Oscillation in Tube Guitar Amps

Parafeed813

Member
2012-08-15 8:24 am
I have thought for years that valve output transformers should come with coaxial cable leads for the high-voltage (primary) side. With the coax shield (outer braid) grounded, there would be much less chance that the signal here would be coupled back into earlier stages in the amp.

-Gnobuddy

Reminds me of the first guitar amp that I build. Nasty squeals when turning up the volume of the lead channel.
The guy that I got the transformers from (an old valve PA amp builder at the time) told me to use coaxial television cable to shield the large voltage swings. Connect the shield to HT so there is no(t much) dc voltage between the core conductor and the shield.
 

Gnobuddy

Member
2016-03-01 4:10 pm
... use coaxial television cable to shield the large voltage swings. Connect the shield to HT so there is no(t much) dc voltage between the core conductor and the shield.
I like the idea of the coax cable, but the idea of HT on the shields scares the bejesus out of me from the safety point of view. It seems like the kind of thing that would fry some poor amp tech some day - who would ever expect to find HT voltage on a coax shield?

-Gnobuddy
 

Gnobuddy

Member
2016-03-01 4:10 pm
The HF resonance of a hi-fi OPT is already usually lower than we would like, given that we want to add decent amounts of feedback. Adding any capacitance will move it lower.
Agree entirely, but I wanted to make an estimate of how much lower - i.e., is it an effect big enough to pay attention to?

Trobbins says transformers may have a few hundred pF of distributed capacitance already. If we assume only 200pF, and an external coax cable adds 20 pF, that is only 10%. Since LC resonance frequency depends on the square root of capacitance, that means only about a 5% change in the
resonant frequency. And that's more or less a worst-case estimate.

Given normal component tolerances, etc, I would consider 5% shift in upper cutoff frequency to be insignificantly small.
Good wire placement and physical design, combined with avoiding huge coupling caps, is the best method. Then, if still in trouble, add some grounded metal between the parts which should not 'see' each other.
I agree with the suggestions you offer. Definitely, layout, physical space, and additional electrostatic shielding are all good things.

But there are circumstances the designer of the original "box" cannot predict. For example, I have seen two different Fender guitar amps (a Super Champ XD, and a '65 Princeton Reverb reissue) become unstable if an external guitar effects pedal was placed within two or three feet of them.

It seems these amps themselves radiate enough AC electric field to be picked up by the external electronics pedal. I suspect the majority of the radiated field came from those output transformer primary leads.

In general, I think it's never a bad idea to minimise the size of the noise/ interference/ instability problem right at the source, when that is possible. After that, make the rest of the system resistant to what is left, and you have the best of both worlds.

One could always cut ouput transformer leads very short (and not use coax), but I'm reluctant to do that, because transformers are relatively expensive and scarce, and I usually hope to re-use them later in some other experiment. I guess I would actually prefer to use audio transformers with solder tags, rather than flying leads, but I've never run across one like this. (Keep in mind, I'm talking guitar amps here.)

No need for either of us to try to convert the other to our viewpoint, or anything - this is just a minor difference of opinion, really!

-Gnobuddy
 

Gnobuddy

Member
2016-03-01 4:10 pm
If so, this would be bad design.
What, bad design from the king of all penny-pinchers, Leo Fender? Amps that use a little aluminium kitchen foil glued to the inside of the particle-board enclosure (and grounded only by hoped-for light physical contact with the metal chassis) for shielding? No, that couldn't possibly happen! :D

In fairness, Fender was an accountant by training, and seems to have learned a little about radio-repair along the way. He wasn't an engineer. Some of his early amp designs (mostly lifted from the back of the RCA catalogues, with minor tweaks) would horrify an audio design engineer from later decades.

There were also almost no external guitar effects units in existence when the '64 Princeton Reverb was designed, so it wouldn't be too surprising if nobody gave that a second thought.

The Super Champ XD is a different story. That one was designed most of a decade into the 2000's. By then, the people involved should have known better.

-Gnobuddy
 

Gnobuddy

Member
2016-03-01 4:10 pm
I suspect that some aspects of some guitar amp designs would have horrified contemporary audio designers too.
Some of the acoustics aspects are almost as bewildering as the electronics ones, and that extends all the way to the present.

One of the head-scratchers for me is that a lot of today's little solid-state practice guitar amps use an open-back "cab" that's a shrunken visual copy of much bigger vintage cabs. What used to be a 3 foot x 3 foot front panel and 14 inches deep circa 1955, is now 8 inches x 8 inches x 4 inches deep.

Now you have all the internal standing-wave resonant frequencies of the cab shifted up by two octaves, into the really obnoxiously "boxy" sounding range. As for bass response - well, what bass response? Two octaves just got lopped off that, too, and it was limited to start with, even in the original full-size beast.

I wonder how many initially enthusiastic young people have been completely turned off guitar-playing by the horrible noises that come from these little amps?

-Gnobuddy
 

colorcat

Member
2011-12-07 3:05 am
What happened to all the Bright Young Talent?

A quick perusal of craigslist musical instruments gives all the indication needed as to, what becomes of "many initially enthusiastic young people" and the instruments they hastily purchase from GC.

