Re: A new servo
thanks for sharing. that's a pretty clever change.
🙂
hopefully scott will chime in on your question ...
mlloyd1
thanks for sharing. that's a pretty clever change.
🙂
hopefully scott will chime in on your question ...
mlloyd1
syn08 said:
Re: A new servo
Clearness of materials and technology?
I remember the story when suddenly reverse currents of devices were increased significantly, nobody could find why, until found that one lady changed a boyfriend and had to change perfume. After that accident usage of perfume saw forbidden.
I like your servo; may will act as a rumble filter.
syn08 said:
Clearness of materials and technology?
I remember the story when suddenly reverse currents of devices were increased significantly, nobody could find why, until found that one lady changed a boyfriend and had to change perfume. After that accident usage of perfume saw forbidden.
I like your servo; may will act as a rumble filter.
Re: Re: A new servo
You've seen nothing; I recall one incident when a diffusion lady operator decided to warm up her luch in the diffusion tube cooling extension.
Or another litography control lady operator, complaining she can't see a damn thing through the microscope; of course it was impossible to see a thing through the thick layer of eye makeup deposited on the oculars.
Or a service techie that left his dirty cleaning sheets in an electron gun metal deposition unit and pressed the vacuum button.
Wavebourn said:
You've seen nothing; I recall one incident when a diffusion lady operator decided to warm up her luch in the diffusion tube cooling extension.
Or another litography control lady operator, complaining she can't see a damn thing through the microscope; of course it was impossible to see a thing through the thick layer of eye makeup deposited on the oculars.
Or a service techie that left his dirty cleaning sheets in an electron gun metal deposition unit and pressed the vacuum button.
That's nasty. A fingerprint is bad enough in an UHV system.
BTW -- thanks for the 4pcb recommendation. I registered and super impressed by their customer service so far.
Wavebourn, I looked up ALPHA in a 1953 book: 'Principles of Transistor Circuits' Edited by R.F. Shea
They don't measure BETA directly at all in the whole book. ALPHA is almost everything and it is difficult to explain here how they defined it, except graphically and through determinant math.
ALPHA was sometimes greater than 1, especially in point contact transistors that were popular up to 1953, and common base was a more stable way to run these devices, as the ALPHA greater than one, would cause instability in common emitter circuits. As we know, point contact transistors are not used much, anymore, and were considered virtually obsolete by 1960. Junction transistors, as they increased current gain, forced BETA to be much more valuable, (and variable) a parameter than ALPHA, by 1960.
Still, voltage drive, while noted, was considered a minor component in the early years. Even in computer models of transistors in ECAP, designed by IBM, ignored voltage drive characteristics.
I still find beta characteristics still important in discrete design. More important than many seem to realize here, it is your choice.
However, I did locate a REAL graph of an IC based output NPN device, made by a competing manufacturer in 1991, RAYTHEON, who was actually making bipolar transistors since the 1950 or so, and it shows a relatively non-linear beta characteristic worth noting. This is the basis of my asking the question about the relative non-linearity of the BETA in AD IC output devices. I still think that Scott's answer in incomplete, and next to worthless, for my needs.
They don't measure BETA directly at all in the whole book. ALPHA is almost everything and it is difficult to explain here how they defined it, except graphically and through determinant math.
ALPHA was sometimes greater than 1, especially in point contact transistors that were popular up to 1953, and common base was a more stable way to run these devices, as the ALPHA greater than one, would cause instability in common emitter circuits. As we know, point contact transistors are not used much, anymore, and were considered virtually obsolete by 1960. Junction transistors, as they increased current gain, forced BETA to be much more valuable, (and variable) a parameter than ALPHA, by 1960.
Still, voltage drive, while noted, was considered a minor component in the early years. Even in computer models of transistors in ECAP, designed by IBM, ignored voltage drive characteristics.
I still find beta characteristics still important in discrete design. More important than many seem to realize here, it is your choice.
However, I did locate a REAL graph of an IC based output NPN device, made by a competing manufacturer in 1991, RAYTHEON, who was actually making bipolar transistors since the 1950 or so, and it shows a relatively non-linear beta characteristic worth noting. This is the basis of my asking the question about the relative non-linearity of the BETA in AD IC output devices. I still think that Scott's answer in incomplete, and next to worthless, for my needs.
Right, John!
Alpha was the main characteristic of the transistor at the dawn of it's era.
Of course, Beta is very important as soon as today transistors are used almost with common emitters, and even when used in common collector (emitter follower) Beta is the parameter to calculate results.
The thinner is base, the higher is beta. But lower is maximal reverse voltage collector-base, due to Zener effect.
However, as Scott said, sometimes it is useful to view transistor as a voltage - controlled device: exponential dependence of voltage between base and emitter on current allows both points of view, which one is more convenient for certain equations.
Teaching electronics, I always stressed power amplification. Such a way students did not think in terms of voltage only amplifications, or current only amplifications. They always assumed that there is a resistance, always complex and non-linear. It is very significant for obtaining desired results in audio: 120 decibels means 1,000,000 times different currents and voltages in a single tune played by a speaker! What happens with currents if we assume that amplify voltages, but forget about exponential dependence of currents on voltages? What happens with voltages, if we think in term of current amplifications without an exponential dependence in mind? What happens with noises when amplifying small signal from low resistance source we load it on high very linear resistance? We are loosing power than need to replenish inevitably adding noises and distortions.
