Homebrew 40M AM Tx Output Power Problem

[KA2DZT]

Thanks for the encouragement Fred. Iv'e put a lot of time into this rig and I really don't want to give up on it. I'll run the test above and see what I get.

I have a 47ohm 1/2w resistor across each amp meter. This should be ok I think... I might have made a mistake in the build process so trying to think of anything that might be the issue.

Joel  

If your current meters are direct reading, meaning the grid meter is about 5ma full scale and the plate meter is say 150-200ma full scale you should not  have any resistors across the meters.  Resistors across the meters will alter the readings.  You can put a cap across the meters but you don't need them.

To add to what Stu suggested,  you have to make sure your grid tank circuit tunes to 7mhz.  You should see max grid current somewhere within the tuning range of the tuning cap, not at min or max capacitance.  Next you have to be sure it's tuned to 7mhz and not a harmonic (14mhz).

Having said all this, the issue you're having is the power drops after a few minutes of operation.  Your problem is not a design issue,  you're looking for a component that is failing under load.  If it was a design issue you wouldn't see higher power at first then dropping off.  You wouldn't see the right power at any time.

Fred

[K1JJ]

Maybe this has been covered, but have you looked at the grid DC and cathode DC levels to see if there are any increases (charging effects)  over time?

The reason I axe, is that over the years I've had several plate modulated rigs using fixed/ grid leak bias that did the same thing - the power output slowly went down after a minute or so.  The grid leak bias circuit (actually the fixed bias capacitor)  kept charging higher and higher with no end in sight. In my case, it took a diode in series with a 10H inductor coming off the fixed bias supply to keep the charging effect from occurring. (blocking effect) The fixed bias supply DC filter cap was simply charging up from rectified RF getting into it. The Tron suggested this fix to me and it worked like a W1GAC charm.

I see you have a different biasing arrangement and if no help to you, maybe this suggestion can help someone else who comes across this problem.

Tom, K1JJ

[KA2DZT]

Looking at your schematic, I agree with Stu, that 5.7uh coil in the osc. plate circuit is a mistake.  It's actually in parallel with the grid tank coil because of the large coupling cap between them.  Replace it, like Stu suggests, with another RFC.  Try not to use the exact same type you have in the cathode circuit to try to avoid additional oscillations at some other frequency.  If you don't have another RFC you can just use a resistor, say 10K.  Check the power dissipated in the plate resistor.  You may need something greater than 2 watts.

Fred

[w4gon]

Thanks Stu! Your detailed explanation was much needed and greatly appreciated! I've printed it out.

I'll switch to an RFC in the oscillator plate and dial in the coupling stage tank values. I'm using an inductor on the oscillator plate now because I'm out of RFCs. I'll have to "borrow" one from... I wound the tank inductor myself so I can redo it for whatever value if needed. The bias explanation was awesome and with it I'll be able to confidently dial in the 6146 bias now.

Thanks again
Joel 

[DMOD]

Is that 2.5 mH ferrite plate choke?

Ferrite core chokes tend to heat up too much and saturate.

Try a 425 uH air core choke.


Phil - AC0OB

[AB2EZ]

Joel
et al.

Recall that it takes two things to make an oscillator

1. Positive feedback

2. Saturation

The positive feedback is controlled by a combination of the 126pF and the 10pf capacitors... working in conjunction with the crystal ... to select the frequency where the phase shift around the feedback loop is 0 degrees (i.e. equivalent to 0 degrees)

The saturation occurs because the plate current of the 2E26 flows through the impedance of the tube's output load... to produce a plate voltage drop.

If the output load impedance that the 2E26 is looking into is too low (as it is with the 5.5uH inductor in parallel with the 7.5uH inductor... and not enough parallel capacitance to resonate the output circuit)... the saturation will occur at a relatively high 2E26 RF output current. For illustration: at 7MHz and 200mA of peak RF plate current, the peak of the RF voltage drop across the output load is only j142 ohms x 0.2A = 28 volts).

