Calculating grid bias for 807

Discussion in 'Technical' started by W6MQI, Nov 8, 2015.

  1. W6MQI

    W6MQI Member

    Okay have to admit I pretty much know nothing about transmitter design, but I'm trying to design a CW transmitter using an 807 running class C. Been going through the step by step calculations using the RCA transmitting tube manual for triodes running class C, and came across one step I don't understand. Calculating grid bias, the formula is Ec= [K3 x (ecmax + ebmin/u) +Eb/u]

    Ec = Grid bias
    ecmax = 10 volts, Peak positive grid voltage
    ebmin = 100 volts, effective minimum plate voltage
    K3 = .520 This value comes from a conduction angle chart.
    u = amplification factor

    My problem is I can't find (u) the amplification factor for an 807 anywhere in the tube data sheet how do I come up with this value?

  2. W5HRO

    W5HRO Administrator

    You are probably trying to make it a bit too complicated. For CW operation with the cathode grounded you want 600V on the plate, 300V on the screen and -200V on the control grid. That's for CCS operation. Those other typical operation specs are more for when the 807 is being used as an oscillator. You just need a fixed supply on the control grid that is always on.

    With -200V on the control grid you set the RF drive to where you get 5mA. You can just use a 30K screen dropping resistor from the HV line for around 10mA at 3 watts dissipation max to be safe. The plate current will run at 100mA.
  3. WD5JKO

    WD5JKO Member

    Dave, Besides what Brian said, a couple of points. Your calculations are for a triode. The 807 is a beam power tube. The addition of a screen grid is a big deal. Also, with a class C RF amplifier, the screen voltage is often obtained from a series dropping resistor off the plate supply. This means that with no RF drive, the screen voltage will rise to the value of the plate voltage. This is why many class C circuits use a clamp tube to knock the screen voltage to a low value when there is no excitation. The grid bias required to cutoff the tube is dependent upon mainly the screen voltage, and to a lesser extent the plate voltage. I'd do this exercise graphically from the 807 plate curves that is included in the datasheet. If the screen voltage is obtained from a dropping resistor, then the curves won't help since they will not have data with a screen voltage as high as the plate could be. If you get the screen voltage from a fixed source, then the curves will be more helpful. Realize that any condition that leaves the screen voltage on without any plate voltage could ruin the tube from excessive screen current.

    data sheet:

  4. W5HRO

    W5HRO Administrator

    Yeah, that's why using a fixed supply on the screen is a usually a bad idea. It's better to use a dropping resistor so when he turns the plate supply on and off the screen switches on and off with it.

    I've seen people even use control grid resistors to ground for that bias in CW mode too, but a stiff fixed supply is much more stable for CW. You don't want weird things happening that might produce chirping, etc. each time you press the key. You want the control grid impedance to stay as stable as possible.
  5. W6MQI

    W6MQI Member

    Attached is two piece schematic of a simple a duel 807 CW transmitter that I cut and pasted different designs together. My voltages are a little different then shown in schematic for example my power supply has a plate voltage of around 700vdc, and screen is around 280vdc.

    The transmitter design is using series resistors off the plate supply for screen voltage, so if it's running class C wheres the protection without a clamp circuit.

    I have do more questions:smile:

    807 finals.jpg osc buff.jpg
  6. W5HRO

    W5HRO Administrator

    With a fixed -200V control grid supply and a 30K screen dropping resistor the screen will probably never raise up very high anyway when no drive is present. That was my whole point. What you can do though is hang a zener diode off of the screen connection to ground for fail-safe protection. You have a 10-ohm cathode resistor to ground which will add some bias anyway, but I'm not a fan of using control grid-leak bias for CW using a resistor You want the load on the CW oscillator stable and it won't be using a grid-leak bias resistor. A fixed supply there is better. You can probably still use a fixed supply and still have just a small bit of cathode resistance like that for protection.

    EDIT: What you could also do on the screen is use a voltage divider with one resistor going to ground. Since you are using two tubes the single 30K a told you is wrong because you want a least 20mA or just a little more for the screens. You could use one resistor from the HV to the screens then a second resistor of the same value from the screens to ground to create the voltage divider and current limit. So, say with a 600V supply the screens would see the 300V and have a current limit.

    P.S. I would use a +600V or +650V max supply instead of +750V. You really want CCS and not ICAS.
  7. W5HRO

    W5HRO Administrator


    I deleted and re-did this post because I had a brainfart moment and I had used 300V instead of the 600V for the voltage divider calculation by mistake ::) If you take your 50K and divide it by two you get the two 25K resistors in series for the divider.

