Strange noise out of an 807

[KC2ZFA]

Hello all,

I've been testing my hb exciter (schematic here: http://amfone.net/Amforum/index.php?action=dlattach;topic=26207.0;attach=23483;image).

It appears to be working fine with a 40M crystal doubling in the oscillator plate
to 20M. I'm sure I can get it to work on 30M and on 17M (the oscillator plate
doesn't tune past 18.5 Mc) with appropriate crystals.

Here is the problem: when using an 80M crystal to tune up on 80M I get a very
low current dip in the PA whatever the output link loading. Also, when I vary the plate
tune cap I get sudden jumps to higher current on either side of the dip. When that
happens the tube screeches and turns blue and I also lose oscillation.

I'm at a loss here and would greatly appreciate any ideas from the collective.

Thanks, Peter

ps. I can't check it on 40M because the oscillator plate coil doesn't tune all the
way up there...I need to tap it, bit that can wait till 80M are sorted out.

[K1ZJH]

Could be a bad tube, or parasitic oscillation on that band. I didn't see a parasitic suppressor on the 807 plate?

[WQ9E]

Maybe RF from the final getting back into your oscillator when operating straight through?  Do you have a 160 meter crystal you can try and let it double in the oscillator plate?

[w3jn]

The 807 is oscillating, perhaps a parasitic or via feedback to its input, or feedback to the oscillator.

Couple of things to try - I don't see any bypassing of your power input leads, including the fils.  Those power leads are acting like antennas and they're only a few inches from the plate coil.  Add plenty of disc caps from every input to ground, with very short leads.  You did a great job twisting the wires, using shielded wire, etc - you need to keep the RF from entering the box thru the power leads in the first place.  The fil leads should be bypassed to ground at each tube socket also (perhaps you've done this and I can't see it).

Also, you might try enclosing the base of the 807 and its associated circuitry in a minibox.

Great workmanship, and looks great though!

Finally, as Pete said, a parasitic suppressor in the plate of the 807 would be a good idea.

[WD5JKO]

Peter,

   I second what the others have said, and also a big Wow on your great construction skills.

I would like to add a few comments.

This two stage exciter has a lot of gain. One area to consider addressing is to tame the gain when the thing is being tuned. The 807 screen fed by a 50K from 600v will allow the screen grid to rise towards 600v when the grid drive is low. This is why many similar exciters use a clamp tube such that a rise in screen voltage when drive is low does not cause problems. In your circuit, maybe take a 100K 2W or greater wattage resistor from screen to ground. With this in there, the maximum 807 g2 voltage will be 400V instead of 600V. A bottom mounted clamp tube is an option too.

You could take a NE2 neon lamp and stick through a rubber grommet up on the front panel. Tie both leads together, and extend to the 807 plate area, but not touching. The idea is to have the bulb as a tuning indicator where max 'orange' color should occur at plate current dip. If VHF parasitics occur, the bulb color will change to violet.

Nice project!
Jim
WD5JKO

[KC2ZFA]

Thanks guys...very precise advise that is paying off !

So I put a parasitic suppressor (see pic). I also bypassed the 807 cathode
at the socket with a .001 (the screen and filament is bypassed with
.01's, see pic). The 6AG7 screen and filament are bypassed with .01's
see previous pic). I tested with the cover to the box on. Using a 7173
xtal and tuning to 14346 works fine (actually smoother than before).

The 807 now does not speak to me on 80M  The plate tunes to a
dip (no loss of oscillation on either side of the dip) but I can't get it to
load over 30mA. I'm using a 25K grid leak and 49V of fixed bias. I set
the drive for 4mA grid current. Can it be I put too much negative volts
on the grid ? In other words, is that 25K too high ? I used a 25K because
the Millen exciter uses 24K with 45 volts fixed bias.

One other thing I noticed is when I key it I see the plate current go above
whatever it loaded to before and then settle back to that value. Is this
indicative of something ?

I should also confess that the exciter is grounded to the power supply
which in turn is grounded to the wall socket ground (the water pipe
for an RF ground is too far from where I'm testing this). Can this be
a contributor to the low loading on 80M ?

If anyone has any extra 160M crystals (around 1940 Kc) for sale please
PM me.

I'm getting close !

73 de Peter

ps. I will be doing the 100K from screen to ground next. And I've started on
boxing the stuff below the 807 socket.

