Homebrew 40M AM Tx Output Power Problem

[w4gon]

I moved the tank circuit to the oscillator plate. I'm using an isolated 17-214pf air variable and a coil wound on a 3/4" form. I lowered the inductance to around 5uh as I wanted a little more play room on the variable. The 75pf variable value is what I'm reading in circuit. Out of circuit it's around 50pf. I'm now able to get well over 3ma as I tune through resonance. It's tuned on the high side right now. Thanks Phil for your schematic. I've applied some of the modifications. For some of the others I don't have the appropriate components right now to implement them. I have a 220pf cap on the cathode-to-ground of the 2E26 and a 47ohm resistor in series with the crystal. I'm able to get accurate grid bias readings now and the values match perfectly with the calculations. Overall I feel that the rig is dialed in nicely. Except that I'm still loosing power! My grid current might drop .25ma during the loss of power, but I can easily bring it back up without any improvement to output power. Per Jon's comment I also can increase power a little bit by adjusting the PA tune capacitor, but it ends up fully open and that's end of story at that point.

I have an Apache waiting to be restored on the bench so I borrowed the 500pf 5kv door knob from it and have it now on the HB rig. I'm still loosing power! What the heck!?!

Fred, as the output power goes down I drop about ~5vdc of B+ voltage. My plate current goes up about 10ma so I guess that makes some sense... My filament voltage is steady at 6vac. I do have the filaments wired as per your description. I added a filament bypass capacitor to both the 2E26 and 6146.  

Could I be having an issue with my PA efficiency? Could my Pi network not be coupling enough power out of the PA and causing a heat build up? The best I've done so far is 66%. 30W in & 20W out. Then down to 10W out... If my Pi tank inductor is not up to the job, could it cause this problem? I was using a rotary coil from an old tuner and taping it, but then wound my own coil with a similar design as the Globe Scout. It's about 20 close wound turns of 22 gauge on a 3/4" form. I was having the same issue before on the larger coil as well.  

When I was building the PA I had a parasitic oscillation around 117MHz. Since adding the suppressor that issue has been resolved.  

My power supply uses a Tektronix transformer 120-0140 from an old Oscope, two 1N5399 diodes as full wave rectifiers, and for the filter a 150uf 450v/choke/150uf 450v combination. I'm truly out of ideas at this point. I really appreciate all the help with this issue.

I've attached the latest schematic with all changes made.

Joel

[AB2EZ]

Joel

Please post a version of the schematic that shows the values of the same voltages and currents at a point in time where the power output has dropped to 10W (without changing the tuning capacitor or the output loading capacitor values).

Of particular interest (to me) are the post-power-drop values of 6146 screen voltage, 6146 cathode voltage, and 6146 grid voltage (measured at the bottom of the grid choke).

Stu

[KA2DZT]

A few volt drop in B+ doesn't mean much, that's not the problem.  This probably also doesn't mean much but the cathode, G3 and internal shield are connected to pins 1,4,6. At the socket pins, wire all three of these pins together with a short wire going right over the socket.  Next, pin 8 is the base sleeve. Do you have this pin grounded??  If not, ground it.

Seems the problem is not the blocking cap.  You're increasing the loading to max when the loading cap plates are open.  Jon said that he would readjust the PA plate tuning cap, not the loading cap.  Try the test over and leave the loading cap as set for the highest power out.  When the power drops readjust the plate tuning cap.

The plate RFC is about the only part left that, if it's breaking down, could cause a loss of power.

Lastly, are you sure your watt meter and dummy load are working correctly.  If the dummy load goes up in resistance you would have to increase the loading to maintain power output.

Fred

[AB2EZ]

Joel

If the PA loading capacitor is not fully open... when it is adjusted for maximum power output and the transmitter is first turned on...

And if the PA loading capacitor is fully open... when (re) adjusted for maximum power output after the output power drops...

Then, that implies that either a) the Q of the LC output circuit (i.e. the Q of the coil) has increased (not likely)... or there is something in parallel with the adjustable loading capacitor that is adding additional loading capacitance after key down...or there is something across the output that is acting like a resistor whose value is decreasing.

