The Class BC Linear

Discussion in 'Technical' started by WD5JKO, Apr 13, 2014.

  1. WD5JKO

    WD5JKO Member

    When using an amplifier for an AM signal, we have always been taught that the amplifier needs to be linear, and if it is not, that distortion products will run up and down the band causing interference to adjacent stations, or even causing out of band emissions.

    Some might remember the CB linears where most have a mode switch. The labeling on the mode switch is usually AM or SSB. With SSB, they bias up the transistors to have increased idle current. With AM, there is no bias at all...this is real close to class C since we are biased at cutoff. I have converted a number of these CB amplifiers to 10m operation over the last 25 years. They all need an attenuator and / or some RF feedback to reduce the gain such that a 3 watt carrier makes about 20 watts RF. Switching the mode switch to SSB is necessary for that mode, but with AM, all it seems to do is increase the gain a bit. Running AM linear at the edge between class B and class C works, does not splatter, and it increases the efficiency of AM linear operation somewhat.

    So today I wanted to put this to the test using my Gonset GSB-201 amplifier. This amplifier runs a quad of 572B tubes at ~ 1500V plate. I recently changed the "too-small" filament chokes for a single bifiliar wound choke. This got rid of the high input SWR problem on 80M, similar to the issue on the Heath SB-200. I also use a small MFJ tuner on the input as this keeps the SWR on the exciter at 1:1 on all bands. This amplifier can do 140 watts AM with headroom for 130% positive peaks. That comes out to around 725 watts PEP. The following test was done on 20M and a RF output of 100 watts.

    What makes this easier to do is the "Kill A Watt" that I have. I plug the Gonset into it, and it can read volts, amps, KVA, and watts. Watts are volts X amps X the cosine of the angle between V&A. If the angle is zero degrees, Cosine of 0 is 1, and therefore VA = W.

    So with the "Kill A Watt", I read 242 watts just to turn on the filaments, and the high voltage. Figure half the total is filament power, and the other half is HV power. This Gonset uses a full wave bridge rectifier with a negative lead choke, and a big bleeder resistor.

    The exciter was a Flex 3000, and was modulated at about 85% in all cases. One consequence of running the "linear" at almost zero idle current is the modulation percentage at the output is greater then at the input. In this case the output seems close to 95% downward modulation when the input was close to 85%. This is a form of distortion, sure...but consider what you gain if the exciter can be backed off to peaking 90% instead of 100%. In a plate modulated exciter, going to 100% with the occasional peak "white-lining" the scope display indeed does cause splatter. Backing off the driver audio, and still getting 100% modulation after the "linear" is a good thing.

    Jim
    Wd5JKO

    Note: I was using two scopes with slightly different sweep rates. In all cases the unmodulated carrier was 4 cm in height.

    Kill A Watt.jpg
    Matrix.jpg
    Zero_PA_Bias_Exciter_Output_12W.jpg
    12V_PA_Bias_Exciter_17W.jpg
    Zero_PA_Bias_PA_Output_100W.jpg
    12V_PA_Bias_PA_Output_100W.jpg
     
  2. WD5JKO

    WD5JKO Member

    Here is a textbook writeup about class BC AM from the 40 year old edition of Electronic Communication by Robert Shrader. Here he describes two forms of "linear" AM amplifiers, one that is somewhere ranging from class A to class B, and the other that is biased at 1.5 times the output tubes cutoff grid bias voltage. Here the carrier is amplified class C, and the audio begins at the edge of class B, hence class BC. In this case, a driver at 70% modulation becomes 100% modulated after the "linear" amplifier.

    It should be noted that it is important to increase the carrier drive level AFTER moving the final stage bias towards cutoff, or beyond. Boost the drive until you get about the same carrier output as before. If this is not done, then the audio will start getting cut off, and will begin to sound like SSB going through an amplifier with no idle current. Using a scope here should not be optional.

    Jim
    Wd5JKO

    Class BC Explained1.jpg
    Class_BC_Explained2.jpg
     
  3. KM1H

    KM1H Guest

    This appears to be a continuation of the olde forum same subject Jim.

    Back then I stated Ive been running some commercial ham amps with increased bias on AM for several years and have been pleased with the results.

