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: Subscribe
ad: L-MFJ
ad: Left-2
ad: MessiPaoloni-1
ad: K5AB-Elect-1
  1. AG6K

    AG6K QRZ Member QRZ Page

     I am not if it's going where I think it's headed.

    • Rich, ag6k
     
    Last edited: Dec 30, 2011
  2. G0HZU

    G0HZU QRZ Member

    How about non linear models for transistors that are a couple of inches across with big fat tab leads that are half an inch across?

    I contacted freescale a few weeks ago for transistor models for a huge PA transistor used at UHF. These were non linear models.

    I've also used models for designing with 'big transistors' at several GHz and had very good agreement with the simulation in terms of harmonics, efficiency and power output once the amplifier was submitted for its formal acceptance test plan.
     
  3. AF6LJ

    AF6LJ Ham Member QRZ Page

    Not that short for sure;
    if there was that grid and filament assembly would look like someone hit it with a combination of acid and a plasma cutter. When an S-Band filter ionizes you can see gold plating evaporated off the filter elements and you can see pitting in the base metal. If the plasma is let go for a time the underlying metal will be melted and combined with some of the gas that made up the plasma.
     
  4. VK4TUX

    VK4TUX Ham Member QRZ Page

    Lou, Seeing above I am curious why you dont direct ground grids on your sb-220 6m conversions?
    They obviously work ok left original, but wondered if you did try that with those?

    cheers

    Adrian ... vk4tux
     
  5. W1QJ

    W1QJ Ham Member QRZ Page

    Good Question Adrian, As many will tell you who have directly grounded the grids on an SB-220 the result is pretty much a slight increase in power output with less drive power. And I do ground the grids on many of my amps. The ones that get directly grounded grids of the many cores that come in are the ones that have been modified with those piles of tiny caps and resistors added to the grids less the RF chokes. (See photo of the similar pile in Tom's post) As you can well attest there is no need for any more power output from the SB-220 on 6 meters. There is enough available power as yours is fitted to blow the transformer into the next century. The increase that you will get if you do ground the grids is just that much more gun power! Since I trust your abilities, go ahead and do the direct grounding on yours and then slightly retune the input and your eyeballs will probably fall on the floor afterwards. I find it hard to compare a mono band amp to an amp that tries to cover 160-10M with similar performace throughout all those bands.
     
  6. W8JI

    W8JI Ham Member QRZ Page

    Right. As I said, I would expect that for compact devices, especially on lower frequencies. A chip capacitor looks very predictably like what we would expect up to several hundred MHz and higher. A standard disk, not so at all. A large air variable, nothiong like we might expect with a lumped model.
     
  7. W8JI

    W8JI Ham Member QRZ Page

    Although it is impossible to tell with a poor quality picture of only one perspective, to me it looks like the grid support is very pitted. The normal place for an arc or plasma is off the sharp points of the anode fins down to the grid support cone, and that cone looks marked in many places.
     
  8. G0HZU

    G0HZU QRZ Member

    As I've said before I'm new to this forum and I now realise that the subject of suppressors and parasitics goes back decades and there's a long standing feud related to this between Tom and Rich.

    So the smart money says this thread is doomed. Which is a shame because I'd like the chance to help explain (independently) how a basic anode suppressor works through basic RF theory and modelling.

    Note that I also feel for the forum moderators who must groan each time this subject kicks off again so can I politely request we put any personal bunfights on hold for this thread (please :) )
    This thread is a nice for me because it deals with a subject I am a bit passionate about. The subject isn't valves and it isn't suppressors. It's much more low level than that.

    The subject is how to use computer modelling to help understand behavioural problems in an RF system.

    So far on the thread there have been plenty of powerful reasons from Tom as to why modelling is a waste of time wrt valves. Now in many ways this is hard to argue against because valves are very different to semiconductors for reasons I'm sure you all know MUCH more about than I will ever know despite all my years of experience in solid state RF design.

    I'm happy to reveal that my valve experience is extremely limited. So why am I still here banging a drum for modelling?

    I guess it's because you can adopt two schools of thought. One is that you can't ever produce a comprehensive model to explain how a valve works because of the basic physics of the device.
    Now I'm not attempting to challenge that school of thought because you already know my limited experience of valves and I would be the wrong person to champion such a challenge.

    The other school of thought wrt modelling parts of a system at RF is to adopt an extremely simplistic approach to modelling. This may appear to fly in the face of the first school of thought but it really doesn't have to.

    The behavioural problem associated with GG amps is a tendency to oscillate at VHF into a typical tank circuit. My argument is that you don't need to use very complex models for an amplifier to model why this instability happens.

    In other words this isn't actually an instability condition that requires a 'school 1' model of a valve to comprehend and study. Of course it would be amazing if the school1 model existed as a download somewhere but it isn't going to happen any time soon if ever.


    Now yesterday I did write a paper on anode suppressors and I now realise I'm going to have to edit it a fair bit to even get it past first base because it uses school2 models that are so simple that I suspect it will be discarded as a pitiful attempt.


    My paper is a word document that adopts an RF designers approach to understanding the need for suppressors for a particular type of oscillation condition. I realise now I'm going to have to add a couple more pages and also adopt a wholly different introduction to simple models in order to gain my target of universal acceptance of its contents :)


    I'll see if I can finish it later today and upload it for polite peer review by anyone who wants to read it :)
     
    Last edited: Dec 30, 2011
  9. W8JI

    W8JI Ham Member QRZ Page

    The small capacitors on the grid have the largest effect on 160 meters, and the least effect on six meters.

    I have S12 or S21 plots of an 811 tube, which is similar to a 572 tube, that show the differences.

    The largest danger of the mica caps comes on multiband amps, and from tube flash overs, and with tubes that have very short very wide grid connections. When the grids are not firmly grounded, an arc can easier get plasma to the filament or cathode, and that can couple out to the exciter or the bias system or antenna relay.

    The largest impedance changes at VHF, and S12 / S21 changes at higher frequencies, are in tubes like the 8877 or 3CX800. The least important
    changes in operating performance are in tubes like 811's or 572's. The grids are already so long, another inch or a few dozen ohms reactance outside the socket hardly matters.

    It is critical, however, in all GG amplifiers, to directly ground the grid for arc fault protection. I initially grounded the grids through equalizing resistors to even-out the grid current in four tube 811 amps and 572 amps. This reduced delta in grid current between parallel tubes. Contrary to what one person claims, this case was not negative RF feedback. I used .01uF capacitors, which are very low reactance, even on 160 meters, compared to grid-cathode impedance. It was entirely to equalize grid current and heating due to tube variations.

    The resistors statistically turned out to be much more costly and less reliable than simply grounding the grids and paying more attention to mixing tubes.

    There are reasons to float grids, but the advantages of those reasons have to exceed the drawbacks. Fusing the grids and negative feedback, or series tuning the grid are bad reasons and can be proven bad.

    I hope this thread does not get locked because some people can't stay on technical topic!!!!

    73 Tom
     
  10. W1QJ

    W1QJ Ham Member QRZ Page

    The small capacitors on the grid have the largest effect on 160 meters, and the least effect on six meters. (TOM W8JI)

    Yes, this is why it really only boils down to gain on 6 meters.
     
Thread Status:
Not open for further replies.

Share This Page