    Anyway..., a simple configuration like the below would work nicely. It just so happens that at +600V a voltage divider consisting of two 25K resistors yields +300V at 7.5W of dissipation. If you regulate the control grid voltage to -180V to -200V you should not have any problem and the oscillator will have a very stable load. If you don't like using 0B3's then two 100V solid-state zeners in series would also work.

    Anyway, I hope this gives you some ideas. I didn't add any of the necessary bypass caps though. I'm just trying to show the basic configuration using the resistors and regulators.

  8. W6MQI

    W6MQI Member

    The supply I have has an output of 700 vdc, 285 vdc, 6.3 vac for filaments, and a spare 12 vac. Sounds like the 700v is a bit to high from what your saying.

    So just for fun the voltage divider might be for 700v first resistor 35k, resistor going to ground 25k would get me around 290 vdc on the screens.

    Also looking at your drawing I assume cathode keying is still employed?

  9. W5HRO

    W5HRO Administrator

    What, key the 807's? You only key the oscillator stage cathode. Did you come up with that drawing from looking at another circuit from somewhere?

    Generally you have a least a keyed oscillator, then a tuned buffer/driver stage, then the final or finals. So basically it consists of at least 3 stages and most often 4 stages. I've seen some really simplified circuits where people were trying to use just one stage or two and I assumed that's maybe what you are trying to do, but you would leave the 807 cathodes grounded. With a fixed grid bias supply no current would really flow until RF excitation was applied from the oscillator. But at the same time the oscillator stage may not be enough to even drive the 807's.

    What you should really do is use 3 stages. The keyed oscillator, the tuned buffer/driver, then the 807 finals. The 700V would probably be ok if that's just the transformer itself. After it goes through a choke input circuit (choke and capacitor) it would be less than 700V anyway. If not there is probably a way to still make it work.
  10. W6MQI

    W6MQI Member

    Key the 807’s directly no.

    The second schematic shows an osc/ buffer stage with the cathode of the 6AG7 being keyed. With the setting of switch SW1B to either "Crystal" or “DDS” position the 6AG7 is an oscillator “crystal position” or a buffer stage “DDS position”. The 0.01 uF capacitor that is switched into the circuit kills the tendency for the 6AG7 to oscillate. The key is connected to both the 6AG7 cathode, and the 807’s cathode via a 10 ohm resistor. I was just asking if this was the same type of keying scheme with your schematic as with the one I have drawn. I wasn’t sure about the arrows off the cathodes if those were grounds or not.

    Yes as I stated earlier this schematic was a cut a paste of two different transmitters that I combined I did not design this transmitter.
  11. W5HRO

    W5HRO Administrator

    No, the circuit I drew has the 807 cathodes grounded. That's what those arrows are, gnd.

    I'm not even sure your buffer stage will drive the 807's, but I'd need to see the original transmitter circuit. I'm assuming since you were trying to figure out the 807 bias though means that transmitter probably didn't use 807's ???

    In my circuit you would probably need more drive due to the higher negative bias voltage. Most transmitters have an oscillator/vfo stage, then a buffer stage, then a driver stage, then and the finals. That’s 4 stages total, but it can be consolidated and simplified if you do it right. What you normally do though is control the screen voltage via a pot or rheostat on the driver stage to set the control grid current with the RF drive to the finals. The amount of drive via the driver's screen voltage will very on different frequencies and/or bands.

    If you can, post the actual circuits you copy and pasted from. I've seen some really simple CW transmitters where only 1 tube was used as the entire oscillator/transmitter, but those are more for experimentation/learning and playing around. Sort of the same thing as building regen receivers.
  12. W6MQI

    W6MQI Member

    Here's one for some reason I can't find the other one I used for the osc/buffer stage it also had different value resistors on the grid, screen was taped off the 750 v B+ through screen dropping resistors it also had only one 807 thus the different value components. Hard to say if this design really works very well I've never seen any write up on it's performance. The single 807 version seems to be every where might be a better way to go I'm really not dead set on duel 807's by any means.

  13. W5HRO

    W5HRO Administrator

    That looks like something someone draw up by hand, but has it been tested?

    They are also using a separate +250V supply for the 807 screens and like Jim said you only want that on when the plate voltage is on. Looks like they are leaving both the plate a screen supply on though an leaving everything turned on in a very low current state just to keep everything warm then the key turns everything on full.

    It might work, but the frequency stability will be very poor if using xtals and will change as you tune the 807's output. If the output of your DDS has a buffer then it might be fine or the 6AG7 might be enough to do it. Only one way to find out.