Scot referenced very useful book: http://www.amazon.com/Trade-Offs-Analog-Circuit-Design-Toumazou/dp/1402080468
It is very rare book. I mean, it is very rare when such topics are discussed in an open press in such concentrated material. After years working with software I regret that I did not keep conspectus of lectures given to us in TIASUR since I returned back to an electronics design and need to refresh a lot... They taught us physics, chemistry, with manufacturing in mind. They taught us design and engineering, with economics in mind. The book Scott is referring to discusses design close to how we were taught: designing thing think how it will be produced; adjust design to available technologies, or drive technologies to afford new design.
Though the book is expensive, it worth money.
Alpha was the main characteristic of the transistor at the dawn of it's era.
Of course, Beta is very important as soon as today transistors are used almost with common emitters, and even when used in common collector (emitter follower) Beta is the parameter to calculate results.
The thinner is base, the higher is beta. But lower is maximal reverse voltage collector-base, due to Zener effect.
However, as Scott said, sometimes it is useful to view transistor as a voltage - controlled device: exponential dependence of voltage between base and emitter on current allows both points of view, which one is more convenient for certain equations.
Teaching electronics, I always stressed power amplification. Such a way students did not think in terms of voltage only amplifications, or current only amplifications. They always assumed that there is a resistance, always complex and non-linear. It is very significant for obtaining desired results in audio: 120 decibels means 1,000,000 times different currents and voltages in a single tune played by a speaker! What happens with currents if we assume that amplify voltages, but forget about exponential dependence of currents on voltages? What happens with voltages, if we think in term of current amplifications without an exponential dependence in mind? What happens with noises when amplifying small signal from low resistance source we load it on high very linear resistance? We are loosing power than need to replenish inevitably adding noises and distortions.
Scot referenced very useful book: http://www.amazon.com/Trade-Offs-Analog-Circuit-Design-Toumazou/dp/1402080468
It is very rare book. I mean, it is very rare when such topics are discussed in an open press in such concentrated material. After years working with software I regret that I did not keep conspectus of lectures given to us in TIASUR since I returned back to an electronics design and need to refresh a lot... They taught us physics, chemistry, with manufacturing in mind. They taught us design and engineering, with economics in mind. The book Scott is referring to discusses design close to how we were taught: designing thing think how it will be produced; adjust design to available technologies, or drive technologies to afford new design.
Though the book is expensive, it worth money.
Allen Wright said:
Thanks. If nobody tried before, devices are expensive, results unpredictable, I am going to stick with my previous idea: to bolt together 11 well matched small transistors (for 10:1 ratio), with input one in the middle.
Syn08,
I assume your new servo OUTPUT is connected the same way as in the previous schematic or am I missing something?
Did you also try the OPA627?
Franklin
I assume your new servo OUTPUT is connected the same way as in the previous schematic or am I missing something?
Did you also try the OPA627?
Franklin
An amplifier must amplify power to be considered an amplifier. It has to be defined this way, otherwise a transformer would be an amplifier.Wavebourn said:
courage said:
Indeed, I think you are missing something; no, the schematic is correct as it is.
The bias current through the up/down resistors is constant, determined by the reference voltage minus a Vbe. And so is the collector current of the first cascode, determined by the JFETs Idss.
Now, imagine a variation of the output bias voltage. The opamp in the servo will convert this variation to a current through the 1kohm load resistor. But where is this current coming from? Obviously from the supply lines. We already know the constant currents in those nodes, so the only currents that may change are the currents through the second cascode transistors. Which has exactly the same effect sinking/sourcing a current in the gain node, as in the previous servo. The result (variation of current through the gain opamp feedback resistors) drags the amp output opposing the initial variation. Take into account the 1Meg/1uF integrator time constant and the net result is the output under 1mV away from null.
Yes I have tried OPA627 and its nothing to call home about. Otherwise, to me, OPA627 is an expensive dinosaur. Much better alternatives exist today, just allow some time until OPA827, AD8599 (JFET) OPA211, ADA4898 (bipolar) and many other will build a reputation among the audiophiles.
Yes, you are right, Mr. Evil, BUT we often forget that power gain is sometimes 'everything' especially in the early days, when there wasn't much gain on any front.
john curl said:
Yep. Additional switch with caps (or resistors) may be used as a selling feature for audiophiles, "Rumble filter frequency".
john curl said:
John - You can't even describe the action of the simplest area ratioed current mirror with these concepts. Or how about the voltage to current transduction of a simple long-tailed pair? It certainly has nothing to do with beta. Even in 1951 Shockley dismissed alpha as so close to 1 that it could be ignored.
We used early simulators too where you hand computed an operating point for each device, and guess what gm. These were equation grinders primitive at best.
Here noises are not specified, though they may be usable:
http://www.datasheetcatalog.org/datasheet/diodes/ds30436.pdf
http://www.datasheetcatalog.org/datasheet/diodes/ds30436.pdf
Transistors can't be made in the single process in a discrete gear, Scott, this is the biggest difference. Also, sometimes I see how IC-specific topologies blindly migrate to the discrete world where they have no meaning...
- Status
- Not open for further replies.