For audio amplifiers do not have to have a THD of less than 0.1%. This is confirmed by the overall amplifier without negative feedback, including hybrid amplifiers in which the driver is made on the tuber (lamp).
It is more important to ensure optimal interaction between the amplifier and the acoustics. To compensate for lead resistance was offered U.S. patent number 4441085, MKI N03F 1/34, publ. 03.04.1984. But this method is not applicable to the amplifiers without feedback. Hawksfords corrector allows you to do is very simple, without unnecessary additional devices and connections! It's enough to unbalance correction so that the output impedance was negative in the range of 0.15 ... 0.35 ohms. This distortion is less than in a simple out stage (current amplifier) is usually not more than 0.05%.
It is more important to ensure optimal interaction between the amplifier and the acoustics. To compensate for lead resistance was offered U.S. patent number 4441085, MKI N03F 1/34, publ. 03.04.1984. But this method is not applicable to the amplifiers without feedback. Hawksfords corrector allows you to do is very simple, without unnecessary additional devices and connections! It's enough to unbalance correction so that the output impedance was negative in the range of 0.15 ... 0.35 ohms. This distortion is less than in a simple out stage (current amplifier) is usually not more than 0.05%.
Hi Davide,
Wow! Congratulations on a magnificent accomplishment. I skimmed through every page of your 100+ page text and am impressed at every step. What you have done is remarkable for someone your age, without a degree, and with limited previous experience. Your ability to read numerous books, learn from them, and put it all together in a compact, wide-ranging and comprehensive way is impressive.
As Mike said, you will make a great engineer, and do try to get a college degree. Going into college having done what you have done here will give you a tremendous head start on most of the other students. If you end up going for a Phd, what you have written here will have given you an idea of how to write a thesis. Engineers who can write and articulate concepts are valuable. Indeed, I encourage you to do some writing, perhaps some magazine articles.
If you haven't already done so, you should get or build some test equipment. Today, Ebay is a good place to start. It is cheap there. Yes, some of it doesn't work, but you could surely fix most of it. The alternative is to build some of your own, where necessary. This is how I got started and learned a lot. When I was young I could not afford a lot of test equipment and so built much of my own. That was quite a learning experience. I can't tell you how many HP and Tek service manuals I read. Even to this day, I build a lot of my own test equipment where a purpose-built piece is needed. Take a look at my old THD analyzer design on my website. It is a perfect example of a piece of high-performance test equipment I built because I could not afford to buy it. Your ability to work with microcontrollers adds a lot of value to your abilities as well.
You are well on your way to a great career.
Cheers,
Bob
I disagree with the used equipment advice – today – really in the last few years - Rigol or other low cost new equipment have become a very high value proposition
knowing the measuring tools in your lab work is a vast time saver - you will have plenty of time to develop debugging skills on you own circuits
I don't know your situation but the inflation adjusted price of my 1st car, as a working class teenager, earned out of summer minimum wage employment - would get a new 100 MHz scope, 20 MHz function generator and bench multimeter today
toss in the cost of auto insurance for a year for a young driver and you're talking Rigol's 4 channel 200 Mhz 'scope
with the modern digital connectivity you can do most of what used to be $5-6 figure specialized test equipment like Vector Impedance in software with modern scope and function gen and your PC
having the base tools that you trust to work then makes used equipment foibles easier to deal with – if you happen to be bent in the old equipment collector direction
knowing the measuring tools in your lab work is a vast time saver - you will have plenty of time to develop debugging skills on you own circuits
I don't know your situation but the inflation adjusted price of my 1st car, as a working class teenager, earned out of summer minimum wage employment - would get a new 100 MHz scope, 20 MHz function generator and bench multimeter today
toss in the cost of auto insurance for a year for a young driver and you're talking Rigol's 4 channel 200 Mhz 'scope
with the modern digital connectivity you can do most of what used to be $5-6 figure specialized test equipment like Vector Impedance in software with modern scope and function gen and your PC
having the base tools that you trust to work then makes used equipment foibles easier to deal with – if you happen to be bent in the old equipment collector direction
Last edited:
Hi Lucchi,
Well done, very nice work!
