Homebrew 40M AM Tx Output Power Problem

[w4gon]

Hello all,

This is my first post on the forum so please take it easy on me. I decided to try building a plate modulated AM transmitter about two months ago. I had never built anything using tubes before so I started by getting a copy of the 1949 ARRL Radio Amateur's Handbook and started on page 1. I've attached the schematic of the rig as it stands now. I have it working and have made a number of 40M contacts with good reports.

So to the problem. For some reason the 6146 is dropping output power after about 30 seconds to a minute of transmitting. With the present screen and cathode resistor values it is initially putting out about 15W and then it drops down to below 10W. I have checked all of the RF stage components and none are overheating. I checked the output of the oscillator and it is generating a steady output during transmission. The B+ value of 380v is under load. The voltage values for the tubes are taken in relation to the cathode. I've tried three different 6146s and even a few 2E26s with no improvement. I noticed that the efficiency of the PA stage is not high enough. It's in the 50%-55% range before power loss.  

I'm thinking that the 6146 plate might be heating up and changing its Rp causing the issue. I've measured the plate voltage and current to try to get a clue as to the issue. While the output power is dropping, the plate voltage goes down and the current goes up. Thinking that I might not be coupling enough power out of the plate I've tried different variable caps and inductors in the PI tank. Currently I'm using a borrowed pair of variables from a Globe Scout. While the power output is dropping I'm able to bring it back up a bit using the load capacitor until it runs out of range in the open position, so 10pf. I'm also thinking that I might not be deep enough in Class C and that this might be causing the extra heat due to an increase in anode current conduction angle. Getting consistently accurate grid voltages via a volt meter has proven difficult though. I've been working on the grid biasing for a few days with no success in fixing the issue yet.    

I'm starting to run out of ideas as to what and how to correct this issue. I don't want to complicate the design by adding a bias supply. I know that this transmitter can be made to work in this current state. I want to keep things simple at all cost. It would be great if someone here can point me in the right direction as to what is causing this problem and possible things to try to correct.

Thanks!
Joel
W4GON

[KA2DZT]

You need to by-pass the RFC in the grid circuit.  Junction of the RFC and the grid resistor add a 0.001 to 0.005 ceramic cap to ground.  Usually you can rely on just grid leak bias to provided enough grid voltage.  So, why the large cathode resistor, usually the cathode of the 6146 is just grounded.  Grid current should be 2-3ma.  Your grid resistor looks about the right value.  I think I have 24K in my HB 6146 xmtr.

2E26 as a crystal oscillator is a bit much.  Sure you're not causing too much crystal current.

Not sure why you have a 4.7uf cap across one of the screen resistors.

Looks like real nice low power rig.

What type of cap are you using for the plate coupling and RFC by-pass??

You can estimate the grid voltage by the grid current through the total DC resistance of the RFC and grid resistor.

Fred

[w4gon]

Thanks Fred! I'll bypass the grid RFC. I don't recall seeing this bypass cap in the ARRL book. Is it to ground any RF that makes it past the RFC?

I added the cathode resistor to lower the current to the 6146 thinking I was overheating it. I actually had the cathode directly grounding before and I was going to go back to that tonight to see if it helps.

Regarding the 2E26 oscillator, good question! In the 6146 datasheet it calls for 100v peak RF Grid No.1 voltage. I was having a hard time getting that much drive with the smaller glass tubes so went with the 2E26. What should the crystal current be for safe operation? I'm assuming I just check the current on the ground side of the crystal?

I'll bypass the RFC and remove the cathode resistor to see if that helps. I have a variable resistor as the grid leak right now while I try to find the correct value for it. Not sure if this is a good idea... I have it set to 28k per the datasheet right now.

The 4.7uf cap across one of the screen resistors is to bypass some of the modulating signal into the screen. I've been reading W8JI's write ups and he recommended experimenting with the plate/screen modulating ratio. It helps with increasing the modulation to 100% on this rig anyway. I don't claim to have the ratio right on it!

The plate coupling cap is also "borrowed" from my Globe Scout so it's a 6kv 1nf cap. The plate RFC by-pass is a ~5nf ceramic cap.  

Very good on the grid bias voltage calculation. I was doing that as well, but it was not matching up with the volt meter so wasn't sure which one to believe. I'll just use the calculation going forward.

