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: Left-2
ad: HRDLLC-2
ad: L-Geochron
ad: Left-3
ad: abrind-2
ad: MessiPaoloni-1
ad: L-MFJ
  1. SM0AOM

    SM0AOM Ham Member QRZ Page

    They certainly can.

    Depending on their topology, they either have a series inductor or a shunt capacitor at their outputs. This affects their impedance behaviour outside the passband.

    It appears that a slight preference is around for the shunt capacitor output topology, which goes towards a short circuit with increasing frequency. A sufficiently long interconnecting cable can then provide almost all phase angles of the source reflection coefficient.

  2. G0HZU

    G0HZU QRZ Member

    Yes, agreed! In case anyone reading this doesn't quite understand this subtle issue yet, see the short youtube video below. The amplifier here is the same one as in the last youtube video and it is a VNA measurement of a real amplifier. It has a K<1 at around 50MHz so it could oscillate under certain conditions. However, in order to get it to oscillate it needs to be able to generate a negative resistance at the output port. At the start of the video with a purely resistive driver it can't do this at 50MHz as the resistance at the amplifier output is still positive at 50MHz. So no amount of tank tuning would ever cause this amplifier to go unstable. So some people might declare it a stable amplifier. But look what happens if the 50 ohm driver is fed through a LPF and a variable length of RG8 coax cable. At certain cable lengths the negative resistance suddenly appears and it also disappears again at some lengths. Once the negative resistance appears with the right length of input cable the amplifier output tank would need to be tuned very carefully to produce the correct inductance to complete the oscillator circuit at around 50MHz where the resistance is negative. Once this is achieved the amplifier would oscillate at 50MHz.

    Could someone really discover subtle instability conditions like this manually? They would need to put something at the input that allowed all angles of the input reflection coefficient to be explored and they would need to do this fluidly for every conceivable combination of tank settings in order to discover subtle weaknesses in the amplifier like this. I'd argue that this would be virtually impossible to do in a thorough manner. The K factor analysis can predict these weak areas in an instant and the designer can then plot stability circles on a smith chart to work out what length of coax would make it generate negative resistance where K is <1.

    That is why it is a common design goal to try and get K>1 and B1>0 at all frequencies. Then all these problems vanish and it won't matter how long the coax is or if there is a LPF at the input. The amplifier will be unconditionally stable regardless of any 'dynamic' tuning or configuration changes like this. It's a much nicer place to be :)

    Last edited: Jul 11, 2019
  3. KM1H

    KM1H Ham Member QRZ Page

    I have a JAN CDZ-811 by Electronic Enterprises Inc that has only one grid wire into a pin. I dont want to take my Collins ATC (prior to the ART-13) apart to see what the original RCA's use unless it becomes important to the discussion.

    Also have several RCA 811A's with both sides individually into the base pin which is obvious when held up to the sun at the right angle. I have used those versions successfully as a pair on 6M as an experiment with the 811A version of the Dentron 10-160.

    The 572B is also an excellent 6M tube, USA or good Chinese as a pair with the basic Gonset neutralization Ive already described

    I'll look at the G-811, Penta, Taylor, and Shuguang labeled versions later.

    As many AL-811 owners have found out and comes as no surprise to those of us that have serviced them. Any instability was denied by MFJ until it became too much of an issue to hide the poor "engineering".
    There must be hundreds of those in the UK and thru out Europe and elsewhere.

  4. KM1H

    KM1H Ham Member QRZ Page

    More likely the young recent hires came right out of college just prior to the 1964 SB-200 introduction. Heath was not a high paying company and Id bet the 30L1 was carefully studied plus they had their own HA-10.

    Squegging was quite common in early grid driven Class C amps with tetrodes/pentodes and low mu triode tubes (up to 30 or so) that fed the DC bias voltage thru a high value choke in the 1 to 2.5 mH range and often some paper foil bypass cap. When the high mu 811A became popular for RF (the 812 was considered the RF tube back in 1939 and very little happened due to the coming war) the cure was to shunt with a low value resistor. The SB-200 used only 28uH and shunted that with 3300 Ohms. Another method was to use just a resistor and a suitable disc ceramic bypass cap which was a 50's development. Having any inductance in that grid path was looking for trouble IMO.

