1. You can now help support WorldwideDX when you shop on Amazon at no additional cost to you! Simply follow this Shop on Amazon link first and a portion of any purchase is sent to WorldwideDX to help with site costs.
    Dismiss Notice
Dismiss Notice
This forum does not allow a single user to have more than one username. If anyone wants to change their username contact an admin and it will be done. Multiple accounts belonging to the same member will be deleted without warning.

The next generation of linear amplifiers has arrived.

Discussion in 'Amplifiers' started by Shockwave, Sep 5, 2019.

  1. nomadradio

    nomadradio Analog Retentive

    Joined:
    Apr 3, 2005
    Messages:
    3,587
    Likes Received:
    4,044
    Cool!

    Just thinking out loud. An old friend had a spurious-trip problem with his 80-meter dipole. The breaker panel was in the adjacent room in his basement. Odd part was that the breaker that was tripping was not the one that served his radio room. Might not have been the ground-fault side of the breaker. Could have been the arc-fault sensing that was causing the nuisance tripping. He changed out the breaker and the problem went away.

    73


     

  2. Shockwave

    Shockwave Sr. Member

    Joined:
    Sep 19, 2009
    Messages:
    3,198
    Likes Received:
    1,949
    Older ones were much more susceptible to nuisance tripping. The longer the wire is after the GFIC, also contributes to false triggering. Mounting at the equipment end solves that issue. Incidentally, one of the first ones I tried was a GFIC cut off the line cord of an old Goldstar air conditioner.

    When it wouldn't trip with a 1 amp load across the line and ground wire (not neutral), I had to crack open the Chinese junk and see why. What a piece of deception that was inside. The entire sensing circuit was absent and replaced with a single transistor connected to the reset button in order to fake the test!
     
    Shadetree Mechanic likes this.
  3. nomadradio

    nomadradio Analog Retentive

    Joined:
    Apr 3, 2005
    Messages:
    3,587
    Likes Received:
    4,044
    Keeping it inside the amplifier cabinet at least allows you to control filtering and bypassing to/from the thing.

    And yeah, it's not "listening" to the safety-ground conductor, only to the hot and neutral for any imbalance.

    Haven't seen a fake one yet. Just a matter of time, looks like.

    Some days it's hard to resist being the devil's advocate.

    73
     
  4. Shockwave

    Shockwave Sr. Member

    Joined:
    Sep 19, 2009
    Messages:
    3,198
    Likes Received:
    1,949
    I had to go back and look at that Goldstar AC unit with the fake GFIC to see if it carried a UL approval sticker. It does! Hopefully the GFIC was "modified" after their inspection...
     
    Shadetree Mechanic and CB590 like this.
  5. Unit 148

    Unit 148 Unit 148 Mobile CYPRUS

    Joined:
    Sep 29, 2010
    Messages:
    66
    Likes Received:
    18
    I read earlier in this thread about holy grail power supplies and efficiency numbers. We use two of these extremely compact and highly efficient in our commercial 1kW FM broadcast transmitter (FCC ID'd but produced in the UK). Have enclosed the pdf. They are hot-swap as well so you're not off-air if one shuts down. Here's the transmitter video: Shockwave, I've designed a 'mobile' chassis for you when you're ready to produce that all-band version of your amplifier + remote head :) Unit 148 mobile - G4ZOW
     

    Attached Files:

    Shockwave and Shadetree Mechanic like this.
  6. Shockwave

    Shockwave Sr. Member

    Joined:
    Sep 19, 2009
    Messages:
    3,198
    Likes Received:
    1,949
    There are some details about the progress of this project I'm now able to share. In other posts I've mention the fact that two Push-Pull (PP) 2SC2879's are unable to produce anymore more power output than one in a single ended (SE) configuration. I've learned the limitation here was NOT in the differences between PP or SE but was the result of the characteristics found in low voltage RF power transistors. High power 12 volt transistors are already operating close to their maximum rated current capacity. Adding a second in PP does not increase the headroom for input current. It does however, double the rated dissipation of the circuit and splits the workload in half between two devices.

    If the input current does not come close to the maximum rating of the transistor and reaches its max dissipation first, this is a prime condition for using PP and one can safely expect up to twice the output power depending on the specific transistor being used. Once you reach an operating voltage of 50 volts, most MOSFET's are drawing a low enough current that they can double output power capability when 2 are used in PP. Once testing revealed this, last minuet changes in design caused delays but took the new circuit many steps closer to complete broadband HF coverage with the full 1500 watt PEP legal limit available from a "two pill"!

    Another nice advantage to PP is that the gate impedance to both transistors are now in series with each other. Being able to double that impedance makes driving the PP circuit much easier on 10 meters. Designing a new broadband output network that provided a suitable impedance match over the entire HF spectrum, without noticeable loss and heating, was another matter. Once that was resolved, the next issue identified was the difference in gain from the lowest frequency to the top of 10 meters. Adjusting the drive level for the different bands was not an option. Gain must be held to 13db across the spectrum. Any fluctuation beyond +/-1db is unacceptable to the builder.