A few minor remarks, if you don't mind.
1. You wrote: 'the second stage also called voltage amplifier stage, VAS, is a transimpedance stage...'
That's right, it is a trans-impedance stage. Therefore, the purists among us call it a TIS instead of the misnomer VAS (courtesy of Mike).
2. How stable is the standing current of the a TIS? I hope it doesn't suffer from the 'Randy Slone syndrome'.
3. I'm afraid the gate protection by means of D5 & D6 isn't full proof. Under rare conditions (e.g. parasitic oscillations of the MOSFETs) the max. gate voltage may still be exceeded. Zener diodes as close as possible to the gate and source are a safer solution.
Nevertheless, the whole project (hard-/software as well as documentation) looks very professional. Congratulations!
Cheers,
E.
Well done, very nice work!
A few minor remarks, if you don't mind.
1. You wrote: 'the second stage also called voltage amplifier stage, VAS, is a transimpedance stage...'
That's right, it is a trans-impedance stage. Therefore, the purists among us call it a TIS instead of the misnomer VAS (courtesy of Mike).
2. How stable is the standing current of the a TIS? I hope it doesn't suffer from the 'Randy Slone syndrome'.
3. I'm afraid the gate protection by means of D5 & D6 isn't full proof. Under rare conditions (e.g. parasitic oscillations of the MOSFETs) the max. gate voltage may still be exceeded. Zener diodes as close as possible to the gate and source are a safer solution.
Nevertheless, the whole project (hard-/software as well as documentation) looks very professional. Congratulations!
Cheers,
E.
Hi Edmond,
It has been called the VAS (Voltage Amplifier Stage for decades. It is only confusing to try to re-name it now. We need not get obsessively purist in semantics that everyone already understands and uses.
The VAS is called the VAS because it is the place where the voltage gets very big. Not all VAS are necessarily transimpedance amplifiers, and maybe not at all frequencies. In the frequency range where the Miller capacitor is in control in a conventional VAS, the function is that of a transimpedance amplifier, and the transimpedance is the impedance of the Miller capacitor.
At very low frequencies, the VAS is more like a current-output amplifier which happens to have some poorly-defined load impedance. A folded cascode VAS may more properly be described as a current-in, current-out amplifier.
We could easily go nuts with this, so let's leave it at VAS.
I believe the stabilizing technique I described in my book to avoid the Randy Slone problem us being used.
Cheers,
Bob
Hi Bob,
>I believe the stabilizing technique I described in my book to avoid the Randy Slone problem us being used.
Indeed, same remedy to avoid the Randy Slone problem (fig. 7.13, R11 & R12). I just wonder how stable it is, as the hot VAS/TIS trannies may spoil the whole concept.
Cheers,
E.
>I believe the stabilizing technique I described in my book to avoid the Randy Slone problem us being used.
Indeed, same remedy to avoid the Randy Slone problem (fig. 7.13, R11 & R12). I just wonder how stable it is, as the hot VAS/TIS trannies may spoil the whole concept.
Cheers,
E.
Its true that one gets a lot for their money with new equipment these days, so it is indeed a matter of choice depending on one's personal preferences and needs. With respect to reliability, HP and Tek equipment, even 30 years old, is very reliable. My nominal preference is for the precision and quality of the Tek and HP equipment. Need a really good audio analog AC voltmeter? Its hard to beat an HP 400EL for about $50. 10MHz, and linear dB scale. I also enjoy the value and variety of the Tek TM500 series of modules.
If you want to spend a few thousand on a REALLY good DSO, you can pick up an Agilent Infinium, vintage ~year 2000, 8 Gs/s 4-channel DSO for about $6k.
Anyway, more good choices, new and used, is a good thing.
