Another weird 6AU6 problem

[WB3JOK]

I've had an HW-101 for a long time as my primary rig...  it's not  Heavy Metal (or even AM ) but I'd like to run this by the tube gurus. I also posted over on the Boatanchors list.

... I usually keep my transmissions short but early this morning on the 7272 Rag Chew there was only three of us so I "gatored" a bit. When I let go of the mike button, the receiver was cut off and the S-meter was nearly pinned. It very slowly returned to normal over about 15 seconds. On subsequent short transmissions, operation was normal, but on any extended key-down it recurred.

I got out the schematic, took the cover off, and quickly discovered that something was driving the AVC line several volts negative immediately on releasing the PTT after a long transmission. Just holding the PTT closed with no modulation did not cause the problem. No one Heath factory service bulletin was a direct clue but I did come across this:

>SEPTEMBER 21, 1984
>HW-101 Bulletin No:
>SSB Transceiver HW-101-82
>S METER DRIFTS; IF OSCILLATES
>
>Check the brand of 6AU6 tubes at V3 and V4. If a brand other than GE is used at these locations, replace them with GE brand tubes. Parts replacement will stock only GE brand of 6AU6 tubes [PN 411-11].

Not very helpful - and I do have GE-made JAN-6AU6-WC tubes in V3 (the 1st IF stage) and V4 (the 2nd IF, which drives V13, the AVC rectifier). I tried swapping some other 6AU6's and it didn't help (including a brand-new 6AU6-WC).

Putting the scope on the output of V4 did indeed show the oscillation at about 3 MHz (at or near the IF frequency), which slowly died out over a period of about 10-20 seconds, while the AVC voltage returned to normal. It would only recur with extended talking. Although I could not see anything coming out of V3, (old scope,10x probe   as a test I quickly pulled V3 from its socket and the oscillation stopped instantly.

Also, V4 is only used during receive and is definitely being cut off during transmit (its screen voltage is removed), and V2 (Isolation Amp between the balanced modulator and V3 which is also the first transmit stage)  is cut off during receive (screen voltage removed and actually biased slightly negative). V3 is "on" during both receive and transmit, so I concluded that that stage has to be the source of the problem.

What I can't figure out is the underlying mechanism of the oscillation. I believe that the stage runs hot during receive and even hotter during transmit (the grid bias while transmitting is set from the ALC voltage that is less negative than the AGC signal).

In my other 6AU6 thread, Richard WA4FUU added:

I wonder if it is caused by grid emission? Some of the free barium from the cathode coating finds its way to the control and/or screen grids; when hot enought, the grid(s) can become another emitting surface.  Just a thought.

The slow speed of onset and decay also suggests a thermal cause. Does the stage gain vary with internal element temperatures (aside from the heater, of course)? Is this something that can be fixed by, for example, redoing the screen bypass? Or do I need to reduce the gain a bit, or move the operating point?

I really don't like designs that are dependent on a particular brand of tube...
thanks for any help.
-Charles

[W3SLK]

Charles, the HW-101 has a fixed ALC. I ran into the same problem with mine. It was especially noticable when I would attempt to run AMTOR. Finally I gave up and got a ricebox, (many years later). I would try to change the R-C time constant for the ALC. I don't have the schematic in front of me but I seem to remember the circuit somewhat. If I get a chance to dig up the schematics, (I have the manual), I'll try to remember what I did to change the recovery time.

[WB3JOK]

Charles, the HW-101 has a fixed ALC.

What does that mean, exactly? Fixed for a particular output level, or mic level?

I ran into the same problem with mine. It was especially noticable when I would attempt to run AMTOR.

Not surprising, as that also has a long transmit time and an even higher duty cycle

I would try to change the R-C time constant for the ALC.

Good point. I need to see if it's working at all, come to think of it 
The ALC doesn't seem to "do" much and I can flat-top the crap out of the RF (observed on the SB610)  by turning up the mic gain. Without a spectrum analyzer I can't tell if it splatters that way (and wouldn't want to do that on the air anyway!)

