The dumbest thing I've ever seen in amp construction....

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I thought the dumb thing was how close the two power tubes are to each other.

That was my first thought as well and probably contributed more problems than the transformers.

GE routinely welds their cores to the frame in many of their Industrial Control Transformers. Yes, they run hot since their core is barely sized big enough for the power rating. The welding may contribute a percent or two to the total losses in the transformer, but it cuts down on the buzzing from the core.
 
Due to cost cutting, welding transformers instead of screwing has become very common during the last decades. Even most of the laminations' alloys are composed to be welded nowadays. Transformer manufacturers even don't swap the E I arrangement with every single lamination when assembling the core, instead they put all the E's from one side and the I's to the other, just to ease welding.

I don't see a significant issue with this. Yes, the lam's still are isolated, preferably by oxide surfaces. Now it is your's to imagine how the flux travels through the core: Isolated lam's instead of a solid iron core has the huge advantage that there isn't closed circle, i.e. a virtual short widing perpendicular to the flux. This remains even with welded cores, as the welding just forms an open straight line instead of a circle. Hence, losses caused by the weldings can be neglected.

One could suspect, though, that weldable laminations are of inferior quality when compared with GOSS, for instance. Hence, increased losses were a matter of laminations quality, not of the welding process per se.

Best regards!
 
What some of you are missing is that the location of the weld across the laminations isn't important. The fact is that the laminations are ELECTRICALLY shorted and as current always follows the easiest path, welding across the laminations means that any current within the lamination is going to flow through the weld.

It reduces efficiency and creates unnecessary heating. If the designer is willing to accept that compromise and has sized the core to withstand the additional heating, that's his choice.

But I'm not under any sort of mistaken impression that Peavey's designers knew what they were doing.

When I see a welded transformer in a Peavey amp I do the only thing that is appropriate: Shake my head and walk away laughing.
 
Welding on the outside of the laminations creates no current loops, so not a problem. If you welded on the inside _and_ outside of the core, that would be a shorted link and the core would overheat and possibly melt. Flux has to cut a full loop to cause any current to flow.

This is the same as when aluminium voice-coil formers have a slit - not a complete loop so no current flows in the former.

Its also why having bolts going through the lam stack isn't a problem.
 
I have seen transformers welded across the laminations, I think it was for a microwave. I found it odd the first time I saw it but thought it might be for vibration. Might be around here somewhere yet. As said, if it does not make a complete loop it is an unshorted conductor.
 
As mentioned earlier, many problems arise from mutually excluding black-white thinking, while Reality is made out of all kinds of grey.

"transformer laminations are isolated" .... well, yes and no :rolleyes:
If you measure resistance pressing both probes on a lamination surface, whether same side or opposite ones, you will *generally* measure lack of continuity.
Yes, all varnish and most oxide treatments are insulating and are applied for that reason.

BUT

if you have an assembled transformer and measure resistance from one lamination to another in the stack, whether it´s the next one or they are at opposite sides of it, you will usually measure continuity :eek:
Such a low value of resistance that it approaches a short :eek:

I guess our friend has never made that simple test which I suggest he carries on for his own peace of mind :)

and the reason for that is that EI laminations are punched out of a sheet of surface treated steel , so surface is insulating, BUT edges are sharp, *naked* and happily touch each other so all laminations in a stack are connected/shorted by the edges.

But then, what´s the point of insulating them at all?
Or even using stacked laminations instead of a solid core?

Point is, and here lies the key, that there are "shorts" and "SHORTS" and everything in between.

Consider a stack of laminations: IF they are unvarnished, lamination to lamination across the full stack resistance may be 0.0001 ohm ... but if they contact just by the edges, same stack resistance is WAY higher, say 0.01 ohm.

Are both the same?
YES to Black/White mentality
NO to a realistic one.