What could that be? Perhaps the RF output connector is becoming leaky when it heats up.

The RF load impedance that the 6146 sees is Q x Q / R-effective... where

Q is the Q of the tuned circuit consisting of the tuning capacitor and the inductor of the pi network.

R-effective is [the transmitter output load impedance that is connected to the output of the pi network (i.e. ideally 50 ohms for a dummy load), taking into account any parallel resistance associated with a leaky connector or a bad piece of coaxial cable] / k

and

k is {1 + [2pi x f x C-loading x the actual antenna output load impedance] x [2pi x f x C-loading x the actual antenna output load impedance]}

Normally, when tuning up... you increase the output loading capacitance in order to reduce R-effective. Reducing R-effective increases the RF load that the tube sees.

If Q increases or if R-effective decreases... then the RF load that the tube sees will get too large... causing the output current to drop.

Stu

[DMOD]

I have an Apache waiting to be restored on the bench so I borrowed the 500pf 5kv door knob from it and have it now on the HB rig. I'm still loosing power! What the heck!?!

Fred, as the output power goes down I drop about ~5vdc of B+ voltage. My plate current goes up about 10ma so I guess that makes some sense... My filament voltage is steady at 6vac. I do have the filaments wired as per your description. I added a filament bypass capacitor to both the 2E26 and 6146.  

Could I be having an issue with my PA efficiency? Could my Pi network not be coupling enough power out of the PA and causing a heat build up? The best I've done so far is 66%. 30W in & 20W out. Then down to 10W out... If my Pi tank inductor is not up to the job, could it cause this problem? I was using a rotary coil from an old tuner and taping it, but then wound my own coil with a similar design as the Globe Scout. It's about 20 close wound turns of 22 gauge on a 3/4" form. I was having the same issue before on the larger coil as well.  



Joel

Joel,

According to my MatLab simulation using the Wingfield Pi-Net equations, and using your Vp and Ip, and F= 7.05 to 7.3 MHz,here are the results:

Qi = 12
Cplatetune = 111 to 109 pF
Lpi = 5 uH (stayed around 5 uH)
Cantune = 588 to 568 pF
Final RF plate choke = 210 uh.
Rplate = 2.14k
Efficiency = 68%

I am with Stu, check the voltages across the screen grid dropping resistors and other dropping resistors and so we can see what stage is dropping.

For power drops, I too have seen the output coupling cap cause this as well (as Jon noted) and if the 6146 screen overheats this can cause this as well.

It would also help if you can notate the components.


Phil  

[AB2EZ]

Joel

Adding to my most recent prior post:

Could it be that the resistance of the dummy load that you are using is getting significantly smaller as it heats up?

Stu

[KA2DZT]

Joel

Adding to my most recent prior post:

Could it be that the resistance of the dummy load that you are using is getting significantly smaller as it heats up?

Stu

Stu,

I questioned his dummy load.  More likely the resistance may be increasing.  

Question, if the load resistance increases would the loading have to be increased or decreased to maintain a certain power??  I think increase the loading.

2nd question,  for an increase in loading the loading cap would have less capacitance or do I have this backwards.  I'm trying to remember which way it works in my xmtr.  I can not see my loading cap and it has turns reduction unit on the shaft.  The loading cap is a four gang 525pfd (each) cap.  Two sections used for 40M and all four sections used on 75M.

Fred

[AB2EZ]

Fred

If the resistance of the dummy load that is connected to the antenna port increases, then the value of the loading capacitance must increase... in order to produce the same RF load impedance on the tube side of the pi network.

Stu

[w4gon]

Sorry guys, when I said PA loading cap I meant PA tuning. The tuning cap is the one that ends up fully open (10pf) as I try to increase power. I've correct my previous post.

I've attached the schematic with pre and post power loss voltage and current values. The bottom one is pre and the top one is post. So for the 6146 plate, the voltage went down and current went up as power was lost. The voltages were taken in respect to the cathode. Let me know if you rather me take the measurements differently. The cathode measurement is to ground. The 2E26 grid measurement caused oscillation to stop so I wasn't able to get an accurate reading there. Currents and voltages stayed pretty constant in the oscillator stage.

