +/- voltage supply for the whole amp. It works very well, and you will find it in most of my cheaper amp designs, for example. It also tracks the average DC output, so you don't lose anything, if you have different AC voltages, while simple fixed Zener regulators must have extra voltage drop in order to work properly with varying input voltages.
For example, what if you are making a small amplifier, without a regulated power supply, and you would like to efficiently regulate, or at least filter much of the residual ripple on the power supply from the input and voltage driver stages? Well, for a relatively efficient filter, you can remove the Zener diode completely, and make a Cap multiplier, instead. Carefully done, you can lose a minimum of volts across the output device so that you can smooth out the power supply ripple, yet lose only a volt or two. This is for when you have only a single +/- voltage supply for the whole amp. It works very well, and you will find it in most of my cheaper amp designs, for example. It also tracks the average DC output, so you don't lose anything, if you have different AC voltages, while simple fixed Zener regulators must have extra voltage drop in order to work properly with varying input voltages.
I think you are all getting too deep into the details too fast. Before the performance details of a regulator become too important perhaps some real insight into what is important in a regulator needs to be addressed.
Different applications have different requirements. the reference voltage for a DAC needs to be very precise-when the output voltage needs to be a precise value. If the absolute value is not that important is absolute accuracy important. I'm using that example to direct toward the issue of supply voltage accuracy on an amp stage. Or is low noise or low output impedance or load regulation or line regulation more important? And are there tradeoff getting one parameter in exchange for another?
The Wenzel active noise reducer idea was very interesting. Its output regulation would not be as good as many other approaches but it may get very low noise on the output, which may be more important on precision oscillators.
Its conceivable that very good load regulation may be more important that low noise on an amplifier supply if the amplifier has reasonable supply rejection.
Different applications have different requirements. the reference voltage for a DAC needs to be very precise-when the output voltage needs to be a precise value. If the absolute value is not that important is absolute accuracy important. I'm using that example to direct toward the issue of supply voltage accuracy on an amp stage. Or is low noise or low output impedance or load regulation or line regulation more important? And are there tradeoff getting one parameter in exchange for another?
The Wenzel active noise reducer idea was very interesting. Its output regulation would not be as good as many other approaches but it may get very low noise on the output, which may be more important on precision oscillators.
Its conceivable that very good load regulation may be more important that low noise on an amplifier supply if the amplifier has reasonable supply rejection.
Sure. And Zener's dynamic resistance is non-zero in respect to the value of the resistor that supply the current for it, so relatively high ripples and voltage variations supply emitter follower's base current that depends on an output current drawn. Also, Beta variations of the transistor with current apply, plus Vbe thermal dependencies...
You are right PMA, it is a boring thread...
Now, what do these simple linear regulators do right?
First, they are fast, and offer a CONSTANT output impedance over a wide frequency range with a Class A load. Second, they do not have rise time overshoot, a problem common with global feedback voltage regulators, and third, they can be very quiet. Especially, without the Zener diode being added.
Now, off hand, what do they do wrong?
Well, they have a fairly high output impedance from a few ohms to a few 10's of milli-ohms. This varies with the series pass device used, added current limiting, and with load current. Sometimes, it is not a bad idea to add a static load, just to keep the series regulator operating within a narrow range of current.
Second, they are not very stable as far as voltage, and can change slightly with load current changes, device selection, etc.
There are certainly more, but I can't think of any more, myself, right now. Any inputs from others?
First, they are fast, and offer a CONSTANT output impedance over a wide frequency range with a Class A load. Second, they do not have rise time overshoot, a problem common with global feedback voltage regulators, and third, they can be very quiet. Especially, without the Zener diode being added.
Now, off hand, what do they do wrong?
Well, they have a fairly high output impedance from a few ohms to a few 10's of milli-ohms. This varies with the series pass device used, added current limiting, and with load current. Sometimes, it is not a bad idea to add a static load, just to keep the series regulator operating within a narrow range of current.
Second, they are not very stable as far as voltage, and can change slightly with load current changes, device selection, etc.
There are certainly more, but I can't think of any more, myself, right now. Any inputs from others?
Last edited:
What does slightly mean? 100uV? 1mV? for a 100mA current step? Besides the Jung regulator I've yet to see another that has very low output impedance in the audio range. Flat amplitude and phase... wishful thinking?
How important is low impedance across the audio range? Its very important if the regulator is used to supply more than a single stage/ channel etc. But would separate followers with a good low Z reference voltage for each load give better results even if the individual regulation were not as good? What about the interaction of the regulator and bypass capacitors?
Many audio circuits don't need tight voltage regulation. You can use a RC filter after a regulator to further reduce noise. And an appropriate sized capacitor at the output reduces the dynamic impedance, like Pass uses with many of his DIY designs.
I think in simple terms. Transients will happen in the audio range. If transients result in enough voltage sag, even separately for each stage, it'll happen because the supply/regulator can't give the current demanded as it is demanded; does this matter? If the voltage modulation is high enough due to a less than ideal output impedance, it will be audible. Would you prefer it to be equally high in magnitude across the audio band?
Ideally there would be either small or no capacitors between the regulator and the powered device. Are you referring to caps in a different position?
Ideally there would be either small or no capacitors between the regulator and the powered device. Are you referring to caps in a different position?
- Status
- Not open for further replies.