The AM Forum

I am building an exciter for my 8000 rig.  6AG7 VFO, 6CL6 buffer and 807 final.

I am plagued with the problem that resonance in the final does not agree with peak power out.

The power peak (watt meter or light bulb) is a considerable twist south of the plate current dip.

Looking at the output on a scope shows nothing strange about the wave shape as the amplitude increases beyond the dip.

I have tried neutralizing with coupled links between the input and output tanks but it makes no difference nor does anything else I have tried.

Any ideas would be welcome.

Jack K9ACT

If peak outpoot does not occur at the dip in plate current (resonance) and you have trouble with it "taking off" it is definately a neutralization problem. If your 8000 rig is an old-buzzard push-pull final, You probably need to use the old style push-pull criss-cross neutralization instead of the link style for optimum effectiveness.

                                                 The Slab Bacon

Not sure...but if you are talking about the exciter....807's like to see a half shield around them and/or have the tube socket mounted an inch or two below the chassis.  This helps with self-oscillation problems by providing some isolation between plate and grid circuits.  Many rigs using the 807 use this arrangement...examples: Knight T-50/Johnson Adventurer, etc.  I think Millen made the "half shields".  They can also be easily home brewed.  The sub-chassis mounting of the tube socket is easily accomplished by using some spacers and of course, cutting a larger hole in the chassis for the tube.

73,  Jack, W9GT

Jack,

It does sound like a neutralization problem but a similar symptom occurs with too small of an output DC blocking/coupling capacitor; see early Viking Valiants for an example.  One of my two Valiants was an early model with not enough "pf" in the coupling cap and on 160 and 80 maximum output did not correspond with plate current dip; increasing the size of the coupling cap took care of that issue.

73, Rodger WQ9E

I agree with Frank,--this problem sounds like a typical final that is not neutralized properly.
The 807 I have found is a good "all purpose" RF amplifier, but like any "hi gain" tube, can be
a "pain" to neutralize, especially if the circuit layout is not good.
                                           73, K1MVP  

Have you tried capacitive feedback, (very small amount) between the plate and the bottom side of
the grid input coil?
This usually works for me, and with B+ removed to the final, I use my GDO to tune for a "null".
The cap will be at B+ potential during normal operation, and can be a real pain to adjust for obvious
reasons.
                                                  73, K1MVP 

The Q of the tank (L/C ratio) could also be wrong if it isn't oscillating.

With proper circuit design and neutralization, 807's can easily work up past 50Mhz. I have an old GE pre-prog base station that uses 2 0f them in the final making something like 75 or 80w on 52.525.
(they are mounted in countersunk sockets)

I tried the half shield as it was easy but making the socket hole bigger at this point is not trivial so I would like the reasoning behind this before I venture forth.  I always thought the  below chassis sockets were to make headroom and never connected it with isolation.

There is nothing but wires in the lower inch of the tube so how does this work?

js

I was using .001 and tried .01 but it made no difference.

Thanks,

js

I opted for the inductive approach because as I see it, there is no bottom side of the grid input coil as it is grounded and not a balanced input.

Am I missing something?

js

               ""There is nothing but wires in the lower inch of the tube so how does this work?"

its late, so i might not get this right...  with the 807, placing the tube base  underneath the chassie keeps the plate from coupiling back into the grid(s) ckt....  IN the tube below, OUT the tube , above....there 's also some advantage with the tank ckt being closer to the tube -- less lead length; the cap closer.......   my under construction PP 813  rig has (2) 813s on a sub chassie; this is more to keep the vertical component a little lower; the 813 has a built in shield,  the grid tank ckt. is under the chassie, the output tank link is above the chassie.......   klc

Normally, the bottom of the grid coil would be held above dc gnd via a capacitor and the neutralizing
cap would be connected at this point, and adjusted for a bit of 'out of phase' feedback from the plate
to the grid circuit, thus stabilizing the amp.,if my memory serves correctly.

I just looked a typical inductive feedback from the 55 handbook of a 5763 driving a 6146 amplifier,
and was wondering if the phasing connections of each coil from the plate to the grid are not correct, could cause you problems?
As I mentioned before,--I have never used inductive feedback to neutralize an amp, so its just a
suggestion.
                                                  73, K1MVP

P.S.,--I JUST thought of something,--it seems to me I once had a problem with the "dip" not "coinciding"
with max power output on a 6146 rig, but only on the 80 meter band, and I tore my hair out,--it turned
out that I needed more inductance in the tank circuit and less capacitance,--the LC ratio was not
right. --Getting late,--gotta get some "shuteye".

