Pair of 813's modulating a pair of 813's

Discussion in 'Technical' started by Dwayne, Feb 9, 2015.

  1. Dwayne

    Dwayne Member

    I am getting together the parts and pieces for a pair of 813 modulating a pair of 813's. . a design from k1jj. . .

    Are there any hints or advice that may give me problems?

    One of the things I saw, it was using two filament transformers. . .well, I have ONE filament transformer. . .20 V Centertapped. . . I was hoping to run all four off this transformer. I did noticed he used a separate transformer for the modulation and separate one for the RF. Will I run into problems here?

    Another thing. . .

    I thought about getting a digital VFO off of Ebay, running it through something like a 12AU7 to amplify it.. .or some other tube to get the RF output of the VFO high enough to run the RF deck. . .Is there any opinion on this too?

    Thanks in advance.

    Dwayne Ka0aam
  2. W5HRO

    W5HRO Administrator

    In Tom's rig below he is using separate filament transformers primarily for metering.


    You could use one transformer for both the AF and RF filaments as long as you bypass the all the connections well with the right caps to keep the RF out of the AF side tubes. Generally it's not a good idea to use one transformer for both though.

    The problem you will run into using one transformer is how will you meter the AF modulator and RF currents? Most everyone meters the currents to set the modulator's static bias and RF plate current from the cathode side via the center taps of those filament transformers to ground. If you use one transformer then you will be forced the meter the currents from the HV plate line side instead. That's the tradeoff/drawback.

    You can use a solid state VFO and just run that into a driver tube. You might need to place a buffer tube or a HV NPN transistor in the middle as a buffer, but if the solid state VFO is well DC regulated then maybe not. I actually would recommend using a solid state VFO. You would just need to make sure you setup the grid leak bias on the driver tube for a lower resistance if possible if no buffer stage is used. It can get a little trickier if fixed bias is used though, but it just depends.
  3. Dwayne

    Dwayne Member

    Thank you very much. . .for the reply. . .

    Yes, I was planning on a Digital VFO for stability and accuracy. . .also, in the long run, it will probably be much cheaper than trying to purchase switches for different bands to attenuate the harmonics associated with a VFO that is free floating compared to a synthesized VFO. If push comes to shove, I can put in a switch for switching a small parallel tuned circuit for each of the bands.

    I guess I had better find another filament transformer. . .

    My idea is this. ..

    I have an OLD Hallicrafters ht-32 that has been sitting in someones basement for decades. . . It is basically worthless because of rust, and was given to me. I think that HT-32 case is big enough to hold everything, including a 20watt Radio Shack PA to drive the grids of the modulators.

    This is my primary thought. . .you know, that thought before Murphy hits. . .
  4. W5HRO

    W5HRO Administrator

    What you could do for the VFO is run it into a HV (like 150V) NPN transistor in emitter follower mode with a 1M resistor to ground. If you use a HV transistor it will work with a large value resistor like that. That would buffer the VFO to the input of a driver tube then you could just use an ordinary RF tuned circuit on the output of the driver tube to eliminate the harmonics.

    Yeah, I would recommend using separate filament transformers. In the old days many transmitters did meter the currents via the HV plate side lines, but they always had a habit of burning up the meters. Not only that, but it also created the safety hazard by possibly placing the HV potential on the face of the meters on the front panels. As a result cathode current metering became the norm and replaced it.
  5. Dwayne

    Dwayne Member

    Thank you very much for all your info and help. . .

    It gives me a lot to work on, as well as excellent idea's to help overcome obstacles that may be Murphy'd into the situation without knowledge.

    One other thing. . .Do you think this project will be able to fit into a ht-32 chassis? or is that being a little bit unreal?

    I think my main problem is making sure the tubes will fit in with safety from the top of the chassis. . .I will have to do some measuring and checking on that.
  6. W5HRO

    W5HRO Administrator

    It should fit if you strip everything out of there, but one thing to consider is recessing the 813 tube bases below the top of the chassis. Have the bases below the top and all the glass above it.