The high value of RF plate current at saturation goes along with a high value of grid-to-cathode voltage. For illustration, with the transconductance of the 2E26 being only 3500 umhos (which is much lower than that of a 6CL6)... and with at least 200mA of peak RF plate current... the peak RF  grid to cathode RF voltage will be greater than 200mA / .0035 mhos = 57V.

This large 2E26 grid to cathode voltage will correspond to a very high crystal current. Also, the plate dissipation associated with the high plate current will cause problems for the tube.

By making the changes needed to present a resonant output load to the 2E26, the peak RF plate current of the 2E26 oscillator will be reduced (i.e. the oscillator will saturate at a much lower amplitude RF plate current), and the amplitude grid-to-cathode voltage will be correspondingly lower... leading to a correspondingly much lower crystal current... and therefore much less crystal heating.

Stu

[KA2DZT]

Stu,

The circuit would work better if the grid tank was in the oscillator plate circuit but probably the tuning cap is grounded to the chassis.

In both my HB 6146 and HB 813 rigs the grid tuning caps are isolated above ground.

Joel, in his last post, said he didn't have another RFC, so he replaced the 5.7uh coil with another inductor of some unknown value.  I suspect this is not going to work any better.

Possibly if he used the slug tuned coil and put the right size cap across it to have it resonant.  Problem is the tuning of the grid tank will greatly affect the resonance.  To reduce these affects the coupling cap between the two tank circuits would have to be greatly reduced.  Right now I think there is a 500pfd coupling cap, reduce it to 100pfd.  The best way to do it is to use link coupling between the two coils.

Fred

[w4gon]

Hello Stu, Fred and rest,

I've made all the changes suggested. Here is the latest schematic. After removing the oscillator plate inductor the oscillator output is way down. I'm measuring 16vp-p out of the oscillator now. I was at over 100v before. With the 6146 online I'm not getting any grid current measurement on the meter. I'm still loosing power also. I put a 10k resistor in place of the plate RFC to make sure that it wasn't due to some interaction between the two RFCs, but same low output from the oscillator. I lowered the blocking cap to 100pf mica and the coupling cap to 1nf mica. I have the tank set to 60pf and 7.9uh which puts me right around 7.290MHz and 364 ohms. What am I missing here?

I disconnected B+ to the plate and screen of the 6146. Grid current went up to about 2ma and stayed there. Reconnecting B+ drops grid current to less than .5ma with accompanying eventual loss of power. I removed the shunt resistors on both meters. The transmitter is a few inches closer to my open window now. 

Joel

[AB2EZ]

Joel

You said you removed the 5.7uH oscillator plate inductor... but what did you put in its place? You need to replace it with a larger inductor (100uH or more) to feed the DC plate current, but to block the 7MHz RF component of the 2E26's plate current. This inductor needs to be rated to handle the 2E26 plate current.

Alternatively, as Fred suggested, you can just move the 7.5uH inductor and the 60pF variable capacitor to the place where the 5.7uH inductor was. That way, the 7.5uH inductor will provide the path through which the DC plate current is fed to the 2E26.

Are you able the resonate the oscillator plate load by adjusting the 60pF tuning capacitor to produce an oscillator output voltage swing that is large enough to produce 2.5 mA of 6146 grid drive?

[w4gon]

Hi Stu,

I put a 2.5mh RFC in its place. I attached the latest schematic as a pdf on my last post since I've made a number of changes.

No, I am no longer able to adjust the tuning capacitor to produce anything close to 2.5ma on the 6146 grid. I'm a bit confused as to why this is the case. I think I'll try moving the tank over to the oscillator plate. I'll have to use a fix value capacitor and rely on the inductor for tuning.

Joel

[AB2EZ]

Joel

Okay

Moving the tuned circuit to the output of the oscillator will allow you to avoid the extra 2.5 mH RFC.

