Grounding the Grids vs. Cap/RFC Grounding

Discussion in 'Amateur Radio Amplifiers' started by W4LAC, Jun 13, 2019.

ad: L-HROutlet
ad: l-rl
ad: HRDLLC-2
ad: L-MFJ
ad: Left-2
ad: L-Geochron
ad: abrind-2
ad: MessiPaoloni-1
ad: Left-3
  1. G0HZU

    G0HZU QRZ Member

    I'm answering my own quote here to keep things more peaceful... A typical suppressor coil will have a fairly squat L/D ratio and for maybe 100 years it has been possible to use this ratio and a classic graph to predict the so called self capacitance of a solenoid. In the case of a suppressor coil I'd expect this self capacitance to be about 0.6pF or so. This would mean the resonant frequency when measured on a classic s11 measurement (against ground) would be up around 600MHz or so if the inductance was just over 100nH.

    However this is fed against ground and when the solenoid is against ground it will show a lower self resonance compared to the case where it isn't grounded at either end. This is because the solenoid is really a complicated transmission line and when it sees a short at one end to ground you get a free 180degree phase shift in the reflection at that end of the solenoid because I think the coil (or transmission line) treats this as a short circuit there. At the other end it will obviously look like a fairly decent open circuit at resonance and many people refer to this as the 1/4 wave resonance mode.

    However, if you place it in free space as you would normally do with a dip meter then it's best to consider the length of the wire when trying to predict the SRF. In this case both ends are unterminated so it is tempting to think the SRF will now leap up to 1.2GHz because the 180degree shift (from the short circuit at one end) is gone. But in reality a regular form factor coil like this would probably show a SRF close to the 1/2 wave frequency of the wire length. 15cm is a 1/2 wavelength at 1GHz. Therefore the SRF doesn't double in frequency for various reasons but if you had a dip meter that could work to 1GHz you would see this resonance mode with the coil in free space and it would be somewhere around 1GHz. This is miles away from 100MHz.

    I'll try and get a few 47R Ohmite OY resistors and see how this affects the SRF when the coil is wound around it. I'm not expecting to see anything dramatic in terms of a change in SRF when in free space.

    Rather than use a GDO I'd recommend a VNA as it can mimic a GDO and cover a huge frequency span in a fraction of 1 second. Try it... the coil will have a 'first' freespace SRF up around 900-1000MHz assuming you make a reasonable copy of a classic suppressor. I'll try and get several 47R resistors and try various coil form factors including a hairpin type. it might not be for a week or so but I'm due to place a Farnell order early next week :)

    There will be higher order SRF modes displayed on the VNA as well and these will be in the 1.5-3GHz region but I guess we aren't concerned about those.
  2. K9AXN

    K9AXN Premium Subscriber QRZ Page

  3. K9AXN

    K9AXN Premium Subscriber QRZ Page

    There are three parts, they should answer your question. There is a good deal more information to follow.

    There is a whole litany of information beyond what the url provides most not published by Collins the genesis of the design.

    Regards Jim
  4. AC0OB

    AC0OB Platinum Subscriber Platinum Subscriber QRZ Page

    I would be interested in your analysis as well.

    As per the PDF, I modeled in MatLab (simplistically) what I considered a typical plate Parasitic choke as a harmonic trap.

    I used the higher order coil equations to derive the Lcoil and the Cselfcoil values and the SRF.

    For Impedance I used the Z equation for a parallel LRC circuit.

    Of course it was modeled in free space and did not account for the tube plate or load impedances at those respective nodes.


    Attached Files:

    Last edited: Jun 21, 2019
  5. G0HZU

    G0HZU QRZ Member

    I've not seen a GDO model that I'd trust to work up at 1GHz but maybe they do exist. I'd much prefer to use a VNA even if it meant using it to mimic a GDO :) A lot of research work was done into this stuff at work nearly 30 years ago because in the early days of the company we had limited test gear and it was very difficult for us to generate proper s2p models of components. So we created a physical model of a solenoid that could be used as a component in an RF simulator. The goal was to be able to have a model that could capture the behaviour of a solenoid as the port impedances are changed and the model would be able to reproduce the various resonance modes at the correct frequencies. We found it wasn't strictly correct to refer to the resonance modes as 1/4 wave or 1/2 wave modes because the EM wave propagation through the solenoid structure isn't that simple. Our model tried to capture all this more accurately than a simple wavelength based model.

    However, if the solenoid is grounded at one end the resonance frequency might be 600MHz. Free it from ground and measure it in free space and the resonance might be up at 1GHz. Ground both ends and the resonance should shoot up even higher. A lot depends on the length vs diameter of the solenoid and so there are no hard and fast rules here. Our model took into account the length and diameter as well as end effects and it worked very well over a wide bandwidth :) However, this old model has been effectively obsolete for 20-25 years since the company grew and we starting filling the labs with decent 2/3/4 port VNAs. We generally model inductors using a 2 port model and the VNA is simply fantastic at this stuff over a huge bandwidth.
  6. G0HZU

    G0HZU QRZ Member

    Interesting read... thanks. I agree that the 200pF caps can help offset the inductance of the grid connection inside the tube. The rest of your page about the negative feedback techniques was very interesting to read. Thanks again. Is any of the old research/theory work by Bruene available to read online?
  7. G0HZU

    G0HZU QRZ Member

    Thanks for that. It looks like you predicted 720MHz as the SRF. One think to be wary of though is that the classic lumped model using a parallel LC isn't very portable when used in anger on an RF simulator. It will misbehave when the model is used in different circuits especially if the port impedances are changed. For circuit board level design a 2 port VNA model is king here but it is possible to make a decent physical model of a solenoid that can work nearly as well as the VNA model.

