4-1000A Grounded Grid Amp for AM

Discussion in 'Technical' started by W5HRO, Jul 28, 2015.

  1. W5HRO

    W5HRO Administrator

    Below is a preliminary drawing for a 4-1000A grounded grid amp using two tubes also with the Eimac specs.. My plan is to drive it on AM with a screen modulated 4-65A driver outputting around 65W to 85W max into it. At +3000V it will increase the needed drive power which is good and also lower the input Z closer to 50-ohms. My goal is to get around 750W of AM carrier output from the 4-1000A's. I'm hoping the control grids and the screens can just remain grounded to where all I have to do is adjust the cathode current and regulate it when static.

    4-1000A_Amp_1.png

    Another thing I'm going to do is connect the filaments in series and double the 7.5V to keep the current on both tubes identical and equalized. The other benefit of doing that is it will greatly reduce the current needed to heat both filaments. It's possible it could alter in input Z just slightly, but I don't forsee any real issue.
    .
     

    Attached Files:

  2. WD5JKO

    WD5JKO Member

    This is a rather lofty goal. This rig will make some soup for sure. Your neighbors will love you!

    I once use a pair of 4-65's in a Globe Champion to replace a bad pair of AX9909's. Since the AX9909's had a 12V filament, I decided to put the 4-65 filaments in series to use the 12VAC filament source. The tubes worked really well, loading up good 160m to 10m. But there was one big problem, and that was 60 hz cathode modulation. With directly heated tubes, in order to prevent the 60 hz cathode modulation, the filament supply must be balanced.

    I had a similar issue with my Gonset GSB-201. On AM, that series of Gonset amplifiers ALL had 60 hz hum on the carrier. Gonset only grounded one side of the filament transformer. They also used two filament chokes instead of one choke bifiliar wound. I had heard that two chokes instead of one as described can cause other issues. So now my GSB-201 has bifiliar wound choke, and a filament center tap made by two 10 ohm 10 watt resistors. The hum is completely gone now.

    Back to the topic of this post. I made these comments to illustrate prior mistakes by me, Gonset, and others in the hope that you won't have to fight the same sort of issues. I would look for an 8V 50 amp transformer, a big cathode choke, and a variac to feed the transformer.

    Jim
    Wd5JKO
     
  3. W5HRO

    W5HRO Administrator

    But wouldn't the same thing happen just using one directly heated cathode tube instead of two if one side of the filament transformer was still off compared to the other side in reference to the grounded center tap? My plan was to find a 15VAC transformer with a center tap. Wouldn't all of the caps like the chopper caps shown in my circuit also help prevent that?
     
  4. WD5JKO

    WD5JKO Member

    Brian, Your example differs from the two examples I gave. Cycle to cycle each 4-1000 filament alternates polarity properly, and should be hum free. If one tube hogs more voltage, then your likely going to have some hum. Building it for half voltage at twice the current with an 8V CT transformer will work. I am not sure about your arrangement. That 304TL G-G amplifier in the 1951 Radio Handbook had a unique way to resonate the cathode choke on multiple bands. Just an idea.

    Good Luck.

    Jim
    Wd5JKO
     
  5. W5HRO

    W5HRO Administrator

    Oh, you are saying that if one tube filament has slightly more resistance than the other meaning a slightly higher voltage drop that it might create an imbalance.

    Maybe…. I guess it's something to consider, but I would think the difference would need to be fairly large for that to happen. My thought was is it would keep the current on both of them identical which it would. I'll have to measure the resistance of each filament and see what they look like. I have four tubes coming, two of which should be pretty well matched. The guy said they were probably manufactured in the 1980's and they were both still new in the boxes. My 4-65A's are also NOS, but 1955 vintage and both have consecutive numbers, meaning one came off of the line right after the other.

    My plan was to try and keep everything lightweight as simple as I can and as efficient as I can like using a smaller 20A transformer instead of a 40A one with a lower voltage. I think I've pretty much simplified the 4-65A driver and it won't even require any interstage or modulation transformers. The 4-1000A amp will be really easy to build with just the one filament transformer underneath. Everything else I can get quickly or just make like the pi-network coil, etc. The only thing that might take me a while is locating a 3000V (or 6000VCT) 1.5A plate transformer. Everything else like the current meters I can still get new in the box because Simpson still makes the round analog meters. There are always a ton of them for sale on eBay too.