Resonating the tuned circuit is critical. As you approach resonance, the 6146 DC grid current will rise dramatically (following the rise in amplitude of the RF component of the oscillator's output voltage)

Stu

[KA2DZT]

OK FB,  I studied your schematic,  everything looks right.  You say you have the grid tuning set at the values stated.  Are you tuning and peaking up the grid tuning.  Do this by peaking up the grid current to max current.

With the B+ removed from the 6146 you have 2ma grid current and it stays steady.  Reconnecting the B+, you must retune the grid back to the max current, are you doing this??

You must tune the 6146 tuning to dip the plate current which may cause you to have to retune the grid again.  Dip the final with light loading to your 50 ohm output watt meter dummy load.  If you then increase loading, you must go back and retune the grid, re-dip the final again.

You should reach some level of power output (whatever 15-20 watts) that stays steady.

I think the grid current drops when you reconnect the B+ because of the cathode bias resistor.  Try your testing with the 6146 cathode grounded.  Just watch the plate current doesn't exceed 130 ma.  You must due all the tuning quickly.

You said earlier that you didn't have another RFC for the osc plate load and put an inductor for the plate load.  What is the final component you're using, coil or RFC??  This part is critical to getting enough voltage (power) to the grid of the 6146.

Fred

[w4gon]

Ok, thanks Fred. Shorting out the cathode resistor is indeed allowing me to generate about 1.2ma of grid current with the B+ connected now. Why is this the case? I'm right around 1.4ma of grid current with B+ removed on the 6146. I followed your tuning process and that's the highest that the grid current will go right now. With B+ connected, grid current stayed at a solid 1.2ma while I watched my power output drop from 20W all the way down to 7W! The cathode resistor is off the "bad component" list since I have it shorted out right now...

One thing I'm noticing about the tank in the coupling stage is that the coil (7.15-13.8uh range) is not giving me the peak in grid current on either side of its sweet spot of 7.9uh (per the resonance calculation). Grid current rises as a go lower in inductance down to its limit and I'm sure I could get higher grid current if I could go down to say 5uh... This doesn't make me feel warm and fuzzy inside about this stage.

The RFC in the oscillator plate is the one I had in the 6146 grid stage. I "borrowed" the grid RFC from the Globe Scout and put it in the grid circuit of this rig. I then moved the spare RFC over to the oscillator plate. So I'm using legit RFCs on all the stages.

I'll try moving the coupling tank over to the oscillator plate to get rid of the RFC and see if I can get good resonance with the 60pf & 7.94uh values.

Disclaimer: No classic vintage rigs where destroyed in the making of this HB rig.

Joel

[AB2EZ]

Joel

Grid current flows when the grid voltage is positive with respect to the cathode voltage.

With the 6146 cathode resistor in place (now bypassed), the cathode voltage is positive with respect to ground by an amount equal to the cathode resistor value (392 ohms) x the average cathode current that is flowing.

If the average cathode current is, for example, 90mA... then the cathode is positive with respect to ground by 0.090mA x 392 ohms = 35.3V.

Therefore, with the 392 ohm cathode resistor in place (not bypassed) and with 90mA of cathode current flowing (i.e. the plate voltage supply and screen voltage supplies are connected), the grid-to-ground amplitude of the RF grid voltage (from the oscillator output) has to be 35.3V higher to produce the same grid-to-cathode voltage (and therefore the same average grid current) as when the plate and screen supplies are disconnected (no 6146 cathode current).

Separately, the 100pF capacitor between the plate of the oscillator and the top of the tuned oscillator tank circuit... that is in your latest schematic... is likely preventing the tuned circuit from providing a resonant (high impedance) load on the oscillator's output... and, therefore, preventing you from getting sufficient peak grid-to-cathode voltage on the 6146. Its reactance at 7MHz is: 227 ohms.