Meanwhile the AVC has a fast attack but a very long decay, perhaps too long. I haven't bothered to play with it but now might be a good time.

-Charles

[WD8BIL]

Charles..... if you ain't using the ALC output to an amp, put a 100 ohm resistor across the jack. Hell... I've even shorted the jack out completely. That triple action ALC circuit is worthless.

Don't need no stinkin' ALC anyway !!

[WB3JOK]

That's a good suggestion... "When in doubt, throw it out" 
Actually I am using it (driving an SB-200) most of the time...

But as you point out, it doesn't work that well anyhow and I do have a monitor scope always in use. I could always add a pot to set the grid bias at a fixed level instead of ALC, and just watch the scope for proper mic gain setting. Another mod would be to leave the ALC but add fixed bias (on transmit-only) for just the 6AU6 stage so it doesn't "cook" on transmit.

-Charles

[Jim, W5JO]

Charles, something strikes me that the SB 200 likes about the 8 volt range for ALC?  Is your HW 101 putting out more than that?  If so you can use a zener the input of the SB 200 to that level.  It also could be the output of the ALC is screwed up as mentioned.

Now that I used my head, the SB 200 puts out more than 8 volts for AGC and I used mine on something that didn't like more than 8 volts.  So I put a zener on the output jack of the SB 200 to keep the voltage down.   Turn on your ignore switch about this possibility.

[WB3JOK]

Charles, something strikes me that the SB 200 likes about the 8 volt range for ALC?  Is your HW 101 putting out more than that? 

No, in fact it seems to be quite a bit smaller than that. I think one or both of the ALC rectumfryers may be blown up. This would also cause the IF stage to run hot (inadequate bias). I'll probe around tonight and see...

Anyone know what the peak-to-peak RF voltage (grid swing) should be for the HW-101 finals (a pair of 6146's in parallel)? The internal ALC circuit looks like a simple voltage doubler driven directly off the final grids. I would think there's at least 30-40 v pp? The ALC output will respond to the average value (about 20% in noncompressed SSB speech).

Turn on your ignore switch about this possibility.

Is that somewhere on the front panel of one of those riceboxes with 1,000 buttons? 

-Charles

[Jim, W5JO]

Turn on your ignore switch about this possibility.

Is that somewhere on the front panel of one of those riceboxes with 1,000 buttons? 

-Charles

Yes it is, but you must transit 4 menus to get to it.

[WB3JOK]

I discovered a bad solder joint on the lug where the ALC diodes connect to the wiring harness... the wires moved when the harness was pulled. Eureka?  Nope.
Diodes are good, too. Still no ALC voltage. So I fired up the scope...

There is about 40 volts p-p of RF at the grids of the 6146's at max drive. The bias is about -60v (the resting cathode current is set at the meter triangle mark which I think is 50 ma). So the grids are nowhere near being driven positive. Consequently there is no grid current, no rectification, and no ALC voltage!

When setting the mode switch to TUNE, the ALC line is forced by the MIC/TUNE LEVEL pot anywhere from 0 to -40 volts depending on the output level, so that works too.

Tell me if this makes sense... there is not enough drive to push the finals into grid current,  so ALC is not being developed, so the 1st IF (among others) is turned on too hard, and therefore overheating. The flattopping at high mic gain is most likely coming from an early (audio) stage, not because an RF stage is being overdriven!
It looks like I should be trying to find why there is NOT excessive drive with the ALC line at zero.
 

-Charles

[W2JBL]

get rid of the ALC alltogether. it actually causes more splatter than if you flat top. shorting the ALC line out the back of the radio only kills it half way. you also have a dirty/intermittent T/R realy. that is why your receive does not allways come back after you transmit. if i worried a bout "grid emmission" in my 6AU6's i'd be in the ground... i have three HW101's and an SB102. they can be made to transmit excellent quality AM by a variety of simple mods.

[Steve - WB3HUZ]

Not clear to me why grid current is needed for the diodes to rectify the 40 volts PP across them (I'm assuming the 40 volts PP is across them).

