Transistors come in different flavors. Tubes used an electric field to regulate current flow, and researchers wanted to find something that worked the same way without the drawbacks like vacuum and filament voltages. However, what they first found — the bipolar transistor — doesn’t work the same way. It uses a small current to modulate a larger current, acting as a switch. What they were looking for was actually the FET — the field effect transistor. These come in two flavors. One uses a gate separated from the channel by a thin layer of oxide (MOSFETs), and the other — a junction or JFET — uses the property of semiconductors to deplete or enhance carriers in the channel. [JohnAudioTech] takes a decidedly practical approach to JFETs in a recent video that you can watch below.
The idea for the FET is rather old, with patents appearing in 1925 and 1934, but there were no practical devices at either time. William Shockley tried and failed to make a working FET in 1947, the same year the first point-contact transistor appeared, which was invented while trying to create practical FETs. In 1948, the bipolar junction transistor hit the scene and changed everything. While there were a couple of working FETs created between 1945 and 1950, the first practical devices didn’t appear until 1953. They had problems, so interest waned in the technology while the industry focused on bipolar transistors. However, FETs eventually got better, boasting both very high input impedance and simplified biasing compared to bipolar technology.
Of course, there are drawbacks, too, so it is important to understand the strengths and weaknesses of each technology. [John’s] video will tell you a lot about the practical aspects of these versatile devices.
We liked that in addition to some theory and graphs, he also wired a FET on a breadboard and showed things like what happens when you cool the device down. There’s a second part of the video forthcoming, and we’re sure it will be worth watching, too.
If you want [Bil Herd’s] take on FET technology, we have videos, too. Because of their high input impedance, FETs are common in things like non-contact voltage sensors, theremins, and guitar pre-amps.
What’s i teresting is that when Jfets came along in the mid-sixties, they were tkuted for their ability to withstand cross-modulation. That’s forradio use.
But in 1971, there was a bipolar transistor preamp for VHF in QST. And that continued. About 1974, Hank Cross wrote about overload in VHF converters, and was talking abiut bipolar transistors designed for CATV use. So pump some current into them. So a lot of well designed bipolar preamps followed for VHF/UHF work. A lot of care put into their design.
I’ve always read thisas meaning a lit of early bipolar re eiver designs optimized for lowest current, which made them prone to overload.
Gaasfets are used in some applications, but Jgfets have mostly disappeared for signal amplifiers. And RF mosfets have become harder to get.
Interesting, thank you very much for the information! 😃👍
What I liked the most about tubes was their electrical robustness. An overload caused them to go softly into saturation merely, for example.
That’s why I still use them sometimes for building direct conversion receivers, for example.
I have that multimeter!
I have the pencil!
and my beard_board. Let’s go for the adventure!
It’s a, um, bread board.
My beard says otherwise.
Let’s don’t forget the Nuvistors.
At their time they fixed the drawbacks of both conventional tubes and FETs.
http://www.r-type.org/articles/art-150.htm
The scratches on the table make it look like the transistor is in the process of blowing up.
It’s weird how when someone is going over something you think you kinda know, in a different way, gives you different ideas.
With him drawing the diode parallels, I’m getting to wondering how to make a copper oxide FET from copper oxide diode tech. If you can do that… maybe you can home lab a logic gate from raw materials…. etc etc.
Then also with showing the temperature effect, I’m wondering if you can thermally bond it to a bipolar and make a temperature compensation gizmo (Matching devices carefully of course)
Diodes have been used as temperature compensating gizmos for decades, so you could almost certainly make JFETs work too – use them like a diode.
I’ve looked at several data sheets for bipolar RF receiver transistors, and it’s often the case that best noise performance occurs at low current. Optimizing for low noise is perhaps a poor choice for CATV systems, but low noise is often what you want for a general purpose receiver
+1
That’s one of the reasons the superhet technology had reached a dead end, also, I think.
Adding more and more transistors (double super, tripple super etc) just increased self-noise, without much (any) benefits in practice.
That’s one of the reasons thermionic tubes were superior, I think.
Except for their heating, self-noise was little.
At least in comparison to the transistors of ~40 years ago.
That seems irrelevant, but it’s remarkable if we think about it.
Tube technology was stuck in the ~1950s, but continued to rival transistors for years to come.
Despite nolonger being improved.
The Nuvistor was the last upcry of tube technology, in some way or another.
The Nuvistor proved that tube technology can be improved, at least.
The noise from the heating, for example, could be lowered a lot. Nuvistors need minimal heating.
Makes me wonder were we could be now, if research of tubes/Nuvistors wasn’t stopped.
I might be wrong, but I think that especially both us western Europeans and the Americans are to blame. The latter for their obsession with solidstate technology, too.
Just look at all the propaganda ads from that time period. It’s beyond rational thinking. Tube technology wasn’t just abandoned. It was intentionally shut down.
And let’s also remember those cheap transistor radios from Japan.. Sure, they were impressive for their size and weight. It waa not wrong praising them. But all these unfair comparisons with bigger tube radios wasn’t cool. Tube radios usually were heavier and more bulky often, but their performance was far better too. The early Germanium transistors weren’t really good. Neither their amplification factor, nor their curve on the oscillogramme.