    Parts have arrived to build and test two different types of gain compensation networks. The results of those tests will determine if either one of these circuits alone is capable of leveling the gain or if varying degrees of both might be needed to achieve the requested specification. Unfortunately the first PP prototype was not laid out with these future gain compensation networks in mind. There is no room left on that PC board to continue to expand on the circuit so, we get to start all over again and the first one will end up being an expensive but worthwhile learning experience. While the deadline for this project is June, I hope to have the first legal limit, broadband unit completed with another video posted by the end of next month.
     
    NZ8N, SockIt2Me, Unit 75 and 10 others like this.
  7. Handy Andy

    Handy Andy Do Your Research First, Then Decide...

    Joined:
    Apr 23, 2018
    Messages:
    3,331
    Likes Received:
    3,687
    Nice to see this moving forward. And to also know you seem to have addressed a problem with matching network issues of the voltage versus current component in the power generation.

    Good job - and to see the test results confirm what I wanted to know in regards to the voltage component of these amps - I look forward to seeing these things in action.
     
    NZ8N, AudioShockwav and CB590 like this.
  8. Staybolt

    Staybolt Well-Known Member

    Joined:
    Mar 20, 2013
    Messages:
    633
    Likes Received:
    182
    Does it sound like dog crap on ssb?
     
  9. Mudfoot

    Mudfoot Elmer

    Joined:
    Jun 17, 2009
    Messages:
    10,712
    Likes Received:
    5,776
    No, it'll be a top notch amp. I'm waiting to try one.
     
  10. kopcicle

    kopcicle Sr. Member

    Joined:
    Feb 17, 2016
    Messages:
    1,888
    Likes Received:
    2,793
    The base-to-base impedance in a push-pull circuit would be four times the base-to-emitter impedance of one transistor.
    However, in Class AB, where the base emitter junction is forward biased and the conduction angle is increased, the impedance becomes closer to twice that of one device.
    Well... Okay so far.
    The MRF454 is specified for a power output of 80 W CW.
    Although the data sheet does not give broadband performance or IMD figures, typical distortion products are [– 31 to – 33 dB below one of the two test tones (7) with a 13.6 V supply. This device has the highest figure of merit(ratio of emitter periphery and base area), which correlates with the highest power gain.
    The maximum dissipation is 250 Watts, and the maximum continuous collector current is 20 A. The minimum recommended collector idling current is 100 mA. This can be exceeded at the expense of collector efficiency, or the device can be operated in Class A at an idling current of approximately one fourth the maximum specified collector current. This rule of thumb applies to most RF power transistors, although not specified for Class A operation can be operated in Class A.
    Specified 12.5 Volt, 30 MHz Characteristics —
    Output Power = 80 Watts
    Minimum Gain = 12 dB
    Efficiency = 50%
    The specified test jig is tuned input and output class "C"
    I have the AN-762 design board. It has been populated over the years with MRF-421, MRF-454, 2SC2290 (better linearity at the expense of some gain and efficiency) and the MRF-455. I have yet to try the SD1446 (ST-Micro)
    I recently acquired another for just $10!
    The MRF-454 would make every bit of 160W class "C" in PP. 140W with 110ma idling current in class "AB".
    The numbers just do not add up. No matter how you spin it 1+1=2 , less broadband impedance matching losses and bias considerations beyond class "C".
    Feel free to ignore the above. It's not even my words and research so it might as well be hearsay. Save for my experiments with the AN-762 reference design the preceding was lifted near word for word from the AN-762 application note authored by Helge O. Granberg.


    I woke in a hell of a mood and it hasn't gone away.
    Either that or I'm firmly rooted in theoretical and experimental data that dates back to the 70's and refuse to acknowledge hype, buzz words, and opinion.
    ~
    One accurate measurement is worth a thousand expert opinions.
     
  11. bob85

    bob85 Supporting Member

    Joined:
    Mar 30, 2005
    Messages:
    3,075
    Likes Received:
    880
    current is not just flow it is flow x time, 1 amp is equal to 1 coulomb per second

    pp means each transistor is operating for only 1/2 of the time, effectively cutting the workload or coulombs per second passing through each transistor,

    two transistors in pp can pull much more current than a single ended amp using the same transistor without either transistor pulling more current than the single ended amp,

    a popular 2 x sd1446 amp can pull 25amp flat balls out on FM,
    a same brand single ended 1446 amp will pull a little less than half that current for a little more than half the output power,

    a 10a fuse will pass 500a for a short time,
    the little gold wires that connect the transistor tabs to the die are just tiny parallel fuses you don't want to blow.
     