[WB3JOK]

Not clear to me why grid current is needed for the diodes to rectify the 40 volts PP across them (I'm assuming the 40 volts PP is across them).

It's not... there is an RF choke that only passes the audio-rate variations. The average value of the resulting AF voltage can only "move" if the stage goes into grid current.. There is also a coupling cap from the screens, but I can't see how the screen voltage could change enough to provide any useful ALC level.

I'm inclined to agree with KD2XA, at least as far as getting rid of the ALC entirely. It looks like a really screwy design the more I study it!

But the problem is not the T/R relay (I've checked to make sure the contacts are really working, too). As posted earlier in the thread, V3 heats enough during prolonged transmissions that it oscillates when returning to receive mode. That oscillation is large enough that it causes the AVC to go way negative, just as if a really strong signal were being received, so the receive gain is basically gone. The S-meter confirms it, as well as directly monitoring the AVC line.. In this particular case it is not an intermittent that sometimes occurs whenever the button is released, although that is another known problem with these rigs. It *only* happens on long, loudly modulated transmissions. Holding the key down for several minutes with no modulation never causes the relay to stick (nor the oscillation to occur).

-Charles


I may just add a bit of additional bias to V3 on transmit, and scrap the ALC.

[Steve - WB3HUZ]

The average value of the resulting AF voltage can only "move" if the stage goes into grid current..

Move to where?

What is done with the output of the rectifiers. I assume there is some sort of cap on the ouput. But where does it go after that? The gain of which stage is being controlled by the ALC voltage?

I'm just curious. I agree with Chris. If you monitor your signal with a scope, ALC is not needed. Many ALC schemes are pretty crappy and when pushed a little create splatter/IMD instead of preventing it.

[WB3JOK]

By "move" I meant, "become less negative". 

There are actually two ALC rectifiers in a half-wave voltage doubler circuit. The output of the doubler is filtered and connected to the grid of V2, the grounded-grid isolation stage after the balanced modulator, and also to the grid resistor of V3, the 1st IF which is subject to the overheating problem that started this thread in the first place... I don't have the schematic in front of me (work PC, no .djvu plugin installation allowed) but I think that's it.

After thinking about this some more, I don't think Heath's ALC was actually designed to be a closed-loop system that continuously adjusts the gain, but just to cut back the big peaks. So the basic design of the 1st IF stage must be flawed because under normal operating conditions (without ALC voltage being generated) its grid bias is insufficient to keep the tube element temps under control on transmit.

I think I am just going to disable the ALC as you are suggesting, and appropriately modify the bias and/or screen voltage on the V3 1st IF to eliminate the problem... I always have a monitor scope running anyhow, and watching the bouncing needles on analog meters just isn't enough to determine if my signal is flat topping!

The bias on V3 is actually switched by the T/R relay. On rx it comes from the AVC line. On tx it's switched to the ALC line. So it'd be simple enough to change the bias only on transmit, and leave the AVC-moderated receive gain alone.

-Charles

[WB3JOK]

A bit more measuring and experimenting tonight...

The plate current on that stage (in either transmit or receive) is about 11 ma. At 290 volts Ep that is a lot of plate dissipation for a 6AU6! Screen voltage is 165v which is also high.

I doubled the screen dropping resistor (from 22k to 44k) and that cooled things off a bit. Ip is now 8 ma, Eg2=125v. The envelope is now cool enough to touch and pull the tube, where before I had to use a rag. Still normal power output in transmit.

However, the problem is still there, but less severe. Even with excessive mic gain and several minutes of nonstop simulated buzzarding (typical 75m SSB use ), the oscillation upon release of the PTT now only blanks the receiver to S6-7 instead of pegging the S-meter with maximum AVC, and it decays more rapidly too.

I may reduce the screen voltage even a bit more... but I think it's time to start playing with the screen bypassing.

-Charles

[WB3JOK]

Stop the presses!

I've been barking up quite the wrong tree, it seems. If I had a more
sensitive scope, and more experience with tube transceivers, I may have figured this out earlier, but at least I am now very familiar with the insides of my HW-101. 

