Parasitic oscillation caused by NEAT lead dress!

[eschertron]

yep, that's right. I spent the entire night troubleshooting what should have been a no-brainer mod in my homebrew 18W amp.

I wanted to tame some of the ice-picky frequencies after the distortion stage in the preamp, so last week or so I jumpered a 500pF (and alternately 1nF) cap from the anode to ground at the preamp stage 3 to see if my math was in the ballpark for a LPF application. Sounded OK, so tonight I put together a 2 pole 6 position switch with 500pF, 1nF and 2.2nF caps to get a range of values. It works now, and I verify that by turning up all the gain and hearing how much the hiss is affected (oh, and by playing through it!).

What took me so long was that version 1 through about version 17 of this circuit plagued me with howl, motorboating, and oscillations that were affected by gain and tone controls, and, well... I was about to throw in the towel until I recognized that the jury-rigged version of the circuit always worked, yet the permanent (neatly-dressed) version of the circuit didn't. So as a last resort, I strung about a foot of 20ga wire between the caps (and their selector switch) and the anode of the gain stage. Doesn't matter where the wire hangs, in fact right now it's just coiled across the other components in the chassis as it came off the spool. But it works like this.

I don't understand why trying to keep my circuit path neat and short would cause problems, when the theory I'm learning tells me the opposite. At normal audio freqs a 1nF capacitance (plus or minus a factor of 2) has a pretty high impedance, so why does a long unshielded wire between the cap and anode not pick up more interference, but actually less?

Any insight will be appreciated!

[R.G.]

When is a wire not a wire? When it's a resistor or an inductor.

Straight wire has a certain inductance per unit length, and coiled wire has more.

I don't think the right observation to draw from this set of experiments is that, in general, long, messy wires work better than short, direct ones. I think rather that it's something like the positioning of your switch for the short, tidy wiring is letting some oscillation happen that is squashed by the extra inductance of the long wires.

I am not sure what about your physical wiring setup lets this happen, but it is not an invalidation of Mother Nature's rules on how voltages and currents work. She is very persnickety about that always being the same. It could be that the grounding on your multi-switch is unfortunately made so the ground currents from the switch let just enough coupling to another section happen unless the phase is hosed around by longer wire. It could be that the long wire is throwing off the unfortunate tuning that the neat, tidy wires set up. Could be other things. But the Rules still apply.

"Neat, tidy, short wires" are not the Rules. It's a simplification that's easy to remember and conceptualize. "Short" is a rule because every unit of length adds some self-inductance to a wire, and some capacitance to other nearby wires. "Direct" is another Rule; actually, a corollary to "short".

"Neat" and "tidy" are sometimes counter to "short". A true point to point chassis has a "star" of components coming off each tube socket for its next destination, and looks very un-tidy. It just happens that "neat" and "tidy" are more often benign than unruly, randomly routed wires are.

The actual Rules are Maxwell's Equations solved for every nanometer of length of every charged or current carrying conductor to every other conductor in the universe, which gets merrily out of hand. The complexity of these solutions exceeds normal human grasp, which is why RF design with non-microscopic components is usually thought of as black magic by RF professionals.

[Chuck H]

What R.G. said ^^^ And very well I think. But just to be more direct and to the point, it's very likely that whatever oscillation caused by "cross talk" (the invisible inductance and capacitances that occur because of lead and component proximities), which in your tidy circuit must be creating positive feedback somewhere, is being countered by the addition of your "messy wire" which just happens to be creating negative feedback that reduces gain at the offending frequencies in the right places. For amplifier types that are more sensitive than even guitar amps there are designers that better understand these principals and commonly use layout as part of the circuit. To be sure, sloppy lead dress and ill considered layouts usually cause more problems than they solve. For our purposes it's best to start clean and proper and then solve for any problems. Which is what you've done without entirely understanding what exactly has happened. What exactly is happening in the invisible circuits of an amp are rarely known because as R.G. said it "gets merrily out of hand". It could be, and probably is, part of the reason some amp designs seem to have a certain mojo. But the mojo isn't magic. It's electronics. Just the part that's unseen and poorly understood by most tinkerers and amateur guitar amp designers. I'd say be happy with your success and enjoy the amp as it is. That's how many guitar amp circuits are achieved. Rote replication of a working example layout. It's often been said that amplifier circuits are more than just schematics. This is a part of what that means.

