Squegging in LC oscillators

(snip)

(my guess) The grid acts as a rectifier that builds a DC grid bias voltage from the rectified AC signal, biasing the tube off. Then the tank dies a natural death. Adding the series resistor reduced the efficiency of the rectification.

Sounds like you have too much positive feedback, to begin with.

But in a "normal-functioning" oscillator, the DC grid bias doesn't cut things off for so long, right?

The P-P amplitude at the grid (as seen by my 10x scope probe) when the circuit is not squegging is in fact larger than when it is squegging.

I suppose it is possible there's some weird kink in tube characteristics for all the pentodes I tried.

Probably, but moving the tap on the tank coil had little effect. The handbook says about a third of the way up from the ground end, but I tried it at a half, two-thirds, one-tenth, etc. It did alter the shape and timing of the squegging a little bit but it was still squegging. If I moved it too far the circuit didn't oscillate at all (too little feedback).

Also, changing the biasing (trying to move it further into class A) by putting a cathode resistor in didn't help much either.

Tim.

Right. If you don't have way too much loop gain, the bias just shifts enough from the grid leak effect to slightly lower the loop gain so that a stable oscillation takes place. The negative feedback loop that adjusts this gain adjustment effect can be stable or unstable.

My concept may be over simplified, and not include everything that is happening. I would look at the screen bias voltage during the squeeging to see if it is also bouncing with it, or remains stable through a cycle.

Spice is your friend. I have found that LT spice is particularly good at predicting squegging, at least in general. There's never a 1:1 correspondence with the real thing, but close is pretty good in RF.

When I have experienced squegging like this it has been because I have built a circuit with a resonance between the circuit capacitances and the RF chokes (you know -- those things you use for biasing that you treat as 'shorts' in your AC analysis?). To the bias circuit, the oscillator looks like a negative resistance, so you get two modes of oscillation at once.

This is complicated by the fact that the circuit wouldn't oscillate at the lower frequency at all if it weren't for the action of the intended oscillation -- it's the current requirements of the active device as oscillation build up that cause the negative resistance action that causes squegging.

I suspect that your grid resistor is killing the Q of the circuit at

30kHz, which kills the squegging without killing the intended mode of oscillation. You may find that loading any bias chokes with carefully selected, unbypassed series resistance will also kill the squegging. You can also sometimes kill it by reducing selected bypass capacitances to lower the circuit Q at the squegging frequency.

Of course all of this also lowers the circuit Q at the desired oscillation frequency -- that's life. Your job is to find a happy medium that gives you good oscillator performance without squegging. Or start building superregenerative receivers!

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google?  See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

Wow, man, if I make it happen again I see the screen voltage motorboating up and down by about 3V (around the nominal 150V) at about

30kc.

Changing the screen bypass capacitor between

680pf/0.001/0.002/0.005/0.010 and changing the current through the 0A2 between 5mA and 10mA and 20mA and 30mA doesn't stop the squegging but it does somewhat alter the timing/amplitude.

Plate can be held at a steadyish 350V (even bypassed) through all this.

So this is something like the textbook squegging which seems to be something like a motorboating of the plate voltage, but in my case I see it in the nominally regulated screen instead. New one for me!

I don't think this is quite like the typical NE-2 relaxation oscillator circuit, because I thought 0A2's were supposed to be stable with these small amounts of capacitance and the behavior seems independent of room lighting, but I could be wrong. The screen voltage waveform sure as hell looks like a relaxation oscillator at 30kc.

Looking at my old schematics I see my Heath HW-16 crystal oscillator puts the crystal between the screen and the grid of a 6CL6. Manual says that the screen is serving as the plate of the oscillator. Probably completely unrelated to the intended operation of my oscillator (where I bypass the screen and the leads are short) but may be related to the unintended mode of operation!

Tim.

What are the three terminals of your oscillator? Your screen is grounded, so I assume that the grid and cathode are both floating at RF

-- is this so?

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google?  See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

I think you are talking yourself out of the problem.

It sounds like you are trying to build an electron-coupled Hartley oscillator. There is no way that a voltage change of 3 volts on the screen (which is functioning as the anode, for Hartley purposes) is going to start and stop the oscillator.

Watch the DC voltage on the grid. You will see it follow the relaxation oscillator waveform when the thing is squegging.

The fundamental problem is two-fold. First, too much feedback. Second, the time constant of the grid circuit is too long. The stored energy in the tank can continue to charge the coupling cap even after the plate current is cut off, and it can't start to oscillate again until the charge on the coupling cap bleeds off.

Start by lowering the grid circuit resistance and lowering the capacitance of the grid coupling capacitor. This will work to cure both evils. Taking a quick look at the 6AH6 curves I would guess that, at 6 MHz, your coupling cap should be about 22pF and the grid return about 10K. The 6AH6 is also a sharp-cutoff pentode, so it will be very sensitive to small grid voltage changes near cut-off.