Also, there existed good radio tubes at the time made for the car radios. Like EF97/EF98. They were roughly the size of a lighter or an eraser. Not very big. They could gave been used to make fine pocket radios, too.
Anyway, my point simply is:
Miniature tubes could have been an ideal partner to solid-state technology.
In a receiver, use a small tube as an RF frontend and do the rest with FETs. That makes the receiver mostly immune against overload, EMPs and lightning strikes.
Same goes for lab power supplies. Let the tubes do the tricky task of stabilization etc.
There are so many applications in which tubes could assist solid-state circuits.
Fail-safe circuits, for example. Tubes can handle over-voltage fir a while (except the heating, but that can be battery powered).
So many possibilities! 😃
You want tubes to be superior, so you view the world that way. Receiver design is a tradeoff. The first superhets weren’t about selectivity, but gain as receivers went up in frequency. When selectivity became an issue, it could only be done at lower frequencies. The fallout was images. So they added double conversion. Collins type receivers of course provided a fixed dial of a small range, easy to calibrate.
But about 1960, there were crystal filters in the HF range. Way easier to get rid of images, and suddenly no need to gang front end tuning with the local oscillator. In the sixties most rigs were made that way.
But then people decided a ham transceiver should have general coverage receive. But a 9MHz IF put a hole in the tuning range. So convert up above the HF range to a roofing filter,then down to a lower frequency for selectivity. Multiple conversion made a comeback, not for image rejection, but for features like passband tuning and variable bandwidth. Some receivers converted down to 50KHz or so, LC circuits providing selectivity, but much easier to vary it.
There’s always been an issue of gain distribution. There’s nothing magical about tubes, some tube receivers were good, others had limitations.
The Kenwood TS-830S came out in 1977, 45 years ago. It’s still considered to be a great receiver. All solid state, except for the driver and output tubes.
SDR is taking over because they make better receivers. Not any old SDR, but a well designed one does. They couldn’t be made in the days of tubes. I remember an article I think forty years ago about SDRs, and it couid onkyge theoretical. Hobbyists didn’t have fast A/D converters, or the computers to process it.
Tubes were noisy. It didn’t matter at HF. But at VHF and above, not many choices. 417s and even better 416s, but expensive if you couldn’t scrounge them. Or build a parametric amplifier after 1958. Few needed one, even fewer built them. Fifty years ago 432 and 1296 converters fed the antenna into a diode mixer. A transistor preamp might be added externally. Better converters have come from solid state.
Nuvistors weren’t to make small tubes to compete with transistors. The small size came from their purpose, good noise figure and gain in the VHF/UHF range. They weren’t the best in terms of noise figure, but they were really close.
Parametric amplifiers lasted about fifteen years, 1958-1973 or so. Too.much work, and transistors were getting good.
Tubes were big, hot and expensive. People used them because there was no alternative, and moved away as soon as they could.
You can stay in the thirties with simple tube gear, but it’s solid state that has made receivers so.much better. And things like digital readouts and digital synthesizers were not feasible in hobby circles with tubes.
If you’re still building reeivers with tubes, you’re living in the past. You’ve not bothered to keep up with fifty year old technology.
Fifty years ago there were hams who didn’t want to transition to solid state. But they grew up with tubes. This fetish for tubes seems to be nostalgia for a past they never experienced. Or led astray by other hams.
But low noise is overrated. For much of the shortwave band, it doesn’t matter. It does become an issue going into VHF. But tube shortwave receivers did not have impressive noise figure. The R390, about the best receiver hobbyists can get ahold of, has somewhere around 10dB noise figure..
Gain should be after the main selectivity, but the need to tune a receiver complicates that. But when receivers dropped to 455KHz for the IF, the need for image rejection caused one or two rf stages before the mixer. Not for gain, but for selectivity. Once crystal filters were available, a lot less front end selectivity was needed. But that ooens things up, a need for better mixers to handle the wider frontend passband.
Look at good receivers. No rf stage, or at least one that can be switched out. Right into the mixer, which is often a balanced diode mixer. Then a broadband amplifier, a 2N5109 or 2N3866, to handle the strong signals. Then as good a crystal filter as possible. If it’s one bandwidth, then put it there. If you need multiple bandwidths for different modes, then it has to be broadened.
The problem with technolog is that it’s often described in terms of history. So subsequent improvements are tacked on, rather than starting from the beginning again. So “you convert down to 455KHz, but maybe have double conversion to get rid of images” continued a long time after there were crystal filters in the MHz range. Peolle are still stuck with the idea of rf stages before the mixer, when other advances has changed that.
I know he said he’d leave theory for another video, but probably the most interesting thing I ever derived from JFET functions is that the intrinsic gain of the device is directly related to the pinch-off voltage if you hold the channel modulation parameter constant. After all, the channel length modulation parameter is intrinsic to the design of the JFET, and can’t really be changed with external components.