[Alan0354]

I work on system layout and pcb design for years. Neat looking and function never goes together. Best layout is always knowing what is important and go point to point. Lately I have been on DIY audio tube power amp forum. Those are ten time more critical because they have to work close to 100KHz to get flat response to 20KHz. Layout is so much more critical. I saw a gut shot of the famous Cary amp, it's like a rats nest just like the picture of my amp. Attached is a picture of my latest build. It's a channel switching with switchable between 3 stages or 4 stages OD channel, power scaling. It is quiet like a mouse, you cannot even make it oscillate. The only time I can make it oscillate was with the Marshall style OT that I had a long thread here.

Guitar amp have very low frequency response, you can get away with a lot, but sometimes it's still oscillates depends on how much gain.

[trobbins]

When high gain amps with lots of feedback first started to be diy constructed, it caused much mayhem, and made almost every diy magazine article for the next 10 years make some comment or two about not observing any parasitic oscillations with their latest build.

The Williamson started it 70 years ago - and it was only the savy builders of that amp that used 'distance' and functional layout to reduce every form of parasitic capacitive coupling between circuit nodes. 'Distance' may not be a tool in your arsenal, but layout that is viewed through 'parasitic coupling' glasses still is.

Any high impedance node (such as input grid circuitry that is separated from a low impedance source like a plate or ground by a lot of resistance) is very prone. This is why shielded cable (one shield end only tethered) is typically used for a signal between a preamp valve base on one side of an amp chassis to a volume or tone pot on the other side of the chassis.

[eschertron]

I think rather that it's something like the positioning of your switch for the short, tidy wiring is letting some oscillation happen that is squashed by the extra inductance of the long wires.

to the point, it's very likely that whatever oscillation caused by "cross talk" (the invisible inductance and capacitances that occur because of lead and component proximities), which in your tidy circuit must be creating positive feedback somewhere, is being countered by the addition of your "messy wire" which just happens to be creating negative feedback that reduces gain at the offending frequencies in the right places.

Inductor? I know the inductors we use in amp building are wires generally coiled around a core, sometimes only air. I haven't thought about a single straight wire as an inductor, but I see now that it is.

How often is self-inductance more critical than mutual inductance (or capacitance)? I still can't seem to clear the hurdle of thinking of every wire as 'antenna'

[Alan0354]

I built 3 high gain amp in a roll and not a single oscillation other than that Marshall style OT that squeal. Still, I need to move the guitar to get it going. I know it's going to start argument again. It's magnetic coupling, not capacitor coupling that is the major problem. It's the long input wires that form a loop with large area in the middle that pick up and emit the field. To show capacitance coupling is not an issue, even if you run two wires in parallel for 12", I doubt the mutual capacitance is 10pF. The standard 50ohm coax is only about 3pF per 12", here you are talking only two wires running parallel, not necessary right up against each other.

In fact, in my Bassman proto platform, I have 3 high gain front end all clamped in, one Dumble 4 stages, one JCM 800 2204 with increased gain, and my own 4 stages. It's always on whenever I turn on the amp. Not a hint of oscillation. It's all about layout. I don't even have to have small caps from signal path to ground to tame anything.

Traditional guitar amp always put input, volume and tone at the front, tubes at the back with components in the middle, this is the formula to get long high impedance input wires that form a loop antenna that receive and emit signal into the air. Guitar amp only run at about 5KHz. It is not even hard to tame a high gain amp. That's the reason when my amp squealed last time, I reach the conclusion it's the OT very fast and was proven true. I can reproduce the problem to the amp I moved the problem OT to. Even that, it's wasn't bad, it's absolutely useable. Just don't stick the guitar 6" from the amp!!!

[Alan0354]

Inductor? I know the inductors we use in amp building are wires generally coiled around a core, sometimes only air. I haven't thought about a single straight wire as an inductor, but I see now that it is.