There are unshielded wires coming out of the plate/grid structure of the tube to the base.  The shielding helps isolate those unshielded leads, which include the grid lead, from the plate.

I have always found 807's (beam power tetrodes) to be squirrelly.  The 802 pentode makes a more stable amplifier, but the power output is somewhat less than that of an 807.

I tried driving the 211 stage in my HF-300 rig with an 807.  Then switched to an 802.  The grid current to the 211, and plate current to the 802 were exactly the same as with the 807.  The rig didn't self-oscillate with the 807, but tuning is much smoother with the 802.

Look for 807's with the little silvery metal shield under the plate structure.  Those are the "good" ones.  Before they were discontinued, the manufacturers started making them on the cheap.  They omitted the little shield, and replaced the internal ceramic insulators with mica sheet.  In other words, the latter-day 807's are little more than a 6L6G with a 5-pin base and the plate lead coming out the top of the envelope.  They have a greater tendency to self-oscillate than the older classic 807's with the shield and ceramic insulation.

I had to neutralise my 802 stage, and the 807 when I used that tube.  It didn't self-oscillate, but on 40m there was enough feed-through to generate a noticeable back-wave on CW.

The ARC-5 Command sets used a pair of 1625's which are essentially 12V 807's.

The design sinks the tube bases below chassis level using a metal tube and socket arrangement. I wonder why? There is no doubt that this was done purposely. These things are full of tricks like this that nobody really pays much attention to.

Even so, they also grid neutralized the stages with a very small fixed capacitor which can not be more than a couple of pF.

The message is: With 807's you need to isolate the grid from the plate as best as possible and you do not need much capacitance to neutralize them. 

Mike WU2D

Mike

Well,  did the half shield help your problem?  That alone could give you the shielding/isolation that could be enough.  As Don 'KYV stated, the problem is with the unshielded leads at the bottom portion of the 807 tube.  When using these tubes in RF applications, you need to ensure that you have good shielding between the grid (input) leads and the plate (output) circuit.  Mounting the tube socket below the chassis is a convenient way to provide additional isolation by keeping the input wiring and tube elements associated with the input below the chassis.  This is certainly not an original idea with me!  This has been used in many rig designs over many years.  Good luck with your project.

73, Jack, W9GT

Try a resistor G1 to cathode to lower the input Z of the tube. Start with a 1 K to 10 k noninductive. You can go lower if you have enough drive. Put across the tube socket not the grid tuning inductor. fc

You might want to look into the design and mechanics of the old Harvey Wells TBS-50. As I recall it used the 807 and did 80 to 10 plus 6 and 2 meters. It might have been unstable, but we used one back in '61 and it always worked.

First of all, let me thank everyone for all the ideas.  Some I had already tried, some were new and tried and all taken seriously and appreciated.  It seems impractical to respond to all.

The shield did not seem to make any difference but I did make it permanent because it makes sense.

So far the only thing that made a difference was moving the osc coil to the under chassis.  Turns out this seems to have fixed the original problem.  The final now peaks at the dip in plate current.  The tuning seems perfectly normal now but what I hear when I swish the receiver around is chaos.... rough notes everywhere, even when using a crystal.

The osc and grid coils are now both on the bottom but at opposite ends of the chassis and cross oriented.

I wound a toroid for the grid thinking this would further de-couple but it made no difference.

Got so disgusted last night that I abandoned it, fired up the 8000 and had a chat with KYV hoping all would go away in the morning.

No such luck,

js

Holy Cow!

We need a photo now and then. I think everybody assumed that the grid components were completely shielded from the plate components.

All of the ideas discussed are in the noise compared to keeping the tank components isolated from one another. You can have grid and tank coils up top, but it takes darn good shielding to get away with it. Sometimes people used to lay the tube horizontal with a vertical wall shield.

 I call this the "emissionary position".

MIke WU2D

Mike,
       I was just thinking the same thing!! A couple of pictures might be worth 1000s of words here!!
 
                                              The Slab Bacon

Time to put that rig on a spectrum analyzer to see what "gives"--and see just how "dirty" the signal
is bandwidth-wise.
I once had a Johnson Navigator transmitter(that was given to me) that looked like it was loading up
great, but when I put it on a spectrum analyzer, the spectral display looked like a christmas tree.

It turned out some guy had "solid stated" the VFO, and I had no documentation as to what he did.

It still sounds like your amp is "finiky" but for what reasons is anybody`s guess,--poor ground returns?
ground loops? due to circuit layout, or  oscillator, amplifier interaction due to poor isolation.

                                                 73, K1MVP

Here is a picture  http://schmidling.com/807a.jpg

It needs a little cleaning up but the good news is that all the trash I was hearing was created by the S38 I use on the bench to listen. 

