How hard is it to move a Broadcast transmitter to 3.880khz from 1570khz?

Discussion in 'Amplitude Modulation' started by K8CCA, Apr 27, 2021.

ad: L-HROutlet
ad: l-rl
ad: L-MFJ
ad: Left-3
ad: L-Geochron
ad: abrind-2
ad: Left-2
  1. K5UJ

    K5UJ Ham Member QRZ Page

    I wonder if the 807's problem has to do with unshielded base leads to the electrodes.
    Did you connect Cn directly to the 802 plate or employ some sort of RF pickup wire?
    Right, but when you don't have that, harmonics pass straight through. Maybe the stock oscillator/driver setup on my 3-400 rig was biased in class A or B.
     
  2. K4KYV

    K4KYV Premium Subscriber Volunteer Moderator QRZ Page

    It was already set up with a balanced tank, because my first attempt at the rig used a triode, an 801-A as I recall, a specifically RF version of a type 10. The triode worked perfectly on 160 and 80 but the preceding stage, a pentode crystal oscillator, wouldn't deliver enough drive for 40 or 20, so I replaced it first with the 807, leaving the balanced tank in place. Cn comes directly off the plate. It's made from a couple of tabs cut from a piece of copper strap, each about 5/8" square, soldered to pieces of solid wire. It's adjusted by bending the solid wire to move one of the tabs closer or farther away from the other. Once set, I have never had to re-adjust it, except for when I replaced the 807 with the 802.

    The 807 has a long-time reputation for being a squirrelly tube. Maybe not shielded well enough for its power sensitivity. After all, it's nothing more than a glorified 6L6G. The 802 was designed from the outset by RCA as an RF transmitting pentode. Although the 807 is rated for higher plate dissipation, the 802 gave exactly the same grid drive to the 211 with the same screen and plate voltages applied. Interestingly, the 807 comes with the familiar dark orange phenolic or occasionally white ceramic base like most other RF tubes, while the 802 has the ordinary black phenolic base like a 6L6G and other receiving tubes.

    Older 807s have a shield under the plate structure that partly covers the leads between the base and electrodes, but later versions made just before RCA discontinued them, omitted the shield, leaving them to more closely resemble the 6L6G. Not surprisingly, the later ones without the little shield structure are even more squirrelly. Another thing RCA did to cheapen the late versions of the 807 was to eliminate the little ceramic insulating posts that supported the plate structure, replacing them with sheet mica, again making them more like a 6L6G.
     
    1 person likes this.
  3. K5UJ

    K5UJ Ham Member QRZ Page

    Ditto for the 804. I have a pair of 804s in the old RCA World Map boxes that I found at Dayton one year. I have nothing that uses them but I have hung on to them because of this notion that they might prove useful some day. The problem with getting anything to use an 804 would be finding replacements if the two I have crap out. But the tube manual specifically says it's a RF tube.

    I have a nice supply of the older JAN style 807s. Anything like that can become the next 2A3 so I wanted to have some before an 807 costs $150.

    Back to the topic of moving a rig to 3880, here are some photos of my work modifying a balanced 3-400 grid input network to transform the calculated grid Z to 50 ohms unbalanced. At this point I'm just fooling around experimenting. The grid current needed is around 200 ma. for two tubes in order to develop the needed class C bias of 100 v. across 1K grid leak resistance. The data sheet RF drive peak v. is 187 v. So at the peak of the RF cycle the grid Z should be 935 ohms. In the photo there is a string of carbon resistors that are close to 900 ohms to simulate the grid Z. The original balanced grid network was a tapped down minidux coil with a CT and an air variable paralleled with the coil. Tap on one side of CT goes to the grids; on the other side the tap goes to Cn. CT goes to junction of RF bypass cap, and grid leak resistors to grid current meter and filament transformer CT. Originally the network transformed the anode Z of a 6146 to the grid Z. To get it down to 50 ohms I turned it into a variable balun, or mini-matchbox (sort of) by adding a link around the middle of the coil using h.v. wire. One end of the link is grounded; the other end goes to the RF drive. grid_link_tests - 1.jpeg grid_link_tests - 2.jpeg grid_link_tests - 3.jpeg
     
  4. K4KYV

    K4KYV Premium Subscriber Volunteer Moderator QRZ Page

    True for any kind of hard-to-find tube, and more and more are falling into that category.