    If worse comes to worse I can always get Hammond to make me up one of the old Peter Dahl 3000V plate transformers. Since its GG a capacitor input supply with two oils filled caps would work best so I don't need any swinging choke. Maybe not even use a choke and just make it all capacitor or just use a high wattage resistor instead of a choke.
     
  6. W5HRO

    W5HRO Administrator

    I did receive my first two 4-1000A's. Brand new and have never been used. Big bottles! They make 4-400A's look like tiny midgets :mrgreen:

    I will measure the filament resistance of each tube when I get home tonight and see how close they are.

    NOS_4-1000As.jpg
     
  7. W5HRO

    W5HRO Administrator

    I've been digging around tonight trying to find my old Drake low pass filters and I can't find them. I know they are here because I unpacked them back in 2009. They are probably out in the garage somewhere in a box at the bottom of one of the stacks that I decided to keep.

    Anyway, I searched around the web and found photos below of the ones I have and the copper one, the TV-1000-LP was still new in the box and never used and it's the older one with the black ink stamp and not the yellow sticker. The copper was still shinny. The TV-3300-LP filter I used for years and it has the yellow sticker. Those were the best low pass filters anyone ever made and the TV-3300 has the sharpest attenuation out of the two.

    My thought was to mount one of them under the amplifier's chassis and connect it to the input. I don't think the amp will need any additional input tuning since the 4-65A driver will have a pi-network. I think I can just coax connect it directly to the amp's input via an SO-239 jack on the back of the amp. I think it would be a good idea though to place one of these filters between the SO-239 jack and the 4-1000A filament just to attenuate any possible harmonics.

    TV-3300-LP
    TV-3300-LP.jpg

    TV-1000-LP
    TV-1000-LP.jpg
     
  8. W5HRO

    W5HRO Administrator

    I picked up a TV-3300-LP filter cheap on eBay and had it delivered to my mom and dads house in Tulsa because we will be making a quick trip there this weekend. My dad said it was just delivered. If I'd known the guy was going to ship it 2nd day priority mail I would have just had it sent to my house instead.

    Anyway, I still cannot find my old TV-3300-LP or TV-1000-LP filters, but I know they are at home somewhere. They will turn up sooner or later, but I can always use two of the TV-3300 models because they have the best attenuation out of any LP filter ever made. The TV-1000 copper model was designed to also be tuned to work up to the bottom edge of 6-meters so that's why the attenuation is not as sharp as the TV-3300.

    Anyway, I will mount the below filter under the chassis and connect it directly to the 4-1000A's filament input connection.

    TV-3300-LP_01.jpg
    TV-3300-LP_02.jpg

    Update: 9/27/15

    I was packing away the 4-1000A's yesterday until I can get around to this project in another year or so and I found my Drake filters in the box with my spare 4-400CG and 810 tubes. Don't know why the heck I had put them in there :icon_wtf:

    drake_filters-jpg.2852.jpg
    .
     
  9. W1VTP

    W1VTP Member

    Are you gonna put a protection RF choke across the output to the ant... in case the plate DC block caps fail?

    Al
     
  10. W5HRO

    W5HRO Administrator

    It's probably not a bad idea. I have 40kV doorknobs that will probably never fail, but... :icon_silent:
     
  11. W1VTP

    W1VTP Member

    Crap(outs) Happen :biggrin:
     
  12. Gary Crandell

    Gary Crandell Member

    I've got a 3kv @1a xformer out of a Harris amp I am not using. Probably won't as I don't have the time, especially now that I've got the 21e. 100 lbs or so, heavy thing.
     
  13. W5HRO

    W5HRO Administrator

    I sent you a PM so if you would like to part with that transformer let me know.
     