Relocating the tuned circuit to replace the 2.5mH inductor that is feeding the plate of the oscillator (or even just increasing the 100pF capacitor value back to a few nF) should allow you to produce the necessary high impedance load for the oscillator

Stu

[KA2DZT]

When you re-connected the B+   the current thru the 6146 created a positive bias on the cathode which put the grid more negative.  This condition requires that you have greater drive voltage, which you don't have.  That's why the grid current dropped to .5ma.  The grid of the 6146 only draws current on positive peaks of the drive voltage.  This grid current creates the neg. grid bias thru the grid resistor.  This pushes the 6146 into class C.  Grid voltage has to be high enough to get about 3ma of grid current.  The 6146 only conducts on about 70 degs of the peak voltage.  The 6146 only puts out pulses of energy into the pi network.  The flywheel effect of the output tank circuit creates the output sine wave to the antenna.

So, you can see that you need more drive than you are now getting.   One thing you need to look at is the coil you are using in the grid circuit.  This coil along with the tuning capacitor must resonant on frequency.  If you are using a slug tuned coil this could be part of your problem.  The grid coil needs to be an air coil of the correct shape.  Something about 3/4"- 1" in diameter is about right,  close space winding with maybe 20ga enamel wire.  This type of coil will have a greater Q that any slug tuned coil.  The Q of the coil is what determines the peak voltage created by the grid tank circuit.

Once we get the drive working correctly it seems you still have the original problem, the power dropping off.  Are you sure your power meter and dummy load are working right.  If you are sure the meter is OK and reading correctly,  I would next look at the plate coupling cap.  If this cap is not up to the job for a plate coupling capacitor it will quickly overheat and probably change value.

You mentioned moving the grid coil to the osc plate with the right cap.  Problem here is that a small cap in the osc. plate tank circuit may also get hot,  an air cap will work better, if you have one.

Fred

 

[KA2DZT]

I think another design change that would again work better is where you have the grid tank circuit.  A better design is to have the grid tank coil and cap in series with grid RFC and grid resistor.  This requires that the grid tuning cap be isolated above ground.  Not sure what type cap you're using.  Why is this design better you ask, I think the grid resistor loads down the grid tank thru the large coupling cap that you have going to the grid.  This lowers the Q of the circuit which lowers the peak voltage developed across the coil.  Not sure how much the RFC lessens this.

So you would have the grid coil and parallel tuning cap connected directly to the grid, the 100pfd coupling cap (from the osc. plate circuit) would also connect to the grid.  At the low end of the coil you would have a by-pass cap to ground with the RFC connected there also.  Another by-pass cap to ground at the low end of the RFC.  Then connect the grid resistor to the low end of the RFC and meter to ground.  Problem is you need to isolate the tuning cap frame above ground.

With this design the grid resistor is no longer across the coil, it is in series with it which removes any loading on the grid tank circuit.  My HB xmtrs are designed this way and it works perfect every time.

Fred

[w4gon]

Thanks Stu and Fred for the detailed explanations, this makes sense to me now. I moved the tank circuit into the oscillator plate and played around with the values a bit. I'm now getting plenty of drive and 3ma on the grid without issue. Problem is that I am using a slug tuned coil and a mica cap... I'll have to wire an air coil of 1" diameter and see how I can isolate a variable cap. I think I have something that might work...

That's correct Fred. After all of this tinkering I still have the original issue. I am confident that the power meter and dummy load are working. I've been using these for a number of years on different gear and all looks good there. The plate coupling cap is the 1nf 6kV coupling capacitor from the Globe Scout which I replaced the old one with when hunting for the source of the problem. I've attached the latest schematic for reference. I'll be working on the correct components for the oscillator tank.

Joel

[AB2EZ]

Joel

Try increasing the 2E26 cathode-to-ground capacitance to 200pF... by tacking on an additional 100pF in parallel with the existing capacitor.
 
This should reduce the feedback... and cause the oscillator to saturate with a smaller plate current RF component when the tank circuit is tuned close to resonance.

That may reduce the heating of the crystal and the 2E26 plate by a significant amount... while still allowing you to obtain sufficient grid drive for the 6146.