    A while back someone asked me about this stuff and I posted up a youtube video of our old physical model vs a VNA model. Both are imported into an RF simulator and then the port impedances are changed and you can see the complex impedance is plotted long with S21 and group delay. The physical model is that of a fairly large airspaced 400nH solenoid. The youtube video is fairly awful to watch because there is no sound (sorry) but it does show how well our old solenoid model could perform. This was the basic version of the model and it is possible to get better performance than this by adding more sections to the model. But this model was usually good enough. It would be good enough for a basic bias tee design for low Z solid state design for example.

    Apologies again for the lack of sound and the dreary nature of the video... the model is the star here, not my awful video editing skills :)

  8. AA5CT

    AA5CT Ham Member QRZ Page


    Since we have advanced, 3D sim tools available today, it would be interesting to "code up" a 3-500Z tube in Ansys/Ansoft HFSS in a typical amplifier circuit (after all, HFSS is used in semi-circuits to model 'bond wires' and the like) and take look at the possible 'resonances' that would present themselves ...

  9. KM1H

    KM1H Ham Member QRZ Page

    I suspect that some are not understanding about the uses of a GDO and are making rather strange jumps of logic and some stranger assumptions.

    Lets stick to the SB-220 and stop drifting off into other areas for awhile. Leave the JFET in a lead box as it only adds confusion for many who have commented to me privately but dont want to get in a never ending p*** contest with Jeremy:eek:

    The GDO in Dip mode is excellent for finding most if not all resonances within the tube cage with everything connected or disconnected. There is no need to have any HV applied! I could care less what happens at 400 MHz and up as the 3-500Z is a dud up there and even the U-310 JFET is gasping

    When in DIODE mode, aka WAVEMETER mode, it can pick up RF a safe distance from any HV or RF. A wavemeter was used for frequency measuring as far back as spark gap days and is still a useful tool. Those not trained or comfortable around HV should either use a SS amp or have someone else do any servicing. Any tube amp is not a place to learn the basics as far as most are concerned....there are exceptions.

    Since a VNA or SA is not necessary to build a good suppressor the first time out for a tube amp it is just an expensive one time toy but some with little to no experience out of a lab are obsessed with mentioning it. Ive lived and worked in home and professional HV/RF environments since 1955 and 1963 respectively; accept my experience, its FREE.

    I do not have a SB-220 handy at the moment but did check out my Alpha 76PA, a 3x 8874 amp from the late 70's. It uses the same 3T coil but over a 100 Ohm 2W carbon resistor that is well within spec. The GDO shows a very weak dip at ~ 235 mHz which is what I would expect with the aluminum cover off. With a SB-220 the steel cover had a big effect and was under 1" from the anode connector and a little blip, barely noticeable on the Wavemeter Mode GDO, was down around 160 MHz if I remember correctly; that blip never had a chance to become an oscillating tiger. This holds no matter what the tuning and band controls were set at or antenna in use here which are often far from Z = 50 +jO

    My conclusion of this thread value so far is that the grid caps are useful and that Heath engineers realized that 115 pf was a better choice; I already suggested 120 pf as a readily available value and close enough for our needs and have used them since the SB-221 came out in 1978. Despite what a certain W8 and his groupies wants us to believe Heath had some well qualified engineers working there.

    A GDO with a Diode Mode switch position is a very cost effective tool for anyone working on a tube amp. Many are in the under $40 range on Fleabay with a full set of coils. Mine is the classic Measurements Corp. Model 59 which was another item picked up at the 1992 IRS National auction and still has its property tag. The one I built as a kid still works but the 59 is easier to use as the tuning head is separate. The 1947 released 59 uses a 955 acorn triode and is good to well beyond 400 MHz with plenty of output. A special UHF head was available with the same tube; I have one. Not bad for a 1935 introduction and still available as NOS WW2 surplus real cheap!

    Solid state versions have less sensitivity but most are small and useful, read the various online reviews and comments. The 1962 Heath Tunnel Dipper, using a negative resistance device, is reported to have poor sensitivity but may be just the thing for probing RF in diode mode. It also uses 3 germanium transistors which may be quite leaky by now.

    Last edited: Jun 21, 2019
  10. G0HZU

    G0HZU QRZ Member

    OK, I have just powered up the VNA and attached some E and H field probes as in the image below. This is a great way to look at resonances in any structure and I've wound a 110nH suppressor coil as a test coil. There are two traces on the VNA and the VNA has a 2GHz span. The yellow trace is S11 and this is connected to the H field probe. This is acting like a GDO as you can see that the coil is sucking energy from port 1 at resonance. In the case of the grounded coil this is happening at about 650MHz. This agrees with general theory that has been around for 100 years or so. This coil is not self resonant anywhere near 100MHz when grounded at one end.
    The blue trace is S21 and this is a very sensitive way to look at resonance. The idea here is that the coil is excited into resonance by the H field probe and the coil then generates a decent E field at resonance and the nearby E field probe can then pick this up at port 2. Not surprisingly this peak in the blue trace happens at the same frequency as the dip in the yellow S11 trace at about 650MHz.

    There are no false dip risks with this approach and this method of testing coils is an established one as well.

    When measured in free space the resonance climbs to about 900MHz as in the other VNA screenshot. This is what you would see with a GDO. You can also see higher resonance modes on the screen as well.

    The point of this? To hopefully demonstrate (finally?) that these coils aren't broadly self resonant at 100MHz.

    Attached Files:

    Last edited: Jun 21, 2019

Share This Page