Thanks for running that simulation Phil. So looks like I need more antenna tune capacitance. I had a feeling that was the case. Overall that doesn't look too bad right? I've noted all components.

Regarding the dummy load. I'm having the same problem on the air using my Delta Loop so this "should" rule out the dummy load. I made the connection between 1-4-6 on the 6146. Pin 8 was grounded already. 

Joel

[DMOD]

Q1: Do you have a Forward/Reverse Power meter between the transmitter and the dummy load? Maybe reflected power is increasing after the dummy load heats up?

Q2: What kind or type of dummy load do you have?

Q3: Have you bypassed the mod transformer T1 and placed your 370V directly to the RF section?

Comment: If the B+ is really 370V, then you have a rather large voltage drop of 70V across the T1 mod transformer.


Suggestion: Add a 220 pF 1 kV ceramic cap across the Cantuner.



Phil - AC0OB

[w4gon]

H Phil,

Q1: Yes, I'm using an MFJ 962D tuner. Reflected power is staying low throughout key down.

Q2: I'm using a cantenna type dummy load with the large resistor immersed in mineral oil.

Q3: The 6146 voltages are taken in reference to the cathode. If I connect to ground then the plate voltage is much closer to 370vdc. I'll have to take this measurement again and report back, but it didn't look alarming to me when I took it last. The power supply idles at around 440v and drops to 370vdc under load with all tubes plugged in. I was having the same issue before adding the mod transformer if I recall. I can bypass it and confirm though.

Joel  

[AB2EZ]

Joel

If the pi network tuning capacitor is fully open (when maximizing the output power) after the power drops.... but not fully open when you first "key down"....

Then, the inductor in the pi network has too much inductance when the power drops (due to heating) ... and probably has more inductance than it should, even when it isn't hot.

If the pi network inductance is too high (on key down, when it is not hot)... you may be able to peak the output power (i.e. resonate the LC tuned circuit)... but the Q will be higher than it should be. This will make the pi-network very sensitive to changes (due to heating) in the inductance of the pi-network inductor.

Stu

[w4gon]

Thanks Stu, interesting. So looks like the inductance should be more in the order of 5uh for the coil. I've never built a large Pi network coil before. What are the best dimensions to avoid this heating business? I'm thinking a trip to Home Depot to get some bare copper wire is in order.

Joel

[AB2EZ]

Joel

If you have room (making sure the coil isn't too close to the chassis or the front panel of the transmitter)... I would suggest a target diameter of approximately 1 inch. The copper wire size you use should be #14 or #12... just to be conservative on skin effect resistive losses.

There are calculators on line that you can use to calculate the inductance of a simple solenoid... given the wire diameter, the coil diameter, the total length of the coil, and the number of turns.

If the coil that you wind turns out to have too much inductance... you can reduce the inductance by stretching out the coil to increase the spacing between turns.

You can wind the coil around a piece of PVC pipe... but make sure to remove the PVC pipe after the coil has been wound. Many plastic materials (including PVC) will absorb RF ... and cause losses.

Stu

[w4gon]

Thanks Stu. Let me work on the coil build then so that we can cross this component off the list. I'll report back.

Joel

[KA2DZT]

To calculate the values for a pi-network you need to figure the plate load.  Plate load is the plate voltage divided by twice the plate current.  370v/200ma= about 1850 ohms.  From the chart in the Radio Handbook your plate tuning cap needs to be about 150pfd, so you need about a 200-250pfd cap.  The inductor needs to be about 4uh.  The loading cap needs to be about 1000pfd.  So you need a cap with about 1500pfd for the loading.

The coil, (with 20 watts) should be about 1.5' to even 2" in diameter.  14ga bare wire is fine,  the coil should not be close spaced.  The coil length should not be more than 1.5 times its diameter to get the right Q.  With 4uh a 2" coil probably would not be even an inch long.

The coil should be away from other metal as much as possible but it can be close to the tuning cap.  The axis of the coil should be parallel with the shaft of the plate tuning cap.  The loading cap can be anywhere.

Another thing,  as the plate load gets lower the value of the inductor goes down and values for the caps go up.