    I havent done any efficiency tests but the old fashioned "hand on the power transformer" plus eyeball on the anodes of glass tubes tells me it works well, PEP is way up compared to as built. For instance a Clipperton L with new direct from China 572B's barely show color on voice peaks at a 300-350W carrier and long evenings of 160-75-40M ragchewing showed no excessive transformer heat.

    The one other change I made was replacing the noisy fan with a new EBM Papst 4600n and removing the series resistor. With the cover on air noise is not excessive. In order to make the bias switching easier I used the later Clipperton QRO circuit which is a TIP-42 transistor, resistor and 1W zener, modified to switch from 9.1V to 18V via a small toggle switch added on the back panel. This saves the high cost of an added 50W zener. Inital experimenting was done with a string of 1N5408's and a rotary switch and I ran out of switch positions to go deeper ::)

    Since twice cutoff for a 572B in Class C CW is about -70V I figured that -18V would put me sufficiently deep in Class B without triggering excessive splatter. In fact the spectrum analyzer looked real good at -9.1 and -18V and others can go deeper and probably fine tune the bias.

    I started this in the mid 80's with a NCL-2000 which had a small anode cooler for 400W dissipation but a good blower and at 300W carrier I could yak for hours: I still use that amp at times with one of the vintage low power exciters. I got the idea from a BCB Chief Engineer non ham friend who used to keep me in 4-1000A's when I was building amps for others for SSB/CW.

    Since AM skyrocketed over the recent decade Ive modified numerous customer amps of all types from a 30L1 to an Alpha 77SX, Commander 2500E, LK-800, etc. My own regular AM linear is a LK-500ZC driven by a TS-950SD with the audio opened up to 3300Hz and a D-104 with impedance converter.

    Carl
     
  4. W5HRO

    W5HRO Administrator

    Monkey Swing !!! :icon_shh:

    monkey_swing.gif
     
  5. WD5JKO

    WD5JKO Member

    No more then KI5SG did from Tulsa! ;)

    I had a long talk Monday night on 20M AM with Tim WA1HLR. He like Carl also has experimented with Class BC AM amplifier. Tim was working on a modified L4B with dual 3-500's, and driving it from a highly modified Ranger. As Tim explained the "biased-up' versus not "biased-up" L4B, I was writing down the numbers he was stating. In his case, at low tap on the L4B, he was doing 125 watts out.

    Stock Bias: 12 watts drive with 30% efficiency
    Biased-Up: 17 watts drive with 35% efficiency

    Just like my situation, the added bias lowered the gain, and raised the efficiency. Both Tim and Carl made mention of excellent positive peaks. A 5% rise in efficiency might not seem like a lot, but in my case that was accompanied by about a 20% drop in standing carrier plate dissipation. This is huge.

    Jim
    Wd5JKO
     
  6. W5HRO

    W5HRO Administrator

    It's a joke among AM'ers because its what CB'er say and do. When you change the bias like that is creates a little thing called "monkey swing" as it becomes more linear. I thought most of us knew that already :lol:
    .
     
  7. KM1H

    KM1H Guest

    I used Tim as my "sounding board" with several amps as he is around often and not hard to find on some band ::) His ears are often better than my spectrum analyzer for the nuances. The last amp I did with him was either an Alpha 77SX 2 holer or a Commander 2500 Magnum which sort of got his attention :lol:

    Also remember that while the carrier efficiency is improved using what I call Deep B, so is the 4X PEP which can reach SSB efficiency which further cools down the tumes and transformer. Unless a peak reading AC line monitor is used the RMS only will be indicated.

    Carl
     
  8. W5HRO

    W5HRO Administrator

    What if you were to do something slightly different and come up with an auto control-grid bias circuit that would automatically vary (shift) the bias level around as the modulated peak-to-peak drive level into the amplifier varied instead?
     
  9. KM1H

    KM1H Guest

    The first time I saw a sliding bias shift article was using a 304TL as a single ended modulator. Bias varied from Class A to B with voice.
    This was in a late 40's to early 50's CQ.

    Others included various small tetrodes/pentodes to switch the same and do away with with a lot of wasted idle current.

    Never tried it and mentioned it on that other site to glassy eyed results several years ago. I still have all the old CQ's.
     
