ad: ProAudio-1

Simulating tube amplifiers

Discussion in 'Amateur Radio Amplifiers' started by W8JI, Dec 29, 2011.

Thread Status:
Not open for further replies.
ad: L-HROutlet
ad: l-rl
ad: abrind-2
ad: QSOToday-1
ad: Subscribe
ad: Left-3
ad: L-MFJ
ad: Left-2
  1. AA5CT

    AA5CT Ham Member QRZ Page

    "Since the tube provides the gain, and the feedback, and primary reactances and phase shifts in the oscillator circuit, how can you model the stability without an accurate representation of the tube? "

    Motorola (now Freescale) used to 'match' a transistor, empirically, then measure into the resulting matching network and figure the conjugate to that measurement being the transistor; not so hard really ....

    The danger I see (me, myself, I, moi, no one else .. get it?) is, getting a tube to the edge of that cliff without going off over the cliff to make measurements in/near an unstable state (I don't have unlimited resources like some here who ship kilowatt-class amps by the truckload off the loading dock) ... I may have to bring out the primary of the Plate transformer separately so as to be able to 'run' plate voltage up slowly while monitoring various parameters like S21 and S12 across a range of frequencies of the SB-220 with its tube while the tube is in an 'active' state ...

    One could probably infer some important factors by some simple lab tests.

    I'm also reminded of the experience I had years back with a crystal osc circuit that used a TTL NAND (or maybe a simple INV gate) as the 'active' element, the performance over temperature was horrid (I did not design it - I simply inherited it).

    Controlling the temperature of the crystal had only a partially good effect, the BIG problem being the TTL gate exhibited varying delay time over temperature translating to frequency changes ... replacing the active osc element with a simple NPN transistor and the 'frequency' of the crystal moved UP by 5 kHz (from 10.24 to 10.245 MHz; this was a synthesizer reference) but at least it was stable! (recall that 'delay' integrated over time =frequency )

    There was THAT much delay in the TTL 'gate' (over several transistor stages if one is familiar with TTL logic internals). Of course, the design required a new crystal at this point ...

    The point being, the 'effect', tube 'delay', due to electron transit time would likely be to shift the frequency of oscillation down in frequency, rather than any really 'new' dramatic effects; as others have pointed out, the primary behavior for a given frequency (re: oscillation) is governed by circuit physics.

    de Chopped Liver

    Happy New Year!
    Last edited: Jan 1, 2012
  2. KL7SG

    KL7SG Platinum Subscriber Platinum Subscriber QRZ Page

    With the advent of the modern high speed computers, we have been modeling virtually everything. While models are not perfect, they are value added. Now days, we model air flow over wings, airframes, engines and electric fields to name a few. The trends observed from these models has resulted vast improvements in the designs and safety. The classical methods used for the last 50 years are still great; but, the models in some areas are superior.

    It sounds to me like there is disagreement concerning the value of modeling RF tube amps. I guess at this stage in the life cycle of the tube, there isn't much to be gained by spending a great deal of time developing models. Unless, your purpose is an enhanced understanding over the devices. Of one thing I am sure, developing a good model of these systems will be a great educational experience.
  3. G0HZU

    G0HZU QRZ Member

    The bit in bold is exactly right!

    My interpretation of the oscillator model is probably very different to how Tom sees the issue in his head. I'm not DEFINITELY saying I'm right and he's wrong because the physics of a valve are still quite unknown and scary to me. Being a bit scared of the unknown is healthy as it make you think harder....

    But my current model is screaming at me to ask an open question that may seem a little odd.

    If you go back a few decades and look at the commercial designs of these amplifiers did they ever do anything 'strange' with the chassis grounding?
    I wonder if they had what may appear to be poor grounding.

    i.e. did any of them have what looks like a poor (wire?) connection between the ground of the tank metalwork and the chunky metalwork around the tube itself?
  4. AG6K

    AG6K Guest

     The unexact /a.k.a. guesswork/ part is coming up with a model that resembles something reasonably close toreality.

     Arriving at a figure for Gm at the frequency of the VHF parasitic anode-resonance would seem to be a sticky wicket.