  12. Shockwave

    Shockwave Sr. Member

    Joined:
    Sep 19, 2009
    Messages:
    3,198
    Likes Received:
    1,949
    I promise, not even a hint of "dog crap" on SSB. Producing a clean signal was a major priority in this design. The class AB bias is double regulated and thermally compensated with the sensors mounted directly on the transistor die. Each output transistor has its own channel of independent adjustment on the bias board. The output of the amplifier also passes through a multi pole filter network to remove harmonics and exceed FCC requirements.
     
    NZ8N and Shadetree Mechanic like this.
  13. Shockwave

    Shockwave Sr. Member

    Joined:
    Sep 19, 2009
    Messages:
    3,198
    Likes Received:
    1,949
    One transistor in single ended is only operating with RF 50% of the time too. Just like in a tube amp, the output tank coil fills in the negative alternation where PP needs the second transistor through a 180 degree phase inversion to form the negative alternation.

    Most of my experience with 12 volt bipolar transistors is with the old 2SC2879 in amplifiers that were designed with impedances to maximize the output per transistor for competition. Most one pill drivers are not designed to maximize the available output from one transistor and still use broadband transformers.

    Perfecting that match using resonant tuned input and output circuits will allow the single 2879 to equal the output of a pair much closer than you might think. This entire scenario disappears once you use transistors that are not being operated close to the maximum current ratings and you can utilize the full double dissipation PP offers.

    There are big differences between average current and the peak current needed to produce PEP watts. I found we were running these transistors right around their maximum collector current in both SE and PP and that a 20 amp fuse could protect one 2879 just as effectively as it did two. Attempting to drive the PP pair past the output a single one could do while using a 25 amps fuse, would burn up one of the 2879's before the 25 amp fuse.

    I sure do hate when the "gold wire fuses" open :(
     
    Shadetree Mechanic likes this.
  14. Shockwave

    Shockwave Sr. Member

    Joined:
    Sep 19, 2009
    Messages:
    3,198
    Likes Received:
    1,949
    Incidentally, the two transistor Push-Pull 1500 watt amp has been fully assembled on a new heatsink with professional looking panels and cover. It almost made it through power testing but just when it passed 1500 watts, caps in the output section arced. That was surprising since they were calculated to easily handle the 1500 watt output. Further testing on the analyzer showed the output impedance is running too high and therefore, so is the RF voltage on these output caps. Pulling the output impedance down closer to 50 ohms will reduce the RF voltage and increase the output efficiency. If the caps still show any abnormal heating on the infrared camera, they will be replaced with doorknobs.

    Once the output impedance is correctly matched, this circuit should be able to produce about 2000 watts PEP without much strain or detectable distortion because that is half of the total output transistor dissipation rating. This will provide the headroom needed to give it a clean, conservative 1500 watt rating while setting the forward power protection channel to trigger somewhere around 1750 to 1800 watts PEP.

    The protection board has been revised for a third time. Some of the original parts had slowed the reaction time to about 15 microseconds. Their values have been changed and one was eliminated with a change of another part. The response time from fault detection to full transistor cutoff is now between 4 and 5 nanoseconds depending on the type and degree of fault detected. Initial testing of the RF board with the wrong values of feedback applied, caused a self sustaining oscillation to instantly occur when RF was applied. The logging current meter recorded a current that was more than twice the value of the fast acting fuse.

    Under most circumstances this would have destroyed the fuse in a fraction of a second, right after one of the output transistors failed. The much faster acting protection circuit saved both output transistors before the fast acting fuse element even had a chance to heat up. This all keeps those expensive "gold wire fuses" inside the output transistors from burning up. Next video still on schedule by the end of the month.
     
  15. bob85

    bob85 Supporting Member

    Joined:
    Mar 30, 2005
    Messages:
    3,075
    Likes Received:
    880
    se has the advantage of been able to optimise load impedance for whatever vcc & drive / bias, you intend using,

    with pp you are stuck with turns ratio of the transformers unless you add extra impedance matching with its associated losses .

    at some vcc & driive level they will be closer in output than others,
    you won't get double the output from pp vs a optimised for max output se stage at any vcc imho.

    i have seen negative bias deep class c keydown amps & multiple tuned se stages combined,

    a lot of people would be surprised what a single genuine 2879 can do when operated beyond datasheet limits & all that matters is power for a 10 second keydown,


    your latest project is much more interesting than 2879's.
     

Share This Page

  • About Us

    The WorldwideDX Radio Forum was originally established in 2001. We pride ourselves on welcoming Radio Hobby enthusiasts of all types, while offering unbiased, informative, and friendly discussion among the members. We are working every day to make sure our community is the best Radio Hobbyist's site.
  • Like us on Facebook

  • Premium VIP Member

    The management works very hard to make sure the community is running the best software, best designs, and all the other bells and whistles. Care to buy us a beer? We'd really appreciate it!

    Donate to us!