  5. KM1H

    KM1H Ham Member QRZ Page

    Many choosing a cheap amp also went for (or already owned) an equally cheap used hybrid transceiver from Drake, Heath, Kenwood and Yaesu using tube finals plus driver and a pi network. Those are still extremely popular over here at least. I use a TS-830 on the customer amp service bench as it is bullet proof from amp flatulence and the 6146B's it arrived with over 20 years ago are still fine; just the 12BY7A driver is often gone soft when pushing 100W on 10-12M for a few years!

    Im really enjoying your recent work and learning much in the process. Thanks.
  6. KM1H

    KM1H Ham Member QRZ Page

    The hundreds/thousands of 30L1's used in USAF MARS stations and elsewhere around the world in military and diplomatic service did not come with matched tubes and of course not for hams either. I have already stated that the 47 Ohm grid resistors were a somewhat half hearted attempt to force drive power sharing; however the two tubes per parasitic suppressor appears to be a part of another design committee.

    That amp was also hard on those RCA tubes in horizontal use as broken filament support springs were fairly common but far from endemic.....but that was only obvious to someone who actually serviced them since the 60's instead of just recent talking via a keyboard.
  7. KM1H

    KM1H Ham Member QRZ Page

    BYE again.
  8. G0HZU

    G0HZU QRZ Member

    Ok thanks for the info because that is the kind of problem (and 'fix') I was expecting.

    I guess you have seen this issue countless times already :) To help others understand this issue I did modify an earlier Simetrix simulation to add the HT choke and a series 1nF blocking cap. To add a bit of impact I opted for a middling value of HT choke at about 200uH.
    The simulation below measures the voltage at the amp output (anode equiv?) and the L match made up of L4 and C4 will transform the tank circuit to look like a high impedance at LF and this will happen in a narrow bandwidth. So a new unwelcome 'resonator' just joined the party and this happens at just under 400kHz in the plot below. With larger HT choke inductances this would happen at lower frequencies. So this creates another 'sleeper' resonance that has the potential to cause problems with squegging. I'd expect to see some form of snubber resistor as you describe and sometimes this can be placed somewhere that might not be obvious.

    The plot below does look scary at LF. The middle response at 7MHz is the wanted response on the 40m band. The parasitic suppressor has damped the VHF resonance above it but the circuit in the simulation below has virtually nothing in it to combat the LF resonance at 400kHz. This is definitely going to be a risk for instability in the system!

    Attached Files:

  9. G0HZU

    G0HZU QRZ Member

    Glad it's useful stuff. I have learned a few things myself from this thread and also from studying the Collins 30L1 and K6SNO designs.

    For example, when stacking 2 or 4 tubes in parallel like this the impact on the system stability (and the suppressor design) may appear counter intuitive at first. I now believe that these basic suppressor/stability design rules needs to be understood at the engineering level because otherwise you can end up with something like the AL811 design where K collapses way below 1 within the HF bands. By contrast, the Collins 30L1 and the K6SNO designs were surprisingly stable when modelled at the system level. The elegance of their designs is hidden in the component values they chose for the suppressors and the grid circuits and also in the neutralisation circuit in the case of the K6SNO amplifier. I really didn't expect to see the K factor behave so well over such a wide bandwidth with these amplifiers once I started stacking 4 tube models in parallel in the system simulator. The excellent performance had me scratching my head for a while because I expected 4 tubes in parallel to be a whole heap of trouble! I think I've worked out a few basic design rules now and this has helped me understand a bit about how they designed these amplifiers.
  10. SM0AOM

    SM0AOM Ham Member QRZ Page

    A very vicious form of parasitic oscillation that manifested itself as a "squegging" was encountered in the push-pull 45oTL triode final stage in the Wilcox 96D 500 kHz transmitters that were in my custody in the early 80s.


    They used a "dynatron rectifier" that was supposed to add positive resistance in parallel with the negative resistance that may occur during parts of the drive voltage cycle.

    However, this only worked if the 35T triode used as diode had its filament lighted. One of the transmitters got a broken filament, which caused an extremely violent oscillation, causing flash-overs in both the plate and grid circuits.

    As the transmitter site was un-manned the first occurrence went along without witnesses.

    After replacing some fuses and chokes, but having overlooked the 35T, my colleagues reapplied power and were met with a blinding flash and a loud bang when the amplifier "took off". After reducing the plate voltage, it was found that the negative resistance together with the grid supply and the modulation transformer permitted a blocking oscillator on a few kHz, which caused very high voltage transients.

    After some head-scratching, a faded note in the manual was found that said: "If instability is found, first check the 35T"...

    KD2ACO likes this.

Share This Page