    Tapping down the grid on a tank coil is an invitation for self oscillation/parasitics, especially in a neutralised stage. The Gates BC1-T did the same thing. I moved the connections that went to the taps over to the ends of the coil, which made it much more stable, and I was still able to get as much grid drive with the same DC input to the pair of 807s driver stage. Less problematic in the early low-power stages, such as a xtal oscillator or VFO buffer stage, if it can't be avoided. I don't like tap down the plate connection either.

    That reminds me of someone I worked a few nights ago who had converted his KW Matchbox into a final amplifier using a pair of triode-connected 813s. It sounded good and seemed to work to his satisfaction, even though he said it had no provisions for a neutralisation circuit, which is usually deemed necessary for a triode power amplifier, even a triode-connected screen grid tube.

    If the driver stage is capable of delivering an excess amount grid drive, you should be able to improve modulation linearity and maybe even PA efficiency by increasing the value of the grid leak resistors, then increasing the coupling to bring the grid current back up to normal. This increases the bias voltage and produces a smaller conduction angle, thus pushing the tube deeper into class-C. I run my HF-300s at normal grid current, but at nearly twice the recommended grid leak/ grid bias voltage. Just be careful not to overdo it, and exceed safe grid dissipation.
     
    N2EY and WZ5Q like this.
  5. K5UJ

    K5UJ Ham Member QRZ Page

    Noted, but it hasn't happened yet. However, I did fashion clips on the tap leads to facilitate experimentation and found the tap pair that provided the best Z transformation and (I hope) power transfer. The rig did need re-neutralizing so I did that which made it more solid. Hard to describe but I've run this baby long enough to know when things aren't right. She makes a high pitched whine sound when I noticed the grid peak and plate dip were not simultaneous. After readjusting Cn, that whining went away.
    I have a few things to try: shorting the unused taps on the input coil; doing the same thing on the output pi net. coil. That coil has only one tap, a clip to copper strap I can move, and I want to short the unused turns. And of course I want to try boosting the grid leak R to maybe 2K and drive that sucker. But first, I want to put it on an antenna with audio and see if anything craps out right away. The Viking 2 driver may not yet be bullet proof because I hurried through it since this odyssey has lasted since late last summer.

    What band were you on? He may have gotten lucky with all the leads and such just so, but if he went to the high bands, or moved the connecting leads around much the final might start acting up.
    The beauty of the 3-400s, is that besides zero bias, each grid can dissipate 20 watts. They're pretty tough grids and 200 ma. Ig for a pair is no problem. What sucks is the need for forced air cooling.

    I'm glad I got the estimate of the grid Z right. The driver came right in and almost didn't need to be re-dipped when I put it into the final PA. driver_grid_link_work - 4.jpeg driver_grid_link_work - 6.jpeg
     
  6. K4KYV

    K4KYV Premium Subscriber Volunteer Moderator QRZ Page

    Your grid coil looks a lot like the one I put in the BC1-T, to replace the original coil wound that crappy mud form. That change increased the grid drive to the PA about 15%, without noticeably altering the DC plate current to the pair of 807s at resonance. I disconnected the RF bypass from the mid-tap on the coil, letting the wire-wound grid resistor serve as an RF choke. The tuned circuit now gets its ground reference through the frame/rotor of the split stator capacitor instead of through a tap on the coil. That greatly improved the stability of the circuit.

    Your high-pitched whine off resonance sounds very similar to the kind of instability I was having. The split-stator modification apparently corrected it.
     
  7. K5MIL

    K5MIL Ham Member QRZ Page

    Not as hard as moving it into the basement I bet.

    Bill - K5MIL
    Advanced Class
     
    W3SLK likes this.

Share This Page

ad: w5yi