    If you are going to do it like that though then I would leave the design exactly like it is and use a separate +250V supply for the screens and just try connecting your DDS to the 6AG7, but you need a buffer or resistor on it to match the DDS's low Z output.
  14. W6MQI

    W6MQI Member

    My intent was to make something simple, but I also didn't want a chirp box either:mrgreen:

    My DDS should drive a design like this it drives my Ranger , and Viking II just fine. The reason I use the DDS on the Ranger is because the stock VFO isn't very stable at all on CW. Yes I have tried all the fixes that everyone seems suggest, but nothing really worked very well for CW op's.

    Here's another design that may look familiar to you I could just copy this and be done with it.

  15. W5HRO

    W5HRO Administrator

    There have been a few write-ups on using the new solid-state DDS's and the need for a low Z to high Z buffer when connecting them to high Z tube input circuits. Even though they may work, a buffer should really be used otherwise the DDS won’t get loaded down correctly and it could possibly create additional harmonics, etc. It’s more than just about having enough pk-to-pk voltage swing. I don’t know what your DDS circuit looks like though and maybe that issue has already been eliminated with yours.

    You can try those simple circuits if you don't care about stability, etc. If it was something you really plan on using though and a lot you might consider an improved design like what I posted. The only difference is that I would use zeners at the screen and at the control grid both to regulate both supplies then have a CW on and off switch to where you just switch the HV on when ready to use the transmitter, but the 807 cathodes would always stay grounded. Then I would use a driver stage with a rheostat to control it’s screen voltage to set the drive level to the 807’s

    Anyway, that’s just me though because the most important thing on CW is stability.
  16. W5HRO

    W5HRO Administrator

    Here's a thought and for anyone thinking of building one of these CW transmitters. A simple pentode buffer with a 51 to 600-ohm resistor from the cathode to ground and you feed your DDS VFO's output across that resistor. Then the plate of that tube gets coupled over to the grid of a 6AG7 driver and then that gets coupled over to the 807 grids. You could control the RF drive level by adjusting the 6AG7's screen voltage from 0V to 150V. The 6AG7 is good for 3 watts so you might need to reduce the negative grid bias on the 807's some, but it wouldn't matter. As long as the 807 screens are between +250V and +300V you would set the control grid bias accordingly to place it just past cutoff. The buffer tube's plate and the 6AG7 plate would both need a tank circuit, but that's easy. At least that way the DDS would have the proper low Z load and would be buffered before reaching the driver stage like it should be.

    Just a thought...

    EDIT: I like playing around and thinking about this kind of stuff so below is the thought in a rough draft. It would depend on how much voltage swing is coming from your DDS, but if you have installed the necessary buffer transistor inside of it to drive a short piece of RG-58A/U then it just might work. It just so happens that there is a spec/mode in the 6AG7 data where it uses a 57-ohm cathode resistor and 51-ohms plus the 10-ohm 2.5mH choke would be close enough.

    The transmitter would only require a +600V or greater supply, a +300V or +250V supply, then a fixed negative bias supply, then a total of 4 zeners. It might be possible to use the +300V or +250V supply and then use a second bridge rectifier to get -300V or -250V negative from it as well and long as it's not the center-tapped half-wave kind. I've seen a few designs in the past where someone got really creative and used one transformer winding to generate both positive and negative voltages, but there are some tricks to making it work. It would be better to find a power transformer with two separate windings though and there are lots of those around. The other option is using a +600V transformer with enough current to power the other +300V section and just using two more zeners for that +300V. A 625V or 650V 250mA transformer would probably work perfect.

    Anyway, all you would need to do is adjust the 807's negative control grid bias supply to just a little beyond cutoff without the DDS connected and all the transmitter supplies would be left on when using the transmitter. Then just pick the closest zener for that necessary negative voltage to hold the 807's just past cutoff. Then simply connecting to DDS to the cathode of the first stage with your key would key the transmitter and current would flow through the 807’s instantly. I've never been a fan of grounding the cathodes with the key because it's too slow. It takes a little time for things to come up and stabilize when doing that. Real time is better :icon_thumbup:

  17. W6MQI

    W6MQI Member

    Boy I guess you do :icon_thumbup: Can't say the transmitter is very simple anymore I guess if it were single band unit it would be. In my transmitter I was going to use plugin coils for the grid, and plate tanks for simplicity, although sort of a PITA to change bands.
    Correct me if I'm wrong ,but for multiband transmitters is it best to use shorting rotary switches in the tanks circuits to short out sections of the coils not being used? So if the transmitter is set to 20m 40, and 80 taps should are shorted.

  18. W6MQI

    W6MQI Member

    Here's the type of switch I was talking about unfortunately these are hard to come by I removed this one from a dead T4XC. Shorting switch maybe the wrong name for this type of rotary switch.