Just to keep it on topic, I now finished reading all 11 pages as I'm about to commission a new build and would like to know as much as possible about tracking down oscillations, should any strange phenomena appear.

colorcat.
 
would like to know as much as possible about tracking down oscillations

Seriously, good luck there. That is like a cook asking why food tastes bad. A zillion reasons. Oscillations are not generic. They occur for many reasons. Often not related. Oh other than in the grand physics idea of signal feeding back somehow. But that happens so many ways. Lack of decoupling, lead dress, mis-wiring, omitting components, etc.
 
Hi Guys

"all 11 pages"?

Oscillation in any amp is caused by poor layout, which in PTP tube amps means porr grounding and poor lead dress. The fact there is oscillation can in itself be construed as a "a good thing" inasmuch as it indicates there is gain in the circuit.

99% of guitar amps use the same power amp circuit and values and negative feedback around the PA. The layout of the Scmitt splitter can be problematic but it has been around for enough decades that stable layouts are plentiful for anyone to copy. Still, with some of the marginal values chosen for the circuits and the variablity of PTP wiring, oscillation can emerge over time with component drift and aging. As The Ultimate Tone vol.2 points out, many high-frequency oscillations in MI PAs are triggered by low-frequency instability. becuase there are two coupling caps involved in the feedback path plus the rolloffs of the OT, it is an easy fix to make one of these capacitances ten-times smaller or larger to get rid of the oscillation. Both fixes work the same, moving one rolloff away from another so that the phase shift is reduced or moved to a nonproblematic frequency.

In many case, the oscillation occurs because gain has been added to the preamp. What was a stable layout for the old circuit is not stable for the new one, so more than just the preamp modification must be done.

When proper grounding and wiring is employed from the start and then continued with the modifications, there will be no fear of oscillation.

When tracking down oscillations, you have to break the circuit or shunt the signal path to ground to see how much of the circuitry is involved in the positive feedback loop causing the oscillation. First thing is simply to manipulate the controls on the amp to see if they affect things. You might add a snubber here or there to see if this stops the oscillation or reduces its amplitude. kepp all bandaids in place until you find where the oscillation stops, then go back to the first bandaid and remove them in the sequence they were installed.

All of the above methods apply to any technology of circuit.

Have fun
 

dotneck335

Member
2012-12-11 10:10 pm
What, bad design from the king of all penny-pinchers, Leo Fender? Amps that use a little aluminium kitchen foil glued to the inside of the particle-board enclosure (and grounded only by hoped-for light physical contact with the metal chassis) for shielding? No, that couldn't possibly happen! :D

In fairness, Fender was an accountant by training, and seems to have learned a little about radio-repair along the way. He wasn't an engineer. Some of his early amp designs (mostly lifted from the back of the RCA catalogues, with minor tweaks) would horrify an audio design engineer from later decades.

There were also almost no external guitar effects units in existence when the '64 Princeton Reverb was designed, so it wouldn't be too surprising if nobody gave that a second thought.

The Super Champ XD is a different story. That one was designed most of a decade into the 2000's. By then, the people involved should have known better.

-Gnobuddy
Yet today, many, many musicians covet the sound of those Fender amps. You sound like the hot-shot (or so they thought) electronic design engineers that CBS brought into Fender in the late 60's. They RUINED that pristine sound; they didn't listen to the musicians' opinions like Leo did. Hundreds of guitar techs make a good living UN-improving those later Fender designs to make them sound like the old ones did, which is what the musicians want. Those old designs are still the gold standard. There are some modern manufacturers (MilkMan) making amps the old-fashioned way---point-to-point wiring and basically the same designs as those old Fenders.
 

Gnobuddy

Member
2016-03-01 4:10 pm
Whoa there, pal! I never said anything about how they sound - many do sound excellent. And you are over-generalizing: just because many of those amps sound good, that doesn't mean there are no engineering problems with them.

Another over-generalization: you seem to be suggesting that *any* engineering improvement would ruin the sound of these amps. That isn't true, either.

I specifically mentioned Fender's use of kitchen foil used for electrostatic shielding, which is a cheap-as-can-be, Mickey Mouse way to do it. Basically, Fender Corp. was saving the cost of a sheet-metal cover for the underside of the chassis. That is extreme penny-pinching, Walmart style.

I also mentioned poor design. There are plenty of examples: for example, the infamous "interactive" controls on the 5E3. That is just plain bad design - it contributes nothing positive to the sound, and makes the controls ridiculously clumsy to use. It is such a stupid design, in fact, that it had to have been either a drafting error when the circuit was originally drawn up, or clear evidence that Leo didn't have a clue how a potentiometer actually works.

It's worth noting that 5E3 clones are still very popular with many guitarists of a certain age - but many of them modify the volume and tone controls to make them more usable, and less idiotic.

The third thing I mentioned was marginal electrical stability. This also comes from bad design (for example, omitting grid stoppers on many of the valves in the amp), and bad layout - no if's or but's about it. No musician wants an amp that doubles as a radio transmitter or squeals like a sick banshee if you put your pedalboard down too close to it - this is not a "feature", it's a bug.