Joel

[KA2DZT]

Not sure about the crystal current,  Make sure the crystal is not getting warm over time.  The problem is you're losing power,  what type cap are you using for the coupling to the pi-network??  Some type caps don't work well with high RF currents, they'll get hot over time.

Fred.

[w4gon]

The coupling cap to the PI network is a ceramic from what I can tell. It's whatever WRL used for that component. I removed it from my Globe Scout 680 and put it in the HB rig on Monday. Actually, I took the parasitic choke, plate RFC and plate coupling capacitor all from the Scout and used them on this rig. I did this to remove any possibility that I had a bad component on that stage. The Scout has been putting out rated output power for years so I know those three components are good now at least...

Joel

[KA2DZT]

You should add a 2.5uh RFC to ground on the output line to the coax connector.  This is a safety feature.  If the plate coupling cap shorts, B+ will be on the antenna.  The choke will short the B+ and blow the fuse.  Hopefully you have some fuses in the rig somewhere.

Fred

[W3GMS]

Hi Joel and welcome to AMfone.  Very friendly folks over hear and a wealth of knowledge. 

My suggestion is to use a scope to find out what is sagging from an RF point.  I would start at the plate of the XTAL oscillator and work your way through the transmitter.   I think with that method you will quickly find out what in the chain is dropping off.  If you don't have a scope, you could use an RF voltmeter.  If you don't have one of them you could make an RF probe and use it with one of your analog VTVM's. 

Since you have no negative fixed bias supply on the final, if the drive drops off the plate current will increase and then your B+ will drop since your pulling a lot more current from your supply. 

73,
Joe-W3GMS         

[AB2EZ]

Joel
Fred

Since the is no fixed bias on the 6146 grid, and no "clamp" circuit... the cathode resistor may be required to protect the tube from run-away plate current if the RF grid drive is not present (or too low in amplitude).

The 392 ohm value seems okay... but, with 87mA of total 6146 cathode current, the dissipation will be 3W. Therefore, a resistor rated at 3W creates (what some call) a "Chernobyl" situation.

Stu

[KA2DZT]

Yes, he can use some cathode bias, especially with no fixed bias or clamp tube.  The problem he posted is he has about 20 watts at first but it drops to about 10 watts after 5 or 10 minutes.

So it seems something is not handling the RF, possibly a bad cap or the type of caps used or something with the crystal oscillator.
Could be a bad resistor or one of the RFC is breaking down.

Hopefully he'll find the problem

Fred

[w4gon]

Correct, the issue is that after 30 seconds to a minute I start loosing output power from about 15W to less than 10W. I do have an Oscope and will do what Joe recommended. I've ran that test on the rig in the past, but will do it again very carefully to see if I can pin it down. Fred, your comments on the bypass caps has me wondering about the plate bypass. I'm going to replace it with a higher voltage orange drop just in case. I replaced the plate RFC already so I don't think that component is the issue.

I'm starting to run out of spare parts for this rig... That's why the cathode resistor is only 3W as that's all I have right now in that range. The power output problem was present when I didn't have a cathode resistor so I don't think this is the issue. I'll be placing an order for more parts once this issue is resolved.

Joel

[KA2DZT]

All the RF by-pass caps should be ceramic disc capacitors.  Don't think orange drop caps are good for use in RF circuits unless the newer modern ones are but I doubt it.  Same with the all important RF coupling cap.  Use a high voltage mica or ceramic disc.   I use old TV high voltage door knob caps, but they're probably hard to find today.  You can buy new RF door knobs.

Fred

[W2JBL]

If the PA plate current is rising and the output dropping what is the PA grid current doing? Sounds like you are slowly losing grid drive. I notice on your schematic the grid current is 2 MA? That's way too low. Try 3.5 MA (the tube manual will say 2.5 but that's not enough). Your efficiency will increase and modulation will improve. I don't think 2MA is enough for the tube to start to run away at that plate voltage though. If in fact the grid current is dropping then start to look at the oscillator for the problem. I also would expect 20 to 25 watts out of that circuit, more with higher plate voltage. I would also suggest getting rid of that cap across the screen resistor. If you can't get 100% modulation without it you may not have enough audio power, or the mod transfomer is mismatched.
 