As different as one having 100 times more losses than the other :eek:

---------------------------------------

As of welding across laminations, corners are almost/practically outside of the magnetic lines path, so welding "short" is not a game changer.
Specially considering lamination sharp edges/burrs are already connected to each other ... with little negative effect.
 
^ This. There are many other dumb construction decisions but I don't think welded transformers is the worst. I'd classify 'dumb', those engineering decisions that endanger life rather than cause inefficiencies.

There is an interesting research paper here that models the eddy current losses caused by welding steel laminations: (PDF) Modeling of Eddy Current Losses of Welded Laminated Electrical Steels

As a non-EE it seems to suggest that at lower frequencies (~50Hz) the losses aren't too significant but the significance raises as the frequency raises. So welded PT not so bad, welded OT probably not so good? In that case, would it be possible that the losses induced on the OT could influence the sound signature of the amp?
 
Last edited:
Account Closed
Joined 2010
The title of this topic and its outcome shows a lot about how biased we get to be because a lot of unprofessional opinions that fill the audio world.I was and still am a victim of many things that i don't understand well enough .

The usual online textbook on Eddy currents is shown a lot as just a function with no variables.
I could say that there are no losses at all when welding the corners on their edge, on the contrary, that there's a gain in doing it. How will you fight me ? :)
 

PRR

Member
Joined 2003
Paid Member
We use laminations for eddy; but also because steel must be *worked* to refine the grain and improve magnetic property. You can't work a bar as much as a thin sheet.

To break-up eddy-loops we only need resistance to be not-small. Actually a prime contributor is Silicon Steel with internal resistivity about 5X that of plain steel. Very little more is needed to make eddy loss negligible. The common treatment for decades has been, not varnish, but RUST. They steam the lams. Hardly a glass-like insulation but ample for making eddy small.

I never thought about it, but they could be using Silicon Steel for the welding rod and mounting plate. Ferro-Silicon is not expensive. Bulk Silicon Steel (rather than lams) is not a big commodity but may exist.

A little shunt-loss is not a terrible thing in an audio transformer. Too high a Q encourages ringing and increases fly-back when clipped.

Here's the real thing. What do transformers die of? Within warranty period? Very few die of smoke-out. Large numbers must come back after drop-tests ripped their iron off. While a maker could cry "abuse!", users never think they are at fault. When drop-kick failures exceed smoke-death, it makes commercial sense to ruggedize the mounting even if it means another dollar and another Watt.
 
I could say that there are no losses at all when welding the corners on their edge, on the contrary, that there's a gain in doing it. How will you fight me ? :)
I think the question you are asking is, how can one tell actual science from total nonsense?

It's a very good question, particularly because of the amount of total nonsense that is everywhere on the 'Net these days. :)

The answer is many-faceted, but the easiest one to understand - especially for non-scientists with limited math skills, who don't have a hope in hell of understanding Maxwell's equations and their implications - is to produce experimental proof.

In this case, if I wanted to prove you wrong, I would take two identical transformers, have the corners of one welded up, and then measure their performance. Apply various input voltages and frequencies, measure input and output currents and voltages, measure core temperature, vary the load, repeat. Eventually you would have a table and a set of graphs that show very clearly what the welds do, or do not do.

For someone who does understand math, you don't even need to perform the experiment. The answer is to be found in Maxwell's equations. They tell you exactly how to calculate eddy currents as well as every other electromagnetic phenomenon that occurs on a macro scale (not inside the atom.) This isn't new, the equations go back to the mid 1800s, and they are the complete and final answer to all things electromagnetic on a macro scale. A transformer is a relatively simple case. I'm quite sure that there is modelling software out there that, using Maxwell's equations, and given the core shapes and coil winding details, can pretty accurately predict what the transformer will do.

In much simpler terms, J.M. Fahey laid it out quite clearly: everyone with a clue understands that an electrically conductive core in a transformer is bad. But the question is, how conductive does it have to be before it creates enough losses to matter? If the losses are insignificant, or the cost-savings from welding substantial, then it's a perfectly reasonable engineering compromise.