Leo got many things right, for sure. And he also got some things wrong. Anyone who takes an objective look at his creations will see that both those statements are true.

Finally, just because a thing is popular with some guitarists, doesn't always mean it's good. For example, many guitarists love blues music, probably because it tends to be utterly (and boringly) predictable, and therefore easy to jam along with. But only about 5% of the general population still says blues are their favourite music - meaning 95% of people aren't particularly interested in it.

I am a guitarist, by the way, and I still have a 64 Princeton Reverb reissue. It too, has some good things about it, and some bad.

-Gnobuddy
 
Hi Guys

When tracking down oscillations, you have to break the circuit or shunt the signal path to ground to see how much of the circuitry is involved in the positive feedback loop causing the oscillation. First thing is simply to manipulate the controls on the amp to see if they affect things. You might add a snubber here or there to see if this stops the oscillation or reduces its amplitude. kepp all bandaids in place until you find where the oscillation stops, then go back to the first bandaid and remove them in the sequence they were installed.

All of the above methods apply to any technology of circuit.

Have fun

What fun ? Over 6 decades I've made several pokey high powered parallel output staged tube amps in my lab, none of which are unstable in any load; (yes, you might say 'what luck') ...My latest project is a triple parallel push pull (stereo 150+150W) and the challenge is invigorating; many seasoned amp makers know too well, that this technology (let alone parallel push pull) is next to ultimate for output transformer bandwidth and quality for the circuit arrangement. It may appear raving mad to go to such lengths of using 3x3x3x3 output stages, esp the heater requirements, but there are ways around this; but I've got loads of those Ruskie boring looking 6550's which can be put to good use at a lovely low B+ voltage. Who wants these bland looking tubes on the 2nd hand market ? However, reading your mail there is one caveat over all the others in design and benchwork that is so appropriate, and that is during the 1950's-60's I went to radio amateur evening school at the local tech. It was here the seeds were sown in what we really need to know the reasons that tube amps, RF and/or AF go haywire and oscillate. Those weaning years got me out of alot of later trouble. There shouldn't be any reason to guess or 'hit and miss'; which is the wrong way. I bet you the circuit will find a way of oscillating against your diagnostic will. Then one can be stumped to the next solution. I may be exacting; but really any well designed amp should take ANY speaker with full power confidence right into clip. You might reply, evening RF courses aren't around these days anymore...too true (but sad). Our cold-war days were a real learning curve. I can assure you that global feedback closed loop induced oscillations require some diagnosis with the poles and that wonderful output transformer phase shift is ready to ruin all. The troublesome issue for many builders is 'break that global nfb connection' and the oscillations cease.! I know it's frustrating but it's here that requires skill.
Your quote :- "First thing is simply to manipulate the controls on the amp to see if they affect things". That's the perfect recipe for peak-clip oscillatory distortion, so typical of the class A stretched B operative zone breaking into oscillation on signal peaks, as peak current is drawn. Be careful which operative class one is using and the levels that changing class breakthrough occur. (this is a big indication and is calculable). No signal or part signal Class A usually ends up going wrong into oscillations displaying a red hot output stage. Thrash a class A operative amp and it can actually sound better as it is at it's most efficient. Play real music on oscillatory class B; there's a good chance of destroying a tweeter as well as it sounding foul. Morgan Jones did right thing not to get bogged down with the oscillatory bit in his Valve amp books; darned right. Never want to be challenged in the text.
More later from the pit.

richy
 
Hi Guys

Richy, the thread is about oscillation in tube guitar amps and these have power amps with an open-loop gain of about 100. The technique I described is appropriate to guitar amps, however, as the Linear technology app writers state, the methods are useful in much more complicated circuitry with orders of magnitude more gain.

The problems you allude to are not of issue in a guitar amp where they may well be in a hifi amp. Guitar PAs are "old-school" hifi amp circuits. Modern tube PAs for hifi are generally more complex, usually with three or more stages rather than just two. Fault-finding in a 2-stage amp is way simpler and safer than for more stages - total agreement there.

Life is about having fun. Otherwise it's not really a life.

Have fun
 

Gnobuddy

Member
2016-03-01 4:10 pm
Fault-finding in a 2-stage amp is way simpler and safer than for more stages - total agreement there.
Having learned about Bode plots, dominant-pole compensation, unity-gain-stable opamps, and PID controllers long before I ever saw my first guitar amp schematic, I'm still amazed at what guitar amp designers have gotten away with.

Transformers and multiple gain stages within the global feedback loop were surprising enough. But even more unexpected was finding active tone controls (such as the misnamed "presence control") within a global negative feedback loop.

Of course, finding just a few dB of global negative feedback was also unexpected. I've had to re-adjust my preconceptions about what constitutes "normal" when it comes to negative feedback!

Life is about having fun. Otherwise it's not really a life.
The Dalai Lama agrees with you, more or less; he used the word "happiness" rather than "fun".

John Cage's mother, on the other hand, disagrees with both of you. In Cage's own words: [special= ...she replied, "Now, John, you know perfectly well I've never enjoyed having a good time."
]%[/special]

-Gnobuddy