Stu

[KA2DZT]

Are you sure the power supply voltage are not dropping under load??  What about the filament voltage??  How did you wire the filament??  Should have one side grounded at the xfmr and only one lead going to the tube with the other filament pin grounded at the tube socket.  Also the filament lead needs a by-pass cap to ground right at the filament pin.  Use at least a .01ufd by-pass for the filament.  This is also needed at the oscillator socket.

Your power loss is starting to get puzzling.  You have to check all the components around the 6146 (this is assuming you're getting a steady 3ma grid drive.

Still wouldn't hurt to try some other coupling cap just to be sure.  Make sure all the connections and solder joints look good.  Checked screws holding the pi-network caps (BTW these screws in the RF tank circuit should be brass).  Your schematic shows a rotary coil in the pi network, make  sure everything is making a good tight connection.  Make sure the 52K screen resistor is good.  Loss of screen voltage level will cause the power to drop.

Check the parasitic coil and resistor.  Your problem could be a parasitic but I don't think it would cause the power to drop after a few minutes of operation.

Only other thing left is the plate RFC.

Fred

[KA2DZT]

Joel

Try increasing the 2E26 screen-to-ground capacitance to 200pF... by tacking on an additional 100pF in parallel with the existing capacitor.
 
This should reduce the feedback... and cause the oscillator to saturate with a smaller plate current RF component when the tank circuit is tuned close to resonance.

That may reduce the heating of the crystal and the 2E26 plate by a significant amount... while still allowing you to obtain sufficient grid drive for the 6146.

Stu

Stu,  Do you mean the cathode to ground capacitance??

[AB2EZ]

Fred

Yes... I meant the cathode to ground capacitance. I will correct this in my earlier post.

Thanks
Stu

[AB2EZ]

Removed duplicate

[DMOD]

Hi Joel,

You have received a lot of good info from Stu and others.

Here is my $0.02 in schematic form.

Regards,

Phil - AC0OB

[ka1tdq]

I had the same problem with a homebrew project some years ago. Timtron and Steve helped me troubleshoot it while we talked on the air. The symptom was the same.  After transmitting, power would drop off dramatically after maybe 10 seconds on my dual 3-500 linear amplifier.  I would have to re-tune the plate cap to keep power up while I transmitted.  

The problem was the plate coupling cap.  I had used a cheap, TV type one and the heat would cause it to change value during key down periods.  I replaced it with a good doorknob type and the problem cleared.

I know you're using a known good one borrowed from another rig, but I thought I would still just throw the idea out there.

If re-adjusting the plate cap brings your power back up, this might be the problem.
 
Jon

[AB2EZ]

I like Phil's (DMOD's) changes... including the changes in design of the oscillator tank circuit to allow the variable capacitor to have the stator side grounded, and the separation of the oscillator screen voltage dropping circuit from the oscillator plate voltage dropping circuit (i.e. both connected directly to the HV supply).

It is important that the 2E26 tuning inductor (shown in Phil's diagram as 10.5uH) not be too large or too small. Phil's 10.5uH value is a good choice... with around 482 ohms of reactance at 7.3 MHz. This will resonate (at 7.3 MHz) with a capacitance of around 45pF. That includes the input capacitance of the 6146 + the output capacitance of the 2E26 + stray capacitance. So the required value of the variable capacitor... to achieve resonance... will be around 25pF. The inductor should be rated to handle at least 100mA of DC, and it should have a self resonant frequency that is well above 7MHz. A fixed, air-would inductor would be a good choice. I.e. no problem with saturation of a ferrite core. With an air-would inductor, you can always reduce the 10.5uH inductor's value by connecting one side to a tap, rather than to the end. Reducing the inductance would allow you to use a higher capacitance value for the variable capacitor... to achieve resonance.

The comments Jon and others have made about the choice of 6146 plate blocking capacitor are (to me) compelling... because they are based on actual past experience that others have had with the problem of slowly decreasing output power on key down.

Stu