Plate stop resistors?

[Enzo]

This is not my area of expertise, but in my mind, "parasitic" oscillations had to be perched on the signal somehow, peaks, zero-crossing, somewhere. Sustained or continuous oscillation would then not qualify as parasitic in my book. Unwanted, yes, but not what I would call parasitic. Now that may be totally wrong, I'd like to hear about it if so.

[soundmasterg]

Thanks for the link, I'll have to check that site out. Nothing is more fun than HV!

I don't think a standard high-voltage flyback transformer will fit in my handheld "trigger" enclosure, unless I get one custom made. It doesn't need to be anywhere near the power a monitor flyback tranny produces, as it is only charging up a 150pF cap to 12kV max. A variable high-voltage flyback controller would certainly be easier to design, with less circuitry. Do you know of any physically small high-ratio flyback transformers?

CCFL transformers, like the CTX210659-R, are very small, efficient, and good for up to around 1.3kV RMS (before the CCW multiplier) and I have designed plenty of switchmode current-fed Royer circuits for CCFL backlight controllers in LCD displays, so I am quite familiar with those. I have already spice'd up my royer/multiplier, so it has a good chance of working. In addition, the CCW multiplier allows use of multiple low-voltage rectifiers and caps, so I can keep them relatively small, except for the final 150pF cap required by the HBM, which will be required to handle the full 4KV - 12KV range.

(apologies to the OP for this continued sidetrack...)


RA

This stuff is interesting....I bet even the OP likes it.

Greg

[Steve Conner]

Enzo:

Yes, you usually see parasitic oscillations like that. The reason is that the gain of tubes and transistors is non-linear. The more current they are flowing, the more transconductance they have. So, in Class-AB amplifiers, parasitics always appear on signal peaks first.

The same mechanism can also cause continuous oscillations in a really bad design, where there's enough gain to excite them even at idle. Or, if the oscillations get started on a signal peak, they can generate enough current to keep themselves going.

So, because it's the same mechanism, I call those parasitics too. But you don't see those designs in the wild as a repair tech, because they catch fire or explode before they leave the prototype stage. Parasitics in MOSFET amps and tube RF amps can literally destroy the whole thing in seconds.

To me, "parasitic" just means oscillations that the circuit designer didn't want. They have other more picturesque names like sproggies, hooting, squegging and snivets.

Randall: Try a flyback from a smaller monitor, then.

[soundmasterg]

Yikes.

A single KT88 on full blast, as part of a push-pull pair, draws about 200mA of plate current. That's average. Because of the peak-to-average ratio of a sine wave, it is around 655mA peak, and because each tube conducts on alternate half-cycles, the RMS value works out to 328mA per tube.

(To get the RMS: divide peak by square root of two because it's a sine wave: and square root of two again because every other half-cycle missing)

328mA RMS through a 100 ohm resistor is 10.7 watts, so no wonder the 3 watt ones burnt up.

Whenever I've seen those plate stopper resistors used, they've been 10 or 22 ohms, not 100. A 22 ohm resistor would dissipate 2.4 watts, so if you want to use a 3 watt type, that would be a reasonable value.

Yes, if a plate resistor goes open circuit, the screen will probably burn out too. (Hopefully the screen resistor will die first.) So maybe using them as fuses isn't so wise.

The Sunn 2000S is using 47 ohm 5 watters on the plates. The Vox AC100 is using 100 ohm's, though I'm not sure of the wattage in the stock ones. I used 5 watters in my project one and they seem to be happy, of course those are KT77/EL34's in there.

Greg

[soundmasterg]

One more thing about parasitics - they don't have to produce sustained oscillations to be troublesome. It is very common to see what is called "boxcars", or small patches of oscillations riding on the sine wave at either the peak or at the crossover point. This is caused typically when there is a small disturbance, or discontinuity, that injects a signal into the circuit, causing it to "ring" at a particular frequency for a short period of time and die out. These are common in output stages where the leakage inductance of the output transformer resonates with stray and other circuit capacitances, and is excited by the zero-crossing switchover of the output stage, or the current reversal at the peak of the sine wave.

RA

Randall, when you see something like what you described as far as a little parasitics riding along at the crossover point or just the peak, is this something that needs to be remedied before the amp can be said to be finished? This is assuming that the amp otherwise sounds good, but these artifacts are seen in certain operation modes? I've seen this problem before on a couple occasions and have not heard anything audible that might be related to that and wondered if I should continue to try to troubleshoot or not.

Greg

[soundmasterg]

Every geek kid wanting a future in electronics should be a ham. I was, until I figured out the girls just didn't go for it, so I switched to guitar.

RA

My grandpa was a ham radio operator, does that count?

Greg

[Steve Conner]

Randall, when you see something like what you described as far as a little parasitics riding along at the crossover point or just the peak, is this something that needs to be remedied before the amp can be said to be finished?

If it were a solid-state amp, I'd say yes, it is a sign of impending doom.

I disagree with Randall's explanation of "boxcars", I prefer my own, that at that point in the waveform the circuit has enough transconductance to start hooting.

Or instantaneously has lost enough bandwidth that you get instability of the NFB loop- "Nyquist oscillations"- which is more likely if you see it at the crossover point, because here we could expect the OPT to be saturated under heavy drive at low frequencies, and that decreases its high frequency bandwidth too.