  14. W5HRO

    W5HRO Administrator

    Well thanks to Gary, N6OYS I now have the plate transformer for this project :icon_thumbup:

    To give you an idea of the size of the transformer that is my other NC-303 receiver behind it covered up with the towel.

    transformer.jpg
    diagram.jpg

    Now I can start thinking about finding a roll-around cabinet for the supply. Maybe something like a tool cabinet base to where I can set the screen modulated driver and the actual 4-1000A amp on top. Sort of like a 3-stack system. Power supply on the bottom on wheels, then the driver on top of that, then the amp next. Now that I have the main hurdle done which was finding the plate transformer everything should be really easy from this point on. The only other issue is finding the filament transformer.
    .
     
  15. W5HRO

    W5HRO Administrator

    I was thinking about the HV supply yesterday and I did run some simulations last week and I am debating on whether to find several oil fill caps with 2 or 3 in series parallel or to just buy a ton of 200uF 500V electrolytics and build a big bank of them. The problem is oil-fills don't really have the capacitance I need because I won't be using a choke. I will also be using a bridge, which I will probably need to build myself using a ton of 1N4007’s in series/parallel with parallel resistors.

    Anyway, below are two different circuits and I’ve determined I need at least 80uF total to make it clean enough under a 1A load. I can get big high wattage Ohmite power resistors to equal about 10 to 20-ohms total in parallel, which should stay nice and cool fairly cheap, but it’s going to require a ton of 200uF 500V caps to make it clean. 100uF works best, but 80uF also works good enough in the simulations. I’m thinking I need to make the capacitor bank handle around 5kV to be safe and that will require 40 caps total with one parallel voltage divider resistor string across them. There is no startup overshoot at 1A or at 100mA because no choke exists in the circuit. I think the second circuit below is best and the resistor should protect the bridge and the caps both in that arrangement. My plan is to turn the primary voltage to the transformer on and off with PTT so I want it to come up super fast with no delay.

    Here is the capacitor input circuit at 1A.
    Supply_40uF.png
    Supply_40uF_Result.png

    Then here is the choke input circuit at 1A, but with no choke.
    Supply_80uF.png
    Supply_80uF_Result.png

    Also, the result outputs are showing a total span of only 1 second in 100mS increments so the tiny bit of ripple shown from the diodes switching should be insignificant I think.
    .
     
  16. W5HRO

    W5HRO Administrator

    I did some searching around and I did find that I can get 10kV and 20kV bridge rectifiers at 10A and higher online so it looks like I won’t need to use a ton of 1N4007's. Or, there are some really super high voltage and high current diodes to where I could use 4 to make one.

    Anyway, I actually think my transformer is a bit more than 1A. The diagram says 3300V at 50-ohms and when I put that into the simulator with a 1A load it yields exactly 3kV. However, when I measure the actual secondary winding it measures 30-ohms (actually 29-ohms).When I put 30-ohms into the simulator instead I can still get about 3kV with a 1.4A load. Not only that, but the two primary windings are going to be connected in series so I can run it off of 220V instead and there will be two additional taps for the connections. Instead of using the two far outside taps I can pick the next two inward or even the next two further inward. That should bring the secondary voltage up some. I’m going to play around with the transformer by inputting a sinewave and see what the ratios are. I might be able to get around 3500V out of it or more under minimal load conditions.

    At 3000Vdc the two 4-1000A’s will draw a maximum of 1.4A during max modulation peaks, but chances are I will never run it that high anyway. Even so, I still think the output will remain at 3kV during those intermittent voice peaks if it ever even reaches them. RMS (mean/avg) power is the only real thing that matters anyway and that should not be an issue.

    I also found 200uF 500V electrolytic caps too for the filter bank. I don't even think I need a 10 to 20-ohm series resistor in the circuit now. I think I can just use the capacitors all by themselves using 10 resistors for the parallel voltage divider across them just for protection and a very light slow bleed.

    .
     
  17. Gary Crandell

    Gary Crandell Member

    Th
    That transformer was out of a Harris commercial transmitter so the 1amp is probably a consevasive rating.
    N6OYS
     
  18. W5HRO

    W5HRO Administrator

    I may decide to just use one 4-1000A instead of two before it's over anyway. It would make the screen modulated driver for it a lot easier doing so.