  10. W5HRO

    W5HRO Administrator

    You could probably just use the center-tap of the Linear's cathode filament transformer(s) and connect it to a control circuit, tube or solid state and use that to auto shift the bias. You could potentially gain a significant pep power increase on AM doing so :eek:

    P.S. I don't mean control the cathode current because some of the big power triodes run off huge amps, but you could install a series sense resistor of some small value between the CT and ground and use the voltage across that resistor to control a circuit that varies the control grid away from 0 volts.
     
  11. WD5JKO

    WD5JKO Member

    I am attaching a bias shift idea as applied to a SE audio output tube stage. This concept could apply to a RF linear amplifier where the transistor stage bias circuit could be applied to the GG amplifier filament CT to ground. This concept would need to be tweaked for up to 1A current, but the idea would be to run the amplifier at the edge of cutoff without modulation, and at AB at full modulation. As the bias slides, the RF stage gain will increase causing some positive carrier shift. Should swing the monkey. :icon_eh:

    Jim
    Wd5JKO

    bias-shift.jpg
     
  12. W5HRO

    W5HRO Administrator

    That would probably work with smaller tubes like sweep tubes, but the big power triodes such as 3-500Z's with directly heated cathodes require 5V @ 14.2A filament transformers each :eek: Not sure you would want to place a transistor inline with the center-taps to ground. The tubes only draw about 400mA each, but it would still require a really hefty circuit.

    What I was thinking was just control the grid bias a little like with CB class BC grounded grid linears that are not really full grounded grid linears by using the voltage coming from a small resistor value (meter shunt value) in series with the CT to ground to control the grid bias via a control circuit. Just move the grid away from 0 volts a little.
     
  13. WD5JKO

    WD5JKO Member

    The amplifier used for testing in this thread has had a steady decline in power capability. It used a quad of Cetron 572B's. I had been waiting for the new 572B's from RF parts, but the availability could be months or years away. Looking at my options, I decided to go back to 811A's. There are more options with these tubes. I chose the cheapest type I could find. These are from the Alpha Amplifiers company. A quad of boxed, and matched 811A's sells for $71.00 + $15 shipping Fed-X.

    http://www.rkrdesignsllc.com/products/transmitting-tubes/

    With these tubes installed I can easily do 150 watts AM with headroom for 100% + modulation. I am using 12V zener cathode bias, which puts the tubes right at cutoff with no RF drive. At this power there is no visible plate dissipation. The extra efficiency of class BC makes the difference.

    I have seen some warnings on different forums. The skeptics say the tubes lose emission fast. I will report back to let everyone know. One thing for sure, my 572B's that I pulled out were as flat as a Kindergarten straw.

    Jim
    Wd5JKO
     
  14. W5HRO

    W5HRO Administrator

    I think I heard something about 572B's becoming scarce, but I really didn't pay any attention to it. Is that true? There are some on eBay being sold as NOS matched pairs for around $140, but who knows if they are any good or not.

    811A's usually don't last very long when used in RF applications, especially when used in overstressed linear RF amps. You have to keep the plate voltage down really low otherwise they won't last. About 1250V max. Most amps I've seen 811A's in run them at 1500V or higher and that won't work for very long. I'm not even sure BC mode will even help all that much. If you could find some of the older NOS USA made 811A's with the U channels they might hold up better, but most of the modern ones are made overseas and are not like the old ones. The newer ones are very cheaply made.
     
  15. WD5JKO

    WD5JKO Member

    On the web there are references that the 572B's seen on Ebay are duds that did not meet the requirements set by Ameritron and other amplifier makers. Here is what RF Parts says about 572B availability, and more:

    https://www.rfparts.com/tubes/tubes-572b.html

    The 811A has been used in many Ham linear amplifiers over the years such as the Collins 30L1, and more recently several Ameritron models. In my case I have the Gonset GSB-201. Variations of this Gonset line have used both 811A's and 572B's. In my situation, Gonset used -4.5V bias when 811A's were used, and zero bias with 572B's. I am not sure of the Gonset B+ voltage, but I have heard it is around 1700v unloaded, and 1500v loaded. Remember that for class B AM modulator use, the Johnson Viking 500 used a pair of 811A's at 2KV plate voltage.