    P.S. You may need to outboard the plate and modulation transformers though. Maybe make the plate supply and modulation transformer a separate unit that plugs in. Sort of like the Johnson 500.
  7. Dwayne

    Dwayne Member

    Ok, I am a little new to this. . .But. . .

    I found a transformer. . .Using a variac, I was able to establish a 1:15 step. . . and was able to get about 1800 volts off of 120 volts input when I was finished.

    1. Is this 1800 volts good enough? Eventhough the schematic says 2000? And this is without a load on it.

    2. In testing this transformer, It had one lead labeled as S0. the other two leads were supposedly the secondary wires.
    But, If I tried to measure across the secondary wires, I got zero. If I measured to the S0 wire, both wires measured 1800.

    This doesn't make any sense to me. The S0 is not a center tap, or I would be able to measure 3600 volts across the two primary wires. But I am able to measure 1800 volts from each of the primary wires to S0.

    Either way, would this transformer do? or will it be to light. I believe it is from an old radio or Microwave transmitter.
  8. W5HRO

    W5HRO Administrator

    I'm assuming the 1800V is VAC and not VDC, right? Are you measuring it with an AC volt meter (DVM) or with a scope using two HV probes?

    First, ohm out the windings to see what's actually connected before trying to check the voltage and if possible attach a photo of the transformer showing all the terminals.

    The other important thing is how much current will it provide. It could be an old microwave pulse transformer and maybe even 3-phase, but I do remember some of the old transformers that were labeled S0 and that was usually the top connection at the top of the secondary winding, but there is no telling.
  9. Dwayne

    Dwayne Member

    Yes, that is 1800V AC measured with 1000X probe on old Balentine scope. I cut the power down, and was able to measure 200V with VOM for verification of stepping.

    I will try to get a picture of it to send to you. It does have a filament winding of 3.3 volts, kinda worthless for my case, I need 10V or a multiple of 10V at 5amps.

    Thanks for your help!

  10. W5HRO

    W5HRO Administrator

    Yeah, upload some photos showing all of the connections, primary, secondary and any labels that might be on it.

    It's just a wild guess at this point, but from your description it could be a 3-phase transformer (Delta) where S0 was tied to ground and/or used as a neutral or it could just be an ungrounded Wye type. The issue would be phase related.
  11. Dwayne

    Dwayne Member

    Ok, here is a picture of the transformer

    The thick Red wires were 3.3 volts.. Transformer said they were filament wires, and that checks out.
    The two white ones are a 1:15 step up. . .if used with the lone S0 wire. . .

    But if you measure between the two white wires, the voltage is zero. . .Which doesn't make any sense to me.

    Maybe if one coil is wound clockwise, and the other coil is wound counter clockwise? And I missed seeing the negative sign? and when reacting together they are zero? Just don't know. . .

  12. W5HRO

    W5HRO Administrator

    So if you ohm (ring-out) between F0 and F1 it shows an open?

    Also dial the variac way back to where you can use two scope probes and see if the outputs are in the same phase. What I mean is do the sinewaves on both channel 1 and channel 2 superimpose over each other? If so then that means each side was designed for a different phase (same phase). Connect both of the probe ground clips to S0 when doing it.

    The issue might be the current, but we don't know yet and I need to look at the 813 data sheet again for the voltages. If both sides do superimpose over each other then you could just parallel them by connecting F0 and F1 together to double the current then use a bridge rectifier to convert it to DC as long as you don't see anything when ohming from F0 to F1.

    Anyway, run those two tests above and let me know what happens.
  13. Dwayne

    Dwayne Member

    Ok, I will scope it out when I get home under a Variac. . .

    I brought the transformer to work. . .and this is what I found.

    Filament wires (3.3V @ 120v) 1.7ohms between the two.
    Noticed there was a color coding on the two HV? wires. . .Possibly to signify different voltages. . .Because I was using a old Ballantine 10mhz scope, I may not have been able to see the voltage differences. . . I will find out tonight..