    With the primary windings in series and connected to 220V on the inside taps I can probably get a bit more than +3000V which would bring the peak power up to maybe around 1600W using only one tube. I'm planning on AM screen scream modulating the thing with high 150% positive peaks and designing it to work on ALL of the bands from 160-meters up to 10-meters.

    I might be able to build the driver and the linear both on one chassis if I use only one 4-1000A. Then the power supply for everything underneath in its own cabinet/box. Chances are with no audio or modulation iron needed it would sound better and be much louder than my HB high-level plate modulated transmitter that I'm working on now.

    LINEAR AM SCREAM..!!! MODULATION :mrgreen:

    specs.png
    .
     
  19. W5HRO

    W5HRO Administrator

    By they way it totally dawned on me today that I had forgotten about an old trick you can do with transformers to double their output voltage. It also isolates the supply from ground a bit which is better for RF decks and safety anyway.

    The below circuit will give me +6000V instead of only +3000V and it's the same thing with a 1A load on it. At +5000V a single 4-1000A will only draw 540mA max and is good for around 1900W out. The current will even be a little less above +5000V.

    I think this is what I will do using only one 4-1000A instead. It also solves other issues too like the required drive power, etc. It will take a lot less.

    double.png

    double_result.png
     
  20. W5HRO

    W5HRO Administrator

    Some additional work on the simulator revealed that it will be better (cleaner) and more linear if I use some big Ohmite power resistor (maybe 2) in parallel and make the supply a real capacitor input supply like the circuit below. Under minimal load it will still exceed 6000V so I think what I can do is just get a big variac and mount it on the back of the bottom supply cabinet and just dial down the output for down for 6000V under minimal load to where it drops down to around 5000V under maximum load. The good thing is it greatly reduces the number of capacitors I will need in the bank for an 8000V rating down to 64 capacitors total. I will only need 25uF on each side if I use the resistor setup in the middle. The supply will only be on during transmit anyway so everything should stay nice and cool.

    The below supply using only one 4-1000A should easily be capable of 2000 watts peak output power.

    Updated supply circuit
    supply_5.png

    Results with 500mA load (Still at 6000V)
    supply_5_results.png
    .
     
  21. WD5JKO

    WD5JKO Member

    Going back to one 4-1 eliminates the series filament sharing problem powered from a 15 vac transformer. Do you have a 7.5VCT filament transformer?

    Jim
    Wd5JKO
     
  22. W5HRO

    W5HRO Administrator

    No, not yet. I'm more concerned about getting the 4-400CG rig going right now. I'm just doing some planning on this project in my spare time like figuring out the supply design and I think I pretty much have it now. The only other thing is maybe trying to find some 5kV or higher oil-fill caps to put in series now that I don't need as much "C". Even only 16uF to 20uF on each side will work, but I prefer at least 20uF minimum. It would be a lot easier only needing 4 or 8 oil-fill caps instead of 64 200uF 500V caps. A lot less work.

    I did look around at the hardware store today while I was there trying to see if they had a bottom roll around tool cabinet that has the doors in front instead of drawers, but they didn't. I can probably find one online though and there are other stores around here that might have one. I could do the interlock thing on the doors to kill the HV when opened.
    .
     
  23. W5HRO

    W5HRO Administrator

    Ok, here is an updated drawing using only one tube. I'm sure it will change a little more before its over, but the below should work. I'll probably use less coil and more capacitance before its over too because 60uH is a big size coil. I'm going to use a vacuum cap anyway which has a low minimum "C" for 10-meters so the coil can be smaller.

    one_tube.png

    I did search around online a bit yesterday and it’s hard to find old-filled caps that are large enough in value that would go in series so I think what I can do it just use the 200uF 500V caps on the transformer side of the resistor, then use one big 20uF or slightly larger oil-filled cap on the output side of the resistor. It's only going to be an issue on the input side to make the full-wave bridge voltage doubler using the two diodes and the caps anyway. The output side won't matter. At least that way it reduces the number of caps needed by half + 1 :icon_thumbup:

    .. supply_7.png
    .
     
  24. K4TQF

    K4TQF Member

  25. W1VTP

    W1VTP Member

    Such a deal!! :mrgreen: