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

    G0HZU QRZ Member

    Hi Karle-Arne, yes that's it. It should oscillate at about 353MHz. I thought you would spot this quickly so I posed the question for a bit of fun...

    For those who don't like maths etc the relevant part of the data file is here.

    352.5885 1.06647 -178.94875
    353.21312 1.06146 -179.54065
    353.83775 1.05679 179.84584
    354.46237 1.05193 179.24806
    355.087 1.04718 178.6498
    355.71162 1.04185 178.05712

    The first column is frequency, the second is the magnitude of the reflection coefficient and the last column is the reflection angle.

    The key thing to be aware of is that when an EM wave (or any kind of wave) hits a short circuit all of the energy gets reflected but the phase changes 180degrees on the reflected wave. So to satisfy oscillation when there is a short at the output port you have to study the file and look for a frequency that has a reflection coefficient above 1.000 and it also has to have a 180degree phase angle. If you combine that 180degrees (from the data file at 353MHz) with another 180degrees caused by the short circuit you get zero degrees overall phase shift in the system when the short is connected to the output. This then sets up the conditions where you have a 1 port device that can top itself up with energy (because the reflection coefficient is above unity) and the top up happens 'with unity phase'. So you satisfy the requirements for an oscillator. The gain overcomes losses and the net phase of the top up is unity. So the JFET amplifier turns into a self fulfilling oscillator at about 353MHz where the phase of the reflection coefficient is closest to 180degrees.

    To demonstrate this I fitted a decent SMD cap (ATC porcelain) as a short circuit at the output of the real circuit . This cap has very low parasitic inductance and makes a very good short circuit. However, the spectrum analyser plot below shows an actual oscillation frequency of 360MHz. This is not at the expected 353MHz probably for two reasons. My clumsy soldering meant I probably disturbed the positioning of the output components slightly and also the 353MHz prediction is based on the small signal startup frequency. Once the JFET oscillates up to its amplitude limit the large signal oscillation frequency often shifts slightly as the device limits. But 360MHz is very close to 353MHz and this is a good demonstration I think!

    Attached Files:

    AI3V likes this.
  2. KM1H

    KM1H Ham Member QRZ Page

    No argument there and Ive been trained in that area since the 60's and finished my career in 2002 as lead engineer on a part of the F-35 avionics package well above 300 GHz.

    Maybe, maybe not, but given a starting point of R and L any decent bench tech could optimize it. The R and L approximations were around long before I worked at National and the HB 6M AM rig I built in 1956 was parasitic free the first time and I didnt even need to use a pencil much less a slide rule which didnt come to much later. I had already built a GDO from a 1949 CQ Magazine article and that along with a very sensitive Knight VSWR bridge were the only tools needed. I was building HF and 6M amps to the legal limit after that without difficulty.
    KISS has always been my first choice.

    That is a no brainer
    353.21312 1.06146 -179.54065
    353.83775 1.05679 179.84584
    And shows what is needed when running a device at or above its parasitic frequency.

    OTOH the engineers realized what was needed back in the ancient days and all it took was a tiny Arnold Engineering ferrite bead over the device input lead. These days Fair-Rite offers ferrite solutions for everything.

    Heath even tried ferrite beads as the sole parasitic suppressor in the SB-230 amp. That caused no end of tube and other failures also helped by the idiotic grid circuit. I picked up a "destroyed" one real cheap and converted to 6M for portable and Rover use using a cheap (at the time) 8874 that I modified the anode cooler; after ~ 30 years it still runs fine at 500-600W out.

    Getting back on subject The SB-220 is barely neutralized to start with on 10-12M by many builders and with ~ 20 dB IN/OUT, or less isolation, just adds to the stability problems. Remember that in a GG amp the I/O is in phase. Comparing to a lab model JFET in the real SB-220 world could become a nightmare to the inexperienced who believes a lab exercise is his answer to everything .....until the Big Bang occurs. Watching a SB-220 really take off with Ip pegged, anodes white, and until the slow circuit breakers trip is something to behold and would make a nice You Tube video. In my case the tubes survived but the barely suitable original PS diodes didnt BUT the transformer did:cool:.

    With about 500 really/truly SB-220 family amps under my belt for HF and 6M conversions for almost 50 years which includes building for two Heathkit Stores and individuals I feel I have a good understanding of what works....and doesnt.

  3. G0HZU

    G0HZU QRZ Member

    This stuff is relevant to the thread topic because these techniques can be used to predict (with fairly good reliability) where a suppressor won't be able to stop oscillation if changes are made to an amplifier at the gate/grid. For example, the data from the attached s parameter file can be used to easily predict where a classic 47R and 85nH suppressor would fail to prevent oscillation with that particular amplifier. The required network to cause that instability at that frequency can be very quickly calculated just by studying that data file again. This is far more powerful than sniffing with a GDO. I could do this prediction in less than a minute.

    I'm tempted to ask you to demonstrate how to do this calculation (and predict the margin of failure for the suppressor) for the benefit of the forum because you do seem to be very confident and this would be another no brainer for you.

