What is "The Norm"?
It is very personal definition. Is "OpAmp" topology "The Norm" for an audio amp? Not for me. But yes, for somebody else.
I was refering to what I thought was a question from you, regarding the slang term "outside the box". It is not a literal box around a circuit, but, a range of thinking considered normal amongst a high percentage of people. Like they are all in a box and think the same
But some, don't go with the status quo and think "outside the box".I would guess you think outside the box. So, I'm waisting my time...
Last edited:
JFET tempco with operating point
Another thing: there is a bias point for JFETs which allows almost zero drift in offset voltage. For long-channel devices it occurs about 626mV more turn-on voltage than the threshold voltage. For shorter-channel devices it shifts, and if the threshold voltage is low enough (like in our friends the BF862, 2SK170, 2SK389) it may even occur for a small forward bias on the gate, hence when biased at more than Idss.
Attached is a page from Richard S. C. Cobbold's book Theory and Applications of Field-Effect Transistors, showing three devices with drain current adjusted to give best results. One of them in particular is very good indeed, but the worst is only about 3mV shift over a wide temperature range.
If one is lucky enough to have a nice dual to begin with, it can further enhance the drift performance to operate both devices near this zero-tempco point.
When used for current generators, a similar enhancement of temperature stability can be had, and if the current be used as a reference in a regulator, some very-low-noise regulators/references result.
Another thing: there is a bias point for JFETs which allows almost zero drift in offset voltage. For long-channel devices it occurs about 626mV more turn-on voltage than the threshold voltage. For shorter-channel devices it shifts, and if the threshold voltage is low enough (like in our friends the BF862, 2SK170, 2SK389) it may even occur for a small forward bias on the gate, hence when biased at more than Idss.
Attached is a page from Richard S. C. Cobbold's book Theory and Applications of Field-Effect Transistors, showing three devices with drain current adjusted to give best results. One of them in particular is very good indeed, but the worst is only about 3mV shift over a wide temperature range.
If one is lucky enough to have a nice dual to begin with, it can further enhance the drift performance to operate both devices near this zero-tempco point.
When used for current generators, a similar enhancement of temperature stability can be had, and if the current be used as a reference in a regulator, some very-low-noise regulators/references result.
Attachments
How is a servo "brute force"? A cubic inch, no heat, less that $2 in parts, reliable, high performance. Perhaps the reason people still use it is because it's an elegant solution that solves the problem completely.
Actually servo in audio amps is nothing more than additional feedback loop with additional amplification. It increases open loop gain on frequencies below audio band. You have to remember that the servo loop has own linearity problems, like any other amplification stages, so on frequencies that are attenuated by servo loop it adds own non-linearities. Compared to balance and self-compensation it is indeed a brute force. And it does not eliminate problems. It minimizes them.
can you proto it and see how well it works? This is a good first attempt IMO of thinking outside the box. As this is a joint opamp design effort forum of a new opamp.... not a rehash of existing opamp designs... it will be nice to see real improvements over those existing designs. Thx-RNM
I can proto it, but I have much more things to do than to prove lessons that I learned 35 years ago. Including design of frequency-stable control loops in systems. Problems of stability of control systems are the same like in opamps, for example. Thermo-stabilization as well can oscillate. I would suggest again to go from basics, instead of stochastic jumps all around "the box" without understanding of details.
Last edited:
Thanks for the compliment on my elegant solution. Brute force refers to cooling and/or heating systems etc. that are being thought about.
Time marches on... the dc servo requires another IC (usually). Now I think it can be integrated into the design. We have one good idea so far From Wavebourn.... I am sure there are others if we brain storm on this.
If you want a clue.... sim the circuit I used for a line stage/headphone amp and figure why IT is so thermally stable without the added servo parts. Then come up with a new opamp design incorp the best idea(s). -Dick Marsh
Last edited:
I am not asking to 'prove' it but to give an example that this thread can take and run with or it will generate more new ideas for various topologies. In the spirit of this thread I am asking. Thx -RNM
Come on... This ideas were implemented long before Wavebourn was born. For example, steam machines contained stabilizers of sped of rotation that used weights on springs.
Right. Because DC is zero frequency, and zero is strictly mathematical term that does not exist in our physical word.
Yes, it was different. And with ideal devices it would be ideal. However, it would require also ideal power sources...
Scott's posting #135 and the related circuitry is a beauty.
Unfortunately the component count is high.
And most DIY enthusiasts would struggle with a discrete built of such high gain at 20kHz. Slowing down the dominant pole most likely would be necessary to ensure DIYability, unfortunately this impacts performance....
Nevertheless, to me it is the most educative circuitry of this thread IMHO,
thanks for that !!
Unfortunately the component count is high.
And most DIY enthusiasts would struggle with a discrete built of such high gain at 20kHz. Slowing down the dominant pole most likely would be necessary to ensure DIYability, unfortunately this impacts performance....
Nevertheless, to me it is the most educative circuitry of this thread IMHO,
thanks for that !!
Let's say sub-0.01 Hz. At least after a minute or two of warmup. This is certainly the case with my preamp's servo, and it uses old fashioned thermionic FETs.
Really, it's no different than an "integrated" approach except it's cheap and easier and potentially far more accurate (higher sub 10 Hz loop gain in an IC).