It is also remarkable what capabilities are available these days with PC-based test equipment, both at based on sound cards and that based on special hardware interfaces, such as PC-based DSOs.
Cheers,
Bob
Good point about temperature of the VAS transistor influencing VAS standing current. A sensitivity analysis can be helpful. We can probably do a little hand-waiving to get a rough idea.
This sensitivity will largely be governed by the amount of degeneration in the VAS. With about 10:1 degeneration of transconductance, the drop across the emitter resistor will be on the order of 300mV. If we assume that the only non-tracking temperature change is that of the VAS junction, then a 40C rise in the VAS junction temperature will correspond to a reduced junction voltage drop of perhaps 90mV, corresponding to an increase in standing current of perhaps 30% from cold to warmed-up. This is not awful for VAS standing current variation. For example, 7mA cold, 10mA warmed up. Once warmed up, VAS transistor temperature does not fluctuate much.
This scheme would not fare well with a VAS stage that has little or no emitter degeneration, but I don't recommend such VAS anyway.
Cheers,
Bob
VAS standing current
Hi Bob,
Agreed, a variation of say 30% of the standing VAS current in itself is not dramatic. BUT... if this current also influences the output voltage of the Vbe-multiplier, it may result in an unacceptable fluctuation of the output stage bias (particular in case of BJTs). One way to avoid the effect of VAS heating is to use cascode transistors, which stay much cooler, of course.
Another point of concern is the base current of the emitter followers in front of the VAS (Q17 & Q21 in Lucchi's schematic) and the offset current of current mirrors. A variation/uncertainty of 10uA for example, causes of voltage drop variation across R68 resp. R76 (33k) of 330mV, corresponding to a VAS current variation of 330mA/60Ohms =5.5mA. Hence my concerns...
Cheers,
E.
Hi Bob,
Agreed, a variation of say 30% of the standing VAS current in itself is not dramatic. BUT... if this current also influences the output voltage of the Vbe-multiplier, it may result in an unacceptable fluctuation of the output stage bias (particular in case of BJTs). One way to avoid the effect of VAS heating is to use cascode transistors, which stay much cooler, of course.
Another point of concern is the base current of the emitter followers in front of the VAS (Q17 & Q21 in Lucchi's schematic) and the offset current of current mirrors. A variation/uncertainty of 10uA for example, causes of voltage drop variation across R68 resp. R76 (33k) of 330mV, corresponding to a VAS current variation of 330mA/60Ohms =5.5mA. Hence my concerns...
Cheers,
E.
These are all good points. The current source that is often used as the load for a single-ended VAS in amplifiers with the classic topology that Doug Self refers to as the Blameless also has a temperature dependence due to its reference being the Vbe of a transistor - however that reference transistor dissipates very little power; it will be affected a bit by ambient. As you point out, the use of cascodes greatly reduces the problem from junction heating due to power dissipation.
The "Slone fix" should always be evaluated for the particular amplifier and the component tolerances of LTP emitter degeneration resistors and expected LTP Vbe matching between like-sex pairs. The resistor value can be reduced at will to achieve the necessary or desired amount of VAS bias current stability, with the proviso that smaller resistances will reduce open-loop gain at low frequencies.
Let's also recognize that the bias of most amplifiers changes more than we would like during warm-up, and that the bias pot is (should) be set when the amplifier has reached thermal equilibrium. The warm or hot VAS transistor will typically reach its thermal equilibrium long before the heat sink does.
Cheers,
Bob
Sahara
Hi Bob,
That's precisely my point. Even expensive so called 'high-end' amplifiers exhibit these anomalies (some people let their amp warm-up for 24hrs, crazy!). In my opinion, a decent amp should operate optimally at any temperature, in the Sahara as well in Antarctica, so to speak. If it does so, no warm-up issues either.