    However, you have already dodged the question where I asked you to explain how you think a regular 85nH suppressor coil is designed to be broadly self resonant at the oscillation frequency. That doesn't tally with the physics of a typical 85nH coil found in these amplifiers. I would suggest that your theory that these coils are 'self resonant' at the oscillation frequency is just plain wrong by a huge margin. Not even close to being correct and it demonstrates the lack of understanding I mentioned earlier.
    Last edited: Jun 20, 2019
  4. G0HZU

    G0HZU QRZ Member

    Of course, to answer this, one could just look at the various plots for the JFET circuit and the suppressor that I posted up earlier in the thread and work it out from them in a matter of seconds. This would give a reasonable answer to the question although it wouldn't be as accurate. No need for a computer or even a calculator or even fingers and toes to count on. That was why I took the time to post them up in the first place. It would just be some simple maths a child could do in their heads. The same approach would work just as well with a tube amplifier with the same depth of information. All based on a single 1 port VNA measurement.

    It would also be wise to measure a real 47R and 85nH suppressor on a VNA to get a better idea as to the quality of the real suppressor as the construction methods and the components used would affect the performance of the suppressor, especially up around 110-150MHz or so.
  5. G0HZU

    G0HZU QRZ Member

    It might have to wait for the weekend, but I could post up a few pictures and words that look at how the feedback inside the device happens and the idea would be to demonstrate where the negative resistance (appearing at the anode) comes from and how the datasheet can be used to give an initial guess as to how much negative resistance to expect at the anode. This would give some idea what amount of negative resistance the parasitic suppressor has to cancel out. I've already posted up a computer generated prediction (based on the datasheet) but the idea here would be to work out the negative resistance by following signals through the device and seeing how it gets generated. This wouldn't produce a very accurate result but it might make it more obvious to more people how the tube generates this nasty negative resistance in the first place (and how to quantify it with pen/paper) and what makes it worse and what reduces it.
    Last edited: Jun 20, 2019
  6. KM1H

    KM1H Ham Member QRZ Page

    What I actually said was:
    I had thought that BROADLY was self evident. The true modern vacuum tube parasitic suppressor as currently used is not a precision device which is obviously not acceptable to those with little experience with vacuum tube amps and in the REAL WORLD. Development started in the 40's with a trimmer cap and various places to connect the resistor to the coil which was hit and miss and time consuming especially in production. It reached the ham world in the 50's. For those that cant get past a lab with expensive test equipment, software, and an obsession about posting I offer my pity.

    Ill repeat again what I said but in bigger print this time which may be easier to fathom:

    The inductor should be BROADLY self resonant around that frequency.

    Which is why a couple of generic designs are used that fits the sole requirements of being universal across a wide range of tubes and amp circuits, requires no overpriced and wasted engineering time for product development, is simple to produce and easy to implement with no tech tweaking it works 100% of the time as long as the resistor value doesnt change excessively as do carbon comps. Ive no idea of the long term stability of the OY resistors into the distant future.

    That alone should put the stopper to your crude insults. IMO it is you who does not understand the actual purpose of the suppressor in a SB-220 and I wonder how many more pages of blathering is forthcoming before it is stopped by a moderator.

    You might want to hire a qualified tech to take your 1980's HP 8566B and adapt it to a SB-220. Also the CRT's in that unit are prone to a limited life and are pretty much unobtainium at a decent price ; a color LCD replacement is available. There are also several tantalum caps waiting to fail. Keeping old gear in proper operation/calibration is also an ongoing issue and a waste of my time which is why Ive sold off my 8566B and others while they still bring some money. A more modern version is readily available for me to borrow in this area of high numbers of RF based companies.
  7. SM0AOM

    SM0AOM Ham Member QRZ Page

    "Arguments lacking, raise your voice"
    - Esaias Tegnér, 19th century bishop, scholar and playwright

  8. KM1H

    KM1H Ham Member QRZ Page

    YAWN, just more nonsense from the junior tag team member who hasnt offered anything useful to the discussion yet.
  9. G0HZU

    G0HZU QRZ Member

    That little 85nH coil will still look like an inductor way up into UHF. A GDO is a risky tool to use if you tried to measure it 'in circuit' because you risk measuring the resonance of the combined tube + suppressor + tank circuit and then attributing this figure to the suppressor coil itself.

    I often use the HP8566B for oscillator demos because it has a very fast screen sweep/refresh on wide spans. This means it is very good at spotting transient behaviour from a wayward amplifier that is trying (somewhere) to be an oscillator. It is my first choice analyser for stuff like this although some of the classic analogue analysers from the 197os were much faster :)
    We can all agree that there is always so much to learn about RF. For this reason every day is a school day for me so I'd appreciate it if you could explain why my negative resistance analysis is wrong and your 'resonant suppressor coil' analysis is the right one?
  10. KM1H

    KM1H Ham Member QRZ Page

    A GDO is a fine tool for those who know how to get the most out of it and use one proper for the application; there are many versions including one with a negative resistance device.

    Please show me where I ever said your analysis was wrong? Are you at the point of having to make things up to continue wasting my time?

    Those 70's and older SA's are getting harder and harder to maintain and I do all my own repair and cal work. I still use a HP 141T coupled off the TX coax to monitor IMD, spurs and other things from my own station and others. It does not run continuously thankfully!

    Please stop trying to spin things and accept that the end result is what counts in the REAL WORLD with tube amps where some of the variables vary that may not be successfully measured for insertion into some programs.

    I understand the SS procedures very well having used Microwave Office for many years in the MM Wave region at the design level.

Share This Page