Fred

[w4gon]

Thanks Fred. Hmm, a 1500pf load capacitor... That's going to be a problem. I don't have a variable with that much capacitance right now. What are my options? Can I put a capacitor in parallel with it maybe?

Joel

[KA2DZT]

Hang in there Joel, when this is all over you'll be a HB xmtr expert.

Well, at least at the 20 watt level.

Fred

[KA2DZT]

Thanks Fred. Hmm, a 1500pf load capacitor... That's going to be a problem. I don't have a variable with that much capacitance right now. What are my options? Can I put a capacitor in parallel with it maybe?

Joel

At a low power level you can use some fixed caps to get that much capacitance.  Add some micas if you have them.  Voltage rating does not have to be high.  1000v caps is more than enough.  Better to use a number of lower value caps than just a single cap.  This will divide the current through a number of caps rather than just one.  Keep in mind that the loading cap has very low voltage on it.  Most of the time a common BC radio tuning cap will work fine, even at much higher power levels.

Fred

[DMOD]

Joel,

For a Plate current of 90 mA, your Pi-Net values are as follows:

Cplatetune = 120 pF

L = 4.4 uH

Cantload = 613 pF;


For coil calculations I use:

http://hamwaves.com/antennas/inductance.html

which is one of the more accurate online coil calculators. Here is the basic data:
 
D   = 25 mm   Mean diameter of the air core coil, measured from wire center to wire center

N = 20            Number of turns

ℓ = 40 mm   Length of the coil, measured from the connecting wires center to center

d = 1.29 mm   Wire diameter (#16; Most Low power rigs such as the HT-40 and DX-60 used this diameter)

from: https://www.tedpella.com/company_html/wire-gauge-vs-dia.htm

f = 7.2  MHz   Design frequency

L = 4.55 uH     Inductance




Phil - AC0OB

[w4gon]

Thanks Phil, Fred and Stu! I'll get to work on this and report back.

Joel

[DMOD]

I know there are many approaches to circuit design, but the schematic you showed appears to be a prototype development schematic due to the non-standard values.


So FWIW, below is a complete set of schematics for your consideration.

Most of all, have fun, which you appear to be doing.


Phil - AC0OB

[w4gon]

Thanks for the schematics Phil. They are most appreciated. That is correct on the prototype dev schematic. I built this rig on an old mobile vhf transmitter chassis I got on eBay for $10 +shipping. The rest of the parts came from my junk drawer. Also, I used the actual values of the components as I measured them. Probably not a good idea, but...

Good news, I found the problem! It was the wire I used between the plate coupling capacitor and the PA tune variable (pic attached). Oops...  Let's just chalk this up to my learning curve ok.

I replaced the bad wire with a short run of enamel copper wire. I'm putting out a steady 24 watt signal after a minute plus of keying down with no loss of power. My efficiency is up to around 76% also.

I built the Pi network coil using #12 copper wire as well. Its D=2.4", N=11, ℓ=2.8" and L=4.6uh on the meter. The Pi variables are both about 50% open so plenty of room for adjustments.

I want to thank Stan, Fred, Phil and all the rest of you for helping me with this problem. I now have a much better working RF stage thanks to your advice and guidance. We started at the crystal and worked our way to the Pi network where the problem was, awesome!

Thanks again!
Joel

[KA2DZT]

Ok great,  you found the problem.  What exactly was wrong with the piece of wire??

I like the size of the coil you made for the pi-network.  Just slightly longer than the diameter.  2.4" diameter is good, much better than a small 1" coil.

Fred

A two tube set and it only took us a week to get it working.  Not too bad, good thing it wasn't a five tube set.

[w4gon]

Tell me about it! And I had been fighting this thing for weeks before finally giving up and asking for help. 

The wire is about 7" long and it is reading 6.5 ohms of resistance and 35pf of capacitance. Really?!?

I have one more question regarding the secondary impedance of the Mod transformer. I read that to calculate the modulating impedance you take the PA DC pate voltage and divide by the DC plate current. That's 370/.090=4100. Is this correct? If so then I'm leaving the secondary at 4200 ohms.

Thanks
Joel