Don't get distracted by the swings in screen volts, you are just drawing large pulses of current and pulling the supply down. Unless the 0A2 is going out of conduction RC oscillation is unlikely.

What you write, which relates the Q (probably very low hundreds) of the tuned circuit and it's frequency to the period the squegging, is in good agreement with the squegging I observe. Will check the DC grid voltage with the scope tonight.

The grid tuned circuit is at 1.8Mc, and the squegging seemed insensitive to the coupling cap. 100pF is the handbook value, but I did play around with it. Too small (39pF) and no oscillation. At 56pF there was squegging.

Did not try playing with the grid return value (still at 47K from the handbook), that may indeed be the key.

Tim.

The circuit is the "tuned plate Hartley" as it appeared in any

50's/60's/70's ARRL handbook.

The grid tuned circuit is at 1.8Mc, the bottom end of the inductor is grounded, there's a tap nominally one third of the way up to the cathode, and the top of the inductor is connected via a 100pF capacitor to the grid, which has a 47K to ground.

The plate circuit in my current incarnation is tuned to the harmonic at

3.6Mc. The "untuned plate" version in the handbook has a RF choke instead of a plate tuned circuit, and indeed I had this on the bench for a little while.

Tim.

I normally don't post, but, since I have done my own amount of torturing of the 6AH6, I think i know what the problem is.

The 6AH6 is slightly misnamed. The 6AH6 isn't really a pentode. It's just a really small beam power tetrode. It has usable gain up to the UHF region. And it suffers a lot of the same personality disorders of it's larger brothers.

It's rated at 3.2 watts plate dissipation. But I have ran it close up to about 12.5 watts for extended periods. At that point, the tube starts to show obvious signs that it's close to the breaking point. The plate starts glow red, but it didn't kill it. Probably hurt it a good amount, but it didn't kill it.

The more current it pulls, and the lower the input and output impedance, the higher the tendency for it to break into oscillation. It will oscillate up to the UHF region easily. It's preferred method of oscillation is the grounded grid form. The cathode is the input, and the anode is the output.

When you have low resistance feedback loop from the plate to the cathode over six or so inches long, and start pulling a good bit of power, then it can go at any time.

Look up parasitic oscillation in your handbook.. You will see the circuit that is causing it. Large power tubes top out at 100Mc or so. But this small one can go up to 400 to 900Mc if it's being driven hard enough.

The loop is most likely from your plate, through the tuning cap of the output tank. The cap will look like a short circuit at UHF. It will then go through the chassis. And then through the cathode tank, which is shorted by the tank cap at UHF.

Ways to solve the problem. Turn down the plate, and screen voltage. It will drop beam current. That will increase the resistance of the tube to the point it can't parasitically oscillate.

Or you could put a 100 ohm or 1000 ohm resistor in the plate circuit. In big amplifiers, where you can't tolerate a large resistance in the plate lead, they use parasitic chokes. The appear as a short to DC, and the desired operating frequency. But they appear as a high resistance to the VHF oscillating frequency. It basically de-Qs the VHF oscillating circuit.

But you can tolerate a little bit of loss in the plate circuit of this unit. So, just put a 100 ohm resistor in series with the plate. If that don't work, then try a 1K resistor. I am pretty sure that it will settle down and behave. It will be a lot easier that winding a parasitic choke for it.

You also want to take a careful look at the screen circuit. It can also act like an output in a parasitic oscillation. So, try putting a 100, or 1K resistor in series with it too.

It's oscillating cycle. It starts it's HF oscillation gain cycle. The signal grows, and the tube bias increases. The current on the positive peaks grown with it. The current on the positive peaks get to the point that the tube breaks into UHF oscillation. The UHF oscillations are several times larger than the intended oscillation. The UHF oscillation only last a fraction of the high frequency cycle. When it starts it's UHF oscillation, it quickly charges the grid capacitor up to a very high level. Way past cutoff. The UHF oscillation quickly extinguishes it's self. But since the tube is in cutoff, the intended HF oscillation dies too. The grid voltage drops back to normal, and the tube starts the cycle again.

Other ideas You could also increase the grid leak resistance. Push it up to 1M or so. It will decrease the tank loading, and allow it to start easier. And it won't pull so much positive peak grid current, and cause the tube to conduct so heavily on the positive peaks.

You can tease out the UHF oscillation into the open if you short the primary grid tank, so it can't oscillate at it's intended frequency. Basically, zero signal condition. Then, push up the plate, and screen voltage. If you have a little signal strength meter that works up to the UHF range, with it's antenna clipped to the chassis, or wiring. You will se it start to come alive. You will be able to move your hand around the unit, and it will break into, and out of oscillation. Sometimes, it will motorboat, or operate continuously.

