10MHz 50dB tube amp ?!?

[clintrubber]

A bit OT this one, but I was wondering what/how many/which TUBES it would take to realize a gain of some 300 (50dB) @ 10MHz with still reasonable (...) phase shift. (Hi-Z in, output-load say a few pF)

OK, it requires a GBW-product of 3GHz & someone mad enough to do it.

But if at all, could it be done with a reasonable amount of parts ??

I'm not sure how silly this question actually is and it's very unlikely I'll be building this, but the thought came to mind and I got curious. I couldn't really find much w.r.t. HF-performance in the tube-databooks.

Bye,

Peter

 

[pstamler]

Take a look at some circuits from tubed oscilloscopes.

Peace,
Paul

 

[NewYorkDave]

See if you can find Vacuum Tube Amplifiers by Valley and Wallman in your local library. It covers the subject of high frequency amplifiers--naturally, since it was a product of WWII radar research.

 

[clintrubber]

[quote author="pstamler"]Take a look at some circuits from tubed oscilloscopes.

Peace,
Paul[/quote]

Oops, of course !
Thanks for the suggestion, I'll get out my Tek 535-schematics.

Regards,

Peter

 

[clintrubber]

[quote author="NewYorkDave"]See if you can find Vacuum Tube Amplifiers by Valley and Wallman in your local library. It covers the subject of high frequency amplifiers--naturally, since it was a product of WWII radar research.[/quote]

It's the Netherlands here, so I doubt they have still tube stuff - let alone specific titles. But while I previously had the impression the combination of 10MHz, 50dB & little phase shift was way too much, I now understand it's not that strange after all. So will browse & google more now I hear it's not so unthinkable.

Thanks,

Peter

 

[bcarso]

I assume you mean d.c. to 10MHz? That is of course the challenging one. If you mean make a tube op amp with 50dB of open loop gain at 10MHz that you can apply global feedback to and be unity-gain stable, that's virtually impossible imo.


But open-loop designs like the scope vertical amplifiers suggested are clearly possible (they exist!). When pushed to the extreme you can do a distributed amplifier, with a great deal of parts and pain. There's a net delay from each stage but the group delay can be well-controlled.

 

[NewYorkDave]

The original Tektronix prototype distributed amplifier from 1948:

Some recommended reading:

"Who Wakes the Bugler?" by Carl Battjes, in The Art And Science of Analog Circuit Design, edited by Jim Williams.

(Both of Jim Williams's analog design books can make for some excellent reading while you're curled up in your favorite easy chair this winter!).

 

[PRR]

> what/how many/which TUBES it would take to realize a gain of some 300 (50dB) @ 10MHz

Hell, that's not even tough.

As Paul says, o'scopes did it right.

TV sets did it cheap. Being cheap, I whomped one up. I have not done this in a looooong time, took more than 10 minutes. Drawing the full schematic with all the pins would take a lot longer than figuring out the basic shape of the amp.

The ideal tube will have very high transconductance and very low capacitance. Rather than compute the merit of every tube in RC-30, I went for 6CL6, a hi-end TV set video output tube that could smack 132V p-p TV video into a big CRT cathode. Eats 30mA plate 7mA screen 6.3V 0.65A heater, currently costs $8 from TheTubeStore.com, and I bet they have a basket-full.

Key parameters are Gm (11,000uMho) and input, output, and feedback capacitance. We can assume plate resistance will be infinite compared to the resistors needed to get low-medium frequency gain equal to gain at 10MHz. A few iterations suggest 3 stages with Rl near 500-1K, but it really wants to dive at 10MHz. Then we whip out some 10uH chokes and fiddle some more:

Hmmmm... forgot stray wiring capacitance. I don't think that kills the idea, nor requires four $8 tubes, but pondering that physical layout, testing, and getting the workers to duplicate it, will be a lot harder than the basic tuning design.

> with still reasonable (...) phase shift. (Hi-Z in, output-load say a few pF)

Oh, fussy, fussy. I gave you 25pFd load (I needed it to flatten the response). What is reasonable??? LF phase is 180 degrees, phase at 10MHz is 55deg or 125deg shift.

> Hi-Z in

Note that the input Z is like 15pFd which is 1KΩ at 10MHz. We can improve that a bit with a buffer, but there are no "high impedances" at 10MHz.

 

[PRR]

There is also the Feedback Pair. As a first-hack, I can't do your spec in 3 tubes, so I blew another $8 for four 6CL6s total. This time I tossed in 10pFd stray capacitance at each node.

49.5dB at LF, a half-dB up at 7MHz, right-on at 10MHz, -3dB at 14MHz. However phase response is 114 degrees at 10MHz (scaled /10 for the graph).

One advantage is that throwing in the 10pFd caps hardly changed the frequency response at all (it did add phase shift). Change of gain with tube Gm variation is slightly reduced, at least in the 2nd and 4th positions, but this is hardly the rock-solid "feedback amp" we are used to from LM741 working at 100Hz.