    Jim
    Wd5JKO
     
  16. W5HRO

    W5HRO Administrator

    When the 30L-1's came out they were using the older USA made 811A's. If I remember correctly though the plates or the grids are really bad about sagging in the 30L-1's because the tubes are mounted horizontal. I think it's worse with the modern made tubes. In the Viking 500 the tubes are in the modulator and not the RF, plus they also used USA made tubes back in the day.

    The Ameritron AL-811 amps are a real POS. The plate voltage runs way too high and will fry the modern tubes in a heartbeat. I went thru two sets in mine operating on SSB until I finally sold it and got the AL-82 instead. That was back when it was still Ameritron and not Mighty Fine Junk like today.

    If you could find some NOS USA made 811A's like the 30L-1's or the other gear used back in the day then they could probably handle the torture a lot better. I'd try to figure out a way to back your B+ down in the Gonset to only 1250V loaded.

    Also, if you could find a NOS set of 812A's that would be even better. 812A's are a little scarce though and have been for a while, but they make better RF tubes than 811A's will. They would just require more bias voltage which is a good thing. You could utilize the -100V bias input connection in the back of the amp.
     
  17. WD5JKO

    WD5JKO Member

    I have a few 812's, but I reserve them as spares for my Central Electronics 600L amplifier. They use one 812 as a screen grid regulator to make 750V for an 813. I have quite a few 808's though with graphite plates. Now those tubes can really take the voltage, and are medium Mu. What I have learned though is that with grounded grid, the stage gain drops as the tubes Mu drops. The 304TL with low Mu might be only good for about 6 db gain, whereas the 811/572B can get 10-13 db gain depending on tuned input or not, and the plate voltage. A guess would be that the 808's might give 8 db of gain...

    If the Chinese 811A's crap out on me, I might try a quad of 808's. It will be a big conversion since the envelopes are bigger around, the grid connection is not in the socket, and the filament voltage is 7.5V instead of 6.3V. The reward though should be decent tube life and linearity. My best ever plate modulator was a quad of those with 2250V on the plates. Made a nice 600 watts of audio when driven by a pair of 6B4's.

    As a side note, I think a trio of 304TL's in G-G would be the perfect amplifier for a 100 watt class transmitter like a Viking II. With 6 DB gain (+/-), a nice 400 watts of AM should be possible. With forced air cooling and chimneys, the 300 watt plate dissipation rating should be quite a bit higher, much like doing the same with an 813 can cool 200 watts of plate dissipation.

    Jim
    Wd5JKO
     
  18. W5HRO

    W5HRO Administrator

    I think my old AL-811 amp was running the plates at 1800V loaded which was way too high. Problems start to happen when you reach and exceed the max 65W plate dissipation per tube at RF frequencies and that rating is ICAS with only 1500V. If you run them at only 1250V it will drop it down to just under 50W which will make the tubes last longer. Some people probably disagree, but it's true. In a linear, I think they would work best with the plate supply right at 1500V under minimum load then dropping down to 1250V under full load.
     
  19. W5HRO

    W5HRO Administrator

    Jim,

    I'm guessing you have already seen this, but I just ran across it.

    http://www.w8ji.com/811a_tube.htm

    "A secondary benefit to the addition of U-channel cooling fins was an increase in infrared emission area. This produced what seems like a meager 5 watt increase in ICAS dissipation, from 60 watts to 65 watts. That was a very conservative "published rating" of 811A anode dissipation. The actual working dissipation was much higher, actually 80 watts or more long term average power."

    "So while the import tubes have the same "book" long-term anode dissipation of 65 watts, they have greatly reduced short-term dissipation capacity. The vertical fold anodes are 65-watt anodes, but the short term dissipation or allowable overload time is reduced to nearly the value of the early non-finned 811 or 812 tubes."


    To summarize the old USA made 811A's could probably handle around 80W of ICAS plate dissipation even though the spec said 65W. The new ones are designed to handle only the 65W of ICAS plate dissipation based on the original spec.
     
  20. WD5JKO

    WD5JKO Member

    Brian,

    That was good info. I attach a photo of a vintage RCA 811A, and the Alpha 811A that I just purchased. Not sure what this means, but my GSB-201 is still set up for either zero bias or 12v bias. As per this thread on AM, I use 12V bias from a 50 watt 12V zener in the filament return path to ground. This 12V almost cuts off the 811A's as I only record about 20ma cathode current. If I switch to zero bias, the idle current comes up to about 150 ma. I am not positive about the anode volts at 150ma load, but if it were 1600V, then 0.15 X 1600 = 240 watts dissipation. Divide that by 4 and we have 60 watts per tube. Under this condition there is no visible plate dissipation.