To take a random example from an entirely different field of engineering: putting bigger tyres and wheels on a car slows it down under normal driving conditions, i.e. it takes longer to accelerate to a given speed. Anyone who paid attention in high-school physics class when the teacher was talking about Newton's laws applied to rotating objects understands why this is so (hint: wheels and tires are heavy, and have a considerable rotational moment of inertia.) Acceleration testing with lasers and stop-watches proves that it is so.

But the difference is usually too small to notice by the "seat of the pants" - you need actual instruments to notice that the car has become slower. And people have been brainwashed to believe that bigger wheels are a good thing, so they will actually pay more for the same identical car if it comes with bigger wheels and tyres. They won't notice the car is slower as a result.

So car manufacturers are happy to oblige. "Here is your slower car, sir. That'll be $5000, please, for the upsized tire and wheel upgrade package that slowed down your car."

So this is an example of a small backwards step in mechanical engineering, which yields a sizable forwards step in the car manufacturer's revenues. Just like welding up one edge of a transformer, which gives a slight backwards step in electrical efficiency, but puts more money in the manufacturer's pocket.

-Gnobuddy
 
But the difference is usually too small to notice by the "seat of the pants"


When I decided to tow an oversize (and weight) trailer with y 4 cylinder Honda Element I installed tires that were 7% smaller in diameter than the recommended size. The mechanical advantage made a big difference with a 1500 pound box following me around especially in the mountains. I am done moving all my stuff, but when it was time for new tires, I opted to stay with the smaller size. Why, my Element will flat out smoke any other stock Element out there and other than the tires, it's totally stock with 110K miles. Definitely a noticeable difference. On the down side I lost about 1 MPG at a steady 70 MPH, but city driving shows no change.
 
True.

One more proof that "you can´t beat Physics Laws" :rolleyes: but on the other side, if you are clever/McGiverish you can use them to your advantage :cool:

In this case Tubelab not only improved his mechanical advantage by shortening the loaded side of a lever which was working against him (Rim diameter/axle diameter ... he decreased rim diameter since axle diameter is fixed) but also lowered rotational inertia (which kills acceleration) by both reducing rotational distance (kess diameter) but also presumably rotational mass, I bet the new tires are "thinner" and in any case weigh less.

My Father used to train Olympic Bicycle racers (he was the official Argentine team Doctor in Japan Olympics 1964 and countless other same level competitions, from Switzerland to France to USA to you-name-it ) and one vivid memory I have is seeing them use "same wheels but Aluminum/Magnesium rim" (diameter was fixed by official Regulations)b but tiny thinnest rubber tubes ... "they would have used toothpick diameter ones if at all possible".
Sadly no Carbon Fiber or similar light composites way back then.
 
Back in my road bike racing days I fitted front chain rings that gave me overlapping and inter-meshing range of ratios that meant I could by juggling front and rear gear wheels go up or down in 2" steps (2.5%) in gearing around my most used flat terrain gearing.

This meant I had a really tall 1st gear (69") and a not very high top gear (100") but I could closely choose optimal effective gear ratio according to road surface, wind etc conditions. This worked out to be of great advantage.

Dan.
 
Thanx avtech23 for the paper link in post #32.

That paper has many graphs showing magnetic flux density in Tesla. The peak working flux of a transformer can depend on many factors - for a power transformer it is sort of set by the mains AC voltage and frequency - but for output transformers the signal level and frequency can span a large range.

For general purpose power transformer steel, a power transformer may be operating well above 1T, up to circa 1.5T, although the inductance is typically starting to fall before 1T.

Distortion in an output transformer is pretty much climbing with peak flux level, and for low distortion performance even with GOSS is likely kept below 1 to 1.2T at rated power at lowest rated frequency. And peak flux density reduces as signal frequency increases, given a set maximum signal voltage (ie. output power).
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.