[WA2IXP]

   Looks excellent Pete, fine job. Maybe you need a few more turns on the output link coil to increase the loading. Add 3 or 4 more turns with hook up wire wound in the same direction and see if that helps. Test it with a coax fed dummy load.  If the screen is fed thru a series dropping resistor you may have the effect you mentioned with the plate curent. Also the power supply caps can hold the plate voltage at the unloaded value momentarily after the load is applied. Don't think its a real problem. I think your bias is ok. The neon light mentioned will tell of parasitics. 807's are not so good in this area. also as mentioned operating straight thru can be a problem without neutralization. Your circuit may have it, I don't know. Keep up the good work . All projects need a few bugs worked out. Its refreshing to see you and others taking an interest in building and understanding whats happening. A quickly fading skill in ham radio and many other things in our world today.

                                                                                                Jay-

[KC2ZFA]

Maybe you need a few more turns on the output link coil to increase the loading. Add 3 or 4 more turns with hook up wire wound in the same direction and see if that helps. Test it with a coax fed dummy load.

thanks for reminding me of the link Jay. I'm using B&W JEL coils and they
have only 3 turns over the end of the plate coil that gets fed with the B+. So
that's why I put in the 350pF air cap from link to ground in order to have some
loading control. So today I experimented by paralleling it with 250pF and 100pF
caps and I got it to load higher (but only 25 mA max) and then I just grounded
the link and that brought the plate current to 30 mA at the dip. That's on 80M
with 80M crystal.

I don't want to butcher the B&W coil so I'll try fashioning my own 80M coil with
the ability to change the link and see what happens.

OTOH, when doubling in the oscillator to 20M (with 40M crystal) I can load
the 807 to whatever current I want. In this case the dip gets broader as loading
is increased and I think this is correct.

And thanks to all who helped me get this far !

73 de Peter

[kc2hgc]

Hi Bro very nice construction job,

Just a though try useing a bigger loading cap two section 465 uuf brodcast type and some additional caps. Also maybe a 2.5 uh plate choke..

73 Steve...

[KC2ZFA]

Just a though try useing a bigger loading cap two section 465 uuf brodcast type and some additional caps. Also maybe a 2.5 uh plate choke..

Geia sou Steve !

I tried a 2.5 mH rfc and the parasitic returned when operating on 80M  ...did
you mean a choke like an Ohmite Z50 ?

I'll try making an 80M coil with more turns on the link...

Maybe the B&W JEL links are not meant to feed coax

73 de Peter

[WU2D]

Very nice construction on the Millen clone! Wow what an effort. If you can't find it - build it! I think bias scheme is a 2.5 mH choke off the grid into a 25K resistor to the bias supply Have you tried a bypass on the 24K so only the choke is hot for RF?

The 807 in Class C Telegraphy according to the book, needs -45 VDC on the grid with 750V on the plate and a couple hundred on the SG and you should get 50 Watts out. If you run lower voltage, the output will be lower. But adjust the bias from -30 to -90 and you will see a place where the tube is most efficient and keys the best.

Even though the Millen apparently did not neutralize the thing, it would be a nice experiment and it would make the stability and keying better. Since you are building your own coils, this simply requires a tap on the output coil for bypassed B+. This does not have to be a center tap. In fact you only need a bit of feedback for the 807 so you can come up just a few turns off the bottom of the coil. The tuning cap remains across the whole coil. The bottom of the coil is then brought to a small trimmer and though a ceramic feed thru back to the grid.       

Mike WU2D

[KC2ZFA]

I think bias scheme is a 2.5 mH choke off the grid into a 25K resistor to the bias supply Have you tried a bypass on the 24K so only the choke is hot for RF?

that's what I have wired in...the bypass (.01) is between the choke and the 25K
and the bottom of the 25K is fed the -49 fixed bias. The bias is off of a 40V
tap on the plate xfmr and has a 3K bleeder.

the problem I'm having is that with the 80JEL coil (3 turns link) and an 80M
crystal I can't load to more than 25-30 mA into the dummy load.

Even though the Millen apparently did not neutralize the thing, it would be a nice experiment and it would make the stability and keying better. Since you are building your own coils, this simply requires a tap on the output coil for bypassed B+. This does not have to be a center tap. In fact you only need a bit of feedback for the 807 so you can come up just a few turns off the bottom of the coil. The tuning cap remains across the whole coil. The bottom of the coil is then brought to a small trimmer and though a ceramic feed thru back to the grid.       

good idea, but the geometry of the chassis setup will involve a long wire (at least
6") to go from the coil to the grid...can't be good ?

Here's the rest of the rig that's waiting to be fed by this exciter (push-pull 812A's
which will also need some ironing-out of kinks once I can drive it):

[w3jn]

Try a 100 pF or so variable cap in SERIES with the output to the dummy load, or from the cold end of the link to ground.