After many experiments with V3, varying the screen voltages and plate current, including trying a 6HS6 (higher transconductance) with lower screen voltage for the same overall gain , I concluded that the problem simply could not be within that stage after all...  all the attempts reduced but did not eliminate the after-transmit IF-frequency signal appearing at the output of V4, the only place it was big enough for me to measure. Listening to the antenna it became obvious that the reason for the decreased signal was the reduced stage gain, so then the receiver got hard of hearing too!

I did increase the plate dropping resistor from 1K to 10K. That reduces the plate voltage by about 100V, (which has negligible effect on the stage operation since it's a pentode and the curves are flat between 150 and 250 Ep), but takes 1 watt of heat out of the tube and into the external resistor. That should prolong the life of the 6AU6 since it runs very close to max power ratings (nearly 3W and 3.2W is max).

So I started looking at this from another angle. I found that the transmit screen voltage bus (white-org-org harness wire) is supposed to cut off when in receive, and in fact there is a 1 meg resistor to the transmit bias pot to make SURE that the transmit stages are cut off when the receiver is operating, by actually biasing their screens negative. That line sits at about -6V in rx, and +230 in tx when all
is behaving. So far so good.

After the next "long talk" test, though, it did not even drop to zero when the PTT released, let alone go negative, and the longer I talked the worse it got (as high as +30 volts which is plenty of screen voltage to keep V2, the isolation stage from the balanced modulator, passing some signal. The carrier null is never perfect, and after going through V2 and the two IF stages (V3 and V4) it's plenty big enough to
confuse the AVC (tens of millivolts). The positive voltage slowly decayed back to 0 and then negative as things cooled off in receive.

Next thing to determine is where the leakage was coming from. Relay is perfectly clean under its dust cover, no carbon tracking anywhere. The diode in the line to the 6146 screens was good (it's a mod to the HW-100 but is already included in the HW-101, apparently to eliminate a similar biasing problem caused by screen emission in the finals).

With some more study of the schematic, it was obvious that only V2 and V7, the 6CL6 driver tube, directly tied their screen grids to that line. I deliberately turned the mic gain too high and read the alphabet into the mic until the cutoff line was being driven positive again... yanked V7 (that tube gets HOT while ratchet-jawing) and lo
and behold, after the hot-socket-removal "pop" the AVC right went to wide-open at its normal decay rate, and the cutoff line went instantly down. Just to verify, I put the DVM across the 4700 ohm resistor to  the  V7 screen and the current flow was indeed OUT of that tube (voltage more positive at the screen end of the resistor).

SO - after all this, it appears I talked myself into the wrong corner and the real problem is more subtle and indirect - the "leaky" 6CL6 runs hot and when heated even more by extended transmissions, its screen starts emitting. This causes the tx cutoff line to go positive, which in turn allows V2 to pass just a trace of carrier oscillator in receive, which when amplified by the two IF stages causes the AVC to
kill the rx gain ...

I sure hope I've made the right diagnosis this time. That line is definitely supposed to go negative and completely cut off the transmit stages, and it should never be positive in receive! I've just ordered a new 6CL6 and that will confirm it for sure.

What a pain in the ass. I'm about to go buy an R-390A   

-Charles

[WB3JOK]

Today I looked up the 6CL6 curves and I think it is run way too hot, which is why even a new one may eventually need replacing, and why there are problems on long transmissions.

The max screen current is supposed to be 7 ma, max diss 1.7W, typically 150V Ec2. But I measure 45 volts drop across the 4.7K screen resistor (which is actually drifted a bit higher in value), and 195v at the screen...  that is over 9 ma and 1.8W both of which are above maximum rating! Supposedly the shielding effect of the suppressor grid prevents secondary emission problems but as you can see it has other effects in this circuit... I could patch it with a diode, just like the screens on the 6146's, or maybe increase the dropping resistor a bit (as long as there's still enough output to swing the final grids).

Hopefully the gain budget is sufficient to allow a more sensible screen voltage. Since the finals run in AB1, very little power should be needed to drive them, so I'm not sure yet why they needed to run the driver stage that hard.

-Charles