    Resistance between S0 and Black striped HV is 35.4 ohms. . .
    Resistance between S0 and Red Striped HV is 38 ohmes.. . .This is probably a higher voltage than the other one.
    Resistance between the two HV wires is 4 ohms.
    All wires are open to ground. .. no shorts.

    Makes me wonder if the secondary has a high tap.

    Using a ratio and the idea that one is a high tape, If I measured the lower line at 1800 volts, then the higher line would probably be around 1900V. If I measured the higher end at 1800 volts, then the lower winding would be around 1600V.

    When I get home, I will stick it on a scope and see if they match up. . . .Or maybe the real questions is whether it is worth messing with, because either way, the max Voltage is 1900 without a load, and with a load, it will drop quite a bit. I wish the filament wire size was the HV wires. The filament wires seem much more robust and stouter than the HV side. My Transformer out of my HXL-1 seems to be more robust on the secondary side.
  14. W5HRO

    W5HRO Administrator

    No need, you already figured it out. One of the F wires is just a tap down the winding about 4-ohms away from the other F wire, but it looks like they are the same winding.

    It may have been done in case the 120V on the primary side was either a little high or low they could just switch taps to bring it up or down to match the requirement. It's normally done with extra taps on the primary side, but it looks like they just did it on the secondary side instead. That or their application needed slightly different voltages for some strange reason.
  15. Dwayne

    Dwayne Member

    Thanks a million for your help. . .

    I scoped it out, and took pictures for you.
    At 20 Volt scale, I couldn't tell any difference. . .
    I pushed it to 200 Volts thereabouts, and one wire showed 215 and the other wire 230. (Used a VOM to measure voltage) Making the lower resistant wire a 1:14 and the higher resistant wire 1:15. Thus 120V is approx 1800 volts max out of this transformer.

    Thus, I am 200 volts short of the schematics 2000 volts. This is without a load on it.
    2000 x .4 = 800 watts input.
    1800 x .4 = 600 watts input, if the voltage does not drop.

    I have no clue at all, how much the voltage will drop from 1800 under a load. My knowledge is not good enough to even guess at it. I also do not know how well the tube will preform at a lower voltage. I seriously doubt if it is linear. . .probably some kind of exponential curve to some degree. I know very little of the 813, just that I have about 14 of them. ..brand new. :biggrin:

    I am uploading the scope pictures of the two secondaries, showing a voltage difference.

    photo 1.jpg photo 2.jpg
  16. W5HRO

    W5HRO Administrator

    Actually 1800VAC is just right for plate modulated class C phone using 813's. Is the 1800VAC an RMS reading from your voltmeter or did you calculate the RMS based on the peak-to-peak scope reading? The voltage will be lower after converting it to DC anyway.

    You will need to use a bridge rectifier and when you use 0.90 X the average VAC (the 1800V) it equals 1620Vdc. I would find a choke for choke input, but you could even do capacitor input and it might be fine when loaded down. The 0.90 X the average VAC will come out the same using either provided you are looking at the loaded down capacitor input method. It will be about the same loaded or not loaded using choke input. That's why choke input is always best. Better regulation.

    Forget what Tom's circuit says, there are some minor mistakes with it. First, you don't run 813's at 2000VDC in class C phone because that's their ICAS (intermittent) ratings. You want 1600VDC max and even 1250VDC will work for lower power. 813's are very durable, but 2000VDC is really too high :icon_thumbdown:

    One of the other mistakes is the 2.5mH RF choke in series with the screens and the 100mA screen current meter. That RF choke does absolutely nothing because there is no bypass cap on the correct side it. Plus that meter needs to be on the other side of the 15H screen choke. He was probably trying to see what the screen current would do during modulation, but its a really bad idea. Put the meter on the other side of the 15H screen choke and measure the static screen current instead. And remove that unnecessary 2.5mH RF choke in series with it because it's not doing anything.

