Fast blow or Time delay fuses?

[ZDL]

I agree.

[R.G.]

Hey! That's a great idea. We'll all vote and Mother Nature will do whatever we decide She should!

Problem solved. Many problems solved, in fact.

[Mike Sulzer]

But if you get a remanence field plus a half cycle, you're guaranteed to push the core over into real saturation.

A transformer that goes into saturation when you plug it in is badly designed or defective. I do not think that this is a good example of how transformers draw transients when you plug them in. I do not know of any case where the transformer itself is responsible for a high transient current. It usually is the capacitors that have to be charged from zero once when you turn the supply on.

[R.G.]

A transformer that goes into saturation when you plug it in is badly designed or defective.

I respectfully disagree.

A transformer that *always* or even *most of the time* goes into saturation when you plug it in (or flip the on-switch, which is what this particular "Variac" brand variac unit used in the test) may well be badly designed or defective.

However, the physics of the situation says that (1) there is a remanent magnetization (2) energizing at a zero crossing gives a full half-cycle of Volt-time integral in one direction. If the design is such that the core remanence plus a half cycle of volt-time causes saturation, it saturates. It is entirely possible to design transformers that will not do this, by simply designing them to run at a low enough peak field in normal operation that a remance-plus-half-cycle will not force them into saturation. But that means giving up some of the normal-operation power available in the iron and copper to allow it to withstand the odd turn-on transient. This can be done by simply keeping the normal flux density down with a bigger core, or by introducing air gaps to make saturation much softer and soggier, which also has the effect of making the core bigger. So a device built to withstand this gets more expensive and heavier.

If there is no law preventing the occasional breaker or fuse popping (and there's not) will a manufacturer willingly build bigger, heavier and more expensive transformers, and/or accept a lower profit margin? Remember that in the absence of laws making certain events illegal, "good design" is an economic concept, not a binary one.

I do not think that this is a good example of how transformers draw transients when you plug them in.

Good example or not, that's how it worked. I did the instrumentation myself, and, working in a power supply design lab at the time, had the instrumentation to document it. At the time, I took pictures on the polaroid oscilloscope camera. (!) And it is what the physics of the situation says happens.

I do not know of any case where the transformer itself is responsible for a high transient current.

I do. I found it. I documented it, and spent some time finding out why it happened.

It usually is the capacitors that have to be charged from zero once when you turn the supply on.

As I said, it's a matter of degree. The smaller the transformer, the more the primary wire resistance limits things compared to the remanence. Big remanent field requires a lot of core. As long as the turn-on transient is small enough not to pop fuses or breakers, no one knows, and all users who do not instrument the setup will agree that "nothing happened". I noticed it because I was P-O'd that the variac I was using to test my power supply bring up made the wall breaker pop about one time in eight.

So we get down to discussing the terms "usually" and "degree". There is no question that a BFC being brought up from zero is a big load. I think I mentioned that. If I didn't, I should have. Locked rotor is another thing that looks just like a dead-flat capacitor. And even the known victim variac only did it when I hit the AC phase just right. They all happened when I started it near a zero crossing.

If I had the big-iron instrumentation still available, I'd go set up a test rig and quantify power-on surge for several sizes of transformer, with and without capacitor load. But I don't still have access to the lab.

[R.G.]

I should probably add that although I'm pig headed when I think I'm right, I try to own up to reality when I'm wrong. If someone has more evidence to see, I'd very much like to see it.

The Variac incident was in 1974, when I was a freshly minted EE, and when I complained about the stupid variac, I was given the teaching-punishment assignment to document it and explain it.

Now that I think about it, I have a 120:120:120:120 2kW transformer in the shop that I intended to use for high-power machine tool isolation and never did. Should be an ideal victim for testing. I'd have to go obtain a 200+A current transformer and some other odds and ends to set up the experiment again.

[loudthud]

I agree with R.G. and add that toroid transformers are the biggest offenders. It could be the low winding resistance, low leakage inductance or the no gap spiral core.

I've got one of those old Polaroid scope cameras but the film packs are crazy expensive these days.

[DRH1958]

For all you doubters out there, read the first six points of this of this: Transformer Inrush Current | Ametherm Notice how it says it will happen even if there is no secondary load!! So the transformer indeed causes heavy inrush current by itself.

This is interesting too. Notice the spike in the unprotected circuit and it's value: http://www.ametherm.com/inrush-curre...revention.html Transformers are very similar to motors and can be included to the list of items here as well. At initial startup, a transformer is like a locked rotor motor, almost a direct short.

[Mike Sulzer]

For all you doubters out there, read the first six points of this of this: Transformer Inrush Current | Ametherm Notice how it says it will happen even if there is no secondary load!! So the transformer indeed causes heavy inrush current by itself.

This is interesting too. Notice the spike in the unprotected circuit and it's value: Surge Current Causes and Prevention | Ametherm Transformers are very similar to motors and can be included to the list of items here as well. At initial startup, a transformer is like a locked rotor motor, almost a direct short.

The argument made in your first link (and discussed in more detail here: Magnetizing Inrush Current in Power Transformer | Electrical4u) is very different from R.G's. He says that the saturation is due to magnetism retained in the core reducing the available range before saturation occurs. Your link says that the problem happens when the waveform crosses zero at the moment of switching. The idea there is that the current builds for the entire first half cycle. It is essentially shifted from steady state operation where the current builds up for only half that time, and so gets to only half the value. I do not think that these two explanations are compatible. I do not know what to think about this, except that I am certainly wrong in stating that there is no such problem!

[DRH1958]

He says that the saturation is due to magnetism retained in the core reducing the available range before saturation occurs.

Do you think your point is related to their(my link) second point which says "inrush current affects the magnetic property of the core"? I don't know exactly what they mean by that. It could be many things. And to think that I was going to make a nice quiet exit from this thread in my post #14.

[R.G.]

The remanent magnetization in the iron either adds to or subtracts from the volt-time integral driving the core's field. Whether it aids or opposes the V-t for a cycle depends on which field polarity the incoming volt-time drives the core. If the remanence in the iron opposes the field generated by the first half-cycle, it can keep the field peak from getting into - or so far into - saturation. If it aids the direction of the field from the first half-cycle, it can make the saturation, and the current peaks worse.

This was in fact what I observed. Some trials, a start near a zero crossing would make huge peaks. Sometimes much less, for the same timing. The only thing that makes sense is the remanent field helping or hindering the startup.

I don't see these as different. I see the volt-time integral from a zero crossing and the remanent field magnitude and polarity as complementary. More importantly, together they describe what I saw.

[Mike Sulzer]

OK, so your study gives the complete explanation, covering what is found on the web in several places plus the effect of the remanent B field. Thank you, R.G., I have learned quite a bit about something I should have know at least a little bit about.

So the important thing for this forum is how much does this affect power transformers in guitar amps. Even if you use a standby switch, you have this effect when turning on the power switch, but if you do not use a standby switch and do use silicon diodes, then you have this and the capacitor charging to consider together. Some measurements might be a good idea unless someone has a reference. Sure, there are rules of thumb, but I tend to question them as I get older and grumpier.

The remanent magnetization in the iron either adds to or subtracts from the volt-time integral driving the core's field. Whether it aids or opposes the V-t for a cycle depends on which field polarity the incoming volt-time drives the core. If the remanence in the iron opposes the field generated by the first half-cycle, it can keep the field peak from getting into - or so far into - saturation. If it aids the direction of the field from the first half-cycle, it can make the saturation, and the current peaks worse.

This was in fact what I observed. Some trials, a start near a zero crossing would make huge peaks. Sometimes much less, for the same timing. The only thing that makes sense is the remanent field helping or hindering the startup.

I don't see these as different. I see the volt-time integral from a zero crossing and the remanent field magnitude and polarity as complementary. More importantly, together they describe what I saw.

[mhuss]

I'm not disagreeing that the remanance effect you describe can be a problem in kVA-sized cores with 00 gauge windings. However (and as you infer), I suspect that the wiring resistance substantially damps this effect on normal guit-amp sized PTs. So, I still think the SLO-BLOs are for the heater surge on 50-100 watt guitar amps.

An interesting (and practical as pertains to this discussion) experiment would be to perform experiments on a twin-reverb-sized PT and see how much the inrush current varies over several hundred turn on attempts.

[R.G.]

So the important thing for this forum is how much does this affect power transformers in guitar amps. Even if you use a standby switch, you have this effect when turning on the power switch, but if you do not use a standby switch and do use silicon diodes, then you have this and the capacitor charging to consider together. Some measurements might be a good idea unless someone has a reference. Sure, there are rules of thumb, but I tend to question them as I get older and grumpier.

Yeah, it would be good to rough this out. A proper setup for it would involve a timing circuit to start the thing at a zero crossing, a BFR (Big Freaking Relay) or contactor so the relay doesn't limit the inrush, a current sensor capable of 200A peaks and a scope to view the results.

I smoked my first o'scope ground lead working on this setup. Turns out you can't connect scope ground/safety ground in the probes to line on the test setup and have the insulation on the scope probe ground not melt and smoke.

I expect the spikes to correlate to the iron mass of the transformer, with bigger/more iron being worse, at a guess. Once the peaks are studied, it should be possible to correlate those to the max-normal power rating for the transformer and make a guess at whether a TD (non-USA term for slow blow) fuse is needed or not from the fuse-maker's curves of I-squared-t.

I'm not disagreeing that the remanance effect you describe can be a problem in kVA-sized cores with 00 gauge windings. However (and as you infer), I suspect that the wiring resistance substantially damps this effect on normal guit-amp sized PTs. So, I still think the SLO-BLOs are for the heater surge on 50-100 watt guitar amps.

An interesting (and practical as pertains to this discussion) experiment would be to perform experiments on a twin-reverb-sized PT and see how much the inrush current varies over several hundred turn on attempts.

Yeah, once you've gone to the trouble to instrument the setup, it would be nice to be able to add/subtract the heater surge and cap fill from this.

I've been designing the tester in my head while typing this. It looks like a metal box with a big power cord to the AC wall. Inside are a small DC power supply to run things (mostly the relay), a relay capable of not limiting the current it's enabling, a current transformer for measuring the current inrushes, and some circuitry for synching to the AC line and letting you offset the relay turn on by a precision amount.

The relay needs to be DC type, not an AC coil, so it's self timing with the AC line gets in the way of the timing. The timer needs to sense the AC line zero crossing and probably polarity/direction to test the add/subtract of the remanence effect. The timer would set up to pull the relay coil one relay-make-time before the next cycle, so as to not add the relay delay to the variation. I'd have to go look at SSRs to see if there are any that can take a few hundred amps of inrush. Those would be better without bouncing.

Finally, an output socket. The case of no heater, no cap can be done by pulling all the tubes and the rectifier, or temporarily opening solid state rectifiers. Then you could add back in just the tubes to introduce heater surge, then the cap link to see the cap fill-up surge.

I'm guessing heater surge is not the big issue with secondaries. Heater draw is about - what? 2-4 times normal heater current? In an amp with 5A of heaters, that's 30W. Multiplying that by 4 gets to 120W, which is about an amp as reflected onto the AC primary side. It's resistive, so multiply by 1.414 again for peak and you get to under 2A on the primary.

Cap inrush, on the other hand, is going to be big. Basically, the caps are going to be a dead short, so the current is only limited by the wire resistance of the primary, the wire resistance of the secondary, and the rectifiers' forward drops. That's not bad for a tube rectified setup, but for solid state diodes it will be big.

Hmm. I could simulate that one. Anyone know the wire resistance of an example tube power transformer's windings?

[trobbins]

PT primary resistance is measurable, and mains impedance is estimable, and leakage inductance is measureble, so peak current can be estimated. But there are dynamic inductive impedances at play. I read one good paper on that a few months ago, with sim and measured results, but won't be able to get the link for a while.

[R.G.]

For sure - simulation won't tell us what the actual results would be. They would only give us the ballpark for the capacitor input case.

Knowing the ballpark size is important. I found 200A hall effect sensors on ebay just now for $20. Lower current is cheaper.