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

    G0HZU QRZ Member

    If the validity of the VCCS model for the JFET is accepted (based on the successful comparison for neg R with the real JFET circuit in post #38) I can have an initial stab at showing what would happen if this circuit was modded to have the gate floated on three 200pF caps. Because this is a VCCS model I don't have to add a bias return choke across this network. So don't worry if you spot this isn't present.

    The top circuit in the simulation below has had this network added at the equivalent of the gate. So it now has three 200pF caps and I've added an extra 12nH inductance per capacitor. So one might think this would be a disaster for stability compared to a direct grounding as in the circuit below it. But look at the prediction for negative resistance. Below 100MHz the floated circuit is much better. This is the dark brown trace. Above 100MHz it is slightly worse. If the caps are reduced in size to 115pF then the floated circuit is better up to about 135MHz. Now this is a JFET model and it isn't the model for a 500Z tube so don't draw too many conclusions yet. But at least this is a first step towards answering the main thread question. It looks like floating the grids may help to (significantly) minimise negative resistance at the lower end of the VHF band. This is the region where the suppressor can't help as much. The suppressor will typically get better with increasing frequency (up to a point).
    The suppressorZ plot below shows that the ESR of the suppressor drops down to just 15R at 60MHz. So it is not able to protect against negative resistance as well down here. But by floating the grids the tube will produce less negative resistance here. So this all helps to broaded the protection against instability.

    What you really want on the thread is one of the original designers of these amplifiers. They may agree with me or they may have other thoughts as to why the grids were floated like this.

    Attached Files:

    Last edited: Jun 19, 2019
  2. HAMHOCK75

    HAMHOCK75 QRZ Member

    I was about to ask if you could model that since you had already done the grounded grid. Is there a reason for three 200 pF caps?
  3. SM0AOM

    SM0AOM Ham Member QRZ Page

    Most interesting discussion.

    Something I would like to learn more about is the modelling of external-anode tetrodes at VHF and UHF operating in grounded cathode.

    There must be models around that were devised when
    multi-octave operating bandwidth linear amplifiers became popular in the 50s, and the need both for stabilising at the operating frequency and to avoid parasitic resonances in and around both the tuned circuits and the operating voltage feeds became pressing.

    Conventional neutralisation schemes at the tuned frequency are covered extensively in the literature, but the parasitic question is not. One such problem occurs when operating high-gain tubes in parallel at HF as there are parasitic push-pull circuits at VHF and UHF that may get inadvertently excited.

    Now, such amplifiers have become "history", but it would be interesting to know if and how the design teams modeled the circuits including the tube in really large frequency ranges in order to find out where to put the suppressors.

    The only tale about this that I can recall from listening to "my betters" as a "wet behind my ears" engineer in the early 80s, was told by Kjell Djurle who was transmitter design manager at ITT-Standard Radio in the 60s and 70s, and responsible for the RF deck in the CT1000 and ST1400 transmitters.

    He told us that the whole amplifier at idle current condition was swept from a few 100 kHz up to above 500 MHz using power signal generators and the shape of the transfer function was examined at all frequencies, except of course the tuned frequency.

    If there should be a "sudden bend" in the function, its frequency was noted, and by using a wavemeter with a probe, locations and types of the resonances were determined one by one.

    But, I never heard about or saw any analytical models trying to pre-calculate the suppressors. It must have been possible to at least estimate the effects of tube capacitances and internal inductances together with the socket properties to find out the amount of gain, phase-shift and regeneration available in the whole frequency range.

    KD2ACO likes this.
  4. KM1H

    KM1H Ham Member QRZ Page

    Since "negative resistance" has been around and understood since before I was was taught and understood even at the technician level in the military and civilian tech schools. Unfortunately this level of training is no longer with us but there is still plenty to read on the subject for those inclined.

    My JFET introduction was around 1966 with the U-310 as a Sr Engineering Aide at National Radio where we had both transistor and vacuum tube Tektronix curve tracers side by side on a table in the R&D lab. I have those same antiques here after purchasing at the IRS auction 1n 1992 and with some basic servicing they work as expected and are used as required. I still rely on the U310 and its cost reduced J310 cousin for many circuits I build and use.

    While the discussion is sometimes interesting I fail to understand how a JFET modeling exercise full of guesswork can have much in common with the performance of a 3-500Z in actual operation in a SB-220. Since it has long been stated that an oscillator is simply a failed amplifier the methods to tame it are long known and understood at the practical level of likely the huge majority on these forums.

    As far as asking about a basic backround of someone who appears to keep it well hidden even to the point of having his ham call info deliberately suppressed I am not prone to believe everything I read from this sort of person. Trying to laugh it off as silly just increases the lack of credibility factor.