    Anyway, the only issues will be the 813 bias voltages, but the main issue is will that transformer provide enough current to run all four 813's? It might.

    813 RF (AM) Data
    Plate = +1600Vdc
    Screen = +300Vdc
    Grid = -160Vdc
    Plate Current = 300mA (2 tubes)
    Power Out 360W (2 tubes)

    813 AF (AB1) Data
    Plate = +1600Vdc
    Screen = +750Vdc
    Grid = -85Vdc
    Plate Current = 300mA (2 tubes)
    Power Out 265W (2 tubes)

    More than enough for exceeding 100% modulation. The transformer just needs to provide around 600mA for instantaneous modulation peaks.
  17. Dwayne

    Dwayne Member

    I found the ratio of the windings on the transformer, and found it to be 1:15 using my VOM.
    Measured my input voltage with VOM divided it into the output voltage(measured with VOM) and got 15.

    Then I did multiplication and took 120 x 15 and got 1800AC.

    My Ballantine scope is not accurate enough to measure voltage reliably. . . It wil measure plus or minus 15 percent probably . . .I usually only use it as an reference of voltage. Scope is so old, it is amazing it still works. but heck, as long as it works, who cares. Frequency measurement is taken by multiplying the timebase by the wavelength and inverting it. . .Accuracy is how well my feeble eyes can distinguish the markings on the scope. :lol:

    Do you think 4 1n4007 will be ok for the bridge? I beleve they are rated at 1amp at 1000V That is about 2000 volts total. which is 200 to 400 volts over operating voltage.
  18. W5HRO

    W5HRO Administrator

    No, I would use at least 3 1N4007's minimum in series x 4. Four sets of 3 to make a bridge. That or 4 sets of 4. 4 sets of 4 might be better.

    The big question is does the transformer have enough current? Get a bag of 1N4007's and place an oil fill cap or smaller series caps across a resistive load and see what it does first. Start out with around a 10K load. Then if the voltage is still there drop the load down to around 5K then 3K. 1600Vdc @ 600mA would be 2.67K. If you still have around 1500V to 1600V with 3K then it should work ok, but if it drops like a brick it won't.

    P.S. The resistor load will need to be high wattage. Just series and/or parallel whatever you can so they don't burn up instantly. 1600Vdc with a 5K load for example will be just over 500 watts :eek:
  19. W2WDX

    W2WDX Member

    I made that drawing years ago for Tom ... the original was a disaster scribbled out by hand. So I made a pretty one! Hehe!

    John, W2WDX

    Note: They took my name and call sign off the drawing on AMFone many years later after I was put on their ever expanding blacklist.
  20. Dwayne

    Dwayne Member

    I have ended up changing this. . .

    I have decided to go for a pair of 813's modulating a 4-400. . .that way I can go higher in frequency and have full power.
  21. W5HRO

    W5HRO Administrator

    The 4-400A is a great tube, but how high in frequency were you talking about?

    I'm asking because there is a bogus myth about 813's not working well up to 30Mc and that's what is, a bogus myth. 813's will put out as much power at 30Mc as they will down at 1.8Mc. You just have to know the trick. On 10-meters for example you have to use a little extra coil (uH) then tap it over a few turns with the plate tuning cap in the pi-network. It's very easy…

    Don't believe everything you hear and/or read on the web. As long as you don't try to use more than 2 tubes in parallel max they will work fine on 10 and 15-meters both and at full power.
  22. KA4KOE

    KA4KOE Guest

    Doesn't this design need a clamp somewhere in case you lose drive or bias?
  23. W5HRO

    W5HRO Administrator

    It should... He is relying solely on the fact that both the grids and the screens are connected to fixed supplies. You really want the screen to pull down well below +400V if the grids all of a sudden become open for any reason and with Tom’s design it won’t do that.

    It’s an 813 meltdown waiting to happen.