Note the Masonite panel is between the osc coil and the final so it is pretty well isolated.


js

p.s. I abhore smileys so work it out.

John,
I am not partial to using wood chassis,(as a rule)for 2 or 3 stage transmitters cause of the potential for "difficulties" i.e., ground return paths, shilelding problems between input and output, etc, etc.
The only time I have used a wooden "chassis" is to build a QRP 6AG7 oscillator rig, where stage
isolation is not a factor.

You might want to consider building using an aluminum chassis, or even an inverted singlesided
PC board mounted on standoffs, which would give a "groundplane" effect where you could keep
bypassing leads real short, etc, etc.
Just a thought, but if your rig works ok as is, you might just leave "as is" and leave well enuff alone.

                                           73`s Rene, K1MVP
       

I agree with Rene. Wooden chassis on multi stage rigs suck. Ground returns and ground loops are always a problem. And masonite offers absolutely no rf shielding!! Metal is always your friend!! :o

                                                    The Slab Bacon

Hi Jack,

Looks like a 6146 in the picture.  I thought you were using an 807.  The 6146 is a little different and doesn't exhibit the same characteristics with regard the the need for shielding around the base as the 807.  It has a metal base and shorter leads connecting to the tube elements.  The 6146 can also be squirrely though, as it is a high gain tube.  It is a good idea to use short leads where possible and keep the input and output circuits away from each other.  Keeping the input below the chassis and the output above the chassis provides good shielding and isolation.

I understand that you might be just "prototyping" and testing out your circuit on a piece of wood with a masonite panel, but a metal chassis is definitely the way to go for your permanent configuration!  Masonite offers no electrical isolation whatsoever, and a metal chassis-common ground/shield will provide a much more stable platform.

73,  Jack, W9GT

Guess I need to reconsider my war against smileys.  That picture was intended as a joke and the Masonite "shield" was to make sure everyone knew it was a joke.

It was my tinkering prototype to develop a layout.  The 807 turned into an 866 and I substituted the 6146 to keep going until I could get a new 807.  It was originally xtal controlled and then I got the urge to add the VFO which is the junk on the left.

A current pic is as http://schmidling.com/807b.jpg

Unfortunately, it looks a lot better than it works.

In case anyone is wondering why the 6CL6, it's an emotional thing.  When I got home from a hamfest with my 6AG7, I was appalled to find a metal tube in the box and to learn that is the only way they were made.  I am thinking of going to a 6L6.

Please also note the 1/2 shield for 807 per your suggestion.

I think I am close enough to solving the original problem or at least know where to concentrate to move on to another subject.... the VFO.  This is afield enough to start a new thread.

Again, thanks to all for the ideas.

js

Jack,
Your metal version looks mucho better than you plywood version, --But there are a couple of things
I question,--Is that that a grid current meter near the plate current meter?
If so I think, I would move that grid meter under the chassis to prevent any coupling between the
grid and plate current meter.
Also, is that variable cap in front of the oscillator, for grid tuning?--I would also move that under
tht chassis to minimize osc-final coupling.
Just a couple of thoughts, that might help.
                                                  73, K1MVP
 :)

Never occurred to me to put it under.  In fact, it never occurred to me that there would be a coupling problem with meters with RF chokes and bypassing .

>Also, is that variable cap in front of the oscillator, for grid tuning?--I would also move that under tht chassis to minimize osc-final coupling.

You're gonna love this... it's the VFO tuning. The grid tuning is underneath.  Both the VFO and grid coils are toroids so the coupling would be minimum between them.  The only thing on top is the cap and it's grounded and the tube is metal.

I suppose the whole VFO should be in a box but this is sort of my Mr Wizzard Science project and it's informative to see the results of not following the rules and fixing things as needed.

The final self-oscillation seems to have been fixed but now I have a self oscillation in the VFO.  That seems like an oxymoron but it's a very unusual one.  It is a ticking at about a 500 ms rate when I unkey. I have the oscillator on all the time and key the grid bias for a nice clean, chirpless key but the ticking between is nutty.  I fixed it with a 9V battery as a fixed bias supply.  Don't ask me how this solved anything but it did.  Now I need to find out what the real cause it and fix it right.

I also added a fixed bias supply for the 807 to cut it off when the key is up.  Works real nice.  Used a backward fil trans on an unused fil winding on the power trans.  Same trick I used on the 8000.

Lot's more problems to resolve yet but it wouldn't much fun if it worked the first time I plugged it in.  Things that do that are called rice boxes and are boring as hell.

>Just a couple of thoughts, that might help.
                                                 
Thanks.  Never too many thoughts,

js