6CL6 may not be the very best tube. If you can keep stray capacitance very small, good old 6AK5 (nearly the first Mini tube) can get into this realm at less than half the power. This figures: the 6CL6 is really a power tube, that happens to have a hunk of gain to keep TV set total cost low. If you are just going a half-inch over, you don't need to grunt so hard.

 

[clintrubber]

Pleasantly overwhelmed :grin:
It's really time for a bed here now otherwise I hear the first birds, but I just caught the new posts, with all this worked out ! Big thanks, I'll print + read + absorb.

FWIW, I did check the Tek 535A circuit this evening, and in the CA-plug the already mentioned 6AK5 did pop up, as did 12AU6 & 6AU6.

FWIW#2, did look shortly for a EU-version of the

but I saw a place here in NL that has it.

FWIW#3, this may not get more serious, otherwise I'm getting tempted to really build it :wink: Background to all this was this gain & phase requirement for some new circuitry at work. Couldn't help thinking that it would raise quite a few eyebrows if the helping amp doing that 50dB turns out to be sans-Si...

Regards,

Peter

 

[clintrubber]

[quote author="bcarso"]I assume you mean d.c. to 10MHz? That is of course the challenging one. If you mean make a tube op amp with 50dB of open loop gain at 10MHz that you can apply global feedback to and be unity-gain stable, that's virtually impossible imo.


But open-loop designs like the scope vertical amplifiers suggested are clearly possible (they exist!). When pushed to the extreme you can do a distributed amplifier, with a great deal of parts and pain. There's a net delay from each stage but the group delay can be well-controlled.[/quote]

Indeed the latter, not a 3GHz GBW-tube opamp.
(W.r.t. DC, no, it all starting from say 100kHz wouldn't matter.)

Bye,

Peter

 

[clintrubber]

[quote author="NewYorkDave"]The original Tektronix prototype distributed amplifier from 1948:

Some recommended reading:

"Who Wakes the Bugler?" by Carl Battjes, in The Art And Science of Analog Circuit Design, edited by Jim Williams.

(Both of Jim Williams's analog design books can make for some excellent reading while you're curled up in your favorite easy chair this winter!).[/quote]

Nice pic & thanks for the suggestion. With the first snow already here that easy chair quote sure sounds cosy :wink:

 

[CJ]

Starting a bootleg station? :thumb:

 

[PRR]

> gain...@ 10MHz... phase shift... {unlikely} I'll be building this,...Background ...was this gain & phase requirement for some new circuitry at work. Couldn't help thinking that it would raise quite a few eyebrows if the helping amp doing that 50dB turns out to be sans-Si...

Gotta be a video amp.

Gotta be a hot new product.

Something that'll sell a gazillion, to cover Clint's salary.

Something with features no other product has.

I forget who here is goofing off on which employer's time,

so sorry if I mock the wrong company.....

 

[Swedish Chef]

:green:
Podzilla, or perhaps GodPod...

 

[clintrubber]

[quote author="PRR"]> gain...@ 10MHz... phase shift... {unlikely} I'll be building this,...Background ...was this gain & phase requirement for some new circuitry at work. Couldn't help thinking that it would raise quite a few eyebrows if the helping amp doing that 50dB turns out to be sans-Si...

Gotta be a video amp. [/quote]
No...

Gotta be a hot new product.

Maybe. Not unlikely. But it'll take a while.

Something that'll sell a gazillion, to cover Clint's salary.

:grin:
The former not unlikely, the latter... we'll see... no, not likely

Something with features no other product has.

No, no new features (too bad, we want new features, more new features, too much new features), but it's about doing existing things in a 'smarter' way. So not that special for the end-user.

Hmm, and I'm afraid that's about all I can say - not to create some mystery, but to avoid spoiling other people's patent-applications.

If we get things more going then it'd be sure a lot of fun to build the 'sans-Si' version. We already went back in history from complete amps to discrete BJTs so let's continue this :wink:

Too bad there aren't any significant advatages of doing it the tube-way i.s.o. with semiconductors (at least not that I see). It'd be great to tell people: "Right, we had to use tubes since you can't do this&that with solid-state here".

That's a hilarious device ! :grin: Truly amazing, I bet there'd be quite some people wanting this.

Best regards,

Peter

 

[bcarso]

I want to see the the accessory line, including the gigantic fanny pack battery-inverter, and some stylized silhouette character jumping around with "My Music" blaring in his headphones. :evil:

 

[PRR]

> that's about all I can say -

Don't apologize. Don't even hint. When Steve Jobs goes on stage to announce a new product, not even the stage manager knows if he is introducing a picoPod smaller than a match-head, or if it's....
ePod: portable for elephants.

2-meter screen.
0.2Hz-500Hz audio response.
batteries not included.

After all.... by next Christmas every human on the planet will have an iPod. The market's gotta expand.

 

[clintrubber]

:grin: :grin: :grin: !