    Ameritron wants $339.80 for a quad of 572B's...That is $84.95 apiece for a 572B versus $16.95 apiece for the Alpha 811A.
    http://www.ameritron.com/Product.php?productid=380-0572B-4M

    Jim
    Wd5JKO

    RCA_811A.jpg Alpha_811A.jpg
     
  21. W5HRO

    W5HRO Administrator

    Yeah, the pic of the old one you posted doesn't have the folds which is good, but you can't tell by the photo if it has the U channels or not. What does that tube look like from the other side? If it doesn't have it then it's probably the original 60W spec version which probably means around 75W.

    The pic of the new one you posted has the 65W folds for sure.

    P.S. I did check and the Collins 30L-1 ran the older tubes at 1600V loaded and the Ameritron AL-811 at 1800V loaded. Not sure how the modern tubes would hold up in the 30L-1, but the AL-811 runs them way too hot for certain. The AL-811 amp I had was when they first came out so it was the original model. I used it with my TS-440S and went through two sets of tubes. I think it came with Penta Lab tubes, but it's been so long ago that I cannot remember for certain.

    penta-labs.jpg
     
  22. W5HRO

    W5HRO Administrator

    I wonder what it would take to use a 4-1000A in class BC mode in grounded grid configuration? That would seem to be the ultimate tube to use in a linear. Both the screen and control grid would be at 0V or close to it. Not only that, but two tubes in parallel would have an input Z close to 50-ohms. At 3000V the input would be exactly 52-ohms. Just drive it with a Super Senior Transmitter dialed back some :mrgreen:

    For one tube...
    Plate Voltage = +3000V
    Screen Voltage = 0V
    Grid Voltage = 0V
    Drive Z = 104-Ohms
    Power Out = 1475W

    Or

    For two tubes...
    Plate Voltage = +3000V
    Screen Voltage = 0V
    Grid Voltage = 0V
    Drive Z = 52-Ohms
    Power Out = 2950W :mrgreen:

    What about Class BC mode with maybe 750W of carrier output power? Attached below are the complete GG specs.
     

    Attached Files:

  23. W5HRO

    W5HRO Administrator

    Below is a quick rough idea using two 4-1000A's in GG for 80-meters on up. I think it would also be best to run the filaments in series to greatly reduce the current required. There are a lot of new transformers around that will provide the needed 15VAC.

    The question is what voltage should be at the center tap of the transformer to place it in BC mode? This amp would basically be the equivalent to the old 750W out AM limit based on 75% efficiency and it would be really easy to make. It could even be driven with a 50-ohm transmitter. I've added the "L" network on the input to provide some simple input tuning to attenuate harmonics and to insure a perfect 50-ohm match.

    4-1000A_GG_Amp.png
     
  24. WD5JKO

    WD5JKO Member

    Something like this?

    PA_Stage2.jpg

    Jim
    Wd5JKO
     
  25. W5HRO

    W5HRO Administrator

    I can't tell what tubes those are. The one in front sort of looks like a 4-1000A, but then again it might be a 3-500Z or a 304TL. I just cannot tell by the pic.

    Anyway, I updated my drawing in the posts a bit. The only issue is that 4-1000A's have gone up in price since the last time I checked. Back in the 1990's I could have gotten them brand new in the box for only $100 each all day long. WTF happened?

    I still think it's a good idea though because it would be really easy to build the amp for just 80 and 40 meters. If it could be placed in BC mode then it could handle a lot more drive power. I'm not sure if those GG specs are for class AB1 or AB2, but for two tubes the max single tone drive would be 260W based on those specs at +3000V. If that drive level could be increased to say like 375W or 400W then around 100 watts of carrier input would be perfect. A Johnson Viking I or II could be used or one of those new Super Senior transmitters as long as the carrier output could be backed down to no more than 90W to 100W max.

    I can see it now, 750W CCS carrier out from a Viking II with astronomical modulation peaks. It would give new meaning to the term peak envelope power :lol: