OT questions

[trem]

Thanks again to everybody.
Enzo, just more more question.... With no load on a OT, that is to say with no speaker(s) connected to a OT as happens when a speaker cable is pulled out of the speaker jack; is there Zero (or very low) resistance, or Infinite (or very high) resistance at the OT secondary.?
Thank You

[Enzo]

Depends on the amp. fender was smart, they used a shorting jack for the main speaker output. If nothing was plugged into it, it shorted across the output winding. That was far preferable to an open. You plug into it, the shunt opens and your speaker gets the power.

Other amps may not have a shorting jack, and in those cases, no speaker means infinite impedance- an open - across the output. You cannot measure that open with an ohm meter, if that is where you are headed. If you connect a meter, you will be measuring the resistance of the transformer wires. That doesn't tell you much. The output winding is a relatively low number of turns of heavy wire, and it will have VERY low resistance.

[trem]

Depends on the amp. fender was smart, they used a shorting jack for the main speaker output. If nothing was plugged into it, it shorted across the output winding. That was far preferable to an open. You plug into it, the shunt opens and your speaker gets the power.

Other amps may not have a shorting jack, and in those cases, no speaker means infinite impedance- an open - across the output. You cannot measure that open with an ohm meter, if that is where you are headed. If you connect a meter, you will be measuring the resistance of the transformer wires. That doesn't tell you much. The output winding is a relatively low number of turns of heavy wire, and it will have VERY low resistance.

Yes, of course -
I cannot begin to make you understand how much I appreciate the knowledge that I gain here from guys like you. You really have no idea. Enzo, and many others, have literally turned the INTERNET into a college that has a very low tuition. All I can do is to keep saying Thank You.
The original quote is in Latin, and I do not know who said it, but a loose translation is "Everything becomes obvious upon explanation". That rings true for me many times on this forum.
Thanks Again

[Bruce / Mission Amps]

Yes, of course - ...
The original quote is in Latin, and I do not know who said it, but a loose translation is "Everything becomes obvious upon explanation". That rings true for me many times on this forum.
Thanks Again

I think that is almost a quote from Sir Arthur Conan Doyle's ... Sherlock Holmes.

[Wilder Amplification]

My "theory" on this is -

In regards to OPT failures, it's not the transformer itself that doesn't like open/shorted loads. It's the tubes that don't like the reactive properties of an OT when the secondary is open or shorted.

Think about a power transformer...you can plug it into a wall and leave it running all day long with no load on the secondary and it will not fry. Hardly any current flows on the primary side and as such the transformer itself could care less.

The OT is very similar to a power transformer in that it is just a simple step down transformer. It transforms high voltage/low current into low voltage/high current.

However...it is a reactive load and as such the tubes feel every reactance that is going on in the transformer itself. As long as the proper load is connected and assuming the amp was built with a transformer with the proper impedance ratio in the first place, the tubes are happy. Open or short the secondary and things get real interesting.

Just like a power transformer with a shorted secondary, an output transformer with a shorted secondary will cause primary current to increase beyond acceptable limits while under signal. This can either burn up the transformer primary all on its own due to the primary current increase, or causes the tubes to go overcurrent/redplating, burns up/shorts the tubes and takes the OT with them.

Open the secondary and the primary becomes an inductor, and inductors oppose changes in current flow. As the magnetic field around the primary builds and collapses over the primary, this induces a rather nice flyback voltage that tubes don't like. On top of this, the primary appears to be an infinite load and electrons see the screens/screen resistor as the easier path back to the power supply and screen current goes up. Excessive screen current is bad and can cause the screens to go into meltdown. Usually it will take out the screen resistor if you're lucky, but if this tube happens to short in just the right way under these conditions, the tube can also take the OT right along with it.

To express this mathematically, Z(Impedance) Ratio = (Turns Ratio^2) * Load Z on secondary

For example, you have a turns ratio of 20:1 (20) and a 16 ohm load on the secondary. The reflected load to the OT primary plate-plate will be -

(20^2) = 400 Z ratio

400 * 16 = 6,400 or 6.4K plate - plate

With no load on the secondary and the secondary open -

400 * Infinite = Infinite plate load...load way too high and current will see the screens as the easier path, therefore increasing screen current beyond acceptable limits

With the secondary shorted -

400 * 0 = Zprimary

Zprimary = 0...plate/OT primary current increases beyond acceptable limits and burns up tubes and very possibly the OT

In regards to whether you can mismatch up or down, I prefer not to mismatch at all if I can avoid it. Too low of a load and the output tubes will be allowed to exceed their maximum plate dissipation at some point in the signal swing given a certain amount of grid drive/volume setting. Too high of a load and again, screen current will increase. As most of us know, tube guitar amps are very hard on output tube screens as it is so anything you can do to minimize screen current is a good thing and will be beneficial to tube life. Too much screen current and you can drastically shorten tube life and eventually short the tube, which again can kill your OT if the tube shorts in just the right way if your HT supply isn't fuse protected (and sometimes even if it is depending on the circumstances).

Again...just my theory.

[Redhat]

I'm not so sure that its the inductive effects that tubes hate so much as it is just excessive source impedance.

Of course with tube amps, the load is inductive but even in a DC application, tubes (in constant current mode) hate open circuits.

If you "get in the tube" its easy to see why.

The filament sits there merrily heating away the cathode. The cathode thermionically boils electrons off its surface. Electrons being negative are extremely horny for positive high voltage things, like anodes and the screen grid and so blast off twoard them.

But if there is no speaker connected, the impedance the anode sees goes from the turns ratio squared times 8 ohms to the turns ratio squared times "infinity".

Actually an open primary can be viewed as a parallel combination of an admittance, G and a negative inductive susceptance, B (to account for energy storage).

G = Poc/(Vpo)^2 : Poc is total power to primary when open, Vpo is the voltage of the primary

B = - 1/Vpo (Ipo^2 - (Poc/Vpo)^2)^.5 : Ipo is the current through the primary when the secondary is open.

The total admittance is small (impedance is large) which can be verified with a wattmeter doing the above test.

So our hot little negatively charged leptons are screaming twoard the high voltage plate but now the plate's impedance is high so the electrons have no where to go -- an electron "traffic jam". No where to go but the good old screen grid which is also sitting there at a high positive voltage.

The problem is that the screen grid is a grid and not a plate. It wasn't designed to carry lots of current. Too many electrons getting out of that hot and crazy tube via the puny screen grid wires causes the wires to get hot and evaporate. Tube gets mad and tube dies....

[R.G.]

Just a bit of commentary.

Electrical fields and magnetic fields are duals of one another, and this is what makes capacitors and inductors duals of one another.

Tranformers are made of iron (and other ferromagnetic materials) because the "resistance" of such materials to the "flow" of a magnetic field is thousands of times less than the "resistance" of free space to magnetic fields. Given a choice, a magnetic field will have a preference by thousands of times to flow inside iron instead of the open space all around it.

That sounds good, but the dual of that is, say, copper conducting current. Copper is many, many millions of times a better conductor than free space, so the tendency of current to stay inside copper is hugely more than the tendency of a magnetic field to stay inside iron. The result is that some of the M-field created by a coil always leaks out of the iron into the more-"resistive" space around it. The more spaced out things are, the worse this gets.

A transformer is even worse. It has two coils, and these cannot occupy the same space, so the M-field leaks between and around them. The worst situation for leakage is two coils side by side. The best is two coils wound bifilar - that is the wires of both wound side by side simultaneously. Alternating layers to interleave the windings is almost as good, and almost as bad as side by side is one coil, then the other coil on top of it.

What's bad about leakage is that it looks like an inductor outside the transformer, in series with the leads. Load current into that inductor and try to change it, and the inductive flyback can only be clamped on the outside of the transformer.

The secondary load is transformed to what appears to be a load in parallel with the primary inductance. If this is small, you can clamp the leakage inductance outside the trannie. If the secondary is open, then as far as the primary side is concerned it does not exist. There is only the primary inductance and the leakages. The primary isn't too bad, because it's getting opposite and theoretically-equal currents from the two sides of the OT. But the leakage currents *can't* by definition be fed equal-and-opposite currents, because they are by definition that M-field that's not going through the other windings.

The secondary load helps with this, but when it's removed, the leakage inductances are free to generate flyback spikes without damping from secondary reflected load. And they do, and it's worst on fast transients, because V= L*di/dt. The faster you try to change the current, the bigger the inductive voltage.

Unloading can also make some amps unstable. If this happens, the amp is trying to drive the output as fast as it can slew, and not being able to do this because of the leakages, and the leakages are free to spit back any flyback the current change rates dictate.

This state of affairs is why the string-of-diodes on OT primaries came about. The string is to get enough voltage, because the voltage on the "off" tube plate can be almost twice the B+ in a pentode output stage. The diodes clamp when the opposing plate flies to megavolts. This drives the lower plate below ground, and clamps that half primary to be between the B+ centertap and ground. This also clamps the other half-primary, by transformer action, to be no more than 2X B+.

The problem there is that leakage inductance can't be clamped by transformer action again. So the high side, if it's been conducting heavily and suddenly turned off, can fly to B+ plus whatever its leakage inductance can add. And that can puncture insulation.

This scenario is why I put a string of MOVs across the Workhorse amps' OT primaries. It clamps the total voltage across the primaries to about 1300V, and since the CT is held at B+, the highest side can't get over about 1100V peak. Not great, but it's way better than you'd get with an unclamped leakage spike. Clamping at the output tube plate does clamp the inductance spike because it's on the outside of the trannie. A clamp to about 2.5x B+ would be great if there was a good, cheap, easy way to do that.

I don't know that leakage inductance is the only culprit, but it's definitely there, and I know it can cause voltages which will make arc-forming sparks.

[Redhat]

[QUOTE=R.G.;150728

The secondary load helps with this, but when it's removed, the leakage inductances are free to generate flyback spikes without damping from secondary reflected load. And they do, and it's worst on fast transients, because V= L*di/dt. The faster you try to change the current, the bigger the inductive voltage.

QUOTE]

Yow! I never ever consider yanking out a speaker cord when any kind of power amp is on. Flyback protection is a good idea (for those who might consider it). Or in case of an accident.

[Steve Conner]

Yow! I never ever consider yanking out a speaker cord when any kind of power amp is on. Flyback protection is a good idea (for those who might consider it). Or in case of an accident.

Well, nobody ever intends to do it. Unless maybe you pissed off the drummer by hiding his drool bucket.

I agree with Wilder's comments. If I had to sum it up, I'd say that the tubes hate the excessive impedance, but the OT (and other insulation) hates the inductive spikes.

RG's point about the diode strings is a good one. They're intended to protect against magnetizing current and reactive load current, returning it to B+ in exactly the same way as the catch diodes on the output of a solid-state power amp.

But it always seemed to me that they protected against leakage inductance kickbacks too, albeit in a ghetto fashion. The leakage inductance kickbacks go positive with respect to ground, so they break the diodes down in the reverse direction, and most diodes can take a little of that. When the energy kicked back into them gets above a certain level, they fail short and blow the HT fuse, which is a way of protecting the OT. The amp stops working and needs repaired, but the diodes only cost pennies to replace.

It might be worth investigating TVS diodes ("transzorbs" is one trade name for these) in place of regular diodes here. Same idea as RG's MOVs, but maybe small transzorbs have less capacitance than MOVs, and being just power zener diodes, they could replace the existing diodes, which helps with the parts count.

[Redhat]

Well, nobody ever intends to do it. Unless maybe you pissed off the drummer by hiding his drool bucket.

...
But it always seemed to me that they protected against leakage inductance kickbacks too, albeit in a ghetto fashion. ....

Some of the places I play have very drunk cowboys. They always seem to land on my equipment when they get too "tipsy".


One thing I see in some of the old Fender schematics is a shunt switch connected across the open speaker output so that the primary "never" sees an open secondary. That would help with accidental inductive spikes provided the switch is clean and makes good contact.

[R.G.]

Well, nobody ever intends to do it. Unless maybe you pissed off the drummer by hiding his drool bucket.

I see you and I have worked with the same drummer.

But it always seemed to me that they protected against leakage inductance kickbacks too, albeit in a ghetto fashion. The leakage inductance kickbacks go positive with respect to ground, so they break the diodes down in the reverse direction, and most diodes can take a little of that. When the energy kicked back into them gets above a certain level, they fail short and blow the HT fuse, which is a way of protecting the OT. The amp stops working and needs repaired, but the diodes only cost pennies to replace.

The reasoning is sound, but the problem is that the actual breakover voltage of high voltage diodes is almost never specified. All you get is '1000V minimum' or the equivalent. The diode makers probably never test the maximum breakover, which is what you're really interested in.

On top of that, you usually have to have more than one diode. Strings of three 1N4007s is common on most amps that have diode strings. That's at least 3kV breakover (if you got good ones that is). Why not use 1N4004, 1n4005, or 1N4006? Because there is no guarantee that the manufacturer didn't just get a batch of 1N4007 diode pellets and label them at lower voltages. They otherwise meet all the same specs. If somebody needs a shipment of 10,000 1N4004s, and all you have is 1N4007 dies, well, they just got extra good ones. In fact, I suspect that it is possible that there are only two or three variations of the 1N400x family, packaged with different numbers, at least at some times and some manufacturers.

Agh! We also have to worry if we get much better parts than we bargained for!

It might be worth investigating TVS diodes ("transzorbs" is one trade name for these) in place of regular diodes here. Same idea as RG's MOVs, but maybe small transzorbs have less capacitance than MOVs, and being just power zener diodes, they could replace the existing diodes, which helps with the parts count.

Although I shouldn't, I regard TVS and MOVs as the same thing - breakover protectors. I went with a string of a couple of 600+ volt ones across the primary. You could replace the diode strings on each plate with TVS easily enough, though.

It was kinda fun - I *tried* to get spikes on the primary at either plate by opening leads, putting in square waves, spikes on top of big sines, whatever I could think of. The worst I ever found was a spike of a couple of hundred volts and some heavily damped ringing. It is difficult to induce this kind of thing in a meaningful way to see if it works or not.

[Enzo]

One thing I see in some of the old Fender schematics is a shunt switch connected across the open speaker output so that the primary "never" sees an open secondary. That would help with accidental inductive spikes provided the switch is clean and makes good contact.

Yep, but that only protects against pulling the plug from the amp. WOn;t protect misdeeds at the speaker end, or a failing cord.