(Probably you guessed it already, this is also a plea for my CMCL circuit
)
Cheers,
E.
edit: CMCL = common mode control loop
Hi Bob,
That's precisely my point. Even expensive so called 'high-end' amplifiers exhibit these anomalies (some people let their amp warm-up for 24hrs, crazy!). In my opinion, a decent amp should operate optimally at any temperature, in the Sahara as well in Antarctica, so to speak. If it does so, no warm-up issues either.
(Probably you guessed it already, this is also a plea for my CMCL circuit
)Cheers,
E.
edit: CMCL = common mode control loop
Last edited:
Actually, you're incorrect. the second stage of the Thompson topology discussed by Douglas Self in his book is, in fact, a transimpedance stage, and was first called a "VAS", as far as I am aware by Douglas; it was not previously called a "VAS" in the literature.
The second stage of the Thompson topology is a transimpedance stage at DC and at all frequencies according to Solomon's tutorial on the monolithic op amp. (the Thompson topology really). Further, I don't think the transimpedance of the second stage "is the impedance of the Miller capacitor". See:
http://ece.wpi.edu/~mcneill/524/handouts/solomon.pdf
A folded cascode front end is, strictly speaking, a single gain stage topology delivering a voltage ouput for a voltage input.
I really think you should consider stripping your book of erroneous allusions to the "VAS" a la Douglas Self, and use the correct terminology: TIS.
P.S. For reference to the Thompson Topology see:
Russell, R. W., and Solomon, J. E., ‘A High-Voltage Monolithic Operational Amplifier’ IEEE
Journal of Solid-state Circuits, Vol. SC-6, NO. 6, December 1971, pg 352
Last edited:
Hi Mike,
I never heard of that classic amplifier topology referred to as the Thompson topology.
VAS makes sense for all the reasons I gave, but I agree that my citing of the folded cascode was probably not a good idea. I'll continue to use the term VAS, as will Doug Self as well, I am sure.
A transimpedance amplifier (TIA) is formed when an impedance element is used as the feedback element in a shunt feedback arrangement and the input is a current into that "virtual ground" node. The transimpedance is the impedance of the feedback element. The most common example is where the feedback element is a resistor, in which case the amplifier is a transresistance amplifier. These are used all the time in optical communications where the input current is a current from the photodetector and is thereby converted to a voltage at the output of the TIA. Thus, over the range where the Miller compensation capacitor dominates, the VAS in a conventional amplifier is a TIA. A lower frequencies, where the Miller capacitor looks more like an open circuit, the VAS acts more like a voltage amplifier.
Cheers,
Bob
> I'll continue to use the term VAS, as will Doug Self as well, I am sure.
Even if he tries to introduce another acronym for TMC?
edit:
>A lower frequencies, where the Miller capacitor looks more like an open circuit, the VAS acts more like a voltage amplifier.
Under this condition, I would say it's a current amplifier, because of current in and current out.
Cheers,
E.
Even if he tries to introduce another acronym for TMC?
edit:
>A lower frequencies, where the Miller capacitor looks more like an open circuit, the VAS acts more like a voltage amplifier.
Under this condition, I would say it's a current amplifier, because of current in and current out.
Cheers,
E.
Last edited:
Hi Bob,
That's precisely my point. Even expensive so called 'high-end' amplifiers exhibit these anomalies (some people let their amp warm-up for 24hrs, crazy!). In my opinion, a decent amp should operate optimally at any temperature, in the Sahara as well in Antarctica, so to speak. If it does so, no warm-up issues either.
(Probably you guessed it already, this is also a plea for my CMCL circuit
Cheers,
E.
edit: CMCL = common mode control loop
But remember, CMCL is only required for certain topologies.
Cheers,
Bob
this is also a plea for my CMCL circuit
If you mean this topology (and its derivatives), then it is yours as much as TMC. It is well known since the early 70's and widely used in integrated bipolar and CMOS opamps (even in a simplified version, since they benefit from the free device matching). I've recently seen this topology in several analog core libraries in our design system.
If you want, I can dig the references and patents in our library.