[KC2ZFA]

Try a 100 pF or so variable cap in SERIES with the output to the dummy load, or from the cold end of the link to ground.

I do have 350pf variable cold-end of link to ground...on 80M that only loaded the PA
to 20 mA max (nice symmetric dip to this low value).

Then I paralleled a 100pf doorknob to it and the dip now is in the range of 25
to 30 mA (QRO !). The JEL has a 3 turn link over the B+ end of the plate coil...

I can't figure out what's going on...can it be that RF is messing up the mA-meter ?
I have a .01 (orange drop) across its terminals.

I'm waiting on three 160M xtals to double to 80 in the osc and see what happens.

73 de Peter

[WU2D]

good idea, but the geometry of the chassis setup will involve a long wire (at least
6") to go from the coil to the grid...can't be good ?

The ARC-5 transmitters (a MOPA with a tank coupled VFO) which use two 1625's uses rice (grid) neutralization and they have a long wire simply run along the chassis. It is a bridge. Once balanced the wire does not matter. This is what the neutralization cap looks like in the ARC-5. That can not be more than a pf! You do not need much feedback for neutralizing beam power tubes. Many argue you don't need any. But the point is they still did it to reduce instability, backwave and to reduce chirp in the keying. My BC696 chirps like a bird without it.

[KC2ZFA]

This is what the neutralization cap looks like in the ARC-5. That can not be more than a pf!

hmmm...I'll try this...where does the neut cap sit ? near the plate or near the
grid ?

[KC2ZFA]

The saga continues !

I wound a new coil on a plug-in form. 40T of 21 awg, 2" long, 1.5" diam,
same as the 80M JEL coil except its on a bakelite form.

On top of the B+ end of the coil I wound 6T of insulated hookup wire
(copper, looks like 21 awg too).

I removed the 100pf I had paralleled to the 350pf variable on the link to ground.

So now I can load up to 55 mA on 3825 Kc but the tuning cap is almost at
min (see second pic) and as a result I get an almost one sided plate tune
dip...so I guess I need to remove turns from the plate coil, yes ?

One other thing I noticed is that the link area of the coil got warm after a few
minutes of key-down. Is that indicative of something out of kilter ?

It seems the 3T link on the B&W 80JEL is wholly inappropriate for coupling to
50 ohm coax.

73 de Peter

[WA2IXP]

Your on the right track now. Take several turns off the plate coil and if the link gets more that a little warm use heavier wire to cut down the losses. That is a few watts of heat thats not heating your antenna.
   An old handbook has charts showing the aprox. amount of tank cap needed for your operating conditions. Not to critical. Just estimate the required cap by eye, ie: cap 1/2 open and remove turns to get resonance point in the ball park. A grid dip meter with a rx to monitor actual frequency while all is hooked up with power off would be helpful. An antenna anylalizer hooked to the output with a resistor from plate of the tube to ground works well. (POWER OFF).  Say 600v/.075A =8000 ohms. Should be able to tune plate and load caps to get a 1:1 swr on meter.
                                                                                                                                   Jay-

[WU2D]

That coil looks almost exactly like one that I use on my 6AG7 6L6 rig. I get 25 Watts out And if I strap a 6DQ6A into the socket, I get close to 40W out. I also use about 40T of # 22 enameled with 5T coupling coil of # 12 enameled also wound over the bottom (B+ RF Cold) third of the coil directly on top of the turns. So my link is fewer turns but perhaps more coupling and heavier. Having too many turns on the coupling is OK if you have a loading cap from the cold end of the link to ground... I think that I use a 100 pF tuning cap which is why my coil came out similar to yours. So yes, I think you need to take some turns off and use a heavier link with tighter coupling and perhaps get some coupling control.

Mike WU2D

[KC2ZFA]

the heaviest wire I have is 14awg so I used it to replace the puny link...

I couldn't get the new wire into the pins from the inside of the coil form
so I had to do it from the outside (see below)...I hope this is not an
issue.

In any case, max loading I can get on 80M is 58 mA (about 700V on the plate).
On 20M I can go up to 85 mA (with the 20JEL coil and doubling from a 40M
xtal).

I should be able to load up to 100 mA...I'm guessing that I have either
a) low screen voltage (my dropping resistor is 50K)
or
b) the bias is incorrect (I have 25K grid leak that is being fed -49V
fixed bias).

You guys have been very generous with your help. Thanks all.

73 de Peter