[w4gon]

I bypassed the grid RFC and ensured all RF stage bypass caps are ceramic per Fred's suggestion. The coupling caps between oscillator and PA are mica.

On closer examination of grid current, I am indeed slowly loosing grid drive. It's slow and only about .5ma, but apparently that's enough at these power levels. I also checked oscillator output using the scope and peak voltage is slowly dropping over time. I measured about a 7v peak to peak drop. I tried three different 2E26s and then rewired for a 6V6 and tried two of those with no change in grid drive loss. I checked for overheating components and the only one that's getting hot is the plate resistor. That would be the case since I'm right at its rating. I replaced it with a 10K 20w resistor as I've done before to drop the output of the oscillator. This resistor drops plate voltage to around 150v and it's also getting hot. No other components in that stage are hot to the touch. I connected the scope to the plate of the oscillator and I'm seeing the voltage drop at this point. I then removed the 6146 and modulator tubes and ran the oscillator without load. After an initial drop in p-p voltage the signal stabilizes and does not drop further.

I'm not sure what to try next to get to the bottom of the issue aside from throwing it out the window.

Joel

[KA2DZT]

What happens if you put a resistor in series with the crystal.  Start with 10K and maybe go higher if the oscillator keeps running.  Problem could be a bad cap that is changing value as it warms up or maybe a bad RFC.  The one in the grid circuit, if it is bad will load down the oscillator.  BTW what type crystals are you using??  The two cap voltage divider in the osc.grid circuit would be the first caps I would check.  What type caps are they??

Another thing, as the voltage drops does the oscillating freq stay stable or is that drifting??

Fred

[AB2EZ]

Attached is the schematic of a low power (CW) transmitter that I used to own. I modified it to add plate modulation... and I used to operate it for long periods of time on key down/AM. It worked fine on 80m. For 40m operation, I had to add a tuned circuit to the output of the oscillator because the input capacitance of the 6L6 was too great a load on the oscillator on 40m.

Comparing to yours...

1. I think that the 50k ohm resistor that is in parallel with the crystal in your transmitter is a significant contributor to crystal heating in your transmitter... and the problem you are having. I also had a DX-20 that I modified for plate modulation (schematic attached)... and I had to reduce the value of the DX-20's 6CL6 oscillator's grid to ground resistor (across the crystal) from 100k to 18k to solve a crystal heating problem that was present on long periods of key down.

2. Separately, the attached "six lima six" transmitter schematic does not employ a tuned circuit at the output of the oscillator. As mentioned, for 40m operation, I had to add a tuned circuit to the "six-lima-six" oscillator's output to resonate out the input capacitance of the 6L6 power amplifier stage

The DX-20 does include a tuned circuit at the output of its oscillator.

I suggest that you reduce the value of the 50k ohm resistor to 10k ohms.

Stu

[w4gon]

Excellent! I have homework to do now. Let me digest all the info provided and apply the mods and I'll report back.

Fred. The two cap voltage divider on the grid uses small ceramics. While I was on the air and watching my signal on a number of websdr receivers I could see my carrier drifting about half a Khz. So yes, there is some drifting of the carrier signal. 

Thanks guys!
Joel

[KA2DZT]

OK great,  Stu thinks the problem is with the crystal current, that's why I suggested adding a resistor in series with it.  I thought it was a crystal heating problem when you first posted the problem, that's why I questioned the 2e26 oscillator.  Try what Stu suggested, reduce the 50K resistor.

He did this with a DX-20, the last time I had a DX-20 was 56 years ago.

Fred

[w4gon]

Will do. Forgot to add that the crystals I'm using are FT-243 types from AF4K.com.

Joel

[AB2EZ]

Joel

I believe the "real" FT-243 crystals are physically larger... and able to handle more crystal current than the AF4K crystals. The AF4K crystals are a newer, smaller type of crystal packaged inside of a reused, surplus FT-243 crystal package.

Stu

[KA2DZT]

Will do. Forgot to add that the crystals I'm using are FT-243 types from AF4K.com.

Joel

OK FB,  You have to hope that they are real FT-243s or a lesser crystal in a FT-243 case, which could be part of the problem.

[KD6VXI]

Brian Carlings crystals are a known problem in some situations.  His FT243 are not the same as a 'old skewl' 243.

Search the archives.  Some people flatly refuse to use them.

I have no experience.  Just relaying what I see about the boards.

