Testing Power Transformers

[catalin gramada]

Hello. I wonder how can I perform a test for a power transformer performance ,please ? Asking because I found one which is not power rated but just nominal secondary currents. The transformer is sell as a Marshall replacement and it have same physical size but claimed to have 600mA instead original 300mA for nominal voltage.(356V C.T./600mA). I really wanna know if I can use it at rated current for an kt88 project. It is a short test over 600 ohm load just enough, please ? Thanks

[J M Fahey]

IF it has same physical size, same voltages, but claims twice the current it simply is not true.

SS electronics advances day by day, tubes are basically same as 50/60 years ago, average silicon steel same as 120 years ago, no new revolutionary advances allowing the doubling of performance for same weight of steel.

Even less performance change for same weight of copper.

Design equations, graphs and tables from 1947 Radiotronīs Designer are perfectly valid today.

If anything, itīs hard to get steel as good as old Armscor and similar, not because "we know less", far from that, but because we donīt want to pay for it, go figure.

Consider that same size same weight iron and copper standard Marshall transformer as a same specs clone, not a miraculously doubled performance improvement.

Now if it were an SMPS, yes, modern Ferrites can reach even a MHZ, some even more, garden variety types are comfortable with a couple hundred kHz, while 1947 flyback cores "just" could be used at 17kHz so there a modern core can easilyprovide 10X the old performance, but .... on laminated silicon steel? ... forget it.

[catalin gramada]

Thanks JMF. The transformer is on the way. I will full load the secondaries with resistors loads at nominal current they claimed leaving for couple of hours under load and post of full review. I have no problem to return a toasted transformer if will be the case.

[Helmholtz]

Thanks JMF. The transformer is on the way. I will full load the secondaries with resistors loads at nominal current they claimed leaving for couple of hours under load and post of full review. I have no problem to return a toasted transformer if will be the case.

Your load resistors will have to absorb/dissipate around 200W in total at nominal current.

[catalin gramada]

Your load resistors will have to absorb/dissipate around 200W in total at nominal current.

This is true. Six pieces of 100 ohm/50w will do the job.

[Helmholtz]

This is true. Six pieces of 100 ohm/50w will do the job.

If used within their power derating temperature limits. This might mean mounting on a large heatsink for some resistor types (e.g. the golden aluminum encased ones).

[catalin gramada]

This is true also.For any power resistor the sheets shows nominal power with and without heat sink relative to different ambient conditions.The aluminium encased resistors are made to use mounted on a certain surface heat sink, exactly like power transistors need to be mounted to a minimum surface heat sink for certain heat dissipation.

[Helmholtz]

This is true also.For any power resistor the sheets shows nominal power with and without heat sink relative to different ambient conditions.The aluminium encased resistors are made to use mounted on a certain surface heat sink, exactly like power transistors need to be mounted to a minimum surface heat sink for certain heat dissipation.

I see you're well prepared.

[catalin gramada]

Does it exist a graph sheet/calculator regards maximum power transfer for a certain core section relative to "standard" lamination in respect of VA capabilities ? Thinking if can use thicker wires you can have an idea by core section if the iron is able to...

[Helmholtz]

Thinking if can use thicker wires you can have an idea by core section if the iron is able to..

A well designed transformer uses the complete coilformer space for the windings. As turns numbers are dictated by the application, increasing wire cross section is typically no option.

[catalin gramada]

A well designed transformer uses the complete coilformer space for the windings. As turns numbers are dictated by the application, increasing wire cross section is typically no option.

This exclude the possibility to get more power even you add lamination to the core and will be necessary to do a step up by lamination size ?

[catalin gramada]

I did a sort of empirical research and found for this standard lamination - EI 114 there are produced standard plastic bobbins. From my point of interest found 50mm deep for stacked EI 114 for power transformers between 200-260 VA and 60mm deep for PT between 250-360 VA. This should be a general indicator if the iron have enough power transfer capabilities in respect with secondary currents states...I think.

[R.G.]