     RF-grounding the grid through a C to cancel some of the L in the grid to increase the grid-resonant freq. - thereby increasing the self-neutralizing region of the g-g circuit - was tried but, as a GDO/Dipmeter shows, it only helps by c. 1%.

     As Terman shows us, there is no such thing as a RF-ground without XL-ohms.
    • Rich, ag6k
  5. AG6K

    AG6K Guest

     Is Mu directly related to VHF-gain?

     Svetlanas do oscillate well Tom. Why?

     Are you saying that a VHF suppressor does not reduce VHF amplification?

     How can one change C-fb Tom? Isn't it a given for each of tube type?
    • Rich, ag6k

  6. AG6K

    AG6K Guest

     And when carbon-crystal semiconductors become available, silicon-crystal semiconductors will be on their way to downtown Oblivionville.

    • Rich, ag6k
  7. G0HZU

    G0HZU QRZ Member

    Hi Rich
    That's not why I ask the question about grounding. My reason is a bit more obscure than that. I think it's too soon to give my reason for the question :)

    To model the valve correctly as a VHF oscillator requires an approach that may appear alien to current thinking in the ham world. My analysis is certainly very different to Tom's.

    I'd like to publish a paper on this (and maybe even gain some credit for it in the ham world). However, to ME the contents of my paper are blindingly obvious and contain nothing new when viewed by an RF designer. So something fundamental in the ham world seems to have got lost over the last 50 years about how to model and analyse a GG valve amplifier for stability.

    I think that's why Tom still isn't on the same page as me wrt the importance of modelling the system correctly.
  8. W8JI

    W8JI Ham Member QRZ Page

    At the operating frequency, that is easy. It is a no-brainer, and even a Chaffee Analysis is close enough. One point, multiple points, at the primary operating frequency is not an issue.

    Since phase delay in the tube up above 30 MHz varies with transit time, and transit time is voltage dependent by any change in electrostatic fields inside the tube, and since the primary oscillation mode is control grid to anode feedback, it becomes impossible to document behavior doing load pulls to characterize the tube in any detail.

    About the best that could be done is to do load pulls to see what impedance makes the tube take off.

    Since the HF system is chassis and socket and wiring dependent on VHF, the data would be useful for only one particular amplifier. That seems like a lot of work for little benefit when the next system comes along.

    They've been done for years.
    Since transit time affects phase of feedback, it can cause a system to start or stop oscillation depending on frequency of the oscillation. or to jump to a new frequency.

    I'm trying to understand the benefits of doing a model when the model will only apply to one system, and the answer wanted is already determined long before the model is complete.

    Clearly there is some VHF printed circuit board thinking going on, and misconceptions, that in 1940 or some year past, there were no layout problems and we knew more about this. We know far more now than we did back then, except some myths and misinformation have made this look like a simple one component Q problem.

    Nearly all people with a wide range of experience in large power amplifiers using tubes are on the same page. It's only in Ham circles we see such mystery and confusion, because so much bad information gets repeated, all of the time is spent debating the bad information. The problem is that someone has decided a single universal cause and cure for all problems, and because people like simple answers that simple square peg is trying to be pounded into varying size and complex-shaped holes.

    73 Tom
  9. G0HZU

    G0HZU QRZ Member

    As you know I am only about '1 week old' in terms of looking at tubes. I recall looking at them very briefly as a student but they were treated as old school tech rather than analysed in depth.

    So I am a newbie :)

    But even as a newbie I'm struggling with why you say the feedback path is Anode-Grid. In my model the grid is ground so feedback here isn't an issue.

    On the contrary. I think the primary issue is the physics of the tank circuit and chassis that create the feedback. No PCB thinking going on from me. Just basic RF fundamentals :)

    Why do you think the feedback is anode/grid?
    Last edited: Jan 2, 2012
  10. G0HZU

    G0HZU QRZ Member


    You see I think the above analysis is fundamentally wrong. I have never been near one of these tube amplifiers but the analysis above dosn't make sense to me.

    I appreciate that there is 50 years+ of knowledge here against my 1 week of casual interest in this subject so I may be missing something very basic.

    Why do you think the feedback is anode/grid via C1?
    Last edited: Jan 2, 2012
Thread Status:
Not open for further replies.

Share This Page