--Shane
KD6VXI

[w4gon]

Well, I was really hoping that we had this thing figured out. I replaced the 50k resistor in parallel with the crystal with a 10K. There was no change to the power loss problem. I have an old school Texas Crystals FT-243 that uses the large crystal wafer so I tested with that one and the problem was still present. I then added an 8k series resistor to the ground side of the crystal. The grid current dropped to less than 1ma, but I still had the expected output power. It then dropped down to less than 10W within a minute. So I then wired my Heathkit VF-1, plugged it in and shorted the oscillator cathode RFC. I got 3ma grid current and output, but it then dropped to less than 10W as it has been doing. I aso replaced the two cap voltage divider caps with micas with no change. I also replaced the grid leak variable resistor with a 33k 1/2w since that's the closest value I had to 28k just in case, but same problem. I'm staring to think that it's not the oscillator now.

Stu, I have a tuned circuit in the coupling stage. Do you think that this is resonating out the input capacitance of the 6146?

Just thinking out loud, but I'd like to load the oscillator without actually having the 6146 in place to see if the grid current drops. I want to take the Rp of the 6146 out of the equation. Is it a bad component or something up with the tube I wonder?

Joel

[KA2DZT]

Remove the plate and screen voltage from the 6146, leave the 6146 in place.  Turn on the oscillator and tune the grid tank for max grid current up to 3-3.5ma.  See if the grid current stays stable.

Also make sure your oscillator plate and screen voltages are not dropping for some reason.

This is a simple straight forward xmtr, you shouldn't be having this much difficulty. Somewhere there is a component that is failing under circuit conditions,  don't give up we'll find it.

Fred

[w4gon]

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 

[AB2EZ]

Joel

According to its specification sheet, a 6146 has an input capacitance (looking into the grid, with the cathode bypassed to ground) of 13.5pF. The 2E26 has an output capacitance of 7pF. If one adds the driver-output 60pF tuning capacitor, one obtains a minimum capacitive reactance at 7MHz of: 282 ohms; and a maximum capacitive reactance (assuming the variable capacitor has a minimum capacitance of around 6.5pf) of: 842 ohms.

If one considers the 5.7uH inductor in parallel with the 7.5uH inductor, one obtains a maximum inductive reactance of 142 ohms.

Therefore... one cannot make the capacitive reactance equal to the inductive reactance... and one cannot achieve the desired resonance at 7MHz.

I suggest that you replace the 5.7uH inductor with a 2.5mH RF choke (or whatever you have on hand that is 100uH or larger).

Put the 5.7uH inductor (now available for other uses) in series with the 7.5uH inductor

This will increase the maximum inductive reactance to that of a 13.2uH inductor ... i.e. a maximum of 580 ohms at 7MHz.


This should allow you the resonate the 6146 input by adjusting the 60pF capacitor.

You should adjust the 60pF capacitor (i.e. not set exactly at resonance) to obtain a target of 2.5mA of grid current on the 6146.

Having (not quite) resonated the 6146 input, you should be able to obtain much more drive voltage. The 6146 grid leak resistor (which should be a value that is reasonably close to 27k ohms) will "self bias" the grid to a DC value of -2.5mA x 27k ohms = -67.5VDC with respect to ground. The cathode resistor will bias the 6146 cathode to 392 ohms x 90mA (total cathode current) =35VDC with respect to ground. The total DC grid-to-cathode bias will be -67.5V - 35V = 102.5V... which will make the 6146 operate very far into class C.

You can always reduce the value of the 6146 grid current if -102.5V turns out to be too large a grid-to-cathode bias. The normal target grid-to-cathode bias is -85V. Taking into account the 392 ohm cathode resistor (and the associated 90mA of DC cathode current) and the 27kohm grid leak resistor... you would have to reduce the 6146 grid current from 2.5mA to 1.9mA if you want the grid-to-cathode bias to be -85V.

As was previously suggested by Fred, you should add a .001uF (or larger value) bypass capacitor between the bottom of the 2.5mH RF choke on the grid of the 6146 and ground. This will allow you to make an accurate measurement of the negative DC bias (relative to ground) on the grid of the 6146 (measured at the point where the bypass capacitor is connected) by removing the RF that would be present at that point in the absence of this bypass capacitor.

Stu