Plate Meter Peaks on Resonance?

[Carl WA1KPD]

Finishing up a new transmitter and I am getting a peak on resonance instead of a dip.
So far everything in the schematic looks fine.

PI network output to 50 ohm dummy load and rather then a dip and load I get a peak. That shows up on the FS meter as max output.

Any ideas?

Thanks and 73

Carl KPD

[Steve - WB3HUZ]

Parasitic?

[ka3zlr]

Carl,

 Any neutralization issues missed, if it's required for your xmitter.

[w3jn]

Parasitic, resonance in the plate choke, neutralization.... post a photo of it and maybe we can make some suggestions.

A scope is your friend - and not one of those crummy Heathkit so-called "signal monitors" either 

[Carl WA1KPD]

Basic 6T9  xmtr.
Will post pics this weekend
No neutralization required. Maybe component placement.
They don't get much easier
Tnx

[Pete, WA2CWA]

Maybe you need a meter scale that reads from right to left

[AB2EZ]

Carl

As others have suggested, it sounds like you are getting some feedback from the output to the input... either inside the tube via inter-electrode capacitance, or outside the tube from the pi-network back into the input circuitry of the tube.

As you tune the pi-network to resonance, you get two effects:

a) the normal effect, in which the impedance of the tank circuit at the input frequency is maximized... which, for most amplifier designs, would result in a dip in the plate current

b) the undesired effect of the increasing r.f. output signal feeding back into the input of the amplifier... with a phase that causes positive feedback... which, in turn, causes the plate current to increase when you tune the pi-network to resonance at the input frequency. [Too much of this effect, and the amplifier will oscillate in the absence of an input... resulting in all sorts of bad consequences]

If the amplifier doesn't normally require neutralization... check to make sure that the pi-network's tank coil is not coupling too strongly into the input circuit of the amplifier (maybe rotate the input tuned circuit inductor, if there is one, by 90 degrees, if possible). Also, if there is a screen grid, make sure that it is properly bypassed to ground. Likewise if there is a suppressor grid. Make sure that the filaments are properly bypassed to ground.

Best regards
Stu

[w3jn]

I see that schematic calls for a center tapped fil xformer, which are fairly hard to come by.  Stu's last comment is pertinent - if you didn';t use a CT xformer (or even if you did) make sure both legs of the fil are bypassed to ground at the tube (not at the transformer!).  If the fil lines are floating above ground, RF-wise, they're gonna act like an antenna and create undesireable feedback.

You should also use a single point ground for all the bypassing associated with the final section - screen bypass, plate bypass, fil bypass, should all be grounded at the point that the cathode is grounded.

[AB2EZ]

Another thing to check:

Looking at the schematic, I wonder if the cathodes of both tubes are well-enough bypassed. One would think that the cathodes would be firmly grounded by the key... but the wiring to the key may provide a loop antenna to pick up some of the r.f.

I suggest that you beef up the capacitance to ground on the cathodes (its hard for me to read the schematic, but if the bypass capacitor is C6, then 470 pF may be too little for operation on the intended frequencies). I would suggest that you use .01 uF (10,000 pF)  to ground at the cathode of the output tube. If the feedback is coming from the keying line, that should provide a big improvement. You might also consider a separate .01 uF capacitor to ground at the cathode of the crystal oscillator.

On my DX-20, which has a lot of similarities to this transmitter, the cathode of the output tube is bypassed to ground with .005 uF (5,000 pF), and the non-grounded side of the filament is also directly bypassed to ground with .005 uF.

Stu

[Carl WA1KPD]

I see that schematic calls for a center tapped fil xformer, which are fairly hard to come by.  Stu's last comment is pertinent - if you didn';t use a CT xformer (or even if you did) make sure both legs of the fil are bypassed to ground at the tube (not at the transformer!).  If the fil lines are floating above ground, RF-wise, they're gonna act like an antenna and create undesireable feedback.

You should also use a single point ground for all the bypassing associated with the final section - screen bypass, plate bypass, fil bypass, should all be grounded at the point that the cathode is grounded.

Hi Stu and Johnny

I did not use a CT xfrmr so I will add in the bypass and check the others as sugessted.
I appreciate your ideas and will let you know what happens.

[WU2D]

Hey Carl,

I wonder if you are getting enough grid drive to the "final"? Can you measure the grid current?

Mike WU2D

[WD8BIL]

You can "create" a centertap with a pair of resistors. 220 Ohms +/- from each filament leg to ground will do it. 6.3v/200= .032 amps     .032 x 6.3 = @.2 watts

Then the differencial cap shown in the scitzo will be less likely to react with the windings inductance.

[WU2D]

Carl,

I remember that circuit from one of my old Handbooks, I think 1973. I still suspect a lack of drive from the Pierce oscillator. you can measure the grid current by lifting the 27K and inserting a 1 mA meter. Are you getting any power out?

Are you using the transmitter right on the crystal frequency? If so you could try to replace the RFC in the plate circuit of the oscillator with a low Q tuned circuit like a variable inductor roughly resonant at the frequency. For 80M you could try a 100 - 140 uH variable inductor for instance. If it is too small, put a small cap across it like 22 pf or 47 pf and grid dip it to 80M.

Mike WU2D

[Carl WA1KPD]

You can "create" a centertap with a pair of resistors. 220 Ohms +/- from each filament leg to ground will do it. 6.3v/200= .032 amps     .032 x 6.3 = @.2 watts

Then the differencial cap shown in the scitzo will be less likely to react with the windings inductance.

I cant get to the rig until tomorrow.

How about grounding one leg of the filament xfrmr and bypassing the other leg with a cap?

[w3jn]

You can "create" a centertap with a pair of resistors. 220 Ohms +/- from each filament leg to ground will do it. 6.3v/200= .032 amps     .032 x 6.3 = @.2 watts

Then the differencial cap shown in the scitzo will be less likely to react with the windings inductance.

I cant get to the rig until tomorrow.

How about grounding one leg of the filament xfrmr and bypassing the other leg with a cap?

Best thing to do is to run both leads to the socket then bypass both leads to ground.  That way you don't have 60 hz AC currents running in the chassis to trip you up.  Twist the wires together to attain a self-shielding effect.

[Carl WA1KPD]

Thanks to all who offered ideas and suggestions on this simple rig. Many of them helped but did not eliminate the problem. However they did lead me down a train of thought and in looking at the Handbook picture of the rig I noticed that the socket was clear of any parts. So I drilled and blasted the circuit doing two things.

1. I kept all of the osc circuit on the lower number pins side of the tube where the osc elements are located.

2. Like the Handbook, picture, I kept all components outside of the socket. In the two cases where pins were connected across the socket via a component I routed them around. Note in particular the small yellow capacitor buried across the tube in the "before" picture. It couples the osc to the PA. Moving that one made a real big difference.It is now located at the bottom of the picture with relatively long leads, but the tube is happier!

3. I moved the rf choke on the output of the PI net from the coax connector to the loading cap. That too made a big difference.

Thanks again to all who offered suggestions. and hopefully my experience will help someone else along the way.

Happy new year and maybe I will wrk you on SKN tonight.