The oscillation frequency can tell you a lot. Proper parasitics are in the 10s to 100s of MHz, and may not even be visible on a cheap scope: "Nyquist" oscillations are in the 100s of kHz. The cures are different: parasitics are cured by stopper resistors on the offending stage, but the Nyquist kind are cleaned up by increasing that little capacitor between the PI plates.

In practice I'm sure quite a few old tube amps do it, and you can probably get away with it in a tube design. The old Crown in my avatar does it when overdriven heavily, but I decided that it wasn't a problem, since it's for hi-fi use.

[R.G.]

Ah, we get back to something I do have experience with. Parasitics are, as described, a place where the changing parasitics of the devices allow it to build up to oscillation somewhere in a cycle. This is a variable and can be a tiny fraction of a cycle or expand to the whole cycle under some environmental change. There's even a term for a common form of it - "bottom side fuzzies", which accurately describes a common manifestation of lower frequency PNP devices in old-school SS amps.

And it is as Steve describes is - your amp is on the edge of oscillation at some frequency and wanders in and out of the oscillation condition during a cycle of signal. I have had MOSFET power amps oscillate at frequencies I could not see with my 20MHz scope. Only when I wheeled in a 100MHz scope could I detect the problem, and it was really above that scope's ability.

[raiken]

Further info, from RDH4:

"Improvments in most cases may be secrured by the use of one or more of the following expedients:

A small condenser from each plate to earth
A condenser from each grid to earth (with transformer input only)
Series stopping resistors in grid, screen and plate circuits, arranged as close as possible to the valve.
Improved layout with short leads.
Input and output transformers with less leakage inductance."

BTW, a "condenser" is what a capacitor used to be called...don't go sticking an air conditioner in your amp.

RA

[Steve Conner]

Is mould from the air conditioner what causes bottom-side fuzzies?

[raiken]

I disagree with Randall's explanation of "boxcars", I prefer my own, that at that point in the waveform the circuit has enough transconductance to start hooting.

I don't disagree with your explanation, I think both are applicable. However, in support of mine, here's an excerpt from RDH4, page 572:

"With Class AB operation, the current in one half of the output transformer is zero for part of the cycle - this causes a very rapid rate of change of current at the cut-off point, tending to cause parasitics due to the leakage inductance of the transformer. This effect may be minimized by the use of a transformer with low leakage inductance - see Chapter 5 Sect. 3(iii)c - or by reducing the rate of change of current."

Also, from page 582:

"Parasitic oscillations in the plate circuit may occur with Class AB1 operation when the plate current is cut off for an appreciable part of the cycle, as a result of the transformer leakage inductance and the rapid change of current at the cut-off point - see Sect. 5(i)C."

If a circuit is on the edge of oscillation, sometimes all it takes is a small transient to push it into oscillation, but once the transient dies out (and you're back in the conducting portion of the cycle), there is not enough gain/feedback to sustain oscillations, so you don't see them in the entire waveform, only in the little "boxcars" riding on the wave at crossover or inflection points, where the rate of change of current is maximized.

RA

[raiken]

Is mould from the air conditioner what causes bottom-side fuzzies?

No, I think that is caused by cheap toilet paper.

[raiken]

Randall, when you see something like what you described as far as a little parasitics riding along at the crossover point or just the peak, is this something that needs to be remedied before the amp can be said to be finished? This is assuming that the amp otherwise sounds good, but these artifacts are seen in certain operation modes? I've seen this problem before on a couple occasions and have not heard anything audible that might be related to that and wondered if I should continue to try to troubleshoot or not.

Greg

Most of the time, they won't hurt anything. However, they can sometimes cause audible artifacts, or worse, what you see on the bench may change when you use the amp in real life hooked up to a different speaker. Different loads can cause the amp to react in different ways, and what is a small "boxcar" on one speaker cabinet may be a full-blown oscillation on another (depending, of course, on what is causing the oscillation - if it is internal and the output stage doesn't affect it, this is not applicable, of course).

As a general rule, whenever I see this in an amp, I fix it, because it shouldn't be there in a proper amp. Usually it occurs in the output stage/phase inverter section, and can be cured with the standard grid/screen/plate stopper type resistors and proper lead dress and shielding. Barring that, the next step is the usual capacitors, like the PI plate-to-plate cap, or other local option caps.

Most of the time when I see an oscillation like that it is easily cured with a grid stopper resistor right on the PI input, usually around 33K or so. All cases are different, because the oscillation can be caused by so many different things, so you have to take each case individually and poke around until you find the cause and the best remedy.

RA

[Wilder Amplification]

Now Randall...these HF oscillations in the RF pass band you speak of...is it typical for those who possess "golden ears" to hear that?

[raiken]

Now Randall...these HF oscillations in the RF pass band you speak of...is it typical for those who possess "golden ears" to hear that?

Of course!

Actually, even if the oscillation is at RF, it can sometimes cause audible artifacts. One such artifact is called "quieting", where the background noise of the amplifier suddenly drops way down at some control point, and you think you've got this nice, quiet, well-behaved amp, but what is really going on is a full-bore oscillation sucking the life out of your tubes and masking any noise amplification. Also, repetitive peaks of noise that occur at a low frequency, like 50/60Hz rectifier noise, can induce audible hash because of the repetition rate, even though the oscillations themselves are above the audible range.

RA