The limit on a transformer's power throughput is difficult to actually pin down. It's determined by the combination of the surface area exposed to the outside air and the temperature withstanding of the interior insulating layers and wire insulation. The copper is good up to its melting temp, and the iron functions fine up to the iron Curie temperature at about... um, 770C? ... both higher than you'd ever get. What fails is the insulation between windings, and that's why hte idea of insulation classes was born.

Unrated materials are assumed to die below 105C, the first step in the insulation temperature classes. No respectable maker (that is a significant limitation, by the way) would use anything less than Class 105. It goes up from there; see what-is-meant-by-class-in-insulation. The higher temperature you can let your insulation run at, the more power you can get out of the transformer before the insulation melts and makes 10,000F arcs inside to melt the copper.

So a transformer's power rating is murky. There is no one good number without careful inspection. The best way to find out whether the transformer can power X watts is to (1) measure the resistance of the copper wires in the windings accurately, (2) load it with X watts and re-measure the copper resistance every so often and plot the temperature rise curve for the copper, (3) stop when you can project from the temperature rise curves that you've either found the top temperature or that you're going to exceed the temperature rating of the wires and internal insulation.

Copper wire makes a great temperature sensor, as its resistance changes by +0.393 percent per degree C. So for example, if the 1.300 ohm wire in a hypothetical winding rises by 10% to 1.430 ohms, the 10% rise is 25.44 times 0.393, or a rise of 25.44C if I did the math right. If you started at 25C outside the transformer with the transformer absolutely at 25C through and through, the copper wire has heated up to 50.44C. So even Class A/105 is fine.

There are some gotchas. You MUST accurately measure the copper wire resistance to be able to tell the rise, and that often means using a low-ohms meter, with four-lead measurement. A cheap DMM may not be up to the task for heater winding. You must also do a good job of watching the temp rise long enough to accurately predict the ending of the (typically) declining exponential temp rise curve AND the thermal time constant of the transformer. A several hundred watt power transformer may have a thermal time constant of an hour or more, and the temperature won't reach its final value for several hours. That's why you want to use the rising temp curve to guess how long you have to watch to get good numbers.

Multiple secondary windings are another gotcha that isn't all that easy to handle. Obviously, if temperature defines the current limit, and you load only one winding, then the heat from other windings not being used isn't heating the inside of the transformer, and it will both take longer for the temperature to settle and be able to withstand much more current because it's the only heat source in there, not one of many. So you need to make a best guess about loading for your testing, and use those loads to get the overall temperature rise. You can test them one at a time, but the resulting data will be substantially worthless. They all have to be loaded to some semblance of the load they will actually be running so you get moderately worthwhile results.

Murky.

[loudthud]

There is a way one vendor could rate a transformer at 300mA and another could rate the same transformer at 600mA. It has various names, Marketing, Specmanship, lying, fraud, you name it. The specs quoted are most commonly incomplete. There are too many unknowns. If they are talking about DC current after a rectifier and cap input filter, the current in the secondary could change by a factor of two to one simply by changing the capacitance. Do they give you this information ? Not a chance. Is a Full Wave Center Tap or Full Wave Bridge (like the one being talked about here) configuration being used ? They usually don't say. That could be another two to one difference in the spec.

Transformers marketed to people building guitar amps are the biggest offenders. You never really know what you are getting. Your best guess (for tube amps) is to get something with a VA rating two to three times the power output in Watts.

[trobbins]

The manufacturer also has the choice of wire enamel thickness to use - typically identified by winding wire being grade 1 or 2. Thicker enamel coating means less turns per layer, so possibly a higher winding resistance.

You also need to factor in your intended amp ambient temperature situation, and if it may typically operate at a significantly higher temperature due to restricted free-air ventilation, or heating from nearby hotter valves, or being in a location where ambient temp is noticeably higher than 20-30C (such as in a rack, or in a hot climate).

If you go down the temperature test path, then I'd recommend testing for as long as possible (hopefully to when winding resistance stabilises) - its amazing how long the upwards creep in temp can continue, and is helped by the positive temp coef of the wire.

Note that resistive loading of the transformer is more benign as far as winding loss than if you loaded the main secondary with rectified and filtered circuitry and then in to a loading resistor.