How often is self-inductance more critical than mutual inductance (or capacitance)? I still can't seem to clear the hurdle of thinking of every wire as 'antenna'

Inductance of a single wire is not important in audio frequency. If you start coiling, then you can have problem.

[g1]

"Neat" and "tidy" are sometimes counter to "short". A true point to point chassis has a "star" of components coming off each tube socket for its next destination, and looks very un-tidy.

Neat looking and function never goes together.

The most "tidy" looking wiring I can think of are the Joyce wired HiWatt's. Great for reliability, but I think there may be some sacrifices with regard to audio (noise etc.) when form is favoured over function in cases like this. I don't think you could follow best audio practices and come out with something that looks so perfect.

[Justin Thomas]

I like a Hiwatt as much as anyone, but you bring up an interesting point. I've never had the privelege of playing a "real" Hiwatt at all, let alone at a "reasonable" volume. Are there any recordings of a guitar into a Hiwatt without pedals or anything? Did Townsend use any pedals on Live at Leeds? Only other Hiwatt players I know of are Gilmour & Lifeson, both who use effects. I guess what I'm saying is, people play Fenders, Marshalls, Voxes, etc all day without "enhancements." Don't even mention Matchless, Dumble, Trainwreck... Maybe that super-clean Hiwatt is a great platform or foundation to build upon. There's nothing wrong with that, and I know a lot of guys who use them wanted loud, clean, and punchy. Maybe it's uninspiring by itself? Like I said, I've never played one, so I'm just guessing. And if I could afford one, I'd get one... only one I ever saw in person was 200W & $4000.

Justin

PS: "best audio practices" to a guitarist is just a wee bit different to a Hifi guy...

[Chuck H]

Not that it's worth a lot, but I hate the tone of HiWatt amps. They have the peaks in all the wrong places, they're noisy and the natural OD tone sounds like radio static. Too scratchy and thin. Just listen to Townshend's tone with a critical ear and you'll hear it. Of course "he" made it work because he's a brilliant player and composer. Finding a way to make any sound musical is the gift of such types. But at face value his tone is just plain bad at it worst and pedestrian at best. And yes, I have played a few HiWatt amps for myself before making these statements. Used as a strictly clean platform may work better. I think Martin Barre of Jethro Tull used HiWatt's in this manor with a pedal board and his tone doesn't bother me so much. I think David Gilmour used them for a while too. His tone is ALL PEDALS! and much better than other HiWatt users. As far as I can tell, plugging into a HiWatt is equal to plugging into a P.A. And I suppose I should site Stan (km6xz) here and mention that REAL musicians just find the useful properties of any tool and start making music. No renown guitar amp deserves this recognition more than HiWatt IMHO. JM2C on HiWatt amps.

[Justin Thomas]

I still see Dave using them... I actually remember hearing Pete click SOMETHING on and off mid-solo on LAL. I actually really like that Leeds sound (I like Syd Barrett's sound too, so you can tell I'm from the Other Tone Planet), so we can agree to disagree there... I don't actually like their studio recprding sound. But that's what I was thinking - the people I know who use them used them with pedals etc. Hmmm... so, they're kinda like a Mesa power amp, except a pantload easier to work on? Sorry, couldn't help it. Working on Mesas seems to be one issue that almost everyone here can agree on. And, even if they CAN be compared to a PA, I'd still rather plug my Tele into a PA-guitar amp than an actual modern PA. Bleck. Tried it. #@$% that $#!+…

Justin

[eschertron]

When is a wire not a wire? When it's a resistor or an inductor.

Straight wire has a certain inductance per unit length, and coiled wire has more.

So the final entry here, tonight. I looked at a few websites that allowed me to play with wire parameters (length, geometry, etc) and calculate an inductive reactance. Turns out for the 'messy' wires that I'd been using (jumpers, or the temporary solution I came up with) at the frequencies that seemed to be involved, the reactance ended up being in the 100R range. so tonight I replaced the wire with a 150R resistor - just what came to hand - and all is good. The noisy interference did not come back, and I can convince myself that 150R in series with a couple nF of capacitance is not going to be audibly different than the theoretical zero-Ohm straight wire that I originally intended.

Thanks, all, for your advice and support.