Replacing tubes with FET/transistor cascodes

It's been tried, never with great success. In spite of the old saw that "A FET looks like a vacuum tube" it doesn't, really -- the bias is different, the impedances are way lower, the heater connection really sucks in a series string, etc.

I've been chewing it over. What have you in mind ?

Many moons ago you could get FETRONS btw. Same idea.

Graham

"Eeyore" wrote in message news: snipped-for-privacy@REMOVETHIS.hotmail.com...

A large collection of old test equipment all with a multitude of different vacuum tubes.

ISTR an article in Wireless World (UK) along those lines.

-- _____ _ _ |_ _| | | | | | | __ _ _ __ ___ | |__| | ___ _ __ ___ ___ _ __ | | / _` | '_ ` _ \\ | __ |/ _ \\| '_ ` _ \\ / _ \\ '__| _| |_ | (_| | | | | | | | | | | (_) | | | | | | __/ | |_____| \\__,_|_| |_| |_| |_| |_|\\___/|_| |_| |_|\\___|_| __ ____ / _| | _ \\ ___ | |_ | |_) | ___ _ __ __ _ / _ \\| _| | _ < / _ \\| '__/ _` | | (_) | | | |_) | (_) | | | (_| |_ \\___/|_| |____/ \\___/|_| \\__, (_) __/ | |___/

A long time ago I'm sure !

So, had you looked into it at all ?

Graham

The Supertex depletion-mode n-ch mosfets are practically tubes. They make great current limiters, too.

John

Do they have the same 'square law' characteristic that tubes and jfets are supposed to share ?

Graham

Dunno... you could look them up! We use them at Idss, as high-voltage current limiters, so we don't care about the transfer curve.

John

"Eeyore" wrote in message news: snipped-for-privacy@REMOVETHIS.hotmail.com...

I came across a reference to the article but can't track down a copy.

-- _____ _ _ |_ _| | | | | | | __ _ _ __ ___ | |__| | ___ _ __ ___ ___ _ __ | | / _` | '_ ` _ \\ | __ |/ _ \\| '_ ` _ \\ / _ \\ '__| _| |_ | (_| | | | | | | | | | | (_) | | | | | | __/ | |_____| \\__,_|_| |_| |_| |_| |_|\\___/|_| |_| |_|\\___|_| __ ____ / _| | _ \\ ___ | |_ | |_) | ___ _ __ __ _ / _ \\| _| | _ < / _ \\| '__/ _` | | (_) | | | |_) | (_) | | | (_| |_ \\___/|_| |____/ \\___/|_| \\__, (_) __/ | |___/

FET

I dare say a Spice model might cast some light on that.

I suspected as much but thought you might just know anyway.

Graham

If you have the date I might be able to help !

Graham

Well, there are a LOT of different tubes out there with a lot of different characteristics and applications.

If you have the ability to go in and rework the design to give you all the bias sources you want then it's not hard at all.

If you want a "drop in" replacement that works in a certain application then it's not too awful but may require some creative thinking to re-bias things where you really want them.

To be pessimistic, exactly reproducing every characteristic of a given tube with solid state stuff isn't trivial. But most tubes are only used over a small fairly linear portion of their characteristic range, they are often used in circuits that self-adjust their bias, and even when they aren't (e.g. AGC action) the exact details are often smooshed into words ("sharp-cutoff" vs "remote cutoff") that except for the exponents are self-adjusting as well!

See Motorola application note AN211A (just google for "AN211A") figure

Tim.

That's doubtful, usually the common MOSFET models fail in the sub-threshold region.

I use the Supertex depletion-mode MOSFETs in the linear region all the time, and have taken careful bench measurements over many orders of drain current. At currents below say Idss/25, they exactly follow the theory we outline and plot in AoE, with an exponential voltage-current relationship, just like a BJT transistor, but with a somewhat lower slope (about 1/2 to 1/5 the transconductance). At higher currents, near Idss, they have a square-law characteristic, like tubes, but you can rarely use individual FETs in this region for continuous linear operation, because for MOSFETs this characteristic is at high currents, where Pd = Id * Vd means you have excessive power dissipation. Now, if you use a small device, with low Idss, just above where you want to operate, and marry it with a cascode power MOSFET, to keep Vd down, and provide a large heatsink, ahem, that's a FETRON. Compare Teledyne's TS-12AT7 to a tube,

In which case my reference to AN211A may not have been so helpful.

Replacement in classic RF/IF/AF stages where exact gain isn't picky and the class-A bias is mostly self-adjusting (the examples in AN211A) is the easy thing.

Using them in test or industrial equipment may be entirely different. It's extremely unlikely that you'll be able to do a simple drop-in replacement in a DC regulator circuit or in a multivibrator or in a sweep generator, for example. It's not that semiconductors are incapable of any of those jobs it's just that those circuits generally depend on cutoff, saturation, and DC characteristics that are not self-adjusting, and substitutions of the active component may very well require changes in the circuit to get it to do anything at all.