  19. W5HRO

    W5HRO Administrator

    Well, I had some time yesterday and I like figuring things out that no one has probably ever thought of before. Actually the circuit is not all that complicated and would be simple to build. Most of the old tube CW transmitters had issues and adding just a bit of newer technology in the right places can make a really big difference.

    Below is an updated version which also eliminates the extra +300V supply. You could probably even just use two of the 50W stud mounted zeners with heat sinks instead of using 4 zeners to get the 300V in both places, but I think using the 4 smaller ones would be easier in the end.

    What you might try doing with the switch is just tap the coil in different places with the Vcap. What I mean is connect it to the Vcap to move it up and down the coil instead of shorting the coil windings. It might be easier.


    P.S. What you would really want to do is have your key plugged into the DDS and switch it on and off. The key in the above drawings were just for reference. I quickly modified the last one so it would make a little more sense. You would key your DDS on and off from inside of the DDS itself somehow with super fast solid state switching. If it had a spot button or switch you could maybe just tie into that.
  20. W5HRO

    W5HRO Administrator

    And, you are probably are going to think I'm really crazy, but if the drive level coming from the DDS wasn't high enough you could just configure the first 6AG7 stage as a grounded grid linear amp :lol:

    What's funny is that it would probably work. You would need to make sure though that it only amplified it to where you get a consistent 4V peak-to-peak max into the grid of the second 6AG7 stage. You would just increase the value of the 47 ohm resistor if it was too hot. I've read where a lot of those DDS's have pretty weak output levels like well under 1V peak-to-peak when loaded down and the drawing below would probably fix it. You would just need a 625V to 650V transformer with at least 250mA to 270mA of current for the whole transmitter. Then the smaller negative one for the 807 control grid bias.

  21. W6MQI

    W6MQI Member

    Yea your crazy:biggrin: The DDS I'm working on is a N3ZI which will have a buffer on the output not 100% sure how much hooch I'll get out that remains to be seen.

    Probably not too hard to do although the DDS buffer gain starts to dropping off above 14 mhz.

    I like this feature, did something similar to the Ranger using the schematic shown.

    I like your first design tweak it in I probably have enough parts laying around to put some thing together just need to find the time my list of projects around here is getting longer, and longer. Lets see extend dog run fence, dog house for two, master bedroom remodel, redo the driveway, finish the NC-303, finish the N3ZI DDS, install the 600hz roofing filter into the R4C then ...............;) it never ends.

  22. W5HRO

    W5HRO Administrator

    You mean to replace the hard to find Johnson 25K rheostat? You can get small Ohmite rheostats that will work fine for home-brew projects like this.

    Anyway, below it a much better update of the first update you liked. You really don't want a resistor on the plate of the first 6AG7 and those little Hammond 2.5mH chokes are really cheap. Plus you don't want noise getting into the control grid so it should be grounded instead. At "0" control grid bias the screen should be 115V and they make 110 and 120V zeners so either one is close enough. I also redid the cathode circuit and that should prevent any additional gain loss.

    This would basically be an RF sensed keyed CW transmitter which would eliminate all of the problems that plagued the old cathode keyed designs.

  23. W5HRO

    W5HRO Administrator

    I played around with the driver screen control circuit and a Bipolar transistor works much better than an FET transistor. I'm assuming you thought because FET's are high Z that it would work better, but the problem is their inputs are high Z and that makes them not well suited for use with voltage divider resistor networks connected to their gates.

    The below circuit seems to work pretty well and the max current the 6AG7 screen would ever need would be 13mA to 15mA. I tried different transistors and either 180V or 200V NPN's work best and they only require just a little over 1mA on their bases to work properly. About 1.4mA to 1.5mA seems works best so the circuit below with a 100K pot would be fine and only dissipate a little over 200mW. I would still always use a 2 watt pot if you can find them though instead of a 1 watt pot.

    Anyway, BiPolar NPN's would work better and be a lot more robust to control screen voltages than FET's would. If you found that circuit on the web then it's just another one of those poor examples someone with misinformation came up with. I tried a couple of FET's just for the hell of it and with a 15mA load the output caved :icon_thumbdown:

    P.S. The circuit would be using the regulated 150V from the zener tap instead of the +300V which is better anyway.

  24. W5HRO

    W5HRO Administrator

    Here's the final circuit using the NPN to control the driver screen voltage. Had just enough room on the drawing to make it fit.

  25. W6MQI

    W6MQI Member

    Fantastic sure wish I had the know how to whip out designs like this so fast I don't work in the electronic industry its just a hobby for me. One question why are you grounding the B+ through the 570K resistor can't this just be left floating on the unused relay contacts when the B+ switch is turned off?