    I have the same thoughts about those with their own cute screen ID's that do not have the nerve to use their ham call.

    Carry on
  5. G0HZU

    G0HZU QRZ Member

    I'm merely following company guidelines. I'm not on Facebook, LinkedIn or Twitter and I try for a minimal digital footprint in terms of personal background.

    The point of the JFET work is that there is minimal guesswork. The model is based on the datasheet and the 1 port model is straight off a VNA measurement of the real amplifier. They both agree very closely in terms of the qty of negative resistance they produce across the VHF band.

    The main point of it all is to demonstrate that this type of amplifier (triode or JFET) is prone to generating negative resistance when the grid/gate is grounded via a small amount of inductance. It's the job of the suppressor to cancel this negative resistance. If it can be demonstrated to have been cancelled then the amplifier will be much more likely to be stable.

    If it was possible (safe?) to measure a 1 port model of the 500Z amplifier output with a VNA then this model would remove a lot of the guesswork associated with the tube version of the model. The VNA would reveal the negative resistance vs frequency on the screen! No need to model ANYTHING once this is done because the VNA gives the information directly.
    Last edited: Jun 19, 2019
  6. G0HZU

    G0HZU QRZ Member

    The 500Z tube has three grid pins on the valve base and the original amplifier design has a 200pF cap fitted to each pin.
    Last edited: Jun 19, 2019
  7. G0HZU

    G0HZU QRZ Member

    Sure, but the s parameter based analysis methods I'm describing are very powerful and they have been around since the 1950s at least. See the attached 1 port data model exported from the VNA. This model covers 300kHz to 500MHz. The VNA has no idea if it is measuring a JFET or a tube amplifier but it is measuring 'something' with a single port that produces negative resistance at that single output port and the VNA provides the data file below. With this data file it would have been possible, even in the 1950s to work out if this device could be tamed by adding a 47R and 85nH suppressor at the output. This could be a pen and paper exercise although a slide rule would have been used back then. A computer can do it in a flash but back then it would have taken a student/engineer a fair time to do it by hand. It should be possible to work out where the suppressor succeeds... and where it fails. All the info required is locked up in that simple 1 port data file below. Just do a few sums to 'add' the suppressor and decide if it can remove the negative resistance across all frequencies. It should be possible to work out at which frequencies it fails and what kind of load would make it oscillate at those frequencies even with the suppressor fitted.
    Repeat the exercise by measuring a tube amplifier and you would get similar insights although great care would be needed to keep the operator away from lethal voltages and there would be a real risk of damaging the VNA as well.

    Attached Files:

    Last edited: Jun 19, 2019
  8. G0HZU

    G0HZU QRZ Member

    Sorry, I probably should have post processed that VNA data file into something more intuitive...The VNA doesn't export it in a basic reflection coefficient format. I've had a quick go at converting the data format to a classic reflection coefficient.

    This is now in the classic mag ang format. Sorry, I should have posted it up like this in the first place as it may look a bit alien to quite a few people. The two files should be equivalent but the one below is more 'human' to look at.

    The VNA has no idea what circuit it is testing but the data file below can instantly reveal so much about the stability of this amp when presented with various loads.

    I could give the file below to a junior engineer and ask them to predict where it will almost certainly oscillate if the amplifier output is presented with an AC short. They should be able to do this by just reading the text file below. No need for sums or a simulator. That's why the format below is much more intuitive to an engineer.

    For a bit of fun, after reading the attached data file for the amplifier output port can anyone predict at what frequency the 'real' JFET amplifier will oscillate if I put an AC short at the output port? No cheating with an RF simulator mind... ;)

    Attached Files:

  9. AA5CT

    AA5CT Ham Member QRZ Page

    re: "What you really want on the thread is one of the original designers of these amplifiers. They may agree with me or they may have other thoughts as to why the grids were floated like this."

    Making use of one's antenna analyzers today ought to making characterizing the (grid) choke-cap combo a cinch, then toss that into the circuit model and observe the effect.

    One thing to look for - any strange resonances ... I have such an SB-220 unmodified, but, it would be some time before I got around to doing this.
  10. SM0AOM

    SM0AOM Ham Member QRZ Page

    Applying one quite "foggy" recollection from the course "Microwave Amplifiers" in 1981,
    this frequency should be the one where both these conditions

    |S11 (device)|.|S11 (short)| > 1 and arg (S11(device)) = - arg (S11(short))

    are fulfilled simultaneously.

    Which occurs at around 353.5 MHz, where the phase angle of the device S11 changes sign.


Share This Page