Tubes aren't exactly hard to come by these days... some are more common than others, but if you do a search on the web you WILL find what you're looking for.

Tim.

I've looked into that myself, and things look pretty dismal.

For really low frequency use, like non-picky audio, you can probably fake it with a FET and a transistor.

You can't use high-voltage MOSFETS, they have really high input capacitance Tens to hundreds of times higher than tubes.

And the transconductances are not too hot, and extremely variable. And you can't find 400-volt transistors with frequency response above a few megahertz.

You'd need a carefully-crafted bootstrapped input stage to minimize the input capacitance, some low-cap diodes to mimic the positive-grid region (for schmitt-triggers and multivibrators), then about ten well-matched 45 volt transistors in series. A complex and ugly arrangement.

-------

On the other hand, the old Collins R-392A radio, which was designed to run with 24 volts on the tubes, can be retrofitted with MOSFETS, just by directly plugging them into all the pentode sockets. The triodes can be replaced with most any silicon NPN transistor. Weird but true.

Ive replaced a couple of tubes directly with just a high voltage T0220 mosfet, these were large expensive high current tubes but just used for power supply regulation, (the 200v rail got shorted and you could see the wire conections inside the tube had fused !) and so mosfet was an ideal replacement.

Colin =^.^=

Got it thanks.

-- _____ _ _ |_ _| | | | | | | __ _ _ __ ___ | |__| | ___ _ __ ___ ___ _ __ | | / _` | '_ ` _ \\ | __ |/ _ \\| '_ ` _ \\ / _ \\ '__| _| |_ | (_| | | | | | | | | | | (_) | | | | | | __/ | |_____| \\__,_|_| |_| |_| |_| |_|\\___/|_| |_| |_|\\___|_| __ ____ / _| | _ \\ ___ | |_ | |_) | ___ _ __ __ _ / _ \\| _| | _ < / _ \\| '__/ _` | | (_) | | | |_) | (_) | | | (_| |_ \\___/|_| |____/ \\___/|_| \\__, (_) __/ | |___/

...

...

IN the old days, the replacements seemed to take three forms.

Creating general replacements that plugged in without any fuss. The Fetrons that someone mentioned were this type, but of course only a handful of types were made, to replace a handful of extremely common tubes. And I recall a detailed letter in QST about them, how they worked well in some applications (I seem to recall limiters in two-way FM receivers) but less well in more linear applications. The exact details may be garbled, but there did seem to be limitations of even such a general replacement.

More common were trying to build to the specific stage. So you look at the circuitry around it, and then try to create somethng that fit in that socket that worked with the surrounding circuitry. This required more fussing, but seemed to be a better choice. I seem to recall that the articles I saw often did adjust external resistors to some extent.

I once solid-stated a surplus Collins PTO by soldering an FET in place across the tube socket, and running it off low voltage, and it was fine (though it was a 1 to 1.5MHz PTO so was pretty simple). I don't think I had to change anything else, but note I wasn't using the PTO in the original piece of equipment.

Then there were the conversions where the chassis and tuned circuits and variable capacitors (or whatever) were basically just the foundation of a new project. It got the hard work out of the way, but of course didn't leave it intact. On the other hand, it was a way of improving equipment that had limitations. In that sort of conversion, the audio amplifier would be tossed out, and replaced with an audio IC amplifier. THe IF strip might be seen more like a function, and then something suitable could replace it. The thing about this was you could tamper with external components, so if that plate resistor was too high for whatever you were doing, you'd just replace it.

But of course, most of this took place in the late sixties and by the end of the seventies it had faded. At the time, the equipment was still seen as functional, but the hope of solid state made people want to modify them, or maybe it was just for the sake of doing it. The equipment was often getting pretty old, as the seventies progressed it was all getting replaced with equipment that was not only solid state (even using those new-fangled ICs) but that switch to solid state allowed for a leap forward in functionality. A tube shortwave receiver with single conversion to 455KHz would often still suffer from image problems, whether the tubes were left in there or replaced with transistors, but a more modern receiver might have moved to a single conversion to an HF IF, or tossed in lots of features that would require extensive work to add to the old receiver.

Michael

Oh, BTW, I love the way Freescale has reworked all the old Mot. data sheets and app-notes. It looks like they paid some junior high school kids to take a rubber stamp and stamp their name/URL over the top of "Motorola." Nothin' but high class for these guys....

BA

Well, all the remaining guys have data sheets that show heritage back through several generations of ownership. I still like the CD4000 series datasheets at TI's website where you can clearly see the Harris/Intersil lineage!

And every time I look at AN211A (every couple of years for the past couple of decades) I get completely thrown for a loop by that AND gate in Figure 22 making -130V :-).

Tim.