|
|
|
|
|
As for using a non-linear vacuum tube to pre-distort the signal into a tube amplifier, maybe it would work. Let's see a schematic and we can run it through SPICE to see what it thinks before actually building it. Next issue I hope to provide a circuit that might accomplish the same thing in a different way.
Subject: Ultra-Linear
Please note that Alan Blumlein patented the screen-grid feedback principle in Britain in 1937, UK Pat. No. 496,883. This patent discloses not only the feedback principle but also the use of taps on the output transformer to obtain the feedback. This pre-dates Hafler and Keroes by 14 years!
I would ask that you update the statement that Hafler and Keroes "created" ultra-linear. They created the name for the concept and products utilizing it.
Blumlein's extraordinary work and inventiveness has only recently been fully brought to the attention of the public. The British government imposed a ban on publication of his contributions following his death in an aircraft crash during radar trial in the second world war.
Richard.
P.S. The true origin of the screen feedback principle was known prior to the recent publication of Blumlein's biography.
Richard, you are right Alan Dover Blumlein, the brilliant inventor of stereo and co-inventor of television, does deserve the credit for the invention of ultra-linear operation. And you are also right that this fact was know prior to the release of his biography, as I knew it, yet when I wrote the article I did Blumlein the great disservice of forgetting that fact.
|
|
|
|
|
|
|
|
|
|
The March issue of Electronics World has a brief article about Blumlein's life and contributions to electronics. (The biography that Richard refers to is titled: "The Inventor of Stereo, The Life and Works of Alan Dover Blumlein" and its author is Robert Charles Alexander.)
Subject: Current Mirror Comment
I enjoyed your latest Tube CAD Journal, tackling the delicate subject of hybrid design. However you glossed-over a factor in current mirror design. Current mirrors using bipolar transistors only work if the devices are matched and are at the same temperature. On an IC, this is easy, so current mirrors are widely used in op-amps, etc. However, they are difficult to implement using discrete transistors. First, they need to be matched for hfe (beta). But they must be *tightly* coupled thermally. I've tried making a discrete transistor current mirror, and found that if the transistors were not tightly coupled, a thermal run-away effect occurs where one transistor would draw much more current than the other. Even with the transistors clamped together, the current matching was not very good. Current mirrors really work the best with monolithic arrays, such as the CA3018 or CA3086, but these don't have very high voltage or power ratings. The quad transistor arrays, such as the MPQ2222, etc. are not monolithic - they are just four 2N2222 dies attached to the same lead-frame. The thermal matching is good (but not perfect), but you would have to use a curve-tracer to search for good matched pairs in a single package.
John
Absolutely right, except when emitter resistors are used. The situation is not that different to when parallel transistors are used in a output stage of an amplifier;
|
|
|
|
|