You want to change the circuit so that you can't get it go parasitic, even when you really push it to it's limits..

Then un-short the tank and you are ready to go.

Tim,

My guess is the OA2 (Neon) and Cap combination is causing the problem. Isn't that an oscillator in itself ?

Try replacing the OA2 with a Zener (or 3 in series). It may be a bit noisy, but it may also be stable.

Or... a pair of resistors to derive 150V from HT. Just to see if squeg. stability improves/changes. 15k + 18k for a 350V HT (just a guess).

-Mark

Tim Sh> I have been playing around with my homebrew VFO, a Hartley oscillator,

OK, I saw the parasitic last night!

I only have a 100MHz scope. Putting a scope probe on the circuit seems to often kill the squegging. But I put a little loop on the end of the scope probe and sniffed around, and indeed there was a 400MHz parasitic that would build up over maybe 0.1usec. Then the impulse from this set the tank ringing at 1.8MHz, and after the ringing mostly decayed (30usec or so) the cycle would repeat.

There seemed to be the most energy at the well-bypassed screen and not at the plate, but remember this is an electron-coupled oscillator so the screen is working as a plate.

I guess even a 0.001uF ceramic cap with short leads isn't a good bypass for 400MHz!

There must've been substantial 400MHz energy to show up on my 100MHz scope :-).

My band-aid of a resistor in series with the grid seems to do the trick.

Tim.

Your resistor in series with the grid would have much more effect on a UHF oscillation than on one at 30kHz, indeed.

Even a 1nF chip cap soldered directly between the screen lead and cathode of the tube would have the internal tube lead inductances to contend with, so you're probably never going to 'get there from here' by that method.

I wonder if a loading resistor in the screen lead instead of the grid would work?

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google?  See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

Nice post. A while ago, I went into "hair-pulling mode" the other way - *trying* to cause squegging in a one transistor DC-DC booster. Finally got it with a bit of fiddling. Maybe it belongs under Murphy's section on oscillators:

When you don't want oscillation, it will oscillate When you want oscillation, it won't When you want to get rid of squegging, you can't When you want squegging, it won't

Do all Parisitics (sic) come from Paris? :-)

Ed

torturing of

Yes. My original reply also mentions that.

.................................... You also want to take a careful look at the screen circuit. It can also act like an output in a parasitic oscillation. So, try putting a 100, or 1K resistor in series with it too. ....................................

He says that the .001uf cap isn't much good at bypassing. The problem is it's too good at bypassing. It acts as a short to the UHF parasitic. That allows enough UHF energy to be conducted to the chassis to allow it to go parasitic. Shorting the energy to the chassis, isn't the cure to the problem, its the cause of the problem.

So I will clarify my previous statement.

You want a 100, or 1000 ohm resister in series with the screen. Between the bypass cap, and the screen. Not between the bypass cap, and the supply.

You want to isolate the capacitance from the screen. That will break the loop required for oscillation.

I have a hard problem making a amplifier that can put out a solid 1.8Mhz signal. But I can build a DC power supply that produces a quality 80Mhz signal.

Maybe, if i want an oscillator, I should try to build a DC supply circuit, and vice versa. I will figure it out one of these days.

They always talk about the French whine. :-)

Indeed this is a working fix, as I verified on the bench.

Actually in my particular case it appears that the UHF parasitic starts first. But probably just the luck of the bias point.

Very true here - this is in fact the only time I've seen 400MHz stuff on my 100MHz scope, so I'm guessing that there is substantial UHF energy there.

This part is very true too. Tiny tiny fraction of a microsecond. My interpretation of the scope trace is that the UHF activity gives the the HF tank a "kick" and it rings for a while.

Tim.

Another couple of points: circa 350 volts sounds far too high for a vfo unless you want significant power from it. It's decades since valves, but iirc, 100 to 150 volts on the anode is the usual level for a high stability valve vfo. From what I remember, keep the anode current and drive level low, overall gain only just high enough to ensure startup and avoid use of rf chokes - resistors are better and won't self resonate anywhere. The tap at 1/3rd sounds too high as well. More usual is about 5/10% as this reduces loading on the tuned circuit and improves stability. Finally, have you measured the anode current to ensure valve dissapation is within ratings ?. At 350 volts, you could almost be at cherry red anode level. Don't have an oa2 data sheet to hand, but I think there is an optimum point for stability in terms of drive current as well.

Old versions of the arrl handbook are full of this sort of stuff. 50's,

60's etc. Abe books is your friend :-)...

Chris

(groups trimmed)

Check for dynatron oscillations; the screen acts as a plate, sort of.

I can't find a schematic of one, but it's just an oscillator with feedback from the screen to the grid.

How long are the leads of your screen